Battam commited on
Commit
58052f7
·
verified ·
1 Parent(s): 85d308b

Release [run release-20261111T143445Z]

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. CHANGELOG.md +18 -0
  2. README.md +41 -41
  3. README.zh.md +41 -41
  4. campaign_files/2026-10-b-disk-images-2/libarchive-iso-malformed/README.md +0 -8
  5. campaign_files/2026-10-b-disk-images-3/libcdio-iso-malformed/README.md +0 -5
  6. campaign_files/2026-10-b-disk-images-3/liu-ctf-2025-corrupted/README.md +0 -3
  7. campaign_files/2026-10-b-disk-images-3/lks-bali-2026-erased-evidence/README.md +0 -3
  8. campaign_files/2026-10-c-repros-2/lampert-2026-sbp-not-enough-repro/README.md +0 -2
  9. campaign_files/2026-10-g-solver-instances-2/lexaire-thermogift/LICENSE +3 -0
  10. campaign_files/2026-10-g-solver-instances-2/lexaire-thermogift/README.md +23 -0
  11. campaign_files/2026-10-g-solver-instances-2/ljiang9-bridges/LICENSE +3 -0
  12. campaign_files/2026-10-g-solver-instances-2/ljiang9-bridges/README.md +23 -0
  13. campaign_files/2026-10-g-solver-instances-2/ljiang9-hitori/LICENSE +3 -0
  14. campaign_files/2026-10-g-solver-instances-2/ljiang9-hitori/README.md +23 -0
  15. campaign_files/2026-10-g-solver-instances-2/ljiang9-killer/LICENSE +3 -0
  16. campaign_files/2026-10-g-solver-instances-2/ljiang9-killer/README.md +23 -0
  17. campaign_files/2026-10-g-solver-instances-2/magictour-top1465/README.md +0 -4
  18. campaign_files/2026-10-g-solver-instances-2/magictour/README.md +0 -16
  19. campaign_files/2026-10-g-solver-instances-2/mayfull-sudoku-4x4/README.md +0 -4
  20. campaign_files/2026-10-gt-browser-tests-2/libkscreen-config-tests/README.md +0 -8
  21. campaign_files/2026-10-gt-browser-tests-2/libreoffice-calc-function-tests/README.md +0 -0
  22. campaign_files/2026-10-gt-browser-tests-2/libreoffice-uitest/README.md +0 -66
  23. campaign_files/2026-10-gt-browser-tests-2/mail-mime-parser-emails/README.md +0 -4
  24. campaign_files/2026-10-gt-browser-tests-2/mail-parser-eml-malformed/README.md +0 -4
  25. campaign_files/2026-10-gt-browser-tests-2/mail-parser-eml-rfc/README.md +0 -4
  26. campaign_files/2026-10-gt-browser-tests-2/mail-parser-mbox/README.md +0 -4
  27. campaign_files/2026-10-gt-browser-tests-2/mailparser-fixtures/README.md +0 -4
  28. campaign_files/2026-10-gt-browser-tests-2/mammoth-test-data-docx/README.md +0 -14
  29. campaign_files/2026-10-gt-browser-tests/lighthouse-smokehouse/README.md +0 -15
  30. campaign_files/2026-10-hw-eda-tests-2/lowrisc-opentitan-ip-block-pattgen/README.md +0 -4
  31. campaign_files/2026-10-hw-eda-tests-2/manifold-robust-mesh-boolean-library/README.md +0 -16
  32. campaign_files/2026-10-m-issue-trackers-2/libvirt-users/README.md +0 -4
  33. campaign_files/2026-10-m-issue-trackers-2/linaro-lava-users/README.md +0 -4
  34. campaign_files/2026-10-m-issue-trackers-2/lineageos-gerrit-changes/README.md +0 -4
  35. campaign_files/2026-10-m-issue-trackers-2/llvm-llvm-dev/README.md +0 -4
  36. campaign_files/2026-10-m-issue-trackers-2/loudoun-county-transit-gtfs/README.md +0 -4
  37. campaign_files/2026-10-m-issue-trackers-2/lttng-dev/README.md +0 -4
  38. campaign_files/2026-10-m-issue-trackers-2/lustre-lustre-discuss/README.md +0 -4
  39. campaign_files/2026-10-m-issue-trackers-2/lustre-whamcloud-gerrit-changes/README.md +0 -4
  40. campaign_files/2026-10-m-issue-trackers-2/lyx-trac/README.md +0 -6
  41. campaign_files/2026-10-m-issue-trackers-2/macports-trac/README.md +0 -8
  42. campaign_files/2026-10-m-issue-trackers-2/mailman3-mailman-users/README.md +0 -6
  43. campaign_files/2026-10-md-media-and-documents/latex3-pdfresources-testfiles-bookmark/README.md +0 -4
  44. campaign_files/2026-10-md-media-and-documents/lepture-mistune-fixtures/README.md +0 -4
  45. campaign_files/2026-10-md-media-and-documents/libaom-test-vectors/README.md +0 -10
  46. campaign_files/2026-10-md-media-and-documents/libass-libass-tests-regression/README.md +0 -4
  47. campaign_files/2026-10-md-media-and-documents/libgd-tests/README.md +0 -6
  48. campaign_files/2026-10-md-media-and-documents/libjpeg-turbo-testimages/README.md +0 -4
  49. campaign_files/2026-10-md-media-and-documents/libjxl-conformance/README.md +0 -6
  50. campaign_files/2026-10-md-media-and-documents/librivox-ozymandias-of-egypt/README.md +0 -4
CHANGELOG.md CHANGED
@@ -3159,3 +3159,21 @@ Changes to the content of this index, oldest first. There is one entry for each
3159
  - hf-training-records: refused
3160
  - npm: refused
3161
  - stackexchange-api-posts: refused
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3159
  - hf-training-records: refused
3160
  - npm: refused
3161
  - stackexchange-api-posts: refused
3162
+
3163
+ ## 2026-11-11: sources held back
3164
+
3165
+ - Each source below keeps its newest version that passed every check; its newer batch waits until it passes them:
3166
+ - borealis-dataverse-workbooks-cc-by: changed pointer group file data/borealis-dataverse-workbooks-cc-by/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 2 key(s) of the old version are in no data file of source borealis-dataverse-workbooks-cc-by after the commit (row 4 from batch 2026-10-a-spreadsheets-with-answers, row 5 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3167
+ - borealis-dataverse-workbooks-cc0: changed pointer group file data/borealis-dataverse-workbooks-cc0/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 1 key(s) of the old version are in no data file of source borealis-dataverse-workbooks-cc0 after the commit (row 7 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3168
+ - cfpb-complaints: refused
3169
+ - data-gov-uk-workbooks: changed pointer group file data/data-gov-uk-workbooks/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 4 key(s) of the old version are in no data file of source data-gov-uk-workbooks after the commit (row 5 from batch 2026-10-a-spreadsheets-with-answers, row 6 from batch 2026-10-a-spreadsheets-with-answers, row 12 from batch 2026-10-a-spreadsheets-with-answers, 1 more); the removal record holds none of them, and none fails a publish check now
3170
+ - data-gov-uk-workbooks-noassertion: changed pointer group file data/data-gov-uk-workbooks-noassertion/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 1 key(s) of the old version are in no data file of source data-gov-uk-workbooks-noassertion after the commit (row 6 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3171
+ - dataversenl-workbooks-cc-by: changed pointer group file data/dataversenl-workbooks-cc-by/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 1 key(s) of the old version are in no data file of source dataversenl-workbooks-cc-by after the commit (row 17 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3172
+ - dataversenl-workbooks-cc0: changed pointer group file data/dataversenl-workbooks-cc0/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 3 key(s) of the old version are in no data file of source dataversenl-workbooks-cc0 after the commit (row 7 from batch 2026-10-a-spreadsheets-with-answers, row 13 from batch 2026-10-a-spreadsheets-with-answers, row 16 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3173
+ - dvn-fvhlgg-repro: changed pointer group file data/dvn-fvhlgg-repro/2026-10-c-repros/part-00000.parquet: 3 key(s) of the old version are in no data file of source dvn-fvhlgg-repro after the commit (row 5 from batch 2026-10-c-repros, row 7 from batch 2026-10-c-repros, row 16 from batch 2026-10-c-repros); the removal record holds none of them, and none fails a publish check now
3174
+ - eu-institution-workbooks-cc-by: changed pointer group file data/eu-institution-workbooks-cc-by/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 3 key(s) of the old version are in no data file of source eu-institution-workbooks-cc-by after the commit (row 2 from batch 2026-10-a-spreadsheets-with-answers, row 4 from batch 2026-10-a-spreadsheets-with-answers, row 5 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3175
+ - hdx-workbooks-cc0: changed pointer group file data/hdx-workbooks-cc0/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 2 key(s) of the old version are in no data file of source hdx-workbooks-cc0 after the commit (row 4 from batch 2026-10-a-spreadsheets-with-answers, row 5 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3176
+ - hdx-workbooks-odbl: changed pointer group file data/hdx-workbooks-odbl/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 1 key(s) of the old version are in no data file of source hdx-workbooks-odbl after the commit (row 4 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
3177
+ - hf-training-records: refused
3178
+ - npm: refused
3179
+ - stackexchange-api-posts: refused
README.md CHANGED
@@ -24,25 +24,25 @@ configs:
24
  <!-- artifact-pointer-layout: v2 -->
25
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
26
 
27
- <img class="not-prose" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/banner.svg" alt="Public web, in two layers (task material in collections; corpora), one row per file (address, size, SHA-256, license), your selection, fetched and verified">
28
 
29
  This dataset is the public index of the **Training Data Foundation**: find public files to build agent tasks and environments from, and fetch them with one command.
30
 
31
  English | [简体中文](README.zh.md) · [Data page](https://huggingface.co/spaces/Battam/training-data-foundation) · [Usage video](https://huggingface.co/spaces/Battam/training-data-foundation#usage-video)
32
 
33
  <picture class="not-prose">
34
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-narrow-light.svg">
35
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-wide-light.svg" alt="1,801 material sets with 4,186 units in 14 collections; 139 corpora with 1,216,362,128 rows; 4.45 PB pointed to, as the index holds addresses, not copies; 99.6% of the 35,277 rows compared match their recorded size.">
36
  </picture>
37
  <picture class="not-prose">
38
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-narrow-dark.svg">
39
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-wide-dark.svg" alt="1,801 material sets with 4,186 units in 14 collections; 139 corpora with 1,216,362,128 rows; 4.45 PB pointed to, as the index holds addresses, not copies; 99.6% of the 35,277 rows compared match their recorded size.">
40
  </picture>
41
 
42
  ## Get data in one command
43
 
44
  ```sh
45
- uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/tools/fetch.py --form spreadsheets --sample 3 --out ./data
46
  ```
47
 
48
  This fetches 3 units of every spreadsheet material set into `./data`, one folder per unit, and checks each file against the size and hashes its row records.
@@ -52,7 +52,7 @@ the same command takes the same units of this release.
52
  <details>
53
  <summary>What the command takes, and in which order</summary>
54
 
55
- The command and the queries below are pinned to the release tag `release-2026-11-11.12`.
56
  [uv](https://docs.astral.sh/uv/) reads the script's dependencies, at the versions published before the day of this release.
57
 
58
  The command returns units: 3 units from each material set whose form is `spreadsheets`, each in a folder of its own named after its `unit_id`.
@@ -79,7 +79,7 @@ Choose material sets from the catalog with [DuckDB](https://duckdb.org/):
79
 
80
  ```sql
81
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
82
- FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/meta/catalog.parquet'
83
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
84
  ORDER BY collection, units DESC;
85
  ```
@@ -90,7 +90,7 @@ with `kind = 'corpus'` it returns corpora instead, counted in rows (`files`).
90
  Choose files in one corpus:
91
 
92
  ```sql
93
- SET VARIABLE files = (SELECT list('hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/meta/catalog.parquet' WHERE source = 'if-archive'));
94
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
95
  FROM read_parquet(getvariable('files'))
96
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
@@ -112,15 +112,15 @@ A data file may hold the rows of several sources, so the query keeps this one's
112
  ### Task material: units by collection
113
 
114
  <picture class="not-prose">
115
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-narrow-light.svg">
116
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-wide-light.svg" alt="Units by collection, longest bar first: Spreadsheets 536; Browser tests 439; Tests and exercises 425; Model serving configurations 397; Media and documents 391; Science reference data 342; Rule calculations 329; Security detection 328; Hardware design (EDA) tests 258; Issue trackers 258; Solver instances 189; Disk images 179; Reproduction packages and runs 106; In no collection 9; SEC filing reviews 0.">
117
  </picture>
118
  <picture class="not-prose">
119
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-narrow-dark.svg">
120
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-wide-dark.svg" alt="Units by collection, longest bar first: Spreadsheets 536; Browser tests 439; Tests and exercises 425; Model serving configurations 397; Media and documents 391; Science reference data 342; Rule calculations 329; Security detection 328; Hardware design (EDA) tests 258; Issue trackers 258; Solver instances 189; Disk images 179; Reproduction packages and runs 106; In no collection 9; SEC filing reviews 0.">
121
  </picture>
122
 
123
- 1,801 material sets with 4,186 units, in 14 collections (1 of them with no material set yet); 1 of the material sets are in no collection.
124
 
125
  <details>
126
  <summary>Task material by collection, as a table</summary>
@@ -131,8 +131,8 @@ A data file may hold the rows of several sources, so the query keeps this one's
131
  | Disk images | 52 | 179 | 324 | 40 | 272 MB | 272 MB |
132
  | Reproduction packages and runs | 88 | 106 | 870 | 88 | 120 MB | 120 MB |
133
  | SEC filing reviews | 0 | 0 | 0 | 0 | 0 B | 0 B |
134
- | Solver instances | 65 | 189 | 342 | 62 | 127 MB | 127 MB |
135
- | Browser tests | 123 | 439 | 852 | 88 | 32.0 MB | 32.0 MB |
136
  | Hardware design (EDA) tests | 81 | 258 | 651 | 75 | 80.8 MB | 80.8 MB |
137
  | Issue trackers | 212 | 258 | 682 | 212 | 196 MB | 196 MB |
138
  | Media and documents | 276 | 391 | 1,088 | 269 | 569 MB | 569 MB |
@@ -150,12 +150,12 @@ Upstreams are the upstream works the material sets of a collection come from.
150
  ### Corpora: rows by main domain
151
 
152
  <picture class="not-prose">
153
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-narrow-light.svg">
154
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-wide-light.svg" alt="Corpus rows by main domain, longest bar first: Data science 750,815,212; System administration 185,826,217; Many domains 109,262,742; Software engineering 45,267,701; Science 44,633,675; Operations 39,404,642; Media 12,751,612; Machine learning 9,636,889; Hardware 8,716,520; Unclassified 5,284,162; Security 4,521,631; Misc 241,125.">
155
  </picture>
156
  <picture class="not-prose">
157
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-narrow-dark.svg">
158
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-wide-dark.svg" alt="Corpus rows by main domain, longest bar first: Data science 750,815,212; System administration 185,826,217; Many domains 109,262,742; Software engineering 45,267,701; Science 44,633,675; Operations 39,404,642; Media 12,751,612; Machine learning 9,636,889; Hardware 8,716,520; Unclassified 5,284,162; Security 4,521,631; Misc 241,125.">
159
  </picture>
160
 
161
  139 corpora with 1,216,362,128 rows.
@@ -188,12 +188,12 @@ Every material set and every corpus, with its counts, size, license tier and log
188
  ## How it is made
189
 
190
  <picture class="not-prose">
191
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-narrow-light.svg">
192
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-wide-light.svg" alt="How the index is made, in four steps. 1, collect: collectors read whole upstreams into corpora, one row per file or record. 2, admit: works the search found upstream and groups of corpus rows become material sets. 3, check: licenses, personal data, overlap with benchmarks and sizes are checked before release. 4, release: one commit with its own tag, release-2026-11-11.12.">
193
  </picture>
194
  <picture class="not-prose">
195
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-narrow-dark.svg">
196
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-wide-dark.svg" alt="How the index is made, in four steps. 1, collect: collectors read whole upstreams into corpora, one row per file or record. 2, admit: works the search found upstream and groups of corpus rows become material sets. 3, check: licenses, personal data, overlap with benchmarks and sizes are checked before release. 4, release: one commit with its own tag, release-2026-11-11.12.">
197
  </picture>
198
 
199
  The terms, each step and how every number is made are folded below.
@@ -247,9 +247,9 @@ only their sizes add up to the size of the whole index.
247
 
248
  | Level | Recorded size | Deduplicated size | Rows read | Rows compared | Rows that match | Rows read, not compared | Rows without a recorded size | Their estimated size |
249
  |---|---:|---:|---:|---:|---:|---:|---:|---:|
250
- | Whole index | 4.45 PB | 4.11 PB | 36,158 | 35,277 | 35,147 (99.6%) | 881: address unreachable 449, file gone 80, login needed 330, no size from the server 19, other error 3 | 312,300,338 | 202 TB, an estimate (95% range 171 TB to 245 TB), for 215,326,513 of them, from 14,161 rows read |
251
  | Corpora | 4.45 PB | 4.11 PB | 31,594 | 30,723 | 30,610 (99.6%) | 871: address unreachable 449, file gone 80, login needed 322, no size from the server 19, other error 1 | 312,300,338 | 202 TB, an estimate (95% range 171 TB to 245 TB), for 215,326,513 of them, from 14,161 rows read |
252
- | Task material | 2.90 GB | 2.82 GB | 4,564 | 4,554 | 4,537 (99.6%) | 10: login needed 8, other error 2 | 0 | none |
253
 
254
  Rows read are the sample of rows the check reads in `meta/sizes.parquet`;
255
  a row compared matched its recorded size or did not, and a row read but not compared never reached its file.
@@ -258,7 +258,7 @@ above one source its range is the sum of its sources' ranges, wider than a joint
258
 
259
  The recorded size is the sum of `size_bytes` over the rows that have one (catalog column `bytes`). The check reads a sample of rows of every data file (`meta/sizes.parquet`): 20 of its rows with a recorded size, and of its rows without one its share of its source's sample, sized for their mean size within 25% at 95% confidence by the spread of the sizes read so far (at most 400 rows per source per round, 20 for a source with none read yet); its share is the rows that match their recorded size out of the rows it compared (status `match`, `size_mismatch` or `hash_mismatch`), and the rows it could not compare, which never reached their file, are counted beside it by status (`unreachable`, `gone`, `needs_auth`, `no_length`, `error`; catalog column `size_check`). The estimate for a source's rows without a recorded size (catalog columns `unsized_rows` and `unsized_estimate`) is their number times the mean measured size of such rows read, with a 95% BCa bootstrap range (`scipy.stats.bootstrap`, its default resample count); a source with fewer than 10 such rows read has none. Above one source, the estimate is the sum of its sources' estimates and its range the sum of their ranges, which is wider than a joint interval. The figures of every level are in `meta/size_figures.json`.
260
 
261
- Distinct content is the sum of `size_bytes` over the rows minus the sizes of the rows that repeat another row's content (the same sha256, else sha1, else md5 and size, within the same hash scope; one copy of each content is kept, in the first source by registry order), counted after each release over the data files it published, per source in the catalog column `distinct_bytes`; archive members whose archive the same source also lists whole have other hashes, so they are not subtracted. `tools/fetch.py` removes duplicates over your own selection when it reads it. The deduplicated sizes here were counted over the data files of release `release-2026-11-11.3`.
262
 
263
  </details>
264
 
@@ -271,7 +271,7 @@ each stage counts what meets its condition before those rules and after each of
271
  Each table below gives what a training stage can take, counted before the rules and after each of them, in their order;
272
  its last row is the stage's count.
273
 
274
- The stage fields (`meta/stage_fields.parquet`) were counted over the data files of release `release-2026-11-11.3`, after it was published.
275
 
276
  #### Post-training (in use)
277
 
@@ -279,11 +279,11 @@ Material set units for building tasks: every file of the unit allowed for traini
279
 
280
  | Step | Material sets | Units | Upstream sources |
281
  |---|---:|---:|---:|
282
- | In the stage's condition, before the rules | 1,425 | 3,477 | 1,349 |
283
- | After the benchmark rule | 1,425 | 3,477 | 1,349 |
284
- | After the rule on rows not compared with the benchmark list | 1,425 | 3,477 | 1,349 |
285
- | After the personal-data rule | 1,267 | 3,083 | 1,201 |
286
- | After deduplication: the stage's count | 1,244 | 2,988 | 1,183 |
287
 
288
  #### Mid-training (provisional: conditions and counts only)
289
 
@@ -404,7 +404,7 @@ Every row field and every column of the catalog: [`docs/fields.md`](docs/fields.
404
  | `none_known` | 3 | 0 |
405
  | `present` | 14 | 0 |
406
  | `suspected` | 1 | 0 |
407
- | Not judged yet | 121 | 1,801 |
408
  | Sample files read per judged source | 9 to 30 | none |
409
  | Days the sample files were read | 2026-10-04 and 2026-10-05 | none |
410
 
@@ -415,10 +415,10 @@ Every row field and every column of the catalog: [`docs/fields.md`](docs/fields.
415
 
416
  | Sources | Corpora | Task material |
417
  |---|---:|---:|
418
- | With a recorded size for every row | 92 of 139 | 1,801 of 1,801 |
419
- | With a summary read from sample files | 0 of 139 | 0 of 1,801 |
420
- | With personal information read from sample files | 18 of 139 | 0 of 1,801 |
421
- | Still being collected | 25 of 139 | 0 of 1,801 |
422
 
423
  - Rows whose size the source does not publish are shown as unknown and left out of the size totals, never counted as 0.
424
  - Fields not judged yet say so;
@@ -444,9 +444,9 @@ Every row field and every column of the catalog: [`docs/fields.md`](docs/fields.
444
  <details>
445
  <summary>Releases</summary>
446
 
447
- - Latest release: `release-2026-11-11.12` (its newest CHANGELOG entry is dated 2026-11-11).
448
  - Releases are listed in [CHANGELOG.md](CHANGELOG.md);
449
- each one is a commit with its own release tag (this one is [`release-2026-11-11.12`](https://huggingface.co/datasets/Battam/training-data-foundation/tree/release-2026-11-11.12)),
450
  so any earlier state stays readable at its tag.
451
  - `meta/projection.json` records the release tag and the version of the program and templates that generated this card.
452
  - Croissant metadata: [generated by the Hub](https://huggingface.co/api/datasets/Battam/training-data-foundation/croissant);
@@ -479,7 +479,7 @@ A removal covers the current version of this repository (the main branch) and th
479
  @misc{training_data_foundation,
480
  title = {training-data-foundation: an index of public materials for building agent environments},
481
  howpublished = {https://huggingface.co/datasets/Battam/training-data-foundation},
482
- note = {Release release-2026-11-11.12},
483
  year = {2026}
484
  }
485
  ```
 
24
  <!-- artifact-pointer-layout: v2 -->
25
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
26
 
27
+ <img class="not-prose" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/banner.svg" alt="Public web, in two layers (task material in collections; corpora), one row per file (address, size, SHA-256, license), your selection, fetched and verified">
28
 
29
  This dataset is the public index of the **Training Data Foundation**: find public files to build agent tasks and environments from, and fetch them with one command.
30
 
31
  English | [简体中文](README.zh.md) · [Data page](https://huggingface.co/spaces/Battam/training-data-foundation) · [Usage video](https://huggingface.co/spaces/Battam/training-data-foundation#usage-video)
32
 
33
  <picture class="not-prose">
34
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-narrow-light.svg">
35
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-wide-light.svg" alt="1,805 material sets with 4,190 units in 14 collections; 139 corpora with 1,216,362,128 rows; 4.45 PB pointed to, as the index holds addresses, not copies; 99.6% of the 34,857 rows compared match their recorded size.">
36
  </picture>
37
  <picture class="not-prose">
38
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-narrow-dark.svg">
39
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-wide-dark.svg" alt="1,805 material sets with 4,190 units in 14 collections; 139 corpora with 1,216,362,128 rows; 4.45 PB pointed to, as the index holds addresses, not copies; 99.6% of the 34,857 rows compared match their recorded size.">
40
  </picture>
41
 
42
  ## Get data in one command
43
 
44
  ```sh
45
+ uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/tools/fetch.py --form spreadsheets --sample 3 --out ./data
46
  ```
47
 
48
  This fetches 3 units of every spreadsheet material set into `./data`, one folder per unit, and checks each file against the size and hashes its row records.
 
52
  <details>
53
  <summary>What the command takes, and in which order</summary>
54
 
55
+ The command and the queries below are pinned to the release tag `release-2026-11-11.13`.
56
  [uv](https://docs.astral.sh/uv/) reads the script's dependencies, at the versions published before the day of this release.
57
 
58
  The command returns units: 3 units from each material set whose form is `spreadsheets`, each in a folder of its own named after its `unit_id`.
 
79
 
80
  ```sql
81
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
82
+ FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/meta/catalog.parquet'
83
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
84
  ORDER BY collection, units DESC;
85
  ```
 
90
  Choose files in one corpus:
91
 
92
  ```sql
93
+ SET VARIABLE files = (SELECT list('hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/meta/catalog.parquet' WHERE source = 'if-archive'));
94
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
95
  FROM read_parquet(getvariable('files'))
96
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
 
112
  ### Task material: units by collection
113
 
114
  <picture class="not-prose">
115
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-narrow-light.svg">
116
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-wide-light.svg" alt="Units by collection, longest bar first: Spreadsheets 536; Browser tests 439; Tests and exercises 425; Model serving configurations 397; Media and documents 391; Science reference data 342; Rule calculations 329; Security detection 328; Hardware design (EDA) tests 258; Issue trackers 258; Solver instances 193; Disk images 179; Reproduction packages and runs 106; In no collection 9; SEC filing reviews 0.">
117
  </picture>
118
  <picture class="not-prose">
119
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-narrow-dark.svg">
120
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-wide-dark.svg" alt="Units by collection, longest bar first: Spreadsheets 536; Browser tests 439; Tests and exercises 425; Model serving configurations 397; Media and documents 391; Science reference data 342; Rule calculations 329; Security detection 328; Hardware design (EDA) tests 258; Issue trackers 258; Solver instances 193; Disk images 179; Reproduction packages and runs 106; In no collection 9; SEC filing reviews 0.">
121
  </picture>
122
 
123
+ 1,805 material sets with 4,190 units, in 14 collections (1 of them with no material set yet); 1 of the material sets are in no collection.
124
 
125
  <details>
126
  <summary>Task material by collection, as a table</summary>
 
131
  | Disk images | 52 | 179 | 324 | 40 | 272 MB | 272 MB |
132
  | Reproduction packages and runs | 88 | 106 | 870 | 88 | 120 MB | 120 MB |
133
  | SEC filing reviews | 0 | 0 | 0 | 0 | 0 B | 0 B |
134
+ | Solver instances | 69 | 193 | 367 | 66 | 127 MB | 127 MB |
135
+ | Browser tests | 123 | 439 | 852 | 88 | 32.0 MB | 31.9 MB |
136
  | Hardware design (EDA) tests | 81 | 258 | 651 | 75 | 80.8 MB | 80.8 MB |
137
  | Issue trackers | 212 | 258 | 682 | 212 | 196 MB | 196 MB |
138
  | Media and documents | 276 | 391 | 1,088 | 269 | 569 MB | 569 MB |
 
150
  ### Corpora: rows by main domain
151
 
152
  <picture class="not-prose">
153
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-narrow-light.svg">
154
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-wide-light.svg" alt="Corpus rows by main domain, longest bar first: Data science 750,815,212; System administration 185,826,217; Many domains 109,262,742; Software engineering 45,267,701; Science 44,633,675; Operations 39,404,642; Media 12,751,612; Machine learning 9,636,889; Hardware 8,716,520; Unclassified 5,284,162; Security 4,521,631; Misc 241,125.">
155
  </picture>
156
  <picture class="not-prose">
157
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-narrow-dark.svg">
158
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-wide-dark.svg" alt="Corpus rows by main domain, longest bar first: Data science 750,815,212; System administration 185,826,217; Many domains 109,262,742; Software engineering 45,267,701; Science 44,633,675; Operations 39,404,642; Media 12,751,612; Machine learning 9,636,889; Hardware 8,716,520; Unclassified 5,284,162; Security 4,521,631; Misc 241,125.">
159
  </picture>
160
 
161
  139 corpora with 1,216,362,128 rows.
 
188
  ## How it is made
189
 
190
  <picture class="not-prose">
191
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-narrow-light.svg">
192
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-wide-light.svg" alt="How the index is made, in four steps. 1, collect: collectors read whole upstreams into corpora, one row per file or record. 2, admit: works the search found upstream and groups of corpus rows become material sets. 3, check: licenses, personal data, overlap with benchmarks and sizes are checked before release. 4, release: one commit with its own tag, release-2026-11-11.13.">
193
  </picture>
194
  <picture class="not-prose">
195
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-narrow-dark.svg">
196
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-wide-dark.svg" alt="How the index is made, in four steps. 1, collect: collectors read whole upstreams into corpora, one row per file or record. 2, admit: works the search found upstream and groups of corpus rows become material sets. 3, check: licenses, personal data, overlap with benchmarks and sizes are checked before release. 4, release: one commit with its own tag, release-2026-11-11.13.">
197
  </picture>
198
 
199
  The terms, each step and how every number is made are folded below.
 
247
 
248
  | Level | Recorded size | Deduplicated size | Rows read | Rows compared | Rows that match | Rows read, not compared | Rows without a recorded size | Their estimated size |
249
  |---|---:|---:|---:|---:|---:|---:|---:|---:|
250
+ | Whole index | 4.45 PB | 4.11 PB | 35,736 | 34,857 | 34,727 (99.6%) | 879: address unreachable 449, file gone 80, login needed 329, no size from the server 19, other error 2 | 312,300,338 | 202 TB, an estimate (95% range 171 TB to 245 TB), for 215,326,513 of them, from 14,161 rows read |
251
  | Corpora | 4.45 PB | 4.11 PB | 31,594 | 30,723 | 30,610 (99.6%) | 871: address unreachable 449, file gone 80, login needed 322, no size from the server 19, other error 1 | 312,300,338 | 202 TB, an estimate (95% range 171 TB to 245 TB), for 215,326,513 of them, from 14,161 rows read |
252
+ | Task material | 2.90 GB | 2.82 GB | 4,142 | 4,134 | 4,117 (99.6%) | 8: login needed 7, other error 1 | 0 | none |
253
 
254
  Rows read are the sample of rows the check reads in `meta/sizes.parquet`;
255
  a row compared matched its recorded size or did not, and a row read but not compared never reached its file.
 
258
 
259
  The recorded size is the sum of `size_bytes` over the rows that have one (catalog column `bytes`). The check reads a sample of rows of every data file (`meta/sizes.parquet`): 20 of its rows with a recorded size, and of its rows without one its share of its source's sample, sized for their mean size within 25% at 95% confidence by the spread of the sizes read so far (at most 400 rows per source per round, 20 for a source with none read yet); its share is the rows that match their recorded size out of the rows it compared (status `match`, `size_mismatch` or `hash_mismatch`), and the rows it could not compare, which never reached their file, are counted beside it by status (`unreachable`, `gone`, `needs_auth`, `no_length`, `error`; catalog column `size_check`). The estimate for a source's rows without a recorded size (catalog columns `unsized_rows` and `unsized_estimate`) is their number times the mean measured size of such rows read, with a 95% BCa bootstrap range (`scipy.stats.bootstrap`, its default resample count); a source with fewer than 10 such rows read has none. Above one source, the estimate is the sum of its sources' estimates and its range the sum of their ranges, which is wider than a joint interval. The figures of every level are in `meta/size_figures.json`.
260
 
261
+ Distinct content is the sum of `size_bytes` over the rows minus the sizes of the rows that repeat another row's content (the same sha256, else sha1, else md5 and size, within the same hash scope; one copy of each content is kept, in the first source by registry order), counted after each release over the data files it published, per source in the catalog column `distinct_bytes`; archive members whose archive the same source also lists whole have other hashes, so they are not subtracted. `tools/fetch.py` removes duplicates over your own selection when it reads it. The deduplicated sizes here were counted over the data files of release `release-2026-11-11.10`.
262
 
263
  </details>
264
 
 
271
  Each table below gives what a training stage can take, counted before the rules and after each of them, in their order;
272
  its last row is the stage's count.
273
 
274
+ The stage fields (`meta/stage_fields.parquet`) were counted over the data files of release `release-2026-11-11.10`, after it was published.
275
 
276
  #### Post-training (in use)
277
 
 
279
 
280
  | Step | Material sets | Units | Upstream sources |
281
  |---|---:|---:|---:|
282
+ | In the stage's condition, before the rules | 1,426 | 3,479 | 1,350 |
283
+ | After the benchmark rule | 1,425 | 3,478 | 1,349 |
284
+ | After the rule on rows not compared with the benchmark list | 1,425 | 3,478 | 1,349 |
285
+ | After the personal-data rule | 1,267 | 3,084 | 1,201 |
286
+ | After deduplication: the stage's count | 1,244 | 2,989 | 1,183 |
287
 
288
  #### Mid-training (provisional: conditions and counts only)
289
 
 
404
  | `none_known` | 3 | 0 |
405
  | `present` | 14 | 0 |
406
  | `suspected` | 1 | 0 |
407
+ | Not judged yet | 121 | 1,805 |
408
  | Sample files read per judged source | 9 to 30 | none |
409
  | Days the sample files were read | 2026-10-04 and 2026-10-05 | none |
410
 
 
415
 
416
  | Sources | Corpora | Task material |
417
  |---|---:|---:|
418
+ | With a recorded size for every row | 92 of 139 | 1,805 of 1,805 |
419
+ | With a summary read from sample files | 0 of 139 | 0 of 1,805 |
420
+ | With personal information read from sample files | 18 of 139 | 0 of 1,805 |
421
+ | Still being collected | 25 of 139 | 0 of 1,805 |
422
 
423
  - Rows whose size the source does not publish are shown as unknown and left out of the size totals, never counted as 0.
424
  - Fields not judged yet say so;
 
444
  <details>
445
  <summary>Releases</summary>
446
 
447
+ - Latest release: `release-2026-11-11.13` (its newest CHANGELOG entry is dated 2026-11-11).
448
  - Releases are listed in [CHANGELOG.md](CHANGELOG.md);
449
+ each one is a commit with its own release tag (this one is [`release-2026-11-11.13`](https://huggingface.co/datasets/Battam/training-data-foundation/tree/release-2026-11-11.13)),
450
  so any earlier state stays readable at its tag.
451
  - `meta/projection.json` records the release tag and the version of the program and templates that generated this card.
452
  - Croissant metadata: [generated by the Hub](https://huggingface.co/api/datasets/Battam/training-data-foundation/croissant);
 
479
  @misc{training_data_foundation,
480
  title = {training-data-foundation: an index of public materials for building agent environments},
481
  howpublished = {https://huggingface.co/datasets/Battam/training-data-foundation},
482
+ note = {Release release-2026-11-11.13},
483
  year = {2026}
484
  }
485
  ```
README.zh.md CHANGED
@@ -11,25 +11,25 @@ tags:
11
  ---
12
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
13
 
14
- <img class="not-prose" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/banner.zh.svg" alt="公开网络,分两层(按合集的任务材料;语料库),每个文件一行(地址、大小、SHA-256、许可证),你的选择,下载并校验">
15
 
16
  本数据集是**数据基座**的公开索引:在这里找到可以用来构建智能体任务和环境的公开文件,并用一条命令把它们取下来。
17
 
18
  [English](README.md) | 简体中文 · [数据页](https://huggingface.co/spaces/Battam/training-data-foundation) · [使用示例视频](https://huggingface.co/spaces/Battam/training-data-foundation#usage-video)
19
 
20
  <picture class="not-prose">
21
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-narrow-light.zh.svg">
22
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-wide-light.zh.svg" alt="1,801 个材料集,共 4,186 个单元,分在 14 个合集;139 个语料库,共 1,216,362,128 行;指向的大小共 4.45 PB,索引存的是地址,不是文件副本;比对过的 35,277 行中,99.6% 与记录的大小一致。">
23
  </picture>
24
  <picture class="not-prose">
25
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-narrow-dark.zh.svg">
26
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/numbers-wide-dark.zh.svg" alt="1,801 个材料集,共 4,186 个单元,分在 14 个合集;139 个语料库,共 1,216,362,128 行;指向的大小共 4.45 PB,索引存的是地址,不是文件副本;比对过的 35,277 行中,99.6% 与记录的大小一致。">
27
  </picture>
28
 
29
  ## 一条命令取数据
30
 
31
  ```sh
32
- uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/tools/fetch.py --form spreadsheets --sample 3 --out ./data
33
  ```
34
 
35
  这条命令从每个电子表格材料集各取 3 个单元放进 `./data`,每个单元一个文件夹,并按每个文件所在行记录的大小和哈希值校验它。
@@ -39,7 +39,7 @@ uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/r
39
  <details>
40
  <summary>命令取什么、按什么顺序取</summary>
41
 
42
- 这条命令和下面的查询都固定在发布标签 `release-2026-11-11.12` 上。
43
  [uv](https://docs.astral.sh/uv/) 会读取脚本声明的依赖,取这个发布当天之前发布的版本。
44
 
45
  命令返回单元:形式为 `spreadsheets` 的每个材料集各 3 个单元,每个单元放在以它的 `unit_id` 命名的单独文件夹里。
@@ -66,7 +66,7 @@ uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/r
66
 
67
  ```sql
68
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
69
- FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/meta/catalog.parquet'
70
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
71
  ORDER BY collection, units DESC;
72
  ```
@@ -77,7 +77,7 @@ ORDER BY collection, units DESC;
77
  在一个语料库里挑选文件:
78
 
79
  ```sql
80
- SET VARIABLE files = (SELECT list('hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.12/meta/catalog.parquet' WHERE source = 'if-archive'));
81
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
82
  FROM read_parquet(getvariable('files'))
83
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
@@ -99,15 +99,15 @@ ORDER BY size_bytes LIMIT 20;
99
  ### 任务材料:按合集的单元数
100
 
101
  <picture class="not-prose">
102
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-narrow-light.zh.svg">
103
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-wide-light.zh.svg" alt="按合集的单元数,条形由长到短:电子表格 536;浏览器测试 439;测试和练习 425;模型服务配置 397;媒体和文档 391;科学参考数据 342;规则计算 329;安全检测 328;硬件设计(EDA)测试 258;问题跟踪系统 258;求解器实例 189;磁盘镜像 179;复现包和运行 106;不属于任何合集 9;SEC 申报审查 0。">
104
  </picture>
105
  <picture class="not-prose">
106
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-narrow-dark.zh.svg">
107
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/material-wide-dark.zh.svg" alt="按合集的单元数,条形由长到短:电子表格 536;浏览器测试 439;测试和练习 425;模型服务配置 397;媒体和文档 391;科学参考数据 342;规则计算 329;安全检测 328;硬件设计(EDA)测试 258;问题跟踪系统 258;求解器实例 189;磁盘镜像 179;复现包和运行 106;不属于任何合集 9;SEC 申报审查 0。">
108
  </picture>
109
 
110
- 1,801 个材料集,共 4,186 个单元,分在 14 个合集中(其中 1 个合集还没有材料集);其中 1 个材料集不属于任何合集。
111
 
112
  <details>
113
  <summary>按合集的任务材料(表格)</summary>
@@ -118,8 +118,8 @@ ORDER BY size_bytes LIMIT 20;
118
  | 磁盘镜像 | 52 | 179 | 324 | 40 | 272 MB | 272 MB |
119
  | 复现包和运行 | 88 | 106 | 870 | 88 | 120 MB | 120 MB |
120
  | SEC 申报审查 | 0 | 0 | 0 | 0 | 0 B | 0 B |
121
- | 求解器实例 | 65 | 189 | 342 | 62 | 127 MB | 127 MB |
122
- | 浏览器测试 | 123 | 439 | 852 | 88 | 32.0 MB | 32.0 MB |
123
  | 硬件设计(EDA)测试 | 81 | 258 | 651 | 75 | 80.8 MB | 80.8 MB |
124
  | 问题跟踪系统 | 212 | 258 | 682 | 212 | 196 MB | 196 MB |
125
  | 媒体和文档 | 276 | 391 | 1,088 | 269 | 569 MB | 569 MB |
@@ -137,12 +137,12 @@ ORDER BY size_bytes LIMIT 20;
137
  ### 语料库:按主要领域的行数
138
 
139
  <picture class="not-prose">
140
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-narrow-light.zh.svg">
141
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-wide-light.zh.svg" alt="按主要领域的语料库行数,条形由长到短:数据科学 750,815,212;系统管理 185,826,217;多领域 109,262,742;软件工程 45,267,701;科学 44,633,675;运营 39,404,642;媒体 12,751,612;机器学习 9,636,889;硬件 8,716,520;未分类 5,284,162;安全 4,521,631;杂项 241,125。">
142
  </picture>
143
  <picture class="not-prose">
144
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-narrow-dark.zh.svg">
145
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/corpora-wide-dark.zh.svg" alt="按主要领域的语料库行数,条形由长到短:数据科学 750,815,212;系统管理 185,826,217;多领域 109,262,742;软件工程 45,267,701;科学 44,633,675;运营 39,404,642;媒体 12,751,612;机器学习 9,636,889;硬件 8,716,520;未分类 5,284,162;安全 4,521,631;杂项 241,125。">
146
  </picture>
147
 
148
  139 个语料库,共 1,216,362,128 行。
@@ -175,12 +175,12 @@ ORDER BY size_bytes LIMIT 20;
175
  ## 怎样做成
176
 
177
  <picture class="not-prose">
178
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-narrow-light.zh.svg">
179
- <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-wide-light.zh.svg" alt="索引分四步做成。1,采集:采集器整批读取完整上游,每个文件或每条记录记一行,成为语料库。2,收录:检索在上游找到的作品和成组的语料库行,收录为材料集。3,检查:发布前核对许可证、个人信息、与基准测试的重叠和大小。4,发布:一个带有自己标签的提交,release-2026-11-11.12。">
180
  </picture>
181
  <picture class="not-prose">
182
- <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-narrow-dark.zh.svg">
183
- <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.12/docs/figures/steps-wide-dark.zh.svg" alt="索引分四步做成。1,采集:采集器整批读取完整上游,每个文件或每条记录记一行,成为语料库。2,收录:检索在上游找到的作品和成组的语料库行,收录为材料集。3,检查:发布前核对许可证、个人信息、与基准测试的重叠和大小。4,发布:一个带有自己标签的提交,release-2026-11-11.12。">
184
  </picture>
185
 
186
  术语、每一步,以及每个数字怎样得出,都折叠在下面。
@@ -234,9 +234,9 @@ ORDER BY size_bytes LIMIT 20;
234
 
235
  | 层级 | 记录的大小 | 去重后大小 | 读过的行 | 比对过的行 | 一致的行 | 读过但未比对的行 | 没有记录大小的行 | 这些行的估计大小 |
236
  |---|---:|---:|---:|---:|---:|---:|---:|---:|
237
- | 整个索引 | 4.45 PB | 4.11 PB | 36,158 | 35,277 | 35,147(99.6%) | 881:地址无法访问 449 行、文件已不存在 80 行、需要登录 330 行、服务器未给出大小 19 行、其他错误 3 行 | 312,300,338 | 202 TB,这是估计值(95% 范围 171 TB 到 245 TB),针对其中 215,326,513 行,依据读过的 14,161 行 |
238
  | 语料库 | 4.45 PB | 4.11 PB | 31,594 | 30,723 | 30,610(99.6%) | 871:地址无法访问 449 行、文件已不存在 80 行、需要登录 322 行、服务器未给出大小 19 行、其他错误 1 行 | 312,300,338 | 202 TB,这是估计值(95% 范围 171 TB 到 245 TB),针对其中 215,326,513 行,依据读过的 14,161 行 |
239
- | 任务材料 | 2.90 GB | 2.82 GB | 4,564 | 4,554 | 4,537(99.6%) | 10:需要登录 8 行、其他错误 2 行 | 0 | 无 |
240
 
241
  读过的行是核对在 `meta/sizes.parquet` 中读取的抽样行;
242
  比对过的行与记录的大小一致或不一致,读过但未比对的行没有取到它的文件。
@@ -245,7 +245,7 @@ ORDER BY size_bytes LIMIT 20;
245
 
246
  记录的大小是有大小的各行 `size_bytes` 之和(目录列 `bytes`)。核对读取每个数据文件中抽取的一组行(`meta/sizes.parquet`):有记录大小的行抽 20 行;没有记录大小的行按其所在来源的样本分摊,样本量按目前读到的大小的离散程度定,使其平均大小在 95% 置信度下误差不超过 25%(每个来源每轮最多 400 行,还没有读数的来源取 20 行);其比例是比对过的行(状态 `match`、`size_mismatch` 或 `hash_mismatch`)中与记录的大小一致的行所占的比例,没能比对的行(没有取到文件)按状态另行计数(`unreachable`、`gone`、`needs_auth`、`no_length`、`error`;目录列 `size_check`)。一个来源没有记录大小的行的估计(目录列 `unsized_rows` 和 `unsized_estimate`)是这些行的行数乘以读过的这类行的平均实测大小,并给出 95% BCa 自助法范围(`scipy.stats.bootstrap`,用其默认重抽样次数);读过的这类行少于 10 行的来源没有估计。在来源之上的层级,估计是其各来源估计之和,范围是各来源范围之和,比联合区间更宽。每个层级的数字都在 `meta/size_figures.json` 中。
247
 
248
- 去重后内容是各行 `size_bytes` 之和,减去与另一行内容重复的那些行的大小(sha256 相同;没有 sha256 时 sha1 相同;再没有时 md5 和大小都相同;并且哈希范围相同;每种内容只保留一份,留在按注册顺序排在最前的来源中),在每次发布之后按它发布的数据文件重新计算,每个来源的值在目录列 `distinct_bytes` 中;同一来源也整个列出的存档,其成员的哈希与存档不同,所以不扣除。`tools/fetch.py` 读取你自己的选择时会去掉其中的重复。 这里的去重后大小是按发布 `release-2026-11-11.3` 的数据文件算���的。
249
 
250
  </details>
251
 
@@ -258,7 +258,7 @@ ORDER BY size_bytes LIMIT 20;
258
  下面每张表给出一个训练阶段能用的材料,在各条规则之前和按顺序每条规则之后计数;
259
  每张表的最后一行是该阶段的计数。
260
 
261
- 阶段字段(`meta/stage_fields.parquet`)是按发布 `release-2026-11-11.3` 的数据文件、在它发布之后算出的。
262
 
263
  #### 后训练(在用)
264
 
@@ -266,11 +266,11 @@ ORDER BY size_bytes LIMIT 20;
266
 
267
  | 步骤 | 材料集数 | 单元数 | 上游来源数 |
268
  |---|---:|---:|---:|
269
- | 符合该阶段的条件,在各条规则之前 | 1,425 | 3,477 | 1,349 |
270
- | 基准测试规则之后 | 1,425 | 3,477 | 1,349 |
271
- | 未与基准测试清单比对的行的规则之后 | 1,425 | 3,477 | 1,349 |
272
- | 个人数据规则之后 | 1,267 | 3,083 | 1,201 |
273
- | 去重之后:该阶段的计数 | 1,244 | 2,988 | 1,183 |
274
 
275
  #### 中训练(暂定:只给条件和计数)
276
 
@@ -393,7 +393,7 @@ ORDER BY size_bytes LIMIT 20;
393
  | `none_known` | 3 | 0 |
394
  | `present` | 14 | 0 |
395
  | `suspected` | 1 | 0 |
396
- | 尚未判读 | 121 | 1,801 |
397
  | 每个已判读的来源读过的样本文件数 | 9 到 30 | 无 |
398
  | 读取样本文件的日期 | 2026-10-04 和 2026-10-05 | 无 |
399
 
@@ -404,10 +404,10 @@ ORDER BY size_bytes LIMIT 20;
404
 
405
  | 来源 | 语料库 | 任务材料 |
406
  |---|---:|---:|
407
- | 每一行都有记录的大小 | 139 个中的 92 个 | 1,801 个中的 1,801 个 |
408
- | 已从样本文件读出摘要 | 139 个中的 0 个 | 1,801 个中的 0 个 |
409
- | 已从样本文件读出个人信息 | 139 个中的 18 个 | 1,801 个中的 0 个 |
410
- | 仍在采集中 | 139 个中的 25 个 | 1,801 个中的 0 个 |
411
 
412
  - 来源没有公布大小的行显示为未知,不计入大小合计,也从不按 0 计。
413
  - 尚未判读的字段会注明;
@@ -433,9 +433,9 @@ ORDER BY size_bytes LIMIT 20;
433
  <details>
434
  <summary>发布</summary>
435
 
436
- - 最新发布:`release-2026-11-11.12`(其最新的 CHANGELOG 条目日期为 2026-11-11)。
437
  - 各次发布列在 [CHANGELOG.md](CHANGELOG.md) 中;
438
- 每次发布是一个带有自己发布标签的提交(本次为 [`release-2026-11-11.12`](https://huggingface.co/datasets/Battam/training-data-foundation/tree/release-2026-11-11.12)),
439
  因此任何较早的状态都可以在它的标签处读到。
440
  - `meta/projection.json` 记录发布标签,以及生成本卡片的程序和模板的版本。
441
  - Croissant 元数据:[由 Hub 生成](https://huggingface.co/api/datasets/Battam/training-data-foundation/croissant);
@@ -476,7 +476,7 @@ ORDER BY size_bytes LIMIT 20;
476
  @misc{training_data_foundation,
477
  title = {training-data-foundation: an index of public materials for building agent environments},
478
  howpublished = {https://huggingface.co/datasets/Battam/training-data-foundation},
479
- note = {Release release-2026-11-11.12},
480
  year = {2026}
481
  }
482
  ```
 
11
  ---
12
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
13
 
14
+ <img class="not-prose" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/banner.zh.svg" alt="公开网络,分两层(按合集的任务材料;语料库),每个文件一行(地址、大小、SHA-256、许可证),你的选择,下载并校验">
15
 
16
  本数据集是**数据基座**的公开索引:在这里找到可以用来构建智能体任务和环境的公开文件,并用一条命令把它们取下来。
17
 
18
  [English](README.md) | 简体中文 · [数据页](https://huggingface.co/spaces/Battam/training-data-foundation) · [使用示例视频](https://huggingface.co/spaces/Battam/training-data-foundation#usage-video)
19
 
20
  <picture class="not-prose">
21
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-narrow-light.zh.svg">
22
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-wide-light.zh.svg" alt="1,805 个材料集,共 4,190 个单元,分在 14 个合集;139 个语料库,共 1,216,362,128 行;指向的大小共 4.45 PB,索引存的是地址,不是文件副本;比对过的 34,857 行中,99.6% 与记录的大小一致。">
23
  </picture>
24
  <picture class="not-prose">
25
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-narrow-dark.zh.svg">
26
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/numbers-wide-dark.zh.svg" alt="1,805 个材料集,共 4,190 个单元,分在 14 个合集;139 个语料库,共 1,216,362,128 行;指向的大小共 4.45 PB,索引存的是地址,不是文件副本;比对过的 34,857 行中,99.6% 与记录的大小一致。">
27
  </picture>
28
 
29
  ## 一条命令取数据
30
 
31
  ```sh
32
+ uv run https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/tools/fetch.py --form spreadsheets --sample 3 --out ./data
33
  ```
34
 
35
  这条命令从每个电子表格材料集各取 3 个单元放进 `./data`,每个单元一个文件夹,并按每个文件所在行记录的大小和哈希值校验它。
 
39
  <details>
40
  <summary>命令取什么、按什么顺序取</summary>
41
 
42
+ 这条命令和下面的查询都固定在发布标签 `release-2026-11-11.13` 上。
43
  [uv](https://docs.astral.sh/uv/) 会读取脚本声明的依赖,取这个发布当天之前发布的版本。
44
 
45
  命令返回单元:形式为 `spreadsheets` 的每个材料集各 3 个单元,每个单元放在以它的 `unit_id` 命名的单独文件夹里。
 
66
 
67
  ```sql
68
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
69
+ FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/meta/catalog.parquet'
70
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
71
  ORDER BY collection, units DESC;
72
  ```
 
77
  在一个语料库里挑选文件:
78
 
79
  ```sql
80
+ SET VARIABLE files = (SELECT list('hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/training-data-foundation@release-2026-11-11.13/meta/catalog.parquet' WHERE source = 'if-archive'));
81
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
82
  FROM read_parquet(getvariable('files'))
83
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
 
99
  ### 任务材料:按合集的单元数
100
 
101
  <picture class="not-prose">
102
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-narrow-light.zh.svg">
103
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-wide-light.zh.svg" alt="按合集的单元数,条形由长到短:电子表格 536;浏览器测试 439;测试和练习 425;模型服务配置 397;媒体和文档 391;科学参考数据 342;规则计算 329;安全检测 328;硬件设计(EDA)测试 258;问题跟踪系统 258;求解器实例 193;磁盘镜像 179;复现包和运行 106;不属于任何合集 9;SEC 申报审查 0。">
104
  </picture>
105
  <picture class="not-prose">
106
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-narrow-dark.zh.svg">
107
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/material-wide-dark.zh.svg" alt="按合集的单元数,条形由长到短:电子表格 536;浏览器测试 439;测试和练习 425;模型服务配置 397;媒体和文档 391;科学参考数据 342;规则计算 329;安全检测 328;硬件设计(EDA)测试 258;问题跟踪系统 258;求解器实例 193;磁盘镜像 179;复现包和运行 106;不属于任何合集 9;SEC 申报审查 0。">
108
  </picture>
109
 
110
+ 1,805 个材料集,共 4,190 个单元,分在 14 个合集中(其中 1 个合集还没有材料集);其中 1 个材料集不属于任何合集。
111
 
112
  <details>
113
  <summary>按合集的任务材料(表格)</summary>
 
118
  | 磁盘镜像 | 52 | 179 | 324 | 40 | 272 MB | 272 MB |
119
  | 复现包和运行 | 88 | 106 | 870 | 88 | 120 MB | 120 MB |
120
  | SEC 申报审查 | 0 | 0 | 0 | 0 | 0 B | 0 B |
121
+ | 求解器实�� | 69 | 193 | 367 | 66 | 127 MB | 127 MB |
122
+ | 浏览器测试 | 123 | 439 | 852 | 88 | 32.0 MB | 31.9 MB |
123
  | 硬件设计(EDA)测试 | 81 | 258 | 651 | 75 | 80.8 MB | 80.8 MB |
124
  | 问题跟踪系统 | 212 | 258 | 682 | 212 | 196 MB | 196 MB |
125
  | 媒体和文档 | 276 | 391 | 1,088 | 269 | 569 MB | 569 MB |
 
137
  ### 语料库:按主要领域的行数
138
 
139
  <picture class="not-prose">
140
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-narrow-light.zh.svg">
141
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-wide-light.zh.svg" alt="按主要领域的语料库行数,条形由长到短:数据科学 750,815,212;系统管理 185,826,217;多领域 109,262,742;软件工程 45,267,701;科学 44,633,675;运营 39,404,642;媒体 12,751,612;机器学习 9,636,889;硬件 8,716,520;未分类 5,284,162;安全 4,521,631;杂项 241,125。">
142
  </picture>
143
  <picture class="not-prose">
144
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-narrow-dark.zh.svg">
145
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/corpora-wide-dark.zh.svg" alt="按主要领域的语料库行数,条形由长到短:数据科学 750,815,212;系统管理 185,826,217;多领域 109,262,742;软件工程 45,267,701;科学 44,633,675;运营 39,404,642;媒体 12,751,612;机器学习 9,636,889;硬件 8,716,520;未分类 5,284,162;安全 4,521,631;杂项 241,125。">
146
  </picture>
147
 
148
  139 个语料库,共 1,216,362,128 行。
 
175
  ## 怎样做成
176
 
177
  <picture class="not-prose">
178
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-narrow-light.zh.svg">
179
+ <img class="not-prose dark:hidden" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-wide-light.zh.svg" alt="索引分四步做成。1,采集:采集器整批读取完整上游,每个文件或每条记录记一行,成为语料库。2,收录:检索在上游找到的作品和成组的语料库行,收录为材料集。3,检查:发布前核对许可证、个人信息、与基准测试的重叠和大小。4,发布:一个带有自己标签的提交,release-2026-11-11.13。">
180
  </picture>
181
  <picture class="not-prose">
182
+ <source media="(max-width: 1023px)" srcset="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-narrow-dark.zh.svg">
183
+ <img class="not-prose hidden dark:block" src="https://huggingface.co/datasets/Battam/training-data-foundation/resolve/release-2026-11-11.13/docs/figures/steps-wide-dark.zh.svg" alt="索引分四步做成。1,采集:采集器整批读取完整上游,每个文件或每条记录记一行,成为语料库。2,收录:检索在上游找到的作品和成组的语料库行,收录为材料集。3,检查:发布前核对许可证、个人信息、与基准测试的重叠和大小。4,发布:一个带有自己标签的提交,release-2026-11-11.13。">
184
  </picture>
185
 
186
  术语、每一步,以及每个数字怎样得出,都折叠在下面。
 
234
 
235
  | 层级 | 记录的大小 | 去重后大小 | 读过的行 | 比对过的行 | 一致的行 | 读过但未比对的行 | 没有记录大小的行 | 这些行的估计大小 |
236
  |---|---:|---:|---:|---:|---:|---:|---:|---:|
237
+ | 整个索引 | 4.45 PB | 4.11 PB | 35,736 | 34,857 | 34,727(99.6%) | 879:地址无法访问 449 行、文件已不存在 80 行、需要登录 329 行、服务器未给出大小 19 行、其他错误 2 行 | 312,300,338 | 202 TB,这是估计值(95% 范围 171 TB 到 245 TB),针对其中 215,326,513 行,依据读过的 14,161 行 |
238
  | 语料库 | 4.45 PB | 4.11 PB | 31,594 | 30,723 | 30,610(99.6%) | 871:地址无法访问 449 行、文件已不存在 80 行、需要登录 322 行、服务器未给出大小 19 行、其他错误 1 行 | 312,300,338 | 202 TB,这是估计值(95% 范围 171 TB 到 245 TB),针对其中 215,326,513 行,依据读过的 14,161 行 |
239
+ | 任务材料 | 2.90 GB | 2.82 GB | 4,142 | 4,134 | 4,117(99.6%) | 8:需要登录 7 行、其他错误 1 行 | 0 | 无 |
240
 
241
  读过的行是核对在 `meta/sizes.parquet` 中读取的抽样行;
242
  比对过的行与记录的大小一致或不一致,读过但未比对的行没有取到它的文件。
 
245
 
246
  记录的大小是有大小的各行 `size_bytes` 之和(目录列 `bytes`)。核对读取每个数据文件中抽取的一组行(`meta/sizes.parquet`):有记录大小的行抽 20 行;没有记录大小的行按其所在来源的样本分摊,样本量按目前读到的大小的离散程度定,使其平均大小在 95% 置信度下误差不超过 25%(每个来源每轮最多 400 行,还没有读数的来源取 20 行);其比例是比对过的行(状态 `match`、`size_mismatch` 或 `hash_mismatch`)中与记录的大小一致的行所占的比例,没能比对的行(没有取到文件)按状态另行计数(`unreachable`、`gone`、`needs_auth`、`no_length`、`error`;目录列 `size_check`)。一个来源没有记录大小的行的估计(目录列 `unsized_rows` 和 `unsized_estimate`)是这些行的行数乘以读过的这类行的平均实测大小,并给出 95% BCa 自助法范围(`scipy.stats.bootstrap`,用其默认重抽样次数);读过的这类行少于 10 行的来源没有估计。在来源之上的层级,估计是其各来源估计之和,范围是各来源范围之和,比联合区间更宽。每个层级的数字都在 `meta/size_figures.json` 中。
247
 
248
+ 去重后内容是各行 `size_bytes` 之和,减去与另一行内容重复的那些行的大小(sha256 相同;没有 sha256 时 sha1 相同;再没有时 md5 和大小都相同;并且哈希范围相同;每种内容只保留一份,留在按注册顺序排在最前的来源中),在每次发布之后按它发布的数据文件重新计算,每个来源的值在目录列 `distinct_bytes` 中;同一来源也整个列出的存档,其成员的哈希与存档不同,所以不扣除。`tools/fetch.py` 读取你自己的选择时会去掉其中的重复。 这里的去重后大小是按发布 `release-2026-11-11.10` 的数据文件算出的。
249
 
250
  </details>
251
 
 
258
  下面每张表给出一个训练阶段能用的材料,在各条规则之前和按顺序每条规则之后计数;
259
  每张表的最后一行是该阶段的计数。
260
 
261
+ 阶段字段(`meta/stage_fields.parquet`)是按发布 `release-2026-11-11.10` 的数据文件、在它发布之后算出的。
262
 
263
  #### 后训练(在用)
264
 
 
266
 
267
  | 步骤 | 材料集数 | 单元数 | 上游来源数 |
268
  |---|---:|---:|---:|
269
+ | 符合该阶段的条件,在各条规则之前 | 1,426 | 3,479 | 1,350 |
270
+ | 基准测试规则之后 | 1,425 | 3,478 | 1,349 |
271
+ | 未与基准测试清单比对的行的规则之后 | 1,425 | 3,478 | 1,349 |
272
+ | 个人数据规则之后 | 1,267 | 3,084 | 1,201 |
273
+ | 去重之后:该阶段的计数 | 1,244 | 2,989 | 1,183 |
274
 
275
  #### 中训练(暂定:只给条件和计数)
276
 
 
393
  | `none_known` | 3 | 0 |
394
  | `present` | 14 | 0 |
395
  | `suspected` | 1 | 0 |
396
+ | 尚未判读 | 121 | 1,805 |
397
  | 每个已判读的来源读过的样本文件数 | 9 到 30 | 无 |
398
  | 读取样本文件的日期 | 2026-10-04 和 2026-10-05 | 无 |
399
 
 
404
 
405
  | 来源 | 语料库 | 任务材料 |
406
  |---|---:|---:|
407
+ | 每一行都有记录的大小 | 139 个中的 92 个 | 1,805 个中的 1,805 个 |
408
+ | 已从样本文件读出摘要 | 139 个中的 0 个 | 1,805 个中的 0 个 |
409
+ | 已从样本文件读出个人信息 | 139 个中的 18 个 | 1,805 个中的 0 个 |
410
+ | 仍在采集中 | 139 个中的 25 个 | 1,805 个中的 0 个 |
411
 
412
  - 来源没有公布大小的行显示为未知,不计入大小合计,也从不按 0 计。
413
  - 尚未判读的字段会注明;
 
433
  <details>
434
  <summary>发布</summary>
435
 
436
+ - 最新发布:`release-2026-11-11.13`(其最新的 CHANGELOG 条目日期为 2026-11-11)。
437
  - 各次发布列在 [CHANGELOG.md](CHANGELOG.md) 中;
438
+ 每次发布是一个带有自己发布标签的提交(本次为 [`release-2026-11-11.13`](https://huggingface.co/datasets/Battam/training-data-foundation/tree/release-2026-11-11.13)),
439
  因此任何较早的状态都可以在它的标签处读到。
440
  - `meta/projection.json` 记录发布标签,以及生成本卡片的程序和模板的版本。
441
  - Croissant 元数据:[由 Hub 生成](https://huggingface.co/api/datasets/Battam/training-data-foundation/croissant);
 
476
  @misc{training_data_foundation,
477
  title = {training-data-foundation: an index of public materials for building agent environments},
478
  howpublished = {https://huggingface.co/datasets/Battam/training-data-foundation},
479
+ note = {Release release-2026-11-11.13},
480
  year = {2026}
481
  }
482
  ```
campaign_files/2026-10-b-disk-images-2/libarchive-iso-malformed/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each unit needs other upstream files is not stated.
10
 
11
- **Answers:** Each unit has one answer file, at answer/<unit path>.json (the unit path plus .json). Reference result kind: corruption_desc.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .uu.
14
 
15
  **Licence (SPDX):** BSD-2-Clause
@@ -20,19 +18,13 @@ Units:
20
 
21
  - `images/test_read_format_iso_rockridge_ce_loop.iso.Z.uu`: Libarchive test image: Rock Ridge ISO with a self-referential CE continuation area.
22
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_ce_loop.iso.Z.uu
23
- Reference result: corruption_desc
24
  - `images/test_read_format_iso_rockridge_ce_overflow.iso.Z.uu`: Libarchive test image: Rock Ridge ISO with a CE location of 0xFFFFFFFF.
25
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_ce_overflow.iso.Z.uu
26
- Reference result: corruption_desc
27
  - `images/test_read_format_iso_rockridge_cl_re_cycle.iso.Z.uu`: Libarchive test image: Rock Ridge ISO whose CL child links form a parent cycle without RE.
28
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_cl_re_cycle.iso.Z.uu
29
- Reference result: corruption_desc
30
  - `images/test_read_format_iso_joliet_utf16be_overflow.iso.uu`: Libarchive test image: Joliet ISO whose UTF-16BE path fills the 1024-byte path buffer.
31
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_joliet_utf16be_overflow.iso.uu
32
- Reference result: corruption_desc
33
  - `images/test_read_format_iso_rockridge_zf_overflow.iso.uu`: Libarchive test image: Rock Ridge ZF entry of length 5 at the very end of the image.
34
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_zf_overflow.iso.uu
35
- Reference result: corruption_desc
36
  - `images/test_read_format_iso_zisofs_overflow.iso.uu`: Libarchive test image: zisofs ZF entry with block-size exponent 2 and a near-4-GiB uncompressed size.
37
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_zisofs_overflow.iso.uu
38
- Reference result: corruption_desc
 
8
 
9
  **Self-contained:** Whether each unit needs other upstream files is not stated.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .uu.
12
 
13
  **Licence (SPDX):** BSD-2-Clause
 
18
 
19
  - `images/test_read_format_iso_rockridge_ce_loop.iso.Z.uu`: Libarchive test image: Rock Ridge ISO with a self-referential CE continuation area.
20
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_ce_loop.iso.Z.uu
 
21
  - `images/test_read_format_iso_rockridge_ce_overflow.iso.Z.uu`: Libarchive test image: Rock Ridge ISO with a CE location of 0xFFFFFFFF.
22
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_ce_overflow.iso.Z.uu
 
23
  - `images/test_read_format_iso_rockridge_cl_re_cycle.iso.Z.uu`: Libarchive test image: Rock Ridge ISO whose CL child links form a parent cycle without RE.
24
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_cl_re_cycle.iso.Z.uu
 
25
  - `images/test_read_format_iso_joliet_utf16be_overflow.iso.uu`: Libarchive test image: Joliet ISO whose UTF-16BE path fills the 1024-byte path buffer.
26
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_joliet_utf16be_overflow.iso.uu
 
27
  - `images/test_read_format_iso_rockridge_zf_overflow.iso.uu`: Libarchive test image: Rock Ridge ZF entry of length 5 at the very end of the image.
28
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_rockridge_zf_overflow.iso.uu
 
29
  - `images/test_read_format_iso_zisofs_overflow.iso.uu`: Libarchive test image: zisofs ZF entry with block-size exponent 2 and a near-4-GiB uncompressed size.
30
  Decompressed from: https://raw.githubusercontent.com/libarchive/libarchive/1c9ad37a33d134772c9a573c0aa34c9f4151b3ba/libarchive/test/test_read_format_iso_zisofs_overflow.iso.uu
 
campaign_files/2026-10-b-disk-images-3/libcdio-iso-malformed/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** Each file with an answer has one answer file, at answer/<unit path>.json (the unit path plus .json). Reference result kind: corruption_desc.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .iso.
14
 
15
  **Licence (SPDX):** GPL-3.0-only
@@ -19,11 +17,8 @@
19
  Files:
20
 
21
  - `test/data/malformed.iso`: Libcdio test image: ISO 9660 primary volume descriptor filled with ASCII zeros (malformed.iso).
22
- Reference result: corruption_desc
23
  Dependencies: only needed to rerun the upstream test
24
  - `test/data/malformed2.iso`: Libcdio test image: ISO 9660 root directory with a 2-byte record and broken both-endian fields (malformed2.iso).
25
- Reference result: corruption_desc
26
  Dependencies: only needed to rerun the upstream test
27
  - `test/data/bad-dir.iso`: Libcdio test image: ISO 9660 root directory whose entries point back to the root (bad-dir.iso).
28
- Reference result: corruption_desc
29
  Dependencies: only needed to rerun the upstream test
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .iso.
12
 
13
  **Licence (SPDX):** GPL-3.0-only
 
17
  Files:
18
 
19
  - `test/data/malformed.iso`: Libcdio test image: ISO 9660 primary volume descriptor filled with ASCII zeros (malformed.iso).
 
20
  Dependencies: only needed to rerun the upstream test
21
  - `test/data/malformed2.iso`: Libcdio test image: ISO 9660 root directory with a 2-byte record and broken both-endian fields (malformed2.iso).
 
22
  Dependencies: only needed to rerun the upstream test
23
  - `test/data/bad-dir.iso`: Libcdio test image: ISO 9660 root directory whose entries point back to the root (bad-dir.iso).
 
24
  Dependencies: only needed to rerun the upstream test
campaign_files/2026-10-b-disk-images-3/liu-ctf-2025-corrupted/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: deleted_files.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .zip.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -20,5 +18,4 @@ Files:
20
 
21
  - `Corrupted/sd.zip`: FAT16 SD-card image with one deleted, header-truncated PNG in /photos (LiU CTF 2025 'Corrupted').
22
  Origin: CTF image made by the challenge author with mkfs.fat and mtools (writeup lines 19 to 32).
23
- Reference result: deleted_files
24
  Dependencies: unzip or Python zipfile to unpack; only needed to rerun the exercise, not to grade the answer
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .zip.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
18
 
19
  - `Corrupted/sd.zip`: FAT16 SD-card image with one deleted, header-truncated PNG in /photos (LiU CTF 2025 'Corrupted').
20
  Origin: CTF image made by the challenge author with mkfs.fat and mtools (writeup lines 19 to 32).
 
21
  Dependencies: unzip or Python zipfile to unpack; only needed to rerun the exercise, not to grade the answer
campaign_files/2026-10-b-disk-images-3/lks-bali-2026-erased-evidence/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** Each file with an answer has one answer file, at answer/<unit path>.json (the unit path plus .json). Reference result kind: deleted_files.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .img.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -20,5 +18,4 @@ Files:
20
 
21
  - `events/bali/Forensics/erased-evidence/dist/usb.img`: FAT16 USB image with one deleted evidence.zip among decoy files (LKS Bali 2026 'Erased Evidence').
22
  Origin: CTF image built by src/build.sh with mkfs.fat and mtools (mcopy, then mdel ::evidence.zip).
23
- Reference result: deleted_files
24
  Dependencies: only needed to rerun the exercise, not to grade the answer
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .img.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
18
 
19
  - `events/bali/Forensics/erased-evidence/dist/usb.img`: FAT16 USB image with one deleted evidence.zip among decoy files (LKS Bali 2026 'Erased Evidence').
20
  Origin: CTF image built by src/build.sh with mkfs.fat and mtools (mcopy, then mdel ::evidence.zip).
 
21
  Dependencies: only needed to rerun the exercise, not to grade the answer
campaign_files/2026-10-c-repros-2/lampert-2026-sbp-not-enough-repro/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each unit needs other upstream files is not stated. 2 of 3 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
 
11
- **Answers:** Each unit with an answer has one answer file, at answer/<package>/reported.json.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .jl; .toml; .md.
14
 
15
  **Licence (SPDX):** MIT
 
8
 
9
  **Self-contained:** Whether each unit needs other upstream files is not stated. 2 of 3 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .jl; .toml; .md.
12
 
13
  **Licence (SPDX):** MIT
campaign_files/2026-10-g-solver-instances-2/lexaire-thermogift/LICENSE ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ Licence: MIT
2
+
3
+ - MIT: https://github.com/Lexaire/ThermoGift/blob/eb9f0e759ed936406d90168681caebd2d79dc96e/LICENSE (publisher: Lexaire)
campaign_files/2026-10-g-solver-instances-2/lexaire-thermogift/README.md ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # lexaire-thermogift
2
+
3
+ **What this is:** 1 unit (5 files) published by Lexaire. Collection theme: Seeded puzzle generators (Simon Tatham puzzle collection class, with generation command, seed and solution).
4
+
5
+ **What one unit is:** Each unit is a few named upstream files, delivered together, kept at its upstream path. Format: JSONL of generated puzzles with solutions, plus a generation record.
6
+
7
+ **Where it comes from:** Upstream: Lexaire/ThermoGift. Kind: generated data.
8
+
9
+ **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
+
11
+ **Tools to open and check:** Check: Unique solution: exact match of the filled grid; equivalently a rule check (each thermometer filled contiguously from the bulb, row and column fill counts equal the clues, README lines 28-29). Wrong-answer controls must fail. The puzzle text holds thermometers and clues only; solution and fills are stored apart.
12
+
13
+ **Licence (SPDX):** MIT
14
+
15
+ **Provenance:** https://github.com/Lexaire/ThermoGift
16
+
17
+ Units:
18
+
19
+ - `/` (5 files): Lexaire/ThermoGift Thermometers generator (ES modules): seeded, uniqueness-checked thermometer puzzles with row and column clues.
20
+ Format: JSONL of generated puzzles with solutions, plus a generation record
21
+ Check: Unique solution: exact match of the filled grid; equivalently a rule check (each thermometer filled contiguously from the bulb, row and column fill counts equal the clues, README lines 28-29). Wrong-answer controls must fail. The puzzle text holds thermometers and clues only; solution and fills are stored apart.
22
+ Companions (not included): src/puzzles/thermometers/solver.js; src/common/util.js; LICENSE; README.md
23
+ Dependencies: Node.js >= 18 with ES modules (no npm packages imported by generator.js, solver.js, util.js; exact version recorded); Lexaire/ThermoGift at git commit eb9f0e75; not packed
campaign_files/2026-10-g-solver-instances-2/ljiang9-bridges/LICENSE ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ Licence: MIT
2
+
3
+ - MIT: https://github.com/ljiang9/bridges/blob/1514a3520590acd45dd32bc09ccdf0fa14ceab9d/LICENSE (publisher: ljiang9)
campaign_files/2026-10-g-solver-instances-2/ljiang9-bridges/README.md ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # ljiang9-bridges
2
+
3
+ **What this is:** 1 unit (4 files) published by ljiang9. Collection theme: Seeded puzzle generators (Simon Tatham puzzle collection class, with generation command, seed and solution).
4
+
5
+ **What one unit is:** Each unit is a few named upstream files, delivered together, kept at its upstream path. Format: JSONL of generated puzzles with answers, plus a generation record.
6
+
7
+ **Where it comes from:** Upstream: ljiang9/bridges. Kind: generated data.
8
+
9
+ **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
+
11
+ **Tools to open and check:** Check: Answers may not be unique, so judge with the upstream check_solution (bridges.py lines 150-170: 0 to 2 bridges per edge, island degree equals its number, no crossing bridges, all islands connected); pass is. Wrong-answer controls must fail.
12
+
13
+ **Licence (SPDX):** MIT
14
+
15
+ **Provenance:** https://github.com/ljiang9/bridges
16
+
17
+ Units:
18
+
19
+ - `generated/` (4 files): ljiang9/bridges: seeded Hashiwokakero generator with solver and rule checker (Python).
20
+ Format: JSONL of generated puzzles with answers, plus a generation record
21
+ Check: Answers may not be unique, so judge with the upstream check_solution (bridges.py lines 150-170: 0 to 2 bridges per edge, island degree equals its number, no crossing bridges, all islands connected); pass is. Wrong-answer controls must fail.
22
+ Companions (not included): __main__.py; LICENSE; README.md
23
+ Dependencies: Python 3 standard library only (version recorded at generation); ljiang9/bridges at git commit 1514a352; checker imported from bridges.py; not packed
campaign_files/2026-10-g-solver-instances-2/ljiang9-hitori/LICENSE ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ Licence: MIT
2
+
3
+ - MIT: https://github.com/ljiang9/hitori/blob/353740eac83c44fc04aecd47a6c8b449cc5afc4e/LICENSE (publisher: ljiang9)
campaign_files/2026-10-g-solver-instances-2/ljiang9-hitori/README.md ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # ljiang9-hitori
2
+
3
+ **What this is:** 1 unit (4 files) published by ljiang9. Collection theme: Seeded puzzle generators (Simon Tatham puzzle collection class, with generation command, seed and solution).
4
+
5
+ **What one unit is:** Each unit is a few named upstream files, delivered together, kept at its upstream path. Format: JSONL of generated puzzles with answers, plus a generation record.
6
+
7
+ **Where it comes from:** Upstream: ljiang9/hitori. Kind: generated data.
8
+
9
+ **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
+
11
+ **Tools to open and check:** Check: Answers are not unique, so judge with the upstream check_solution (hitori.py lines 24-66): no two shaded cells adjacent, no repeated number among unshaded cells in a row or column, unshaded cells connected; pass is. Wrong-answer controls must fail.
12
+
13
+ **Licence (SPDX):** MIT
14
+
15
+ **Provenance:** https://github.com/ljiang9/hitori
16
+
17
+ Units:
18
+
19
+ - `generated/` (4 files): ljiang9/hitori: seeded Hitori generator with backtracking solver and rule checker (Python).
20
+ Format: JSONL of generated puzzles with answers, plus a generation record
21
+ Check: Answers are not unique, so judge with the upstream check_solution (hitori.py lines 24-66): no two shaded cells adjacent, no repeated number among unshaded cells in a row or column, unshaded cells connected; pass is. Wrong-answer controls must fail.
22
+ Companions (not included): __main__.py; LICENSE; README.md
23
+ Dependencies: Python 3 standard library only (version recorded at generation); ljiang9/hitori at git commit 353740ea; checker imported from hitori.py; not packed
campaign_files/2026-10-g-solver-instances-2/ljiang9-killer/LICENSE ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ Licence: MIT
2
+
3
+ - MIT: https://github.com/ljiang9/killer/blob/67320c2d329926c953c40e0e4fc591ac372356ec/LICENSE (publisher: ljiang9)
campaign_files/2026-10-g-solver-instances-2/ljiang9-killer/README.md ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # ljiang9-killer
2
+
3
+ **What this is:** 1 unit (4 files) published by ljiang9. Collection theme: Seeded puzzle generators (Simon Tatham puzzle collection class, with generation command, seed and solution).
4
+
5
+ **What one unit is:** Each unit is a few named upstream files, delivered together, kept at its upstream path. Format: JSONL of generated puzzles with answers, plus a generation record.
6
+
7
+ **Where it comes from:** Upstream: ljiang9/killer. Kind: generated data.
8
+
9
+ **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes. Dependencies named upstream are listed per unit in the notes.
10
+
11
+ **Tools to open and check:** Check: Answers may not be unique, so judge with the upstream check_solution (killer.py lines 204-230: rows, columns and boxes 1-9, cage sums and no repeat inside a cage, cages cover all 81 cells); pass is. Wrong-answer controls must fail.
12
+
13
+ **Licence (SPDX):** MIT
14
+
15
+ **Provenance:** https://github.com/ljiang9/killer
16
+
17
+ Units:
18
+
19
+ - `generated/` (4 files): ljiang9/killer: seeded killer sudoku generator with cage-aware solver and checker (Python).
20
+ Format: JSONL of generated puzzles with answers, plus a generation record
21
+ Check: Answers may not be unique, so judge with the upstream check_solution (killer.py lines 204-230: rows, columns and boxes 1-9, cage sums and no repeat inside a cage, cages cover all 81 cells); pass is. Wrong-answer controls must fail.
22
+ Companions (not included): __main__.py; LICENSE; README.md
23
+ Dependencies: Python 3 standard library only (version recorded at generation); ljiang9/killer at git commit 67320c2d; checker imported from killer.py; not packed
campaign_files/2026-10-g-solver-instances-2/magictour-top1465/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Unique solution grid. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: No solver is needed at judge time.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `g140-07/top1465`: top1465: a list of 1,465 hard sudoku puzzles from magictour.free.fr.
22
  Format: plain text, 1,465 lines of 81 puzzle characters ('.' for an empty cell) and a trailing empty line, LF; a case is one line.
23
  Origin: a long-standing public list of hard sudoku puzzles on Guenter Stertenbrink's magictour.free.fr, cited by Sudoku-classif as [Magictour-top1465]
24
- Reference result: unique solution grid
25
- Reference source: No solutions ship with the file.
26
  Check: No solver is needed at judge time.
27
  Dependencies: not packed; path and commit filled when the script is committed, before the run; not packed
 
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: No solver is needed at judge time.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `g140-07/top1465`: top1465: a list of 1,465 hard sudoku puzzles from magictour.free.fr.
20
  Format: plain text, 1,465 lines of 81 puzzle characters ('.' for an empty cell) and a trailing empty line, LF; a case is one line.
21
  Origin: a long-standing public list of hard sudoku puzzles on Guenter Stertenbrink's magictour.free.fr, cited by Sudoku-classif as [Magictour-top1465]
 
 
22
  Check: No solver is needed at judge time.
23
  Dependencies: not packed; path and commit filled when the script is committed, before the run; not packed
campaign_files/2026-10-g-solver-instances-2/magictour/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Any valid completion of the grid; any valid completion of the grid under the stated rules. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required. Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules: every given is kept, and each row and each column holds each of the 25 symbols once. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules stated at the top of the file: every given is kept, and no two cells that share a letter in any of the 6 constraint figures hold the same digit. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules: every given is kept, and each row, column and 5x5 box holds the 25 symbols once each. Any grid that passes is correct (rule check, SPEC line 283).
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,49 +19,35 @@ Files:
21
  - `g258-06/top2365`: top2365: 2,365 hard 9x9 sudoku puzzles.
22
  Format: plain text, LF, one 81-character puzzle per line ('.' for an empty cell), 2,365 puzzles; delivered whole as the upstream file
23
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; rule judging accepts any valid completion
24
- Reference result: any valid completion of the grid
25
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
26
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required.
27
  Dependencies: not packed; not packed
28
  - `g258-07/subig20`: subig20: 17,445 9x9 sudoku puzzles.
29
  Format: plain text, LF, one 81-character puzzle per line ('.' for an empty cell), 17,445 puzzles, 1.43 MB; the sudoku.htm entry says the list also holds some easy puzzles; delivered whole as the upstream file
30
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; result in REPORT.md; rule judging accepts any valid completion
31
- Reference result: any valid completion of the grid
32
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
33
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required.
34
  Dependencies: not packed; not packed
35
  - `g258-08/top44`: top44: 44 hard 16x16 sudoku puzzles.
36
  Format: plain text, LF, one 256-character puzzle per line (symbols 1 to 9 and A to G, '.' for an empty cell), 44 puzzles; delivered whole as the upstream file
37
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; uniqueness not required for rule judging; probe result in REPORT.md where it finished
38
- Reference result: any valid completion of the grid
39
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
40
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283).
41
  Dependencies: not packed; not packed
42
  - `g258-09/su16`: su16: 16 16x16 sudoku puzzles.
43
  Format: plain text, LF, one 256-character puzzle per line (symbols 1 to 9 and A to G, '.' for an empty cell), 16 puzzles; delivered whole as the upstream file
44
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; uniqueness not required for rule judging; probe result in REPORT.md where it finished
45
- Reference result: any valid completion of the grid
46
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
47
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283).
48
  Dependencies: not packed; not packed
49
  - `g258-10/qwhm25`: qwhm25: 50 25x25 Latin square completion puzzles.
50
  Format: plain text, LF, one 625-character puzzle per line (symbols 1 to 9 and A to P, '.' for an empty cell), 50 puzzles; Latin square completion (quasigroup with holes), no boxes; delivered whole as the upstream file
51
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the sudoku.htm entry describes the list as random minimal Latin square completion puzzles
52
- Reference result: any valid completion of the grid under the stated rules
53
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
54
  Check: Check the agent grid by the rules: every given is kept, and each row and each column holds each of the 25 symbols once. Any grid that passes is correct (rule check, SPEC line 283).
55
  Dependencies: not packed; not packed
56
  - `g258-11/sudoku6`: sudoku6: 100 9x9 sudoku puzzles with 6 constraint families.
57
  Format: plain text, LF; the file starts with the rule statement (6 constraint figures: rows, columns, blocks, positions inside a block, minirows, minicolumns, each drawn as a 9x9 letter grid) and an example solution, then 100 puzzles as 9 lines of 9 characters each; delivered whole as the upstream file
58
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the 6 figures are parsed from the file header at judge time
59
- Reference result: any valid completion of the grid under the stated rules
60
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
61
  Check: Check the agent grid by the rules stated at the top of the file: every given is kept, and no two cells that share a letter in any of the 6 constraint figures hold the same digit. Any grid that passes is correct (rule check, SPEC line 283).
62
  Dependencies: not packed; not packed
63
  - `g258-12/min25`: min25: 2 locally minimal 25x25 sudoku puzzles.
64
  Format: plain text, LF; a short note, then 2 puzzles drawn as 25 rows of boxed text (symbols 1 to 9 and A to P, '.' for an empty cell); the note says each has a unique solution and removing any clue gives several; delivered whole as the upstream file
65
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the file states uniqueness
66
- Reference result: any valid completion of the grid under the stated rules
67
- Reference source: No solutions ship with the file. judging is by the rules, so the reference is not needed to judge.
68
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 5x5 box holds the 25 symbols once each. Any grid that passes is correct (rule check, SPEC line 283).
69
  Dependencies: not packed; not packed
 
8
 
9
  **Self-contained:** Each file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required. Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules: every given is kept, and each row and each column holds each of the 25 symbols once. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules stated at the top of the file: every given is kept, and no two cells that share a letter in any of the 6 constraint figures hold the same digit. Any grid that passes is correct (rule check, SPEC line 283). Check the agent grid by the rules: every given is kept, and each row, column and 5x5 box holds the 25 symbols once each. Any grid that passes is correct (rule check, SPEC line 283).
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `g258-06/top2365`: top2365: 2,365 hard 9x9 sudoku puzzles.
20
  Format: plain text, LF, one 81-character puzzle per line ('.' for an empty cell), 2,365 puzzles; delivered whole as the upstream file
21
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; rule judging accepts any valid completion
 
 
22
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required.
23
  Dependencies: not packed; not packed
24
  - `g258-07/subig20`: subig20: 17,445 9x9 sudoku puzzles.
25
  Format: plain text, LF, one 81-character puzzle per line ('.' for an empty cell), 17,445 puzzles, 1.43 MB; the sudoku.htm entry says the list also holds some easy puzzles; delivered whole as the upstream file
26
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; result in REPORT.md; rule judging accepts any valid completion
 
 
27
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 3x3 box holds the digits 1 to 9 once each. Any grid that passes is correct (rule check, SPEC line 283), so uniqueness is not required.
28
  Dependencies: not packed; not packed
29
  - `g258-08/top44`: top44: 44 hard 16x16 sudoku puzzles.
30
  Format: plain text, LF, one 256-character puzzle per line (symbols 1 to 9 and A to G, '.' for an empty cell), 44 puzzles; delivered whole as the upstream file
31
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; uniqueness not required for rule judging; probe result in REPORT.md where it finished
 
 
32
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283).
33
  Dependencies: not packed; not packed
34
  - `g258-09/su16`: su16: 16 16x16 sudoku puzzles.
35
  Format: plain text, LF, one 256-character puzzle per line (symbols 1 to 9 and A to G, '.' for an empty cell), 16 puzzles; delivered whole as the upstream file
36
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; uniqueness not required for rule judging; probe result in REPORT.md where it finished
 
 
37
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 4x4 box holds the 16 symbols 1 to 9 and A to G once each. Any grid that passes is correct (rule check, SPEC line 283).
38
  Dependencies: not packed; not packed
39
  - `g258-10/qwhm25`: qwhm25: 50 25x25 Latin square completion puzzles.
40
  Format: plain text, LF, one 625-character puzzle per line (symbols 1 to 9 and A to P, '.' for an empty cell), 50 puzzles; Latin square completion (quasigroup with holes), no boxes; delivered whole as the upstream file
41
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the sudoku.htm entry describes the list as random minimal Latin square completion puzzles
 
 
42
  Check: Check the agent grid by the rules: every given is kept, and each row and each column holds each of the 25 symbols once. Any grid that passes is correct (rule check, SPEC line 283).
43
  Dependencies: not packed; not packed
44
  - `g258-11/sudoku6`: sudoku6: 100 9x9 sudoku puzzles with 6 constraint families.
45
  Format: plain text, LF; the file starts with the rule statement (6 constraint figures: rows, columns, blocks, positions inside a block, minirows, minicolumns, each drawn as a 9x9 letter grid) and an example solution, then 100 puzzles as 9 lines of 9 characters each; delivered whole as the upstream file
46
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the 6 figures are parsed from the file header at judge time
 
 
47
  Check: Check the agent grid by the rules stated at the top of the file: every given is kept, and no two cells that share a letter in any of the 6 constraint figures hold the same digit. Any grid that passes is correct (rule check, SPEC line 283).
48
  Dependencies: not packed; not packed
49
  - `g258-12/min25`: min25: 2 locally minimal 25x25 sudoku puzzles.
50
  Format: plain text, LF; a short note, then 2 puzzles drawn as 25 rows of boxed text (symbols 1 to 9 and A to P, '.' for an empty cell); the note says each has a unique solution and removing any clue gives several; delivered whole as the upstream file
51
  Origin: a public puzzle list on Guenter Stertenbrink's magictour.free.fr, linked from sudoku.htm; the file states uniqueness
 
 
52
  Check: Check the agent grid by the rules: every given is kept, and each row, column and 5x5 box holds the 25 symbols once each. Any grid that passes is correct (rule check, SPEC line 283).
53
  Dependencies: not packed; not packed
campaign_files/2026-10-g-solver-instances-2/mayfull-sudoku-4x4/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Solution grid (may not be unique). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Compare the agent grid with the shipped solution cell by cell (equivalently: a valid 4x4 grid, rows, columns and 2x2 boxes each 1 to 4, that keeps every given; the train puzzles all have one solution).
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `data/train-00000-of-00001.parquet`: 4x4 sudoku, 1,000,000 puzzles with solutions (train split of the d1 / d2 / SPG 4x4 sudoku data).
22
  Format: Parquet: puzzle, solution (16-character strings, 0 = blank); delivered whole as the upstream file; a case is one row; the 44 rows in exclusions are removed when the pool is built
23
  Origin: 4x4 sudoku set published as-is by Black-Phoenix (2019, made with Arel Cordero's sudoku code) and re-uploaded to Hugging Face by Mayfull (was procedurally_generated)
24
- Reference result: solution grid (may not be unique)
25
- Reference source: Upstream README: generated with Arel Cordero sudoku code; "the solutions to these puzzles may not be unique". So the shipped solution is the unique answer.
26
  Check: Compare the agent grid with the shipped solution cell by cell (equivalently: a valid 4x4 grid, rows, columns and 2x2 boxes each 1 to 4, that keeps every given; the train puzzles all have one solution).
27
  Dependencies: not packed; pyarrow 21.0.0 (the version this re-check read the file with) to read the Parquet unit; not packed
 
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Compare the agent grid with the shipped solution cell by cell (equivalently: a valid 4x4 grid, rows, columns and 2x2 boxes each 1 to 4, that keeps every given; the train puzzles all have one solution).
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `data/train-00000-of-00001.parquet`: 4x4 sudoku, 1,000,000 puzzles with solutions (train split of the d1 / d2 / SPG 4x4 sudoku data).
20
  Format: Parquet: puzzle, solution (16-character strings, 0 = blank); delivered whole as the upstream file; a case is one row; the 44 rows in exclusions are removed when the pool is built
21
  Origin: 4x4 sudoku set published as-is by Black-Phoenix (2019, made with Arel Cordero's sudoku code) and re-uploaded to Hugging Face by Mayfull (was procedurally_generated)
 
 
22
  Check: Compare the agent grid with the shipped solution cell by cell (equivalently: a valid 4x4 grid, rows, columns and 2x2 boxes each 1 to 4, that keeps every given; the train puzzles all have one solution).
23
  Dependencies: not packed; pyarrow 21.0.0 (the version this re-check read the file with) to read the Parquet unit; not packed
campaign_files/2026-10-gt-browser-tests-2/libkscreen-config-tests/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 3 of 3 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_display_configuration. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
14
 
15
  **Licence (SPDX):** LGPL-2.1-or-later
@@ -21,8 +19,6 @@ Files:
21
  - `autotests/configs/singleoutput.json`: A laptop panel alone: LVDS1 enabled at (0,0) in its 1280x800 mode at 59.9 Hz, no rotation, scale 1, as the KDE Display Configuration page shows and sets it, written as a libkscreen screen configuration (JSON).
22
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
23
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
24
- Reference result: expected_display_configuration
25
- Reference source: testScreenConfig::singleOutput in autotests/testscreenconfig.cpp lines 63-103: screen min 320x200, max 8192x8192, current 1280x800; 1 output LVDS1, type Panel, 3 modes, pos (0,0), geometry 1280x800, current and preferred mode 3, rotation none, scale 1.0, connected, enabled, no clones; mode 3 1280x800 at 59.9 Hz
26
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The file-level judge checks the stored fields the assertions read: screen sizes, one output named LVDS1 of type LVDS (shown as Panel), three modes, pos (0,0), currentModeId 3 with mode 3 1280x800 at refreshRate 59.9, rotation 1 (none), no scale key or scale 1, connected and enabled true, no clones.
27
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
28
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
@@ -30,8 +26,6 @@ Files:
30
  - `autotests/configs/multipleoutput.json`: Laptop panel plus an external HDMI monitor extended to its right at x=1280, the HDMI monitor at 1920x1080 60 Hz with 140 percent scaling, the panel as primary display, written as a libkscreen screen configuration (JSON).
31
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
32
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
33
- Reference result: expected_display_configuration
34
- Reference source: testScreenConfig::multiOutput in autotests/testscreenconfig.cpp lines 118-154: screen current 3200x1880; 2 outputs; HDMI1 type HDMI, 4 modes, pos (1280,0), geometry (1280,0,ceil(1920/1.4),ceil(1080/1.4)), current and preferred mode 4, rotation none, scale 1.4, connected, enabled, priority 2, no clones; mode 4 1920x1080 at 60 Hz
35
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The geometry and priority 2 are derived by libkscreen from pos, scale, the mode size and the primary flag; the file-level judge checks the stored fields they rest on: HDMI1 pos x=1280 y=0, scale 1.4, currentModeId 4 with that mode 1920x1080 at refreshRate 60, enabled and connected true, and primary false (the test reads priority 2).
36
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
37
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
@@ -39,8 +33,6 @@ Files:
39
  - `autotests/configs/multipleclone.json`: Laptop panel and HDMI monitor mirrored: both at 1024x768, the panel cloning the HDMI output, written as a libkscreen screen configuration (JSON).
40
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
41
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
42
- Reference result: expected_display_configuration
43
- Reference source: testScreenConfig::clonesOutput in autotests/testscreenconfig.cpp lines 156-176: screen current 1024x768; outputs 1 and 2 in modes of the same size; output 1 clones exactly one output, output 2; output 2 has no clones
44
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The file-level judge checks output 1 "clones" equal to [2], output 2 with no clones, and both currentModeId entries naming modes of the same size (1024x768).
45
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
46
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 3 of 3 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
12
 
13
  **Licence (SPDX):** LGPL-2.1-or-later
 
19
  - `autotests/configs/singleoutput.json`: A laptop panel alone: LVDS1 enabled at (0,0) in its 1280x800 mode at 59.9 Hz, no rotation, scale 1, as the KDE Display Configuration page shows and sets it, written as a libkscreen screen configuration (JSON).
20
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
21
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
 
 
22
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The file-level judge checks the stored fields the assertions read: screen sizes, one output named LVDS1 of type LVDS (shown as Panel), three modes, pos (0,0), currentModeId 3 with mode 3 1280x800 at refreshRate 59.9, rotation 1 (none), no scale key or scale 1, connected and enabled true, no clones.
23
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
24
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
 
26
  - `autotests/configs/multipleoutput.json`: Laptop panel plus an external HDMI monitor extended to its right at x=1280, the HDMI monitor at 1920x1080 60 Hz with 140 percent scaling, the panel as primary display, written as a libkscreen screen configuration (JSON).
27
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
28
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
 
 
29
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The geometry and priority 2 are derived by libkscreen from pos, scale, the mode size and the primary flag; the file-level judge checks the stored fields they rest on: HDMI1 pos x=1280 y=0, scale 1.4, currentModeId 4 with that mode 1920x1080 at refreshRate 60, enabled and connected true, and primary false (the test reads priority 2).
30
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
31
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
 
33
  - `autotests/configs/multipleclone.json`: Laptop panel and HDMI monitor mirrored: both at 1024x768, the panel cloning the HDMI output, written as a libkscreen screen configuration (JSON).
34
  Format: libkscreen fake-backend configuration JSON (screen object with id, minSize, maxSize, currentSize, maxActiveOutputsCount; outputs array with id, name, type, modes, pos, currentModeId, rotation, scale, connected, enabled, primary, clones, edid)
35
  Origin: fake-backend configuration fixture of libkscreen autotests (autotests/configs), loaded by testscreenconfig.cpp through the TEST_DATA backend argument (tool_fixture)
 
 
36
  Check: Final configuration state, no screenshot needed. Parse the JSON the agent leaves with a JSON parser (or load it into libkscreen at the pinned commit through the fake backend and read the Config back) and compare the judged values with the values the test asserts, all exactly (integers, strings, booleans and the refresh rate as the literal written in the test, compared after rounding both to single precision as the test does with (float)). The file-level judge checks output 1 "clones" equal to [2], output 2 with no clones, and both currentModeId entries naming modes of the same size (1024x768).
37
  Tools: write the JSON with a text editor or a script (Python json), or on a running Plasma session apply the same layout with kscreen-doctor (`kscreen-doctor output.HDMI-1.position.1280,0 output.HDMI-1.scale.1.4 output.HDMI-1.mode.1920x1080@60`) and read it back with `kscreen-doctor -j`, which prints the libkscreen configuration as JSON
38
  Companions (not included): autotests/testscreenconfig.cpp; LICENSES/LGPL-2.1-or-later.txt
campaign_files/2026-10-gt-browser-tests-2/libreoffice-calc-function-tests/README.md CHANGED
The diff for this file is too large to render. See raw diff
 
campaign_files/2026-10-gt-browser-tests-2/libreoffice-uitest/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** 13 of 32 files are self-contained; the others are used together with other upstream files. 32 of 32 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: asserted_document_state. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** UNO Python API headless: open goalSeek.ods, call XGoalSeek.seekGoal on the sheet (formula cell B4, variable cell B1, target 15000) and write the result into B1, then store; or solve directly and write B1 with odfpy and recalculate in soffice --headless. Other files: see the notes.
14
 
15
  **Licence (SPDX):** MPL-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `sc/qa/uitest/goalSeek/goalSeek.py`: LibreOffice Calc UI test goalSeek.py: Tools > Goal Seek on goalSeek.ods (find the principal B1 so that interest B4 = 15000).
22
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
23
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
24
- Reference result: asserted_document_state
25
- Reference source: assertEqual lines of goalSeek.py on cell values: after Goal Seek B1 = 200000 and B4 = 15000 (lines 37, 38); after Undo B1 = 100000 and B4 = 7500 (lines 42, 43); after reopening the dialog and answering No, unchanged at 100000 and 7500 (lines 60, 61). Kept: these 6 cell assertions. Dropped: 4 dialog-field assertions (lines 24, 51, 52, 53: formula cell, target value and variable cell fields)
26
  Check: Final file state: after the agent saves, B1 must equal 200000 and B4 must equal 15000 on recalculation (read with the UNO API or odfpy after soffice --headless recalculation). Dialog-field assertions in the script are ignored
27
  Tools: UNO Python API headless: open goalSeek.ods, call XGoalSeek.seekGoal on the sheet (formula cell B4, variable cell B1, target 15000) and write the result into B1, then store; or solve directly and write B1 with odfpy and recalculate in soffice --headless
28
  Companions (not included): sc/qa/uitest/data/goalSeek.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -30,8 +26,6 @@ Files:
30
  - `sc/qa/uitest/textToColumns/textToColumns.py`: LibreOffice Calc UI test textToColumns.py: Data > Text to Columns on six documents with comma, dot, pipe, pipe plus space, semicolon and space separators.
31
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
32
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
33
- Reference result: asserted_document_state
34
- Reference source: 210 assertEqual lines on cell values (getValue/getString) after Text to Columns, lines 44 to 385, one block per separator document. Kept: all 210 (all are on cells). Dropped: none (the script asserts nothing on dialog state)
35
  Check: Final file state: open the saved document and compare every asserted cell (sheet 0, given row and column) with the asserted value; all 210 must match. No screenshot needed
36
  Tools: UNO Python API headless (read column A, split, write cells, store) or odfpy/pandas on the .ods; soffice --headless --convert-to csv to read back
37
  Companions (not included): sc/qa/uitest/data/text_to_columns_comma.ods; sc/qa/uitest/data/text_to_columns_dot.ods; sc/qa/uitest/data/text_to_columns_pipe.ods; sc/qa/uitest/data/text_to_columns_pipe_space.ods; sc/qa/uitest/data/text_to_columns_semicolon.ods; sc/qa/uitest/data/text_to_columns_space.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -39,8 +33,6 @@ Files:
39
  - `sc/qa/uitest/sort/subtotals.py`: LibreOffice Calc UI test subtotals.py: Data > Subtotals (group by, sum, page breaks, sort options) on four bug-report workbooks.
40
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
41
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
42
- Reference result: asserted_document_state
43
- Reference source: 8 assertEqual lines on cell values after the subtotal operation (lines 39, 40, 70, 84, 85, 121, 137, 138; for example tdf88792 D6 = 28000, tdf88735 A9 = "z" and B9 = 8, tdf56958 C2 = -0.25). Kept: all 8 (all are on cells). Dropped: none
44
  Check: Final file state: after the agent saves, read the asserted cells and compare exactly; subtotal rows and formulas must sit where the assertions expect them. No screenshot needed
45
  Tools: UNO Python API headless: XSubTotalCalculatable.createSubTotalDescriptor and applySubTotals on the data range, then store; check with odfpy or soffice --headless --convert-to csv
46
  Companions (not included): sc/qa/uitest/data/tdf55734.ods; sc/qa/uitest/data/tdf56958.ods; sc/qa/uitest/data/tdf88735.ods; sc/qa/uitest/data/tdf88792.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -48,8 +40,6 @@ Files:
48
  - `sw/qa/uitest/table/textToTable.py`: LibreOffice Writer UI test textToTable.py: Table > Convert > Text to Table with semicolon and custom ':' separators and a heading row.
49
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
50
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
51
- Reference result: asserted_document_state
52
- Reference source: assertEqual lines on the document object (lines 27, 29, 30, 33, 34, 51, 53, 54, 57, 58): number of text tables, rows and columns after conversion (1 table of 1x3; 1 table of 4x3) and text after Undo. Kept: all 10 (all are on the document). Dropped: none
53
  Check: Final file state: open the saved .odt and count tables, rows and columns (odfpy or UNO), compare with the asserted values. No screenshot needed
54
  Tools: UNO Python API headless: select the text range and call XTextConvert (convertToTable) or build the table with XTextTable, then store; check with odfpy (count table:table, rows and columns) or python-docx after soffice --headless --convert-to docx
55
  Companions (not included): sw/qa/uitest/data/textToTable.odt; COPYING.MPL; readlicense_oo/license/license.xml
@@ -57,8 +47,6 @@ Files:
57
  - `sw/qa/uitest/findReplace/findReplace.py`: LibreOffice Writer UI test findReplace.py: Edit > Find and Replace, including regular-expression replace with back references and diacritic replacement.
58
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
59
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
60
- Reference result: asserted_document_state
61
- Reference source: assertEqual lines on document text (Text.String slices such as "Replace first first", "replacedest number1 testnot"). Kept: 10 document-text assertions (lines 51, 61, 115, 136, 168, 178, 183, 193, 208, 212). Dropped: 12 assertions on dialog fields or the current page (lines 26, 33, 39, 80, 81, 86, 111, 119, 151, 154, 158, 234)
62
  Check: Final file state: read the saved document text and compare the asserted prefixes. Only the 10 document-text assertions count; dialog-field and current-page assertions are dropped. No screenshot needed
63
  Tools: UNO Python API headless: XReplaceable.createReplaceDescriptor with SearchRegularExpression, then replaceAll and store; or edit content.xml text with odfpy
64
  Companions (not included): sw/qa/uitest/data/findReplace.odt; sw/qa/uitest/data/tdf116242.odt; COPYING.MPL; readlicense_oo/license/license.xml
@@ -66,8 +54,6 @@ Files:
66
  - `sc/qa/uitest/statistics/anova.py`: LibreOffice Calc UI test anova.py: Data > Statistics > Analysis of Variance (single factor) (.uno:AnalysisOfVarianceDialog) on typed data, input $Sheet1.$A$2:$C$13 grouped by columns, alpha 0.05, output $F$1.
67
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
68
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
69
- Reference result: asserted_document_state
70
- Reference source: assertEqual lines of anova.py on cell strings and values after the dialog (45 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
71
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Analysis of Variance (single factor) table with the settings given (input $Sheet1.$A$2:$C$13 grouped by columns, alpha 0.05, output $F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
72
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Analysis of Variance (single factor) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
73
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -75,8 +61,6 @@ Files:
75
  - `sc/qa/uitest/statistics/chiSquare.py`: LibreOffice Calc UI test chiSquare.py: Data > Statistics > Chi-square Test (.uno:ChiSquareTestDialog) on typed data, input $Sheet1.$A$1:$B$13 grouped by columns, output F1.
76
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
77
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
78
- Reference result: asserted_document_state
79
- Reference source: assertEqual lines of chiSquare.py on cell strings and values after the dialog (11 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
80
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Chi-square Test table with the settings given (input $Sheet1.$A$1:$B$13 grouped by columns, output F1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
81
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Chi-square Test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
82
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -84,8 +68,6 @@ Files:
84
  - `sc/qa/uitest/statistics/correlation.py`: LibreOffice Calc UI test correlation.py: Data > Statistics > Correlation (.uno:CorrelationDialog) on typed data, test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7.
85
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
86
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
87
- Reference result: asserted_document_state
88
- Reference source: assertEqual lines of correlation.py on cell strings and values after the dialog (28 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
89
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Correlation table with the settings given (test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
90
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Correlation output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
91
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -93,8 +75,6 @@ Files:
93
  - `sc/qa/uitest/statistics/covariance.py`: LibreOffice Calc UI test covariance.py: Data > Statistics > Covariance (.uno:CovarianceDialog) on typed data, test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7.
94
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
95
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
96
- Reference result: asserted_document_state
97
- Reference source: assertEqual lines of covariance.py on cell strings and values after the dialog (28 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
98
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Covariance table with the settings given (test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
99
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Covariance output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
100
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -102,8 +82,6 @@ Files:
102
  - `sc/qa/uitest/statistics/descriptiveStatistics.py`: LibreOffice Calc UI test descriptiveStatistics.py: Data > Statistics > Descriptive Statistics (.uno:DescriptiveStatisticsDialog) on typed data, input $Sheet1.$A$1:$C$13 grouped by columns, output $Sheet1.$F$1.
103
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
104
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
105
- Reference result: asserted_document_state
106
- Reference source: assertEqual lines of descriptiveStatistics.py on cell strings and values after the dialog (64 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
107
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Descriptive Statistics table with the settings given (input $Sheet1.$A$1:$C$13 grouped by columns, output $Sheet1.$F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
108
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Descriptive Statistics output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
109
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -111,8 +89,6 @@ Files:
111
  - `sc/qa/uitest/statistics/exponentialSmoothing.py`: LibreOffice Calc UI test exponentialSmoothing.py: Data > Statistics > Exponential Smoothing (.uno:ExponentialSmoothingDialog) on typed data, input $Sheet1.$A$1:$B$13, smoothing factor 0.5, output $F$1, grouped by columns (test 1) and by rows (test 2).
112
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
113
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
114
- Reference result: asserted_document_state
115
- Reference source: assertEqual lines of exponentialSmoothing.py on cell strings and values after the dialog (44 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
116
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Exponential Smoothing table with the settings given (input $Sheet1.$A$1:$B$13, smoothing factor 0.5, output $F$1, grouped by columns (test 1) and by rows (test 2)), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
117
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Exponential Smoothing output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
118
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -120,8 +96,6 @@ Files:
120
  - `sc/qa/uitest/statistics/fTest.py`: LibreOffice Calc UI test fTest.py: Data > Statistics > F-test (.uno:FTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
121
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
122
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
123
- Reference result: asserted_document_state
124
- Reference source: assertEqual lines of fTest.py on cell strings and values after the dialog (32 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
125
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the F-test table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
126
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the F-test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
127
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -129,8 +103,6 @@ Files:
129
  - `sc/qa/uitest/statistics/movingAverage.py`: LibreOffice Calc UI test movingAverage.py: Data > Statistics > Moving Average (.uno:MovingAverageDialog) on typed data, input $Sheet1.$A$1:$B$13, interval 3, output $F$1, grouped by columns (test 1) and by rows (test 2).
130
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
131
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
132
- Reference result: asserted_document_state
133
- Reference source: assertEqual lines of movingAverage.py on cell strings and values after the dialog (35 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
134
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Moving Average table with the settings given (input $Sheet1.$A$1:$B$13, interval 3, output $F$1, grouped by columns (test 1) and by rows (test 2)), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
135
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Moving Average output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
136
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -138,8 +110,6 @@ Files:
138
  - `sc/qa/uitest/statistics/regression.py`: LibreOffice Calc UI test regression.py: Data > Statistics > Regression (.uno:RegressionDialog) on regression.ods, linear with and without intercept, logarithmic and power, data grouped by columns and by rows.
139
  Format: Python UI test (unittest class driving the Calc Regression dialog through the UNO UITest API) with the ODF data document it loads; the document carries its own expected values and check formulas
140
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
141
- Reference result: asserted_document_state
142
- Reference source: regression.ods carries its own expected tables (sheets ExpectedValuesLinear, ExpectedValuesLinearNoIntercept, ExpectedValuesLog, ExpectedValuesPower) and check sheets (ChecksLinear and so on: ABS(Expected - Actual) < $J$4) summed up in CombineChecks B7 "Final Status"; regression.py asserts CombineChecks B7 is "FALSE" before (line 30) and "TRUE" after the four regressions (line 37), for data grouped by columns (test line 20) and by rows (test line 17). Dropped: test_regression_cancel (only closes the dialog)
143
  Check: Final file state: the agent receives regression.ods and must fill the four Actual sheets with the regression output for the data in DataInColumns (or DataInRows); after soffice --headless recalculation of the saved file, CombineChecks B7 must read TRUE, which holds when every expected cell matches its actual cell within the tolerance of ChecksLinear J4 (text cells compared exactly)
144
  Tools: soffice --headless with the UNO Python API: open regression.ods, write the regression output tables into sheets ActualLinear, ActualLinearNoIntercept, ActualLog and ActualPower at A1 (formulas such as LINEST or computed values), store; or compute the fits with numpy and write the cells with odfpy, then recalculate in soffice --headless
145
  Companions (not included): sc/qa/uitest/data/regression.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -147,8 +117,6 @@ Files:
147
  - `sc/qa/uitest/statistics/sampling.py`: LibreOffice Calc UI test sampling.py: Data > Statistics > Sampling (periodic) (.uno:SamplingDialog) on typed data, input $Sheet1.$A$1:$C$9, periodic method with period 2, output $Sheet1.$F$1.
148
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
149
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
150
- Reference result: asserted_document_state
151
- Reference source: assertEqual lines of sampling.py on cell strings and values after the dialog (24 assertEqual lines in the kept tests; only test_statistic_sampling (lines 18 to 105) is kept, test_tdf142986 (line 106) is dropped because it samples at random and asserts only nonzero values (line 144)); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
152
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Sampling (periodic) table with the settings given (input $Sheet1.$A$1:$C$9, periodic method with period 2, output $Sheet1.$F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
153
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Sampling (periodic) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
154
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -156,8 +124,6 @@ Files:
156
  - `sc/qa/uitest/statistics/tTest.py`: LibreOffice Calc UI test tTest.py: Data > Statistics > t-test (paired) (.uno:TTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
157
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
158
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
159
- Reference result: asserted_document_state
160
- Reference source: assertEqual lines of tTest.py on cell strings and values after the dialog (35 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
161
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the t-test (paired) table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
162
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the t-test (paired) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
163
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -165,8 +131,6 @@ Files:
165
  - `sc/qa/uitest/statistics/zTest.py`: LibreOffice Calc UI test zTest.py: Data > Statistics > z-test (.uno:ZTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
166
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
167
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
168
- Reference result: asserted_document_state
169
- Reference source: assertEqual lines of zTest.py on cell strings and values after the dialog (24 assertEqual lines in the kept tests); expected numbers are written in the script; the assertions after .uno:Undo (output cell empty again) are dropped
170
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the z-test table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
171
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the z-test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
172
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -174,8 +138,6 @@ Files:
174
  - `sw/qa/uitest/table/tableToText.py`: LibreOffice Writer UI test tableToText.py: Table > Convert > Table to Text on tableToText.odt with separators tab, paragraph, semicolon and a custom ":".
175
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF document it loads
176
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
177
- Reference result: asserted_document_state
178
- Reference source: assertEqual lines of tableToText.py after each conversion: document text starts with "a\ta" and no tables remain (lines 23, 24); paragraphs "a\na\n" (lines 34, 35); semicolons "a;a\n;\n" (lines 45, 46); custom separator ":" gives "a:a\n:\n" (lines 60, 61). Kept: these 8 assertions. Dropped: the 4 Undo assertions (lines 27, 38, 49, 64)
179
  Check: Final file state: the agent converts the table of tableToText.odt to text with the separator named in the task and saves; read the saved .odt with the UNO API or odfpy: no text table remains and the document text (CRLF normalised to LF) starts with the expected string of the script for that separator
180
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:ConvertTableToText) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
181
  Companions (not included): sw/qa/uitest/data/tableToText.odt; COPYING.MPL; readlicense_oo/license/license.xml
@@ -183,8 +145,6 @@ Files:
183
  - `sw/qa/uitest/table/splitTable.py`: LibreOffice Writer UI test splitTable.py: Table > Split Table on splitTable.odt at row 3 with copy heading, custom heading and no heading.
184
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF document it loads
185
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
186
- Reference result: asserted_document_state
187
- Reference source: assertEqual lines of test_split_table: copy heading gives 2 tables of 2 and 5 rows (lines 25, 27, 28); custom heading gives 2 and 4 rows (lines 42, 44, 45); no heading gives 2 and 4 rows (lines 59, 61, 62). Kept: these 9 assertions. Dropped: the Undo assertions (lines 31, 48, 65) and test_tdf115572_remember_split_table_option (line 67), which asserts only a dialog radio state (line 85)
188
  Check: Final file state: the agent splits the table of splitTable.odt before its third row (the cursor position after two GoDown from the document start) with the heading mode named in the task and saves; read the saved .odt: exactly 2 text tables with the row counts the script asserts for that mode
189
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:SplitTable) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
190
  Companions (not included): sw/qa/uitest/data/splitTable.odt; COPYING.MPL; readlicense_oo/license/license.xml
@@ -192,8 +152,6 @@ Files:
192
  - `sw/qa/uitest/writer_tests/insertCaption.py`: LibreOffice Writer UI test insertCaption.py: Insert > Frame, then Insert > Caption three times with separator and position options.
193
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API); starts from a new empty document, no data file
194
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
195
- Reference result: asserted_document_state
196
- Reference source: assertEqual lines of insertCaption.py: one text frame (line 29); frame text "\nText 1: Caption" (line 37), then "\nText 1: Caption\nText 2-: Caption2" (line 46), then "\nText 1: Caption\nText 2-: Caption2\nText 3--: Caption3" (line 57). Kept: all 4
197
  Check: Final file state: the agent creates a Writer document with one frame and adds the three captions with category Text, the separators and the position of the script, then saves; read the saved .odt: one text frame whose text (CRLF normalised to LF) equals the string of line 57
198
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:InsertCaptionDialog) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
199
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -201,8 +159,6 @@ Files:
201
  - `sd/qa/uitest/impress_tests2/tdf139511.py`: LibreOffice Impress UI test tdf139511.py: insert a 4 x 4 table on a new slide and set its size to 10 cm x 5 cm through Format > Position and Size.
202
  Format: Python UI test (unittest class driving Impress dialogs through the UNO UITest API); starts from a new empty presentation, no data file
203
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
204
- Reference result: asserted_document_state
205
- Reference source: assertEqual lines of tdf139511.py on the table shape BoundRect in 1/100 mm: before resizing 3885 x 14136 (lines 26, 27), after setting width 10 cm and height 5 cm 5037 x 10037 (lines 47, 48). Kept: lines 47 and 48; lines 26 and 27 describe the starting state
206
  Check: Final file state: the agent inserts a 4 x 4 table on the first slide of a new presentation and resizes it to 10 cm wide and 5 cm high, then saves as .odp; reopen in soffice --headless and read DrawPages[0] table shape BoundRect: Height 5037 and Width 10037 (1/100 mm, the bound rectangle includes the border line); the svg:width and svg:height of the table frame in content.xml (10cm and 5cm) are an equivalent file check
207
  Tools: soffice --headless with the UNO Python API: create the presentation, insert the table shape (dispatcher .uno:InsertTable?Columns:short=4&Rows:short=4 or a com.sun.star.drawing.TableShape), set its Size to 10000 x 5000 (1/100 mm), store as .odp; or write the table frame size in content.xml with odfpy
208
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -210,8 +166,6 @@ Files:
210
  - `sw/qa/uitest/writer_tests3/sort.py`: LibreOffice Writer UI test sort.py: Tools > Sort on three typed paragraphs (descending) and on two ";"-separated number lines (descending, numerical, column separator ";").
211
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
212
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
213
- Reference result: asserted_document_state
214
- Reference source: assertEqual lines of sort.py on the new document text: after the descending sort the text starts with "v" (line 34); after the descending numerical sort with separator ";" the text starts with "2;8;3" (line 58). Kept: both assertions (both read the document text). Dropped: none
215
  Check: Final file state: the agent types or is given the paragraphs, sorts them as the task says and saves; read the saved .odt with the UNO API or odfpy and compare the leading characters of the document text with the expected string. No screenshot needed
216
  Tools: soffice --headless with the UNO Python API: XSortable.createSortDescriptor / sort on the text cursor (or the dispatcher call .uno:SortDialog with arguments), then store; or reorder the paragraphs of content.xml with odfpy or odfdo
217
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -219,8 +173,6 @@ Files:
219
  - `sw/qa/uitest/writer_tests3/insertFootEndnote.py`: LibreOffice Writer UI test insertFootEndnote.py: Insert > Footnote and Endnote with automatic numbering and with the custom character "A", each as footnote and as endnote.
220
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
221
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
222
- Reference result: asserted_document_state
223
- Reference source: assertEqual lines of insertFootEndnote.py on the document: one footnote after the automatic insert (line 22), one endnote (line 30), one footnote with the custom mark "A" (line 40), one endnote with the custom mark (line 54). Kept: these 4 counts. Dropped: the 4 Undo assertions (lines 24, 32, 42, 56) and the cancel-button case (line 59, no assertion)
224
  Check: Final file state: the agent inserts the footnote or endnote named in the task (automatic or with the character mark) and saves; read the saved .odt with the UNO API (XFootnotesSupplier.Footnotes, XEndnotesSupplier.Endnotes) or count text:note elements by text:note-class in content.xml, and for a character-mark task check the text:note-citation label. No screenshot needed
225
  Tools: soffice --headless with the UNO Python API: createInstance("com.sun.star.text.Footnote") or ("com.sun.star.text.Endnote"), set Label for a character mark, insertTextContent, then store; or add text:note elements to content.xml with odfpy or odfdo
226
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -228,8 +180,6 @@ Files:
228
  - `sw/qa/uitest/table/insertTableDialog.py`: LibreOffice Writer UI test insertTableDialog.py: Table > Insert Table with name "Test3", 2 columns and 2 rows.
229
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
230
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
231
- Reference result: asserted_document_state
232
- Reference source: assertEqual lines of insertTableDialog.py on the document after the dialog closes: the first text table is named "Test3" (line 46) and has 2 rows (line 47) and 2 columns (line 48). Kept: these 3 assertions. Dropped: the 3 dialog field assertions (lines 39 to 41) and the cancel-button case (test_cancel_button_insert_table_dialog, line 60: no table after Cancel, a dialog behaviour check)
233
  Check: Final file state: the agent inserts the table the task names and saves; read the saved .odt with the UNO API (getTextTables) or odfpy (table:table name, table:table-row and table:table-column counts) and compare name, rows and columns exactly. No screenshot needed
234
  Tools: soffice --headless with the UNO Python API: createInstance("com.sun.star.text.TextTable"), initialize(2, 2), setName("Test3"), insertTextContent, then store; or write a table:table element into content.xml with odfpy or odfdo
235
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -237,8 +187,6 @@ Files:
237
  - `sw/qa/uitest/writer_tests3/pageNumber.py`: LibreOffice Writer UI test pageNumber.py: Insert > Page Number wizard: number in the footer (default), then in the header with format "A, B, C, ..."; and right-aligned mirrored footer numbers on a two-page document (tdf#165852).
238
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
239
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
240
- Reference result: asserted_document_state
241
- Reference source: assertions of pageNumber.py on the page style: no header and footer text "1" after the default wizard run (lines 38 to 40); header text "A" with the footer kept after choosing the header and letters (lines 54, 55); on a two-page document no header and footer texts "1" (default), "1" (first page), "2" (left pages), "1" (right pages) (lines 83, 87 to 90). Kept: these 9 assertions. Dropped: the 16 dialog field assertions (lines 20 to 36, 73, 77, 80), the 4 assertions after Undo (lines 44, 46, 60, 61) and the Pages count of the edit window (line 69, view state)
242
  Check: Final file state: the agent inserts page numbers as the task says and saves; read the saved .odt with the UNO API (StyleFamilies.PageStyles.Standard: HeaderText, FooterText, FooterTextFirst, FooterTextLeft, FooterTextRight) or the style:master-page header and footer of styles.xml with odfpy, and compare the rendered field text exactly. No screenshot needed
243
  Tools: soffice --headless with the UNO Python API: set HeaderIsOn or FooterIsOn on the Standard page style, insert a com.sun.star.text.TextField.PageNumber with NumberingType, set the paragraph adjust, then store; or edit style:master-page in styles.xml with odfpy or odfdo
244
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -246,8 +194,6 @@ Files:
246
  - `sc/qa/uitest/textCase/textCase.py`: LibreOffice Calc UI test textCase.py: change case of typed cells A1, A2, A4 (one empty cell A3/A5 in the selection) to title case, sentence case, toggle case, upper case and lower case, first on one range then on a two-part selection.
247
  Format: Python UI test script (LibreOffice uitest framework) that builds its document by typing cells, runs .uno:ChangeCase* commands and asserts cell strings
248
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
249
- Reference result: asserted_document_state
250
- Reference source: assertEqual lines of textCase.py on cell strings of column A after each case command: lines 31-34 (title), 39-42 (sentence), 46-49 (toggle), 53-56 (upper), 60-63 (lower) on A1:A4; lines 79-82 (lower), 87-90 (title), 95-98 (sentence), 102-105 (toggle), 109-112 (upper), 116-119 (lower) on the selection A1:A2 plus A4:A5. Kept: these 40 assertions. Dropped: the 8 rotate-case assertions (lines 67-70 and 123-126), because .uno:ChangeCaseRotateCase gives a result that depends on the command run before it (UI state, not a rule on the cell text)
251
  Check: Final file state: the agent is given the typed cells, applies the case change the task names and saves; read the saved cells with odfpy or the UNO API and compare each string with the expected string exactly (empty cells stay empty). No screenshot needed
252
  Tools: LibreOffice headless with the UNO Python API (read each cell string, apply the case rule, setString, save), or openpyxl/odfpy with Python str.title / str.capitalize / str.swapcase / str.upper / str.lower; judge by reading the saved cells
253
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -255,8 +201,6 @@ Files:
255
  - `sc/qa/uitest/sort/naturalSort.py`: LibreOffice Calc UI test naturalSort.py: Data > Sort with natural sort on 5 typed part-number strings (rows), on 4 "MW-n" strings in D1:D4, and left to right on 4 "MW-n" strings in A1:D1.
256
  Format: Python UI test script (LibreOffice uitest framework) that types cells, runs Data > Sort with natural sort and asserts cell strings
257
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
258
- Reference result: asserted_document_state
259
- Reference source: assertEqual lines of naturalSort.py on cell strings after each natural sort: lines 42-46 (A1:A5 top to bottom), lines 74-77 (D1:D4), lines 108-111 (A1:D1 left to right). Kept: these 13 assertions. Dropped: lines 50-54 (state after Undo, a UI history check)
260
  Check: Final file state: the agent is given the typed strings, sorts them naturally as the task says (rows, or columns left to right) and saves; read the saved cells and compare each with the expected string exactly, in order. No screenshot needed
261
  Tools: LibreOffice headless with the UNO Python API (sort descriptor with IsNaturalSort true; for the column case IsSortColumns false / SortColumns), or read the cells with odfpy/openpyxl and sort with a natural sort key (digit runs compared as numbers), write back and save
262
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -264,8 +208,6 @@ Files:
264
  - `sc/qa/uitest/sort/sorting.py`: LibreOffice Calc UI test sorting.py: toolbar ascending/descending sort on a typed two-column list (Number; Misc), with header detection and the sort key taken from the active column.
265
  Format: Python UI test script (LibreOffice uitest framework) that types a two-column list, sorts it with the toolbar commands and asserts cell values
266
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
267
- Reference result: asserted_document_state
268
- Reference source: assertEqual lines of sorting.py on cells after toolbar sorts: header row stays in place (lines 40-41, 49-50, 68-69), no header detected so row 1 moves (lines 91-92), column A sorted descending 6, 4, 3, 2 (lines 126-130), column B sorted ascending d, f, g, s (lines 138-142), sort key column A after TAB (lines 168-172), sort key column B after TAB (lines 197-201). Kept: these 28 assertions. Dropped: line 61 and line 103 (state of the header check box in the Sort dialog) and lines 228-270 (8 dialog-control assertions of test_Sorting_sort_criteria)
269
  Check: Final file state: the agent is given the typed list, sorts it as the task says (key column, direction, first row as header or not) and saves; read the saved cells and compare: strings exactly, numbers by value. No screenshot needed
270
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
271
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
@@ -273,8 +215,6 @@ Files:
273
  - `sc/qa/uitest/sort/stableSorting.py`: LibreOffice Calc UI test stableSorting.py with stableSorting.ods: four successive toolbar sorts (Sales, Product, Salesman, Region) on a 500-row sales list must keep earlier orders (stable sort), checked by the CheckOrder column F.
274
  Format: Python UI test script plus one ODF spreadsheet (.ods, 500 data rows in columns A to F, F = CheckOrder)
275
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
276
- Reference result: asserted_document_state
277
- Reference source: assertEqual lines of stableSorting.py: after ascending sorts on E, D, C, B, CheckOrder F2:F501 = 1..500 (line 38, loop over range(1, 501)); after descending sorts on E, D, C, B, F2:F501 = 500..1 (line 59). Kept: both loops, 1000 cell checks. Dropped: none
278
  Check: Final file state: the agent opens the .ods, sorts it by the keys in the order the task gives (stable) and saves; read column F of the saved sheet and check it equals 1..500 (ascending task) or 500..1 (descending task). No screenshot needed
279
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
280
  Companions (not included): sc/qa/uitest/data/stableSorting.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -282,8 +222,6 @@ Files:
282
  - `sc/qa/uitest/calc_tests/autosum.py`: LibreOffice Calc UI test autosum.py with autosum.ods: AutoSum on three selected rows (B126:D126, B128:D128, B130:D130) with no empty result cell selected, sums written to column E.
283
  Format: Python UI test script plus one ODF spreadsheet (.ods) holding 13 AutoSum test layouts; this script tests layout 8
284
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
285
- Reference result: asserted_document_state
286
- Reference source: assertEqual lines of autosum.py on E126, E128, E130: values 30, 90, 150 and formulas =SUM(B126:D126), =SUM(B128:D128), =SUM(B130:D130) (lines 29-34). Kept: all 6. Dropped: none
287
  Check: Final file state: the agent opens the .ods, adds row sums for rows 126, 128 and 130 in column E and saves; read value and formula of E126, E128, E130. No screenshot needed
288
  Tools: LibreOffice headless with the UNO Python API (setFormula("=SUM(B126:D126)") on E126 etc., calculateAll, save), or odfpy/openpyxl to write the formulas and a headless recalculation with soffice --convert-to; judge by reading formula and value of each result cell
289
  Companions (not included): sc/qa/uitest/data/autosum.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -291,8 +229,6 @@ Files:
291
  - `sc/qa/uitest/calc_tests/mergedRowsColumns.py`: LibreOffice Calc UI test mergedRowsColumns.py (method test_calculations_in_merged_cells_tdf51368) with tdf51368.ods: merge cells that feed formulas, choosing to move hidden contents into the first cell or keep them, then check the merged text and the dependent formula results.
292
  Format: Python UI test script plus one ODF spreadsheet (.ods) with formulas that read the cells to be merged
293
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
294
- Reference result: asserted_document_state
295
- Reference source: assertEqual lines of test_calculations_in_merged_cells_tdf51368 after each merge: move contents of A11:A12 gives A11 "2 3" and B12 "0" (lines 72-73, again 97-99 with A12 ""); keep contents gives A11 "2", A12 "3", B12 "3" (lines 84-86); A21:A22 kept (lines 110-112); A30:A32 kept, B32 "thisisbad" (line 123); J12:K12 kept (lines 135-136); J22:K22 kept (lines 147-148). Kept: these 16 assertions. Dropped: the 13 assertions after Undo in this method, and all assertions of the other 4 methods (they need tdf105412.ods or check undo state or cursor position)
296
  Check: Final file state: the agent opens the .ods, merges the named range with the named option (move hidden contents to the first cell, or keep them) and saves; read the cell strings (after recalculation) and compare exactly with the expected strings. No screenshot needed
297
  Tools: LibreOffice headless with the UNO Python API (XMergeable.merge on the range after concatenating the contents into the first cell with a space when moving, or leaving them when keeping; calculateAll; save), judged by reading cell strings with odfpy or the UNO API
298
  Companions (not included): sc/qa/uitest/data/tdf51368.ods; COPYING.MPL; readlicense_oo/license/license.xml
@@ -300,8 +236,6 @@ Files:
300
  - `sc/qa/uitest/conditional_format/tdf81696_sort_cells_with_conditional_formatting.py`: LibreOffice Calc UI test tdf81696: Data > Sort (column A ascending, header row) on A1:B3 that carries conditional formatting; check the sorted values.
301
  Format: Python UI test script plus one ODF spreadsheet (.ods) with a 3 x 2 range under conditional formatting
302
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark (regression test for a crash, bug tdf#81696)
303
- Reference result: asserted_document_state
304
- Reference source: assertEqual lines on A1:B3 after the sort: A1 "A", B1 "B", A2 1, B2 2, A3 2, B3 1 (lines 31-36). Kept: all 6. Dropped: none
305
  Check: Final file state: the agent opens the .ods, sorts A1:B3 by column A ascending with the first row as header and saves; the file must still open and the six cells must match (strings exact, numbers by value). No screenshot needed
306
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
307
  Companions (not included): sc/qa/uitest/data/tdf81696.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
8
 
9
  **Self-contained:** 13 of 32 files are self-contained; the others are used together with other upstream files. 32 of 32 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** UNO Python API headless: open goalSeek.ods, call XGoalSeek.seekGoal on the sheet (formula cell B4, variable cell B1, target 15000) and write the result into B1, then store; or solve directly and write B1 with odfpy and recalculate in soffice --headless. Other files: see the notes.
12
 
13
  **Licence (SPDX):** MPL-2.0
 
19
  - `sc/qa/uitest/goalSeek/goalSeek.py`: LibreOffice Calc UI test goalSeek.py: Tools > Goal Seek on goalSeek.ods (find the principal B1 so that interest B4 = 15000).
20
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
21
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
 
 
22
  Check: Final file state: after the agent saves, B1 must equal 200000 and B4 must equal 15000 on recalculation (read with the UNO API or odfpy after soffice --headless recalculation). Dialog-field assertions in the script are ignored
23
  Tools: UNO Python API headless: open goalSeek.ods, call XGoalSeek.seekGoal on the sheet (formula cell B4, variable cell B1, target 15000) and write the result into B1, then store; or solve directly and write B1 with odfpy and recalculate in soffice --headless
24
  Companions (not included): sc/qa/uitest/data/goalSeek.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
26
  - `sc/qa/uitest/textToColumns/textToColumns.py`: LibreOffice Calc UI test textToColumns.py: Data > Text to Columns on six documents with comma, dot, pipe, pipe plus space, semicolon and space separators.
27
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
28
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
 
 
29
  Check: Final file state: open the saved document and compare every asserted cell (sheet 0, given row and column) with the asserted value; all 210 must match. No screenshot needed
30
  Tools: UNO Python API headless (read column A, split, write cells, store) or odfpy/pandas on the .ods; soffice --headless --convert-to csv to read back
31
  Companions (not included): sc/qa/uitest/data/text_to_columns_comma.ods; sc/qa/uitest/data/text_to_columns_dot.ods; sc/qa/uitest/data/text_to_columns_pipe.ods; sc/qa/uitest/data/text_to_columns_pipe_space.ods; sc/qa/uitest/data/text_to_columns_semicolon.ods; sc/qa/uitest/data/text_to_columns_space.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
33
  - `sc/qa/uitest/sort/subtotals.py`: LibreOffice Calc UI test subtotals.py: Data > Subtotals (group by, sum, page breaks, sort options) on four bug-report workbooks.
34
  Format: Python UI test (unittest class driving Calc dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
35
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
 
 
36
  Check: Final file state: after the agent saves, read the asserted cells and compare exactly; subtotal rows and formulas must sit where the assertions expect them. No screenshot needed
37
  Tools: UNO Python API headless: XSubTotalCalculatable.createSubTotalDescriptor and applySubTotals on the data range, then store; check with odfpy or soffice --headless --convert-to csv
38
  Companions (not included): sc/qa/uitest/data/tdf55734.ods; sc/qa/uitest/data/tdf56958.ods; sc/qa/uitest/data/tdf88735.ods; sc/qa/uitest/data/tdf88792.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
40
  - `sw/qa/uitest/table/textToTable.py`: LibreOffice Writer UI test textToTable.py: Table > Convert > Text to Table with semicolon and custom ':' separators and a heading row.
41
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
42
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
 
 
43
  Check: Final file state: open the saved .odt and count tables, rows and columns (odfpy or UNO), compare with the asserted values. No screenshot needed
44
  Tools: UNO Python API headless: select the text range and call XTextConvert (convertToTable) or build the table with XTextTable, then store; check with odfpy (count table:table, rows and columns) or python-docx after soffice --headless --convert-to docx
45
  Companions (not included): sw/qa/uitest/data/textToTable.odt; COPYING.MPL; readlicense_oo/license/license.xml
 
47
  - `sw/qa/uitest/findReplace/findReplace.py`: LibreOffice Writer UI test findReplace.py: Edit > Find and Replace, including regular-expression replace with back references and diacritic replacement.
48
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF data documents it loads; each test method is one scenario: open document, perform the dialog action, assert the document state
49
  Origin: UI regression tests written by LibreOffice developers and QA; data documents are bug-report attachments or test author files (tool_fixture); the uitest framework (uitest/, the libreoffice Python helpers) is named for provenance only: not bundled, not run
 
 
50
  Check: Final file state: read the saved document text and compare the asserted prefixes. Only the 10 document-text assertions count; dialog-field and current-page assertions are dropped. No screenshot needed
51
  Tools: UNO Python API headless: XReplaceable.createReplaceDescriptor with SearchRegularExpression, then replaceAll and store; or edit content.xml text with odfpy
52
  Companions (not included): sw/qa/uitest/data/findReplace.odt; sw/qa/uitest/data/tdf116242.odt; COPYING.MPL; readlicense_oo/license/license.xml
 
54
  - `sc/qa/uitest/statistics/anova.py`: LibreOffice Calc UI test anova.py: Data > Statistics > Analysis of Variance (single factor) (.uno:AnalysisOfVarianceDialog) on typed data, input $Sheet1.$A$2:$C$13 grouped by columns, alpha 0.05, output $F$1.
55
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
56
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
57
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Analysis of Variance (single factor) table with the settings given (input $Sheet1.$A$2:$C$13 grouped by columns, alpha 0.05, output $F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
58
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Analysis of Variance (single factor) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
59
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
61
  - `sc/qa/uitest/statistics/chiSquare.py`: LibreOffice Calc UI test chiSquare.py: Data > Statistics > Chi-square Test (.uno:ChiSquareTestDialog) on typed data, input $Sheet1.$A$1:$B$13 grouped by columns, output F1.
62
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
63
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
64
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Chi-square Test table with the settings given (input $Sheet1.$A$1:$B$13 grouped by columns, output F1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
65
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Chi-square Test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
66
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
68
  - `sc/qa/uitest/statistics/correlation.py`: LibreOffice Calc UI test correlation.py: Data > Statistics > Correlation (.uno:CorrelationDialog) on typed data, test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7.
69
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
70
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
71
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Correlation table with the settings given (test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
72
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Correlation output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
73
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
75
  - `sc/qa/uitest/statistics/covariance.py`: LibreOffice Calc UI test covariance.py: Data > Statistics > Covariance (.uno:CovarianceDialog) on typed data, test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7.
76
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
77
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
78
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Covariance table with the settings given (test 1 input $Sheet1.$A$1:$C$13 to $F$1, test 2 input $Sheet1.$A$1:$M$3 grouped by rows to $A$7), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
79
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Covariance output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
80
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
82
  - `sc/qa/uitest/statistics/descriptiveStatistics.py`: LibreOffice Calc UI test descriptiveStatistics.py: Data > Statistics > Descriptive Statistics (.uno:DescriptiveStatisticsDialog) on typed data, input $Sheet1.$A$1:$C$13 grouped by columns, output $Sheet1.$F$1.
83
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
84
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
85
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Descriptive Statistics table with the settings given (input $Sheet1.$A$1:$C$13 grouped by columns, output $Sheet1.$F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
86
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Descriptive Statistics output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
87
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
89
  - `sc/qa/uitest/statistics/exponentialSmoothing.py`: LibreOffice Calc UI test exponentialSmoothing.py: Data > Statistics > Exponential Smoothing (.uno:ExponentialSmoothingDialog) on typed data, input $Sheet1.$A$1:$B$13, smoothing factor 0.5, output $F$1, grouped by columns (test 1) and by rows (test 2).
90
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
91
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
92
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Exponential Smoothing table with the settings given (input $Sheet1.$A$1:$B$13, smoothing factor 0.5, output $F$1, grouped by columns (test 1) and by rows (test 2)), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
93
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Exponential Smoothing output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
94
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
96
  - `sc/qa/uitest/statistics/fTest.py`: LibreOffice Calc UI test fTest.py: Data > Statistics > F-test (.uno:FTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
97
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
98
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
99
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the F-test table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
100
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the F-test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
101
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
103
  - `sc/qa/uitest/statistics/movingAverage.py`: LibreOffice Calc UI test movingAverage.py: Data > Statistics > Moving Average (.uno:MovingAverageDialog) on typed data, input $Sheet1.$A$1:$B$13, interval 3, output $F$1, grouped by columns (test 1) and by rows (test 2).
104
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
105
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
106
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Moving Average table with the settings given (input $Sheet1.$A$1:$B$13, interval 3, output $F$1, grouped by columns (test 1) and by rows (test 2)), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
107
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Moving Average output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
108
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
110
  - `sc/qa/uitest/statistics/regression.py`: LibreOffice Calc UI test regression.py: Data > Statistics > Regression (.uno:RegressionDialog) on regression.ods, linear with and without intercept, logarithmic and power, data grouped by columns and by rows.
111
  Format: Python UI test (unittest class driving the Calc Regression dialog through the UNO UITest API) with the ODF data document it loads; the document carries its own expected values and check formulas
112
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
113
  Check: Final file state: the agent receives regression.ods and must fill the four Actual sheets with the regression output for the data in DataInColumns (or DataInRows); after soffice --headless recalculation of the saved file, CombineChecks B7 must read TRUE, which holds when every expected cell matches its actual cell within the tolerance of ChecksLinear J4 (text cells compared exactly)
114
  Tools: soffice --headless with the UNO Python API: open regression.ods, write the regression output tables into sheets ActualLinear, ActualLinearNoIntercept, ActualLog and ActualPower at A1 (formulas such as LINEST or computed values), store; or compute the fits with numpy and write the cells with odfpy, then recalculate in soffice --headless
115
  Companions (not included): sc/qa/uitest/data/regression.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
117
  - `sc/qa/uitest/statistics/sampling.py`: LibreOffice Calc UI test sampling.py: Data > Statistics > Sampling (periodic) (.uno:SamplingDialog) on typed data, input $Sheet1.$A$1:$C$9, periodic method with period 2, output $Sheet1.$F$1.
118
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
119
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
120
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the Sampling (periodic) table with the settings given (input $Sheet1.$A$1:$C$9, periodic method with period 2, output $Sheet1.$F$1), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
121
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the Sampling (periodic) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
122
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
124
  - `sc/qa/uitest/statistics/tTest.py`: LibreOffice Calc UI test tTest.py: Data > Statistics > t-test (paired) (.uno:TTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
125
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
126
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
127
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the t-test (paired) table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
128
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the t-test (paired) output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
129
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
131
  - `sc/qa/uitest/statistics/zTest.py`: LibreOffice Calc UI test zTest.py: Data > Statistics > z-test (.uno:ZTestDialog) on typed data, variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns.
132
  Format: Python UI test (unittest class driving a Calc Data > Statistics dialog through the UNO UITest API); the input data is typed into a new spreadsheet by the enter_text_to_cell lines of the script itself, so no data document is needed; each test method is one scenario: type data, run the dialog, assert the output cells
133
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
134
  Check: Final file state: build the starting sheet from the script's enter_text_to_cell lines, let the agent produce the z-test table with the settings given (variable 1 $Sheet1.$A$1:$A$13, variable 2 $Sheet1.$B$1:$B$13, output F1, grouped by columns), save as .ods; open the saved file after soffice --headless recalculation (UNO API or odfpy) and compare each asserted numeric cell with the expected value of the script using the rounding of the same assertion; label strings are compared exactly when the output language is en-US and are otherwise optional. Dialog-field and Undo assertions are ignored
135
  Tools: soffice --headless with the UNO Python API: create the sheet from the enter_text_to_cell values of the script, write the z-test output table (labels and formulas or values) at the stated output cell, store as .ods; or compute the statistics in Python (statistics module or numpy) and write the cells with odfpy, then recalculate in soffice --headless
136
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
138
  - `sw/qa/uitest/table/tableToText.py`: LibreOffice Writer UI test tableToText.py: Table > Convert > Table to Text on tableToText.odt with separators tab, paragraph, semicolon and a custom ":".
139
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF document it loads
140
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
141
  Check: Final file state: the agent converts the table of tableToText.odt to text with the separator named in the task and saves; read the saved .odt with the UNO API or odfpy: no text table remains and the document text (CRLF normalised to LF) starts with the expected string of the script for that separator
142
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:ConvertTableToText) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
143
  Companions (not included): sw/qa/uitest/data/tableToText.odt; COPYING.MPL; readlicense_oo/license/license.xml
 
145
  - `sw/qa/uitest/table/splitTable.py`: LibreOffice Writer UI test splitTable.py: Table > Split Table on splitTable.odt at row 3 with copy heading, custom heading and no heading.
146
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) with the ODF document it loads
147
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
148
  Check: Final file state: the agent splits the table of splitTable.odt before its third row (the cursor position after two GoDown from the document start) with the heading mode named in the task and saves; read the saved .odt: exactly 2 text tables with the row counts the script asserts for that mode
149
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:SplitTable) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
150
  Companions (not included): sw/qa/uitest/data/splitTable.odt; COPYING.MPL; readlicense_oo/license/license.xml
 
152
  - `sw/qa/uitest/writer_tests/insertCaption.py`: LibreOffice Writer UI test insertCaption.py: Insert > Frame, then Insert > Caption three times with separator and position options.
153
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API); starts from a new empty document, no data file
154
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
155
  Check: Final file state: the agent creates a Writer document with one frame and adds the three captions with category Text, the separators and the position of the script, then saves; read the saved .odt: one text frame whose text (CRLF normalised to LF) equals the string of line 57
156
  Tools: soffice --headless with the UNO Python API (XTextDocument, XTextTable, dispatcher calls such as .uno:InsertCaptionDialog) on the document, then store; or edit content.xml of the .odt with odfpy or odfdo
157
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
159
  - `sd/qa/uitest/impress_tests2/tdf139511.py`: LibreOffice Impress UI test tdf139511.py: insert a 4 x 4 table on a new slide and set its size to 10 cm x 5 cm through Format > Position and Size.
160
  Format: Python UI test (unittest class driving Impress dialogs through the UNO UITest API); starts from a new empty presentation, no data file
161
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
162
  Check: Final file state: the agent inserts a 4 x 4 table on the first slide of a new presentation and resizes it to 10 cm wide and 5 cm high, then saves as .odp; reopen in soffice --headless and read DrawPages[0] table shape BoundRect: Height 5037 and Width 10037 (1/100 mm, the bound rectangle includes the border line); the svg:width and svg:height of the table frame in content.xml (10cm and 5cm) are an equivalent file check
163
  Tools: soffice --headless with the UNO Python API: create the presentation, insert the table shape (dispatcher .uno:InsertTable?Columns:short=4&Rows:short=4 or a com.sun.star.drawing.TableShape), set its Size to 10000 x 5000 (1/100 mm), store as .odp; or write the table frame size in content.xml with odfpy
164
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
166
  - `sw/qa/uitest/writer_tests3/sort.py`: LibreOffice Writer UI test sort.py: Tools > Sort on three typed paragraphs (descending) and on two ";"-separated number lines (descending, numerical, column separator ";").
167
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
168
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
169
  Check: Final file state: the agent types or is given the paragraphs, sorts them as the task says and saves; read the saved .odt with the UNO API or odfpy and compare the leading characters of the document text with the expected string. No screenshot needed
170
  Tools: soffice --headless with the UNO Python API: XSortable.createSortDescriptor / sort on the text cursor (or the dispatcher call .uno:SortDialog with arguments), then store; or reorder the paragraphs of content.xml with odfpy or odfdo
171
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
173
  - `sw/qa/uitest/writer_tests3/insertFootEndnote.py`: LibreOffice Writer UI test insertFootEndnote.py: Insert > Footnote and Endnote with automatic numbering and with the custom character "A", each as footnote and as endnote.
174
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
175
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
176
  Check: Final file state: the agent inserts the footnote or endnote named in the task (automatic or with the character mark) and saves; read the saved .odt with the UNO API (XFootnotesSupplier.Footnotes, XEndnotesSupplier.Endnotes) or count text:note elements by text:note-class in content.xml, and for a character-mark task check the text:note-citation label. No screenshot needed
177
  Tools: soffice --headless with the UNO Python API: createInstance("com.sun.star.text.Footnote") or ("com.sun.star.text.Endnote"), set Label for a character mark, insertTextContent, then store; or add text:note elements to content.xml with odfpy or odfdo
178
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
180
  - `sw/qa/uitest/table/insertTableDialog.py`: LibreOffice Writer UI test insertTableDialog.py: Table > Insert Table with name "Test3", 2 columns and 2 rows.
181
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
182
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
183
  Check: Final file state: the agent inserts the table the task names and saves; read the saved .odt with the UNO API (getTextTables) or odfpy (table:table name, table:table-row and table:table-column counts) and compare name, rows and columns exactly. No screenshot needed
184
  Tools: soffice --headless with the UNO Python API: createInstance("com.sun.star.text.TextTable"), initialize(2, 2), setName("Test3"), insertTextContent, then store; or write a table:table element into content.xml with odfpy or odfdo
185
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
187
  - `sw/qa/uitest/writer_tests3/pageNumber.py`: LibreOffice Writer UI test pageNumber.py: Insert > Page Number wizard: number in the footer (default), then in the header with format "A, B, C, ..."; and right-aligned mirrored footer numbers on a two-page document (tdf#165852).
188
  Format: Python UI test (unittest class driving Writer dialogs through the UNO UITest API) on a document it creates itself
189
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the dialog writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
190
  Check: Final file state: the agent inserts page numbers as the task says and saves; read the saved .odt with the UNO API (StyleFamilies.PageStyles.Standard: HeaderText, FooterText, FooterTextFirst, FooterTextLeft, FooterTextRight) or the style:master-page header and footer of styles.xml with odfpy, and compare the rendered field text exactly. No screenshot needed
191
  Tools: soffice --headless with the UNO Python API: set HeaderIsOn or FooterIsOn on the Standard page style, insert a com.sun.star.text.TextField.PageNumber with NumberingType, set the paragraph adjust, then store; or edit style:master-page in styles.xml with odfpy or odfdo
192
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
194
  - `sc/qa/uitest/textCase/textCase.py`: LibreOffice Calc UI test textCase.py: change case of typed cells A1, A2, A4 (one empty cell A3/A5 in the selection) to title case, sentence case, toggle case, upper case and lower case, first on one range then on a two-part selection.
195
  Format: Python UI test script (LibreOffice uitest framework) that builds its document by typing cells, runs .uno:ChangeCase* commands and asserts cell strings
196
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
197
  Check: Final file state: the agent is given the typed cells, applies the case change the task names and saves; read the saved cells with odfpy or the UNO API and compare each string with the expected string exactly (empty cells stay empty). No screenshot needed
198
  Tools: LibreOffice headless with the UNO Python API (read each cell string, apply the case rule, setString, save), or openpyxl/odfpy with Python str.title / str.capitalize / str.swapcase / str.upper / str.lower; judge by reading the saved cells
199
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
201
  - `sc/qa/uitest/sort/naturalSort.py`: LibreOffice Calc UI test naturalSort.py: Data > Sort with natural sort on 5 typed part-number strings (rows), on 4 "MW-n" strings in D1:D4, and left to right on 4 "MW-n" strings in A1:D1.
202
  Format: Python UI test script (LibreOffice uitest framework) that types cells, runs Data > Sort with natural sort and asserts cell strings
203
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
204
  Check: Final file state: the agent is given the typed strings, sorts them naturally as the task says (rows, or columns left to right) and saves; read the saved cells and compare each with the expected string exactly, in order. No screenshot needed
205
  Tools: LibreOffice headless with the UNO Python API (sort descriptor with IsNaturalSort true; for the column case IsSortColumns false / SortColumns), or read the cells with odfpy/openpyxl and sort with a natural sort key (digit runs compared as numbers), write back and save
206
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
208
  - `sc/qa/uitest/sort/sorting.py`: LibreOffice Calc UI test sorting.py: toolbar ascending/descending sort on a typed two-column list (Number; Misc), with header detection and the sort key taken from the active column.
209
  Format: Python UI test script (LibreOffice uitest framework) that types a two-column list, sorts it with the toolbar commands and asserts cell values
210
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
211
  Check: Final file state: the agent is given the typed list, sorts it as the task says (key column, direction, first row as header or not) and saves; read the saved cells and compare: strings exactly, numbers by value. No screenshot needed
212
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
213
  Companions (not included): COPYING.MPL; readlicense_oo/license/license.xml
 
215
  - `sc/qa/uitest/sort/stableSorting.py`: LibreOffice Calc UI test stableSorting.py with stableSorting.ods: four successive toolbar sorts (Sales, Product, Salesman, Region) on a 500-row sales list must keep earlier orders (stable sort), checked by the CheckOrder column F.
216
  Format: Python UI test script plus one ODF spreadsheet (.ods, 500 data rows in columns A to F, F = CheckOrder)
217
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
218
  Check: Final file state: the agent opens the .ods, sorts it by the keys in the order the task gives (stable) and saves; read column F of the saved sheet and check it equals 1..500 (ascending task) or 500..1 (descending task). No screenshot needed
219
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
220
  Companions (not included): sc/qa/uitest/data/stableSorting.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
222
  - `sc/qa/uitest/calc_tests/autosum.py`: LibreOffice Calc UI test autosum.py with autosum.ods: AutoSum on three selected rows (B126:D126, B128:D128, B130:D130) with no empty result cell selected, sums written to column E.
223
  Format: Python UI test script plus one ODF spreadsheet (.ods) holding 13 AutoSum test layouts; this script tests layout 8
224
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
225
  Check: Final file state: the agent opens the .ods, adds row sums for rows 126, 128 and 130 in column E and saves; read value and formula of E126, E128, E130. No screenshot needed
226
  Tools: LibreOffice headless with the UNO Python API (setFormula("=SUM(B126:D126)") on E126 etc., calculateAll, save), or odfpy/openpyxl to write the formulas and a headless recalculation with soffice --convert-to; judge by reading formula and value of each result cell
227
  Companions (not included): sc/qa/uitest/data/autosum.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
229
  - `sc/qa/uitest/calc_tests/mergedRowsColumns.py`: LibreOffice Calc UI test mergedRowsColumns.py (method test_calculations_in_merged_cells_tdf51368) with tdf51368.ods: merge cells that feed formulas, choosing to move hidden contents into the first cell or keep them, then check the merged text and the dependent formula results.
230
  Format: Python UI test script plus one ODF spreadsheet (.ods) with formulas that read the cells to be merged
231
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark
 
 
232
  Check: Final file state: the agent opens the .ods, merges the named range with the named option (move hidden contents to the first cell, or keep them) and saves; read the cell strings (after recalculation) and compare exactly with the expected strings. No screenshot needed
233
  Tools: LibreOffice headless with the UNO Python API (XMergeable.merge on the range after concatenating the contents into the first cell with a space when moving, or leaving them when keeping; calculateAll; save), judged by reading cell strings with odfpy or the UNO API
234
  Companions (not included): sc/qa/uitest/data/tdf51368.ods; COPYING.MPL; readlicense_oo/license/license.xml
 
236
  - `sc/qa/uitest/conditional_format/tdf81696_sort_cells_with_conditional_formatting.py`: LibreOffice Calc UI test tdf81696: Data > Sort (column A ascending, header row) on A1:B3 that carries conditional formatting; check the sorted values.
237
  Format: Python UI test script plus one ODF spreadsheet (.ods) with a 3 x 2 range under conditional formatting
238
  Origin: unit test of the LibreOffice project itself, written by its developers to check that the command writes the right result into the document; expected values are the ones LibreOffice must produce, not values from a GUI agent benchmark (regression test for a crash, bug tdf#81696)
 
 
239
  Check: Final file state: the agent opens the .ods, sorts A1:B3 by column A ascending with the first row as header and saves; the file must still open and the six cells must match (strings exact, numbers by value). No screenshot needed
240
  Tools: LibreOffice headless with the UNO Python API (XSortable.createSortDescriptor / sort on the cell range, then save and read the cells), or rewrite the rows with openpyxl/odfpy and a stable Python sort; judge by reading the saved file with odfpy or the UNO API
241
  Companions (not included): sc/qa/uitest/data/tdf81696.ods; COPYING.MPL; readlicense_oo/license/license.xml
campaign_files/2026-10-gt-browser-tests-2/mail-mime-parser-emails/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: inline_expected_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** zbateson/mail-mime-parser (PHP), or Python email.parser with policy.default plus get_body and iter_attachments; a script decodes the RFC 2047 headers, writes the text body and saves the three PNG attachments
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,8 +19,6 @@ Files:
21
  - `tests/_data/emails/github-115.txt`: Decode an Outlook message with encoded headers and save its text body and three image attachments.
22
  Format: RFC 5322 message; text and PNG files as answers
23
  Origin: Hunnysoft MIME sample used as a mail-mime-parser fixture (tool_fixture)
24
- Reference result: inline_expected_values
25
- Reference source: EmailFunctionalTest.php lines 1752-1775: From Doug Sauder doug@example.com, To Heinz Müller mueller@example.com, Subject, text body = files/hareandtortoise.txt, 3 attachments, part tree multipart/* > text/plain, image/png x3
26
  Check: Final file state: header values exact, attachment count 3 and part types exact, body text equal to hareandtortoise.txt ignoring whitespace (assertStringEqualsIgnoreWhiteSpace after converting the expected file from ISO-8859-1), saved PNGs byte-equal to the ball images. No screenshot
27
  Tools: zbateson/mail-mime-parser (PHP), or Python email.parser with policy.default plus get_body and iter_attachments; a script decodes the RFC 2047 headers, writes the text body and saves the three PNG attachments
28
  Companions (not included): tests/_data/emails/files/hareandtortoise.txt; tests/_data/emails/files/blueball.png; tests/_data/emails/files/greenball.png; tests/_data/emails/files/redball.png
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** zbateson/mail-mime-parser (PHP), or Python email.parser with policy.default plus get_body and iter_attachments; a script decodes the RFC 2047 headers, writes the text body and saves the three PNG attachments
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `tests/_data/emails/github-115.txt`: Decode an Outlook message with encoded headers and save its text body and three image attachments.
20
  Format: RFC 5322 message; text and PNG files as answers
21
  Origin: Hunnysoft MIME sample used as a mail-mime-parser fixture (tool_fixture)
 
 
22
  Check: Final file state: header values exact, attachment count 3 and part types exact, body text equal to hareandtortoise.txt ignoring whitespace (assertStringEqualsIgnoreWhiteSpace after converting the expected file from ISO-8859-1), saved PNGs byte-equal to the ball images. No screenshot
23
  Tools: zbateson/mail-mime-parser (PHP), or Python email.parser with policy.default plus get_body and iter_attachments; a script decodes the RFC 2047 headers, writes the text body and saves the three PNG attachments
24
  Companions (not included): tests/_data/emails/files/hareandtortoise.txt; tests/_data/emails/files/blueball.png; tests/_data/emails/files/greenball.png; tests/_data/emails/files/redball.png
campaign_files/2026-10-gt-browser-tests-2/mail-parser-eml-malformed/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: golden_output_files. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
14
 
15
  **Licence (SPDX):** MIT AND (Apache-2.0 OR MIT)
@@ -21,8 +19,6 @@ Files:
21
  - `resources/eml/malformed/000.eml`: Parse a malformed multipart message (boundary ":foo") from the Dovecot test set into the mail-parser JSON message structure.
22
  Format: RFC 5322 message (.eml) with malformed MIME as input; pretty-printed serde JSON of the parsed Message as answer
23
  Origin: Dovecot test message carried into mail-parser with its snapshot (tool_fixture); original source dovecot/core
24
- Reference result: golden_output_files
25
- Reference source: resources/eml/malformed/000.json (tests/integration_test.rs lines 84-105 compare every .eml of the snapshot suites with its .json)
26
  Check: Final file state: parse 000.eml and serialize with serde_json::to_string_pretty; the output must equal 000.json byte for byte (tests/integration_test.rs lines 90-93). Other parsers: compare parsed JSON on the fields the task names. No screenshot
27
  Tools: mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
28
  Companions (not included): resources/eml/malformed/000.json (golden JSON snapshot; hide); resources/eml/malformed/COPYING (licence of the malformed messages); tests/integration_test.rs (parse_full_messages); tests/support/mod.rs (snapshot_messages, snapshot_inputs); Cargo.toml, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
12
 
13
  **Licence (SPDX):** MIT AND (Apache-2.0 OR MIT)
 
19
  - `resources/eml/malformed/000.eml`: Parse a malformed multipart message (boundary ":foo") from the Dovecot test set into the mail-parser JSON message structure.
20
  Format: RFC 5322 message (.eml) with malformed MIME as input; pretty-printed serde JSON of the parsed Message as answer
21
  Origin: Dovecot test message carried into mail-parser with its snapshot (tool_fixture); original source dovecot/core
 
 
22
  Check: Final file state: parse 000.eml and serialize with serde_json::to_string_pretty; the output must equal 000.json byte for byte (tests/integration_test.rs lines 90-93). Other parsers: compare parsed JSON on the fields the task names. No screenshot
23
  Tools: mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
24
  Companions (not included): resources/eml/malformed/000.json (golden JSON snapshot; hide); resources/eml/malformed/COPYING (licence of the malformed messages); tests/integration_test.rs (parse_full_messages); tests/support/mod.rs (snapshot_messages, snapshot_inputs); Cargo.toml, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
campaign_files/2026-10-gt-browser-tests-2/mail-parser-eml-rfc/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: golden_output_files. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
14
 
15
  **Licence (SPDX):** Apache-2.0 OR MIT
@@ -21,8 +19,6 @@ Files:
21
  - `resources/eml/rfc/003.eml`: Parse the RFC 2046 multipart/alternative example message into the mail-parser JSON message structure.
22
  Format: RFC 5322 message (.eml) as input; pretty-printed serde JSON of the parsed Message as answer
23
  Origin: test fixture of mail-parser built from the RFC 2046 example message (tool_fixture)
24
- Reference result: golden_output_files
25
- Reference source: resources/eml/rfc/003.json, the expected serde_json::to_string_pretty output for 003.eml (tests/integration_test.rs lines 84-105); the CRLF variant would be 003.crlf.json, absent here, so the LF snapshot is the answer
26
  Check: Final file state: parse 003.eml and serialize the Message with serde_json::to_string_pretty; the output must equal 003.json byte for byte (tests/integration_test.rs lines 90-93 `if fs::read(&expected).ok().as_deref() != Some(json.as_bytes())`). For an agent not using the crate, compare its JSON to 003.json as parsed JSON on the fields the task names (headers, part tree, part bodies). No screenshot
27
  Tools: mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
28
  Companions (not included): resources/eml/rfc/003.json (golden JSON snapshot; the answer, hide from the agent); tests/integration_test.rs (parse_full_messages, the comparing test); tests/support/mod.rs (snapshot_messages, snapshot_inputs: LF and CRLF variants); src/parser/tests.rs and src/scan/tests.rs (parser unit tests, context for the JSON shape); Cargo.toml, README.md, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
12
 
13
  **Licence (SPDX):** Apache-2.0 OR MIT
 
19
  - `resources/eml/rfc/003.eml`: Parse the RFC 2046 multipart/alternative example message into the mail-parser JSON message structure.
20
  Format: RFC 5322 message (.eml) as input; pretty-printed serde JSON of the parsed Message as answer
21
  Origin: test fixture of mail-parser built from the RFC 2046 example message (tool_fixture)
 
 
22
  Check: Final file state: parse 003.eml and serialize the Message with serde_json::to_string_pretty; the output must equal 003.json byte for byte (tests/integration_test.rs lines 90-93 `if fs::read(&expected).ok().as_deref() != Some(json.as_bytes())`). For an agent not using the crate, compare its JSON to 003.json as parsed JSON on the fields the task names (headers, part tree, part bodies). No screenshot
23
  Tools: mail-parser Rust crate (MessageParser::parse, serde JSON), or Python email.parser with policy=email.policy.default and mailbox.mbox; a script writes the parsed structure to a JSON file
24
  Companions (not included): resources/eml/rfc/003.json (golden JSON snapshot; the answer, hide from the agent); tests/integration_test.rs (parse_full_messages, the comparing test); tests/support/mod.rs (snapshot_messages, snapshot_inputs: LF and CRLF variants); src/parser/tests.rs and src/scan/tests.rs (parser unit tests, context for the JSON shape); Cargo.toml, README.md, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
campaign_files/2026-10-gt-browser-tests-2/mail-parser-mbox/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** mail-parser MessageIterator over an mbox file, or Python mailbox.mbox with the From_ line parsed; a script writes each message (sender, internal date, body with >From unquoted) to a JSON file
14
 
15
  **Licence (SPDX):** Apache-2.0 OR MIT
@@ -21,8 +19,6 @@ Files:
21
  - `resources/mbox/sample.mbox`: Split a four-message mbox file and recover each message sender, From_ line date and unquoted body.
22
  Format: mbox (mboxrd quoting of >From lines) as input; expected Message list in Rust source as answer
23
  Origin: hand-written mbox test fixture of mail-parser (tool_fixture)
24
- Reference result: expected_values_in_test
25
- Reference source: src/mailbox/mbox.rs fn expected() (lines 266-290): 4 messages, internal_date 820631134, 1027453163, 1028640874, 1516030200, from god@heaven.af.mil, cras@irccrew.org, test@test.com, other@domain.com, and bodies "Message 1 ", "Message 2 ", "Message 3 From hello >From world >>From test ", "Message 4 > From >F " (one level of >From quoting removed)
26
  Check: Final file state: the agent writes one record per message (from, internal_date as Unix seconds UTC, contents); compare with expected() of mbox.rs exactly (the parse_mbox test, lines 339-343, asserts equality of the whole list). No screenshot
27
  Tools: mail-parser MessageIterator over an mbox file, or Python mailbox.mbox with the From_ line parsed; a script writes each message (sender, internal date, body with >From unquoted) to a JSON file
28
  Companions (not included): src/mailbox/mbox.rs (fn expected(), lines 266-290, holds the four expected messages; parse_mbox test lines 339-343; the answer, hide); src/mailbox/maildir.rs (sibling Maildir reader, context); Cargo.toml, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** mail-parser MessageIterator over an mbox file, or Python mailbox.mbox with the From_ line parsed; a script writes each message (sender, internal date, body with >From unquoted) to a JSON file
12
 
13
  **Licence (SPDX):** Apache-2.0 OR MIT
 
19
  - `resources/mbox/sample.mbox`: Split a four-message mbox file and recover each message sender, From_ line date and unquoted body.
20
  Format: mbox (mboxrd quoting of >From lines) as input; expected Message list in Rust source as answer
21
  Origin: hand-written mbox test fixture of mail-parser (tool_fixture)
 
 
22
  Check: Final file state: the agent writes one record per message (from, internal_date as Unix seconds UTC, contents); compare with expected() of mbox.rs exactly (the parse_mbox test, lines 339-343, asserts equality of the whole list). No screenshot
23
  Tools: mail-parser MessageIterator over an mbox file, or Python mailbox.mbox with the From_ line parsed; a script writes each message (sender, internal date, body with >From unquoted) to a JSON file
24
  Companions (not included): src/mailbox/mbox.rs (fn expected(), lines 266-290, holds the four expected messages; parse_mbox test lines 339-343; the answer, hide); src/mailbox/maildir.rs (sibling Maildir reader, context); Cargo.toml, LICENSES/MIT.txt, LICENSES/Apache-2.0.txt (licence)
campaign_files/2026-10-gt-browser-tests-2/mailparser-fixtures/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** mailparser (MailParser stream) from node, or Python email.parser plus html conversion; the script writes the plain text and the html body the mail client would show
14
 
15
  **Licence (SPDX):** MIT-0
@@ -21,8 +19,6 @@ Files:
21
  - `test/fixtures/nested.eml`: Flatten a nested multipart message into the single text body and single html body a reader shows.
22
  Format: RFC 5322 message with nested multipart/mixed and alternative parts as input; expected text and html strings in a nodeunit test as answer
23
  Origin: mailparser test fixture (tool_fixture)
24
- Reference result: expected_values_in_test
25
- Reference source: test/mail-parser-test.js lines 1670-1671: text equals '\nDear Sir,\n\nGood evening.\n\n\n\n\n\n\n\nThe footer\n' and html equals '<p>Dear Sir</p>\n<p>Good evening.</p>\n<p></p><br/>\n<p>The footer</p>\n'
26
  Check: Final file state: the agent writes text and html bodies to two files; each must equal the expected string exactly. The joining rule (text parts separated, html parts joined with <br/>) is mailparser specific; the task statement must give it. No screenshot
27
  Tools: mailparser (MailParser stream) from node, or Python email.parser plus html conversion; the script writes the plain text and the html body the mail client would show
28
  Companions (not included): test/mail-parser-test.js (Advanced nested HTML test; hide); LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** mailparser (MailParser stream) from node, or Python email.parser plus html conversion; the script writes the plain text and the html body the mail client would show
12
 
13
  **Licence (SPDX):** MIT-0
 
19
  - `test/fixtures/nested.eml`: Flatten a nested multipart message into the single text body and single html body a reader shows.
20
  Format: RFC 5322 message with nested multipart/mixed and alternative parts as input; expected text and html strings in a nodeunit test as answer
21
  Origin: mailparser test fixture (tool_fixture)
 
 
22
  Check: Final file state: the agent writes text and html bodies to two files; each must equal the expected string exactly. The joining rule (text parts separated, html parts joined with <br/>) is mailparser specific; the task statement must give it. No screenshot
23
  Tools: mailparser (MailParser stream) from node, or Python email.parser plus html conversion; the script writes the plain text and the html body the mail client would show
24
  Companions (not included): test/mail-parser-test.js (Advanced nested HTML test; hide); LICENSE
campaign_files/2026-10-gt-browser-tests-2/mammoth-test-data-docx/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 6 of 6 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test_asserts. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
14
 
15
  **Licence (SPDX):** BSD-2-Clause
@@ -21,8 +19,6 @@ Files:
21
  - `test/test-data/tables.docx`: Convert a Word document with a 2x2 table between two paragraphs to semantic HTML.
22
  Format: Word document (.docx) as input; HTML string as answer
23
  Origin: test fixture of mammoth.js (tool_fixture)
24
- Reference result: expected_values_in_test_asserts
25
- Reference source: test/mammoth.tests.js lines 345-357: <p>Above</p><table><tr><td><p>Top left</p></td><td><p>Top right</p></td></tr><tr><td><p>Bottom left</p></td><td><p>Bottom right</p></td></tr></table><p>Below</p>, messages empty
26
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 345-357 exactly, with the options the test passes; the conversion must raise no messages.
27
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
28
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
@@ -30,8 +26,6 @@ Files:
30
  - `test/test-data/footnotes.docx`: Convert a Word document with two footnotes to HTML with linked footnote references (id prefix doc-42-).
31
  Format: Word document (.docx) as input; HTML string as answer
32
  Origin: test fixture of mammoth.js (tool_fixture)
33
- Reference result: expected_values_in_test_asserts
34
- Reference source: test/mammoth.tests.js lines 359-374, expected string at lines 366-370: two <sup> links [1] and [2] to doc-42-footnote-1 and -2 and an <ol> with the footnote texts and back links, messages empty
35
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 359-374 exactly, with the options the test passes; the conversion must raise no messages.
36
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
37
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
@@ -39,8 +33,6 @@ Files:
39
  - `test/test-data/endnotes.docx`: Convert a Word document with two endnotes to HTML with linked endnote references (id prefix doc-42-).
40
  Format: Word document (.docx) as input; HTML string as answer
41
  Origin: test fixture of mammoth.js (tool_fixture)
42
- Reference result: expected_values_in_test_asserts
43
- Reference source: test/mammoth.tests.js lines 376-390: two <sup> links [1] and [2] to doc-42-endnote-2 and -3 (the Word endnote ids) and an <ol> with the endnote texts and back links, messages empty
44
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 376-390 exactly, with the options the test passes; the conversion must raise no messages.
45
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
46
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
@@ -48,8 +40,6 @@ Files:
48
  - `test/test-data/comments.docx`: Convert a Word document with two review comments to HTML that keeps the comments as a definition list.
49
  Format: Word document (.docx) as input; HTML string as answer
50
  Origin: test fixture of mammoth.js (tool_fixture)
51
- Reference result: expected_values_in_test_asserts
52
- Reference source: test/mammoth.tests.js lines 406-423: comment references [MW1] and [MW2] as <sup> links and a <dl> with Comment [MW1] / Comment [MW2] and their texts, messages empty
53
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 406-423 exactly, with the options the test passes; the conversion must raise no messages.
54
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
55
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
@@ -57,8 +47,6 @@ Files:
57
  - `test/test-data/embedded-style-map.docx`: Convert a Word document that embeds its own style map to HTML, under four option settings.
58
  Format: Word document (.docx) as input; HTML string as answer
59
  Origin: test fixture of mammoth.js (tool_fixture)
60
- Reference result: expected_values_in_test_asserts
61
- Reference source: test/mammoth.tests.js lines 83-122: default gives <h1>Walking on imported air</h1>; with styleMap "p => p" gives <p>Walking on imported air</p>; with "r => strong" gives <h1><strong>Walking on imported air</strong></h1>; with includeEmbeddedStyleMap false gives <p>Walking on imported air</p>
62
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 83-122 exactly, with the options the test passes; the conversion must raise no messages.
63
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
64
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
@@ -66,8 +54,6 @@ Files:
66
  - `test/test-data/underline.docx`: Convert a Word paragraph with bold and underlined runs to HTML, with and without an underline style mapping.
67
  Format: Word document (.docx) as input; HTML string as answer
68
  Origin: test fixture of mammoth.js (tool_fixture)
69
- Reference result: expected_values_in_test_asserts
70
- Reference source: test/mammoth.tests.js lines 433-445: default gives <p><strong>The Sunset Tree</strong></p> (underline dropped), at lines 436 and 443; with styleMap "u => em" gives <p><strong>The <em>Sunset</em> Tree</strong></p>
71
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 433-445 exactly, with the options the test passes; the two tests do not check messages (they assert result.value only).
72
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
73
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 6 of 6 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
12
 
13
  **Licence (SPDX):** BSD-2-Clause
 
19
  - `test/test-data/tables.docx`: Convert a Word document with a 2x2 table between two paragraphs to semantic HTML.
20
  Format: Word document (.docx) as input; HTML string as answer
21
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
22
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 345-357 exactly, with the options the test passes; the conversion must raise no messages.
23
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
24
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
26
  - `test/test-data/footnotes.docx`: Convert a Word document with two footnotes to HTML with linked footnote references (id prefix doc-42-).
27
  Format: Word document (.docx) as input; HTML string as answer
28
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
29
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 359-374 exactly, with the options the test passes; the conversion must raise no messages.
30
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
31
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
33
  - `test/test-data/endnotes.docx`: Convert a Word document with two endnotes to HTML with linked endnote references (id prefix doc-42-).
34
  Format: Word document (.docx) as input; HTML string as answer
35
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
36
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 376-390 exactly, with the options the test passes; the conversion must raise no messages.
37
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
38
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
40
  - `test/test-data/comments.docx`: Convert a Word document with two review comments to HTML that keeps the comments as a definition list.
41
  Format: Word document (.docx) as input; HTML string as answer
42
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
43
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 406-423 exactly, with the options the test passes; the conversion must raise no messages.
44
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
45
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
47
  - `test/test-data/embedded-style-map.docx`: Convert a Word document that embeds its own style map to HTML, under four option settings.
48
  Format: Word document (.docx) as input; HTML string as answer
49
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
50
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 83-122 exactly, with the options the test passes; the conversion must raise no messages.
51
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
52
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
 
54
  - `test/test-data/underline.docx`: Convert a Word paragraph with bold and underlined runs to HTML, with and without an underline style mapping.
55
  Format: Word document (.docx) as input; HTML string as answer
56
  Origin: test fixture of mammoth.js (tool_fixture)
 
 
57
  Check: Final file state: the agent's HTML file must equal the expected string in test/mammoth.tests.js lines 433-445 exactly, with the options the test passes; the two tests do not check messages (they assert result.value only).
58
  Tools: mammoth (npm; mammoth.convertToHtml or the mammoth CLI with --style-map) or python-mammoth (mammoth.convert_to_html); the output HTML is written to a file
59
  Companions (not included): LICENSE (licence text); package.json (licence line 54 and version 1.13.0); test/mammoth.tests.js (the comparing test; the expected HTML lives in its asserts, hide from the agent)
campaign_files/2026-10-gt-browser-tests/lighthouse-smokehouse/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 7 of 7 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_audit_results. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** or inspect the HTML with a parser and write the findings as JSON
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,7 +19,6 @@ Files:
21
  - `cli/test/smokehouse/test-definitions/a11y.js`: Lighthouse accessibility audit of a deliberately broken test page.
22
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
23
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
24
- Reference result: expected_audit_results
25
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
26
  Tools: or inspect the HTML with a parser and write the findings as JSON
27
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
@@ -29,8 +26,6 @@ Files:
29
  - `cli/test/smokehouse/test-definitions/forms-autocomplete.js`: Lighthouse form autocomplete artifact extraction on a payment and address form.
30
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer); the answer here is the extracted artifact, not an audit score
31
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
32
- Reference result: expected_audit_results
33
- Reference source: `expectations.artifacts.Inputs` in forms-autocomplete.js (forms, inputs such as name_cc1 and CCNo1, labels, autocomplete attribute and prediction values); the commented-out `lhr.audits.autocomplete` block is excluded
34
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
35
  Tools: or inspect the HTML with a parser and write the findings as JSON
36
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules); core/config/experimental-config.js (imported on line 7 of forms-autocomplete.js; the test definition's `config` is built from it, so the judge reads it to run Lighthouse with the right audits; it extends 'lighthouse:default', which the Lighthouse package itself resolves); types/lh.js (imported by core/config/experimental-config.js; an empty ES module that only carries the type import, but Node needs the file to load the config)
@@ -38,8 +33,6 @@ Files:
38
  - `cli/test/smokehouse/test-definitions/seo-failing.js`: Lighthouse SEO audit of a page with many SEO faults.
39
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
40
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
41
- Reference result: expected_audit_results
42
- Reference source: `expectations.lhr.audits` in seo-failing.js (8 audits); the requestedUrl carries extra_header parameters that the static server turns into response headers (x-robots-tag: none and two Link headers for hreflang and canonical)
43
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
44
  Tools: or inspect the HTML with a parser and write the findings as JSON
45
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); the requestedUrl in seo-failing.js, whose query string sets the response headers; shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
@@ -47,8 +40,6 @@ Files:
47
  - `cli/test/smokehouse/test-definitions/seo-passing.js`: Lighthouse SEO audit of a mostly passing page: 8 SEO audits pass, crawlable-anchors fails on 2 links.
48
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
49
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
50
- Reference result: expected_audit_results
51
- Reference source: `expectations.lhr.audits` in seo-passing.js (9 audits: 8 with score 1; crawlable-anchors with score 0 and 2 items)
52
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
53
  Tools: or inspect the HTML with a parser and write the findings as JSON
54
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); cli/test/fixtures/robots.txt (served by static-server.js at /robots.txt; the robots-txt audit reads it; 25 bytes, "User-agent: *" and an empty "Disallow:"); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
@@ -56,8 +47,6 @@ Files:
56
  - `cli/test/smokehouse/test-definitions/perf-diagnostics-unsized-images.js`: Lighthouse unsized-images audit: find images without explicit width and height.
57
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
58
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
59
- Reference result: expected_audit_results
60
- Reference source: `expectations.lhr.audits["unsized-images"]` in perf-diagnostics-unsized-images.js
61
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
62
  Tools: or inspect the HTML with a parser and write the findings as JSON
63
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
@@ -65,8 +54,6 @@ Files:
65
  - `cli/test/smokehouse/test-definitions/perf-diagnostics-animations.js`: Lighthouse non-composited-animations audit: find animations the compositor cannot run.
66
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
67
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
68
- Reference result: expected_audit_results
69
- Reference source: `expectations.lhr.audits["non-composited-animations"]` in perf-diagnostics-animations.js
70
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
71
  Tools: or inspect the HTML with a parser and write the findings as JSON
72
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
@@ -74,8 +61,6 @@ Files:
74
  - `cli/test/smokehouse/test-definitions/clickjacking-missing-headers.js`: Lighthouse clickjacking-mitigation audit of a page served without frame control headers.
75
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
76
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
77
- Reference result: expected_audit_results
78
- Reference source: `expectations.lhr.audits["clickjacking-mitigation"]` in clickjacking-missing-headers.js
79
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
80
  Tools: or inspect the HTML with a parser and write the findings as JSON
81
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 7 of 7 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** or inspect the HTML with a parser and write the findings as JSON
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `cli/test/smokehouse/test-definitions/a11y.js`: Lighthouse accessibility audit of a deliberately broken test page.
20
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
21
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
22
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
23
  Tools: or inspect the HTML with a parser and write the findings as JSON
24
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
26
  - `cli/test/smokehouse/test-definitions/forms-autocomplete.js`: Lighthouse form autocomplete artifact extraction on a payment and address form.
27
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer); the answer here is the extracted artifact, not an audit score
28
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
29
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
30
  Tools: or inspect the HTML with a parser and write the findings as JSON
31
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules); core/config/experimental-config.js (imported on line 7 of forms-autocomplete.js; the test definition's `config` is built from it, so the judge reads it to run Lighthouse with the right audits; it extends 'lighthouse:default', which the Lighthouse package itself resolves); types/lh.js (imported by core/config/experimental-config.js; an empty ES module that only carries the type import, but Node needs the file to load the config)
 
33
  - `cli/test/smokehouse/test-definitions/seo-failing.js`: Lighthouse SEO audit of a page with many SEO faults.
34
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
35
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
36
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
37
  Tools: or inspect the HTML with a parser and write the findings as JSON
38
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); the requestedUrl in seo-failing.js, whose query string sets the response headers; shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
40
  - `cli/test/smokehouse/test-definitions/seo-passing.js`: Lighthouse SEO audit of a mostly passing page: 8 SEO audits pass, crawlable-anchors fails on 2 links.
41
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
42
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
43
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
44
  Tools: or inspect the HTML with a parser and write the findings as JSON
45
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); cli/test/fixtures/robots.txt (served by static-server.js at /robots.txt; the robots-txt audit reads it; 25 bytes, "User-agent: *" and an empty "Disallow:"); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
47
  - `cli/test/smokehouse/test-definitions/perf-diagnostics-unsized-images.js`: Lighthouse unsized-images audit: find images without explicit width and height.
48
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
49
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
50
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
51
  Tools: or inspect the HTML with a parser and write the findings as JSON
52
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
54
  - `cli/test/smokehouse/test-definitions/perf-diagnostics-animations.js`: Lighthouse non-composited-animations audit: find animations the compositor cannot run.
55
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
56
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
57
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
58
  Tools: or inspect the HTML with a parser and write the findings as JSON
59
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
 
61
  - `cli/test/smokehouse/test-definitions/clickjacking-missing-headers.js`: Lighthouse clickjacking-mitigation audit of a page served without frame control headers.
62
  Format: HTML fixture page (input) plus a JS module whose `expectations.lhr.audits` object is the expected Lighthouse result (answer)
63
  Origin: Lighthouse smoke-test fixture and expectation written by the Lighthouse maintainers (tool_fixture)
 
 
64
  Check: Final file state: the agent writes the Lighthouse report JSON (or an audit findings JSON in the same shape); compare it with the hidden `expectations.lhr` by partial deep match, as the smokehouse runner does (listed keys must match, unlisted keys are ignored, regex and length matchers as written). No screenshot needed
65
  Tools: or inspect the HTML with a parser and write the findings as JSON
66
  Companions (not included): cli/test/fixtures/static-server.js (local server on port 10200; serves the fixtures and honours the extra_header query parameter); shared/root.js (imported by static-server.js; sets LH_ROOT, from which the server takes its fixtures directory and its allowed-path check, and reads package.json when loaded); shared/esm-utils.js (imported by shared/root.js; finds the module directory that LH_ROOT is built from); package.json (repo root; shared/root.js reads it when loaded, and its "type": "module" makes Node load static-server.js, root.js and esm-utils.js as ES modules)
campaign_files/2026-10-hw-eda-tests-2/lowrisc-opentitan-ip-block-pattgen/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: register_map_header_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Task: from pattgen.hjson produce the register package and documentation. Judge: lay out registers from the hjson (automatic interrupt registers, multireg compaction) and compare every register offset in the package and every field bit position and width in registers.md by exact integer equality.
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `hw/ip/pattgen/data/pattgen.hjson`: OpenTitan pattgen.hjson: generate the SystemVerilog register package and register documentation with matching offsets and field bits.
22
  Format: OpenTitan reggen hjson; SystemVerilog package; Markdown register tables
23
  Origin: register description of a production open silicon IP block with its committed generated outputs
24
- Reference result: register_map_header_values
25
- Reference source: hw/ip/pattgen/rtl/pattgen_reg_pkg.sv line 5 `Register Package auto-generated by `reggen``; e.g. pkg line 162 `PATTGEN_SIZE_OFFSET = 6'h 2c` against pattgen.hjson line 223 `name: "SIZE"`
26
  Check: Task: from pattgen.hjson produce the register package and documentation. Judge: lay out registers from the hjson (automatic interrupt registers, multireg compaction) and compare every register offset in the package and every field bit position and width in registers.md by exact integer equality.
27
  Companions (not included): LICENSE; hw/ip/pattgen/rtl/pattgen_reg_pkg.sv; hw/ip/pattgen/doc/registers.md
28
  Dependencies: OpenTitan util/regtool.py (reggen) at the pinned commit: only needed to regenerate; Python 3.9 or later standard library for the comparison judge (hjson read by a small stdlib parser)
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Task: from pattgen.hjson produce the register package and documentation. Judge: lay out registers from the hjson (automatic interrupt registers, multireg compaction) and compare every register offset in the package and every field bit position and width in registers.md by exact integer equality.
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `hw/ip/pattgen/data/pattgen.hjson`: OpenTitan pattgen.hjson: generate the SystemVerilog register package and register documentation with matching offsets and field bits.
20
  Format: OpenTitan reggen hjson; SystemVerilog package; Markdown register tables
21
  Origin: register description of a production open silicon IP block with its committed generated outputs
 
 
22
  Check: Task: from pattgen.hjson produce the register package and documentation. Judge: lay out registers from the hjson (automatic interrupt registers, multireg compaction) and compare every register offset in the package and every field bit position and width in registers.md by exact integer equality.
23
  Companions (not included): LICENSE; hw/ip/pattgen/rtl/pattgen_reg_pkg.sv; hw/ip/pattgen/doc/registers.md
24
  Dependencies: OpenTitan util/regtool.py (reggen) at the pinned commit: only needed to regenerate; Python 3.9 or later standard library for the comparison judge (hjson read by a small stdlib parser)
campaign_files/2026-10-hw-eda-tests-2/manifold-robust-mesh-boolean-library/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 7 of 7 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: cad_geometry_assert. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `test/boolean_test.cpp`: Manifold boolean tests: unions, differences and intersections of cubes, spheres and meshes, checking the resulting volume and surface area.
22
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
23
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
24
- Reference result: cad_geometry_assert
25
- Reference source: upstream expected values written in the test file itself: line 158 volume 0.75, line 159 area 5.5; lines 168-169 volume 1.6 within 0.001 and area 9.2 within 0.01; line 192 volume 0.5 within 1e-6; closed-form.
26
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
27
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
28
  Companions (not included): test/test.h; LICENSE
@@ -30,8 +26,6 @@ Files:
30
  - `test/hull_test.cpp`: Manifold convex hull tests: hulls of point sets, spheres and hollow shapes, checking hull volume and area.
31
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
32
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
33
- Reference result: cad_geometry_assert
34
- Reference source: upstream expected values written in the test file itself: line 84 `EXPECT_FLOAT_EQ(hollow.Hull().Volume(), sphere_vol);` and cube, tetrahedron and sphere hull volumes and areas at lines 84-119; closed-form.
35
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
36
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
37
  Companions (not included): test/test.h; LICENSE
@@ -39,8 +33,6 @@ Files:
39
  - `test/measurement_test.cpp`: Manifold measurement tests: ray casting, volume and area queries on spheres, cubes and transformed solids.
40
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
41
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
42
- Reference result: cad_geometry_assert
43
- Reference source: upstream expected values written in the test file itself: line 66 `EXPECT_NEAR(la::length(hits[0].position), 1.0, 1e-4);` (ray hit on the unit sphere) and the other hit positions and normals; closed-form.
44
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
45
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
46
  Companions (not included): test/test.h; LICENSE
@@ -48,8 +40,6 @@ Files:
48
  - `test/properties_test.cpp`: Manifold property tests: volume, surface area, curvature and tolerance queries on cubes, spheres and scaled solids.
49
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
50
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
51
- Reference result: cad_geometry_assert
52
- Reference source: upstream expected values written in the test file itself: lines 29-30 cube volume 1.0 and area 6.0, lines 33-34 the same after mirroring by scale -1; volumes and areas of tolerance-simplified spheres and cylinders compared with bounds 0.002 to 0.02 (lines 97-128); closed-form.
53
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
54
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
55
  Companions (not included): test/test.h; LICENSE
@@ -57,8 +47,6 @@ Files:
57
  - `test/samples_test.cpp`: Manifold sample model tests: knots, scallops, sponges and other sample solids, checking genus, volume and surface area.
58
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
59
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
60
- Reference result: cad_geometry_assert
61
- Reference source: upstream expected values written in the test file itself: lines 47-49 knot13 genus 1, volume 20786 within 1, area 11177 within 1; lines 84-85 scallop volume 37.3 within 0.1 and area 82.7 within 0.1; recorded kernel results.
62
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
63
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
64
  Companions (not included): test/test.h; LICENSE
@@ -66,8 +54,6 @@ Files:
66
  - `test/sdf_test.cpp`: Manifold signed-distance-field tests: mesh level sets of spheres, cubes and gyroids and check bounds, genus and volume.
67
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
68
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
69
- Reference result: cad_geometry_assert
70
- Reference source: upstream expected values written in the test file itself: lines 77-82 bounds equal +-size/2 within epsilon; line 46 gyroid sphere genus 14235 within 1000.
71
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
72
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
73
  Companions (not included): test/test.h; LICENSE
@@ -75,8 +61,6 @@ Files:
75
  - `test/smooth_test.cpp`: Manifold smoothing tests: subdivide and smooth tetrahedra, cylinders and spheres and check volume and surface area.
76
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
77
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
78
- Reference result: cad_geometry_assert
79
- Reference source: upstream expected values written in the test file itself: line 30 volume 17.0 within 0.1, line 31 area 32.9 within 0.1; line 55 cylinder volume 2*pi within 0.003, line 56 area 6*pi within 0.004.
80
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
81
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
82
  Companions (not included): test/test.h; LICENSE
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 7 of 7 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `test/boolean_test.cpp`: Manifold boolean tests: unions, differences and intersections of cubes, spheres and meshes, checking the resulting volume and surface area.
20
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
21
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
22
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
23
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
24
  Companions (not included): test/test.h; LICENSE
 
26
  - `test/hull_test.cpp`: Manifold convex hull tests: hulls of point sets, spheres and hollow shapes, checking hull volume and area.
27
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
28
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
29
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
30
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
31
  Companions (not included): test/test.h; LICENSE
 
33
  - `test/measurement_test.cpp`: Manifold measurement tests: ray casting, volume and area queries on spheres, cubes and transformed solids.
34
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
35
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
36
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
37
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
38
  Companions (not included): test/test.h; LICENSE
 
40
  - `test/properties_test.cpp`: Manifold property tests: volume, surface area, curvature and tolerance queries on cubes, spheres and scaled solids.
41
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
42
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
43
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
44
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
45
  Companions (not included): test/test.h; LICENSE
 
47
  - `test/samples_test.cpp`: Manifold sample model tests: knots, scallops, sponges and other sample solids, checking genus, volume and surface area.
48
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
49
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
50
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
51
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
52
  Companions (not included): test/test.h; LICENSE
 
54
  - `test/sdf_test.cpp`: Manifold signed-distance-field tests: mesh level sets of spheres, cubes and gyroids and check bounds, genus and volume.
55
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
56
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
57
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
58
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
59
  Companions (not included): test/test.h; LICENSE
 
61
  - `test/smooth_test.cpp`: Manifold smoothing tests: subdivide and smooth tetrahedra, cylinders and spheres and check volume and surface area.
62
  Format: C++ googletest file of the manifold test suite, built with test_main.cpp and test.h from the same tree
63
  Origin: test written by the Manifold authors; expected numbers are closed-form solids (cubes, spheres, cylinders) or recorded kernel results for sample models
 
 
64
  Check: Task: given each test's construction (primitives, booleans, hulls, smoothing, SDF meshing) with the expected numbers hidden, build the solid and report its volume, surface area, genus or bounds, or derive them in closed form. Judge: by the same rule as the upstream assertion: EXPECT_NEAR passes when abs(got - want) <= the stated bound; EXPECT_FLOAT_EQ within 4 ULPs as float; EXPECT_EQ exact.
65
  Tools: cmake builds the manifold_test binary and ./manifold_test --gtest_filter=<Suite>.* runs the suite headless; volumes and areas read from Manifold::Volume() and SurfaceArea() in the C++ API (Python bindings manifold3d give the same values)
66
  Companions (not included): test/test.h; LICENSE
campaign_files/2026-10-m-issue-trackers-2/libvirt-users/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m209-11/thread_MJN2VTYL3KWGAFDI33EJJFQWES5Y6SHL_emails.json`: Virtual Network API for QEMU (9 messages, 2021-03-27 to 2021-04-03).
22
  Format: HyperKitty REST API JSON (?format=json): list of the thread's messages, each with message_id, message_id_hash, subject, sender (address obfuscated with '(a)', mailman_id), sender_name, date (with zone), parent (URL of the parent message, null for the root), children and thread.
23
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one thread (all its messages, across months), saved as the JSON the API returns; the listing is the pinned pointer.
24
- Reference result: facts computed from the thread headers
25
- Reference source: message_id, parent, sender (mailman_id), sender_name and date fields of the thread listing, read from the pinned JSON.
26
  Check: Exact match against values computed from the pinned JSON: count the entries; sort by date converted to UTC; a direct reply is an entry whose parent URL ends with the root's message_id_hash; the longest chain and branch points follow the parent links (cross-checked against the children lists, which agree). People are counted by sender.mailman_id, falling back to the obfuscated sender address when mailman_id is null (messages imported from the old archive); on this thread that count equals the count by sender_name after removing accents and letter case. The first-reply interval is the exact difference in whole seconds between the dates of the root and the earliest reply. Deterministic. Hand-checked on 2 threads of this host (general rule 5): the chosen thread and '[libvirt-users] [virtual interface] detach interface during boot succeed with no changes' (thread PCVVWANNUJCRO6VO57D74B6EF6YHTIJT, 5 messages); counts, last sender and time, first-reply seconds, direct replies and parent-against-children links read by hand agree with the computed facts.
27
  Tools: Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m209-11/thread_MJN2VTYL3KWGAFDI33EJJFQWES5Y6SHL_emails.json`: Virtual Network API for QEMU (9 messages, 2021-03-27 to 2021-04-03).
20
  Format: HyperKitty REST API JSON (?format=json): list of the thread's messages, each with message_id, message_id_hash, subject, sender (address obfuscated with '(a)', mailman_id), sender_name, date (with zone), parent (URL of the parent message, null for the root), children and thread.
21
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one thread (all its messages, across months), saved as the JSON the API returns; the listing is the pinned pointer.
 
 
22
  Check: Exact match against values computed from the pinned JSON: count the entries; sort by date converted to UTC; a direct reply is an entry whose parent URL ends with the root's message_id_hash; the longest chain and branch points follow the parent links (cross-checked against the children lists, which agree). People are counted by sender.mailman_id, falling back to the obfuscated sender address when mailman_id is null (messages imported from the old archive); on this thread that count equals the count by sender_name after removing accents and letter case. The first-reply interval is the exact difference in whole seconds between the dates of the root and the earliest reply. Deterministic. Hand-checked on 2 threads of this host (general rule 5): the chosen thread and '[libvirt-users] [virtual interface] detach interface during boot succeed with no changes' (thread PCVVWANNUJCRO6VO57D74B6EF6YHTIJT, 5 messages); counts, last sender and time, first-reply seconds, direct replies and parent-against-children links read by hand agree with the computed facts.
23
  Tools: Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
campaign_files/2026-10-m-issue-trackers-2/linaro-lava-users/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread listing. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,8 +19,6 @@ Files:
21
  - `m027-16/thread_RSCJLEGSYZYZDNCSMBZBVYRT2Z6XGJUA_emails.json`: [Lava-users] lavacli: adding remote worker requires superuser (8 messages, 2018-08-21 to 2018-09-10).
22
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children.
23
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2018-09-10; checked by reading the listing twice).
24
- Reference result: facts computed from the thread listing
25
- Reference source: sender_name, date and parent fields of the listing.
26
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): RSCJLEGS and DLGVYBM2 (a 4-message thread used only as a check); dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
28
  Companions (not included): Per-message bodies at https://lists.linaro.org/archives/api/list/lava-users@lists.linaro.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
 
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m027-16/thread_RSCJLEGSYZYZDNCSMBZBVYRT2Z6XGJUA_emails.json`: [Lava-users] lavacli: adding remote worker requires superuser (8 messages, 2018-08-21 to 2018-09-10).
20
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children.
21
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2018-09-10; checked by reading the listing twice).
 
 
22
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): RSCJLEGS and DLGVYBM2 (a 4-message thread used only as a check); dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
24
  Companions (not included): Per-message bodies at https://lists.linaro.org/archives/api/list/lava-users@lists.linaro.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
campaign_files/2026-10-m-issue-trackers-2/lineageos-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.lineageos.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/431541/detail.json`: LineageOS/android_packages_apps_FMRadio change 431541: FMRadio: Add base support for media buttons.
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/431541/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: none in this record, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://review.lineageos.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.lineageos.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/431541/detail.json`: LineageOS/android_packages_apps_FMRadio change 431541: FMRadio: Add base support for media buttons.
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: none in this record, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://review.lineageos.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/llvm-llvm-dev/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m209-01/2019-March.txt.gz`: [llvm-dev] GSoC19: Debug Info should have no effect on codegen (6 messages, 2019-03-05 to 2019-03-05).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC (a Date header without zone is read as UTC) and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The list rewrites many From headers for DMARC to 'llvm-dev at lists.llvm.org (Name via llvm-dev)'; for such messages the person is the name in parentheses with ' via llvm-dev' removed, and the address is not used for merging. The chosen thread's five senders are all distinct by that rule. The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[llvm-dev] Meaning of `sub nsw`' (4 messages, root <CABrM6wkQ45cvXuW56Cc0LzUSuSAW1rbhBf7Fip_WA-Wpk7cExw@mail.gmail.com>); message count, last message time, first-reply seconds and the file's distinct message count agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m209-01/2019-March.txt.gz`: [llvm-dev] GSoC19: Debug Info should have no effect on codegen (6 messages, 2019-03-05 to 2019-03-05).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC (a Date header without zone is read as UTC) and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The list rewrites many From headers for DMARC to 'llvm-dev at lists.llvm.org (Name via llvm-dev)'; for such messages the person is the name in parentheses with ' via llvm-dev' removed, and the address is not used for merging. The chosen thread's five senders are all distinct by that rule. The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[llvm-dev] Meaning of `sub nsw`' (4 messages, root <CABrM6wkQ45cvXuW56Cc0LzUSuSAW1rbhBf7Fip_WA-Wpk7cExw@mail.gmail.com>); message count, last message time, first-reply seconds and the file's distinct message count agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/loudoun-county-transit-gtfs/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the service calendar itself (active services on a date, removed or added dates, day counts). Where the reference result comes from and how a solution is checked is written per unit in the notes.
12
-
13
  **Tools to open and check:** Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -22,8 +20,6 @@ Units:
22
  Read out of the archive: https://lfportal.loudoun.gov/LFPortalinternet/0/edoc/1946650/LCTActiveFeed.zip
23
  Format: text/csv (GTFS Schedule calendar.txt; companions calendar_dates.txt, agency.txt, feed_info.txt)
24
  Origin: Live static GTFS feed published by Loudoun County Transit (Loudoun County, Virginia) at https://www.loudoun.gov/loudountransitgtfs. calendar.txt and its companions were read by member out of the zip (spec general rule 4: by member); kept as a snapshot pinned by sha256.
25
- Reference result: facts computed from the service calendar itself (active services on a date, removed or added dates, day counts)
26
- Reference source: calendar.txt weekday flags with start_date and end_date, calendar_dates.txt exception rows, agency_timezone of agency.txt, all from the pinned members.
27
  Check: deterministic, no model judgement. Hand-checking 2 samples per source is pending for the delivery step (spec general rule 5).
28
  Tools: Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
29
  Companions (not included): calendar_dates.txt; agency.txt; feed_info.txt
 
8
 
9
  **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes.
10
 
 
 
11
  **Tools to open and check:** Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
12
 
13
  **Licence (SPDX):** NOASSERTION
 
20
  Read out of the archive: https://lfportal.loudoun.gov/LFPortalinternet/0/edoc/1946650/LCTActiveFeed.zip
21
  Format: text/csv (GTFS Schedule calendar.txt; companions calendar_dates.txt, agency.txt, feed_info.txt)
22
  Origin: Live static GTFS feed published by Loudoun County Transit (Loudoun County, Virginia) at https://www.loudoun.gov/loudountransitgtfs. calendar.txt and its companions were read by member out of the zip (spec general rule 4: by member); kept as a snapshot pinned by sha256.
 
 
23
  Check: deterministic, no model judgement. Hand-checking 2 samples per source is pending for the delivery step (spec general rule 5).
24
  Tools: Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
25
  Companions (not included): calendar_dates.txt; agency.txt; feed_info.txt
campaign_files/2026-10-m-issue-trackers-2/lttng-dev/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m089-08/2019-March.txt.gz`: [lttng-dev] Question related to LTTng (9 messages, 2019-03-27 to 2019-03-29).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing double quotes (one archive writes a quoted local part that replies cite unquoted; found in the hand check). People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). Deterministic. The program was checked by hand on 2 threads of this source before giving answers: the chosen thread (read message by message) and 'Improvements in sessiond-comm' (a patch series: 30 messages by headers, 29 carry their own patch titles and are linked to the cover letter by In-Reply-To); it agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m089-08/2019-March.txt.gz`: [lttng-dev] Question related to LTTng (9 messages, 2019-03-27 to 2019-03-29).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing double quotes (one archive writes a quoted local part that replies cite unquoted; found in the hand check). People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). Deterministic. The program was checked by hand on 2 threads of this source before giving answers: the chosen thread (read message by message) and 'Improvements in sessiond-comm' (a patch series: 30 messages by headers, 29 carry their own patch titles and are linked to the cover letter by In-Reply-To); it agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/lustre-lustre-discuss/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m141-06/2019-March.txt.gz`: [lustre-discuss] Lustre 2.12.0 and locking problems (11 messages, 2019-03-05 to 2019-03-06).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[lustre-discuss] ZFS tuning for MDT/MGS' (7 messages, root <DM6PR09MB30528E5411A135A210E7350CA74A0@DM6PR09MB3052.namprd09.prod.outlook.com>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m141-06/2019-March.txt.gz`: [lustre-discuss] Lustre 2.12.0 and locking problems (11 messages, 2019-03-05 to 2019-03-06).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[lustre-discuss] ZFS tuning for MDT/MGS' (7 messages, root <DM6PR09MB30528E5411A135A210E7350CA74A0@DM6PR09MB3052.namprd09.prod.outlook.com>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/lustre-whamcloud-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.whamcloud.com/changes/67869/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/67869/detail.json`: fs/lustre-release change 67869: LU-20507 tests: do not need IPv6 on the fake interface.
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote, messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/67869/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: Gerrit AI review for Lustre, Lustre Gerrit Janitor, Maloo, Misc Code Checks Robot (Gatekeeper helper), jenkins, wc-checkpatch, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler|maloo|janitor|autotest|gerrit). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://review.whamcloud.com/changes/67869/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.whamcloud.com/changes/67869/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/67869/detail.json`: fs/lustre-release change 67869: LU-20507 tests: do not need IPv6 on the fake interface.
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote, messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: Gerrit AI review for Lustre, Lustre Gerrit Janitor, Maloo, Misc Code Checks Robot (Gatekeeper helper), jenkins, wc-checkpatch, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler|maloo|janitor|autotest|gerrit). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://review.whamcloud.com/changes/67869/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/lyx-trac/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the ticket's own change feed (status transitions, owner changes, comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,14 +19,10 @@ Files:
21
  - `trac/ticket/9317.rss`: Lyx-trac Ticket #9317: No document classes, Yosemite (final status closed, fixed).
22
  Format: RSS 2.0 export of one Trac ticket: channel description holds the ticket text; one item per change with dc:creator, pubDate, title (changed field names) and description (HTML change list with trac-field-old and trac-field-new, then the comment text). The ticket's reporter and creation time are not in the feed.
23
  Origin: Trac RSS export of a closed LyX ticket, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the ticket's own change feed (status transitions, owner changes, comments)
25
- Reference source: The pinned response: trac/ticket/9317.rss.
26
  Check: Exact match against values read from the pinned RSS: per item dc:creator and pubDate, and the change list entries for 'status', 'owner' and 'resolution' (trac-field-old, trac-field-new); comment text is the description after the change list. Deterministic, no model judgement. Human comment count excludes bot accounts, recognised by dc:creator against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: none matched. Timeline events = RSS items (each a dated change or comment). Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
28
  - `trac/ticket/10136.rss`: Lyx-trac Ticket #10136: LyX crashes with input files (final status closed, fixed).
29
  Format: RSS 2.0 export of one Trac ticket: channel description holds the ticket text; one item per change with dc:creator, pubDate, title (changed field names) and description (HTML change list with trac-field-old and trac-field-new, then the comment text). The ticket's reporter and creation time are not in the feed.
30
  Origin: Trac RSS export of a closed LyX ticket, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
31
- Reference result: facts computed from the ticket's own change feed (status transitions, owner changes, comments)
32
- Reference source: The pinned response: trac/ticket/10136.rss.
33
  Check: Exact match against values read from the pinned RSS: per item dc:creator and pubDate, and the change list entries for 'status', 'owner' and 'resolution' (trac-field-old, trac-field-new); comment text is the description after the change list. Deterministic, no model judgement. Human comment count excludes bot accounts, recognised by dc:creator against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: none matched. Timeline events = RSS items (each a dated change or comment). Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
34
  Tools: curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
 
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `trac/ticket/9317.rss`: Lyx-trac Ticket #9317: No document classes, Yosemite (final status closed, fixed).
20
  Format: RSS 2.0 export of one Trac ticket: channel description holds the ticket text; one item per change with dc:creator, pubDate, title (changed field names) and description (HTML change list with trac-field-old and trac-field-new, then the comment text). The ticket's reporter and creation time are not in the feed.
21
  Origin: Trac RSS export of a closed LyX ticket, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned RSS: per item dc:creator and pubDate, and the change list entries for 'status', 'owner' and 'resolution' (trac-field-old, trac-field-new); comment text is the description after the change list. Deterministic, no model judgement. Human comment count excludes bot accounts, recognised by dc:creator against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: none matched. Timeline events = RSS items (each a dated change or comment). Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
24
  - `trac/ticket/10136.rss`: Lyx-trac Ticket #10136: LyX crashes with input files (final status closed, fixed).
25
  Format: RSS 2.0 export of one Trac ticket: channel description holds the ticket text; one item per change with dc:creator, pubDate, title (changed field names) and description (HTML change list with trac-field-old and trac-field-new, then the comment text). The ticket's reporter and creation time are not in the feed.
26
  Origin: Trac RSS export of a closed LyX ticket, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
27
  Check: Exact match against values read from the pinned RSS: per item dc:creator and pubDate, and the change list entries for 'status', 'owner' and 'resolution' (trac-field-old, trac-field-new); comment text is the description after the change list. Deterministic, no model judgement. Human comment count excludes bot accounts, recognised by dc:creator against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: none matched. Timeline events = RSS items (each a dated change or comment). Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
28
  Tools: curl the Trac ticket RSS export (/trac/ticket/N?format=rss) on www.lyx.org and parse it with Python xml.etree or xmllint; field changes sit in each item's description as Trac's change list (trac-field-old, trac-field-new).
campaign_files/2026-10-m-issue-trackers-2/macports-trac/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the ticket's own timeline (status transitions, assignee changes, comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,21 +19,15 @@ Files:
21
  - `ticket/60000.rss`: MacPorts Trac #60000: git @2.25.0 version conflict for package Tcl (closed, fixed).
22
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
23
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the ticket's own timeline (status transitions, assignee changes, comments)
25
- Reference source: The pinned response(s): ticket/60000.rss.
26
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
28
  - `ticket/63000.rss`: MacPorts Trac #63000: py27-gdal fails to build: 'encoding' is an invalid keyword argument (closed, fixed).
29
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
30
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
31
- Reference result: facts computed from the ticket's own timeline (status transitions, assignee changes, comments)
32
- Reference source: The pinned response(s): ticket/63000.rss.
33
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
34
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
35
  - `ticket/63500.rss`: MacPorts Trac #63500: hdf5 1.12.1_0 build failure due to implicit declaration of clock_gettime (closed, fixed).
36
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
37
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
38
- Reference result: facts computed from the ticket's own timeline (status transitions, assignee changes, comments)
39
- Reference source: The pinned response(s): ticket/63500.rss.
40
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
41
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
 
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `ticket/60000.rss`: MacPorts Trac #60000: git @2.25.0 version conflict for package Tcl (closed, fixed).
20
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
21
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
24
  - `ticket/63000.rss`: MacPorts Trac #63000: py27-gdal fails to build: 'encoding' is an invalid keyword argument (closed, fixed).
25
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
26
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
 
 
27
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
28
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
29
  - `ticket/63500.rss`: MacPorts Trac #63500: hdf5 1.12.1_0 build failure due to implicit declaration of clock_gettime (closed, fixed).
30
  Format: Trac ticket RSS 2.0 export: one <item> per ticket change (dc:creator, pubDate, title naming the changed fields, HTML description with the field changes and the comment text).
31
  Origin: Live tracker interface of a closed ticket, collected as a NOASSERTION pointer row under spec general rule 4: copies fetched apart were byte-identical. Unit files kept only outside the repository for checking and then deleted.
 
 
32
  Check: Exact match against values read from the pinned RSS: each <item> is one timeline entry (dc:creator, pubDate, title naming changed fields, description with the field list 'status changed from X to Y' / 'owner set to Z' and the comment text after it). Deterministic, no model judgement. Human comments = items with comment text after the field list, excluding attachment-only entries, empty or placeholder text and commit-hook notices (comment text starting 'In <changeset>' or items without dc:creator); no bot accounts were seen on this tracker. Timeline events = RSS items. Before answers are issued, the extraction program must pass on 2 hand-checked tickets of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
33
  Tools: curl the Trac ticket RSS export (/ticket/N?format=rss) and parse it with xmllint, xmlstarlet or Python xml.etree; Trac also offers ?format=csv and ?format=tab for the current field values.
campaign_files/2026-10-m-issue-trackers-2/mailman3-mailman-users/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread listing. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,16 +19,12 @@ Files:
21
  - `m027-17/thread_TTP4FHL42FLZKGLSZ5YVKL6YSXEUUW5X_emails.json`: [MM3-users] How to identify the exact reason behind the shunting (7 messages, 2021-06-17 to 2021-06-20).
22
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children. Returned as a paginated object (count, next, results).
23
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2021-06-20; checked by reading the listing twice).
24
- Reference result: facts computed from the thread listing
25
- Reference source: sender_name, date and parent fields of the listing.
26
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): TTP4FHL4 and RX6425LU; dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
28
  Companions (not included): Per-message bodies at https://lists.mailman3.org/archives/api/list/mailman-users@mailman3.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
29
  - `m027-18/thread_RX6425LU4KKHP52DQFL4AOKSLJVPIJIA_emails.json`: [MM3-users] New Mailman 3 installation issues (12 messages, 2021-05-18 to 2021-05-21).
30
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children. Returned as a paginated object (count, next, results).
31
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2021-05-21; checked by reading the listing twice).
32
- Reference result: facts computed from the thread listing
33
- Reference source: sender_name, date and parent fields of the listing.
34
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): TTP4FHL4 and RX6425LU; dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
35
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
36
  Companions (not included): Per-message bodies at https://lists.mailman3.org/archives/api/list/mailman-users@mailman3.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
 
8
 
9
  **Self-contained:** Each file is self-contained. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m027-17/thread_TTP4FHL42FLZKGLSZ5YVKL6YSXEUUW5X_emails.json`: [MM3-users] How to identify the exact reason behind the shunting (7 messages, 2021-06-17 to 2021-06-20).
20
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children. Returned as a paginated object (count, next, results).
21
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2021-06-20; checked by reading the listing twice).
 
 
22
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): TTP4FHL4 and RX6425LU; dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
24
  Companions (not included): Per-message bodies at https://lists.mailman3.org/archives/api/list/mailman-users@mailman3.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
25
  - `m027-18/thread_RX6425LU4KKHP52DQFL4AOKSLJVPIJIA_emails.json`: [MM3-users] New Mailman 3 installation issues (12 messages, 2021-05-18 to 2021-05-21).
26
  Format: HyperKitty REST API JSON: list of the thread's messages, each with message_id_hash, subject, sender_name, date, parent (URL of the parent message) and children. Returned as a paginated object (count, next, results).
27
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one closed thread (no message after 2021-05-21; checked by reading the listing twice).
 
 
28
  Check: Exact match against values computed from the pinned JSON (count entries; group sender_name by person; sort by date in UTC; follow parent links). Deterministic. Hand-checked on 2 threads of this host (general rule 5): TTP4FHL4 and RX6425LU; dates, Re: subjects and parent links read by hand agree with the computed facts. Body-level questions need a judge and are not offered on this row.
29
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers. For HyperKitty, curl the archive REST API (?format=json) and walk the parent links.
30
  Companions (not included): Per-message bodies at https://lists.mailman3.org/archives/api/list/mailman-users@mailman3.org/email/<message_id_hash>/?format=json (field 'content'); not fetched twice and not pinned, so only header-level questions are offered.
campaign_files/2026-10-md-media-and-documents/latex3-pdfresources-testfiles-bookmark/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected normalised PDF (.tpf), one per engine. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** l3build check -c config-bookmark bookmark-001 (headless LaTeX run; the produced PDF is normalised and compared with the .tpf)
14
 
15
  **Licence (SPDX):** LPPL-1.3c
@@ -21,8 +19,6 @@ Files:
21
  - `testfiles-bookmark/bookmark-001.pvt`: LaTeX pdfmanagement bookmark test 001: PDF outline entries at levels -10 to 10, grouped, italic and coloured, checked on the produced PDF.
22
  Format: LaTeX PDF test file (.pvt) plus expected normalised PDFs (.tpf)
23
  Origin: regression test of the LaTeX Project team; the .tpf files are the accepted normalised PDFs per engine
24
- Reference result: expected normalised PDF (.tpf), one per engine
25
- Reference source: testfiles-bookmark/bookmark-001.tpf, .luatex.tpf and .xetex.tpf, saved by the maintainers from checked runs
26
  Check: run the .pvt with l3build under config-bookmark on each engine; l3build normalises the produced PDF and the check passes when it equals the matching .tpf; hide the .tpf files when posing the task
27
  Tools: l3build check -c config-bookmark bookmark-001 (headless LaTeX run; the produced PDF is normalised and compared with the .tpf)
28
  Companions (not included): build.lua; config-bookmark.lua; LICENSE
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** l3build check -c config-bookmark bookmark-001 (headless LaTeX run; the produced PDF is normalised and compared with the .tpf)
12
 
13
  **Licence (SPDX):** LPPL-1.3c
 
19
  - `testfiles-bookmark/bookmark-001.pvt`: LaTeX pdfmanagement bookmark test 001: PDF outline entries at levels -10 to 10, grouped, italic and coloured, checked on the produced PDF.
20
  Format: LaTeX PDF test file (.pvt) plus expected normalised PDFs (.tpf)
21
  Origin: regression test of the LaTeX Project team; the .tpf files are the accepted normalised PDFs per engine
 
 
22
  Check: run the .pvt with l3build under config-bookmark on each engine; l3build normalises the produced PDF and the check passes when it equals the matching .tpf; hide the .tpf files when posing the task
23
  Tools: l3build check -c config-bookmark bookmark-001 (headless LaTeX run; the produced PDF is normalised and compared with the .tpf)
24
  Companions (not included): build.lua; config-bookmark.lua; LICENSE
campaign_files/2026-10-md-media-and-documents/lepture-mistune-fixtures/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected reStructuredText per example. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Hide the expected parts; the agent converts each Markdown input to reStructuredText; pass per example when the output equals the expected text exactly (tests/__init__.py line 34), with the arrow character read as a tab on both sides (tests/fixtures/__init__.py lines 53 to 54).
14
 
15
  **Licence (SPDX):** BSD-3-Clause
@@ -21,7 +19,5 @@ Files:
21
  - `tests/fixtures/renderer_rst.txt`: Mistune RST renderer fixtures: 22 Markdown snippets converted to reStructuredText.
22
  Format: fixture file of 22 examples (Markdown, a line with a dot, expected reStructuredText, inside 32-backtick example fences)
23
  Origin: Mistune RST renderer fixtures: 22 Markdown inputs with the reStructuredText the renderer must produce
24
- Reference result: expected reStructuredText per example
25
- Reference source: the part of each example after the line with a single dot
26
  Check: hide the expected parts; the agent converts each Markdown input to reStructuredText; pass per example when the output equals the expected text exactly (tests/__init__.py line 34), with the arrow character read as a tab on both sides (tests/fixtures/__init__.py lines 53 to 54)
27
  Dependencies: mistune
 
8
 
9
  **Self-contained:** The file is self-contained. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Hide the expected parts; the agent converts each Markdown input to reStructuredText; pass per example when the output equals the expected text exactly (tests/__init__.py line 34), with the arrow character read as a tab on both sides (tests/fixtures/__init__.py lines 53 to 54).
12
 
13
  **Licence (SPDX):** BSD-3-Clause
 
19
  - `tests/fixtures/renderer_rst.txt`: Mistune RST renderer fixtures: 22 Markdown snippets converted to reStructuredText.
20
  Format: fixture file of 22 examples (Markdown, a line with a dot, expected reStructuredText, inside 32-backtick example fences)
21
  Origin: Mistune RST renderer fixtures: 22 Markdown inputs with the reStructuredText the renderer must produce
 
 
22
  Check: hide the expected parts; the agent converts each Markdown input to reStructuredText; pass per example when the output equals the expected text exactly (tests/__init__.py line 34), with the arrow character read as a tab on both sides (tests/fixtures/__init__.py lines 53 to 54)
23
  Dependencies: mistune
campaign_files/2026-10-md-media-and-documents/libaom-test-vectors/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: MD5 of every decoded frame (raw planes, no row padding). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Tools are named in the check written per file in the notes.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,32 +19,24 @@ Files:
21
  - `md221-10/av1-1-b8-01-size-16x16.ivf`: Libaom AV1 test vector av1-1-b8-01-size-16x16.ivf with its per-frame MD5 list.
22
  Format: IVF (AV1, 16x16, Main profile, 4:2:0 8-bit, 318 bytes) plus a text list of 2 frame MD5s
23
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 152; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
24
- Reference result: MD5 of every decoded frame (raw planes, no row padding)
25
- Reference source: av1-1-b8-01-size-16x16.ivf.md5: one line per frame, "<md5> <name>-<w>x<h>-<frame>.i420", 2 frames; checked this round: FFmpeg 7.1 (libaom-av1 decoder) decoding to yuv420p gives the same 2 MD5 values in order (film grain applied, as libaom applies it by default); the SHA-1 of both files equals test/test-data.sha1 lines 208 to 209
26
  Check: hide av1-1-b8-01-size-16x16.ivf.md5; the agent decodes av1-1-b8-01-size-16x16.ivf (16x16, Main profile, 4:2:0 8-bit) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
27
  Companions (not included): av1-1-b8-01-size-16x16.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
28
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
29
  - `md221-11/av1-1-b8-04-cdfupdate.ivf`: Libaom AV1 test vector av1-1-b8-04-cdfupdate.ivf with its per-frame MD5 list.
30
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 8-bit, CDF update modes, 36088 bytes) plus a text list of 2 frame MD5s
31
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 255; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
32
- Reference result: MD5 of every decoded frame (raw planes, no row padding)
33
- Reference source: av1-1-b8-04-cdfupdate.ivf.md5: one line per frame, "<md5> <name>-<w>x<h>-<frame>.i420", 2 frames; checked this round: FFmpeg 7.1 (libaom-av1 decoder) decoding to yuv420p gives the same 2 MD5 values in order (film grain applied, as libaom applies it by default); the SHA-1 of both files equals test/test-data.sha1 lines 551 to 552
34
  Check: hide av1-1-b8-04-cdfupdate.ivf.md5; the agent decodes av1-1-b8-04-cdfupdate.ivf (352x288, Main profile, 4:2:0 8-bit, CDF update modes) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
35
  Companions (not included): av1-1-b8-04-cdfupdate.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
36
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
37
  - `md221-12/av1-1-b8-23-film_grain-50.ivf`: Libaom AV1 test vector av1-1-b8-23-film_grain-50.ivf with its per-frame MD5 list.
38
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 8-bit, film grain synthesis on, 28634 bytes) plus a text list of 10 frame MD5s
39
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 262; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
40
- Reference result: MD5 of every decoded frame (raw planes, no row padding)
41
- Reference source: av1-1-b8-23-film_grain-50.ivf.md5: one line per frame, "<md5> <name>-<w>x<h>-<frame>.i420", 10 frames; checked this round: FFmpeg 7.1 (libaom-av1 decoder) decoding to yuv420p gives the same 10 MD5 values in order (film grain applied, as libaom applies it by default); the SHA-1 of both files equals test/test-data.sha1 lines 558 to 559
42
  Check: hide av1-1-b8-23-film_grain-50.ivf.md5; the agent decodes av1-1-b8-23-film_grain-50.ivf (352x288, Main profile, 4:2:0 8-bit, film grain synthesis on) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
43
  Companions (not included): av1-1-b8-23-film_grain-50.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
44
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
45
  - `md221-14/av1-1-b10-23-film_grain-50.ivf`: Libaom AV1 test vector av1-1-b10-23-film_grain-50.ivf with its per-frame MD5 list.
46
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 10-bit, film grain synthesis on, 28906 bytes) plus a text list of 10 frame MD5s
47
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 149; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
48
- Reference result: MD5 of every decoded frame (raw planes, no row padding)
49
- Reference source: av1-1-b10-23-film_grain-50.ivf.md5: one line per frame, "<md5> <name>-<w>x<h>-<frame>.i420", 10 frames; checked this round: FFmpeg 7.1 (libaom-av1 decoder) decoding to yuv420p10le gives the same 10 MD5 values in order (film grain applied, as libaom applies it by default); the SHA-1 of both files equals test/test-data.sha1 lines 560 to 561
50
  Check: hide av1-1-b10-23-film_grain-50.ivf.md5; the agent decodes av1-1-b10-23-film_grain-50.ivf (352x288, Main profile, 4:2:0 10-bit, film grain synthesis on) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
51
  Companions (not included): av1-1-b10-23-film_grain-50.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
52
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Tools are named in the check written per file in the notes.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `md221-10/av1-1-b8-01-size-16x16.ivf`: Libaom AV1 test vector av1-1-b8-01-size-16x16.ivf with its per-frame MD5 list.
20
  Format: IVF (AV1, 16x16, Main profile, 4:2:0 8-bit, 318 bytes) plus a text list of 2 frame MD5s
21
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 152; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
 
 
22
  Check: hide av1-1-b8-01-size-16x16.ivf.md5; the agent decodes av1-1-b8-01-size-16x16.ivf (16x16, Main profile, 4:2:0 8-bit) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
23
  Companions (not included): av1-1-b8-01-size-16x16.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
24
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
25
  - `md221-11/av1-1-b8-04-cdfupdate.ivf`: Libaom AV1 test vector av1-1-b8-04-cdfupdate.ivf with its per-frame MD5 list.
26
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 8-bit, CDF update modes, 36088 bytes) plus a text list of 2 frame MD5s
27
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 255; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
 
 
28
  Check: hide av1-1-b8-04-cdfupdate.ivf.md5; the agent decodes av1-1-b8-04-cdfupdate.ivf (352x288, Main profile, 4:2:0 8-bit, CDF update modes) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
29
  Companions (not included): av1-1-b8-04-cdfupdate.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
30
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
31
  - `md221-12/av1-1-b8-23-film_grain-50.ivf`: Libaom AV1 test vector av1-1-b8-23-film_grain-50.ivf with its per-frame MD5 list.
32
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 8-bit, film grain synthesis on, 28634 bytes) plus a text list of 10 frame MD5s
33
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 262; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
 
 
34
  Check: hide av1-1-b8-23-film_grain-50.ivf.md5; the agent decodes av1-1-b8-23-film_grain-50.ivf (352x288, Main profile, 4:2:0 8-bit, film grain synthesis on) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
35
  Companions (not included): av1-1-b8-23-film_grain-50.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
36
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
37
  - `md221-14/av1-1-b10-23-film_grain-50.ivf`: Libaom AV1 test vector av1-1-b10-23-film_grain-50.ivf with its per-frame MD5 list.
38
  Format: IVF (AV1, 352x288, Main profile, 4:2:0 10-bit, film grain synthesis on, 28906 bytes) plus a text list of 10 frame MD5s
39
  Origin: AV1 decoder conformance test vector of libaom, stored on the aom-test-data bucket and listed in libaom test/test_vectors.cc line 149; the .md5 file holds the MD5 of every decoded frame; the picture content is a camera clip (or synthetic noise for the 16 by 16 vector) whose original source the bucket does not name
 
 
40
  Check: hide av1-1-b10-23-film_grain-50.ivf.md5; the agent decodes av1-1-b10-23-film_grain-50.ivf (352x288, Main profile, 4:2:0 10-bit, film grain synthesis on) with any open-source conforming AV1 decoder and reports the MD5 of each output frame computed over the Y, U and V planes in that order, each plane row by row at the displayed width and height without padding (2 bytes per sample, little-endian, for 10-bit); pass when the list equals the .md5 file frame for frame, as TestVectorTest MD5Match checks it
41
  Companions (not included): av1-1-b10-23-film_grain-50.ivf.md5; test/test_vector_test.cc; test/md5_helper.h; test/test_vectors.cc; test/test-data.sha1; LICENSE; PATENTS
42
  Dependencies: FFmpeg 7.1 with the libaom-av1 decoder; libaom (C; commit a024e0fb06e99f2c44e705a29fbe1074fa96774d, test_vector_test) only to rerun the upstream test.
campaign_files/2026-10-md-media-and-documents/libass-libass-tests-regression/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Reference rendering (PNG) compared with libass compare at pass level 2. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** regression/run-single.sh <compare> regression/token-parsing (libass compare tool, headless render and image compare)
14
 
15
  **Licence (SPDX):** NOASSERTION AND Bitstream-Vera
@@ -21,8 +19,6 @@ Files:
21
  - `regression/token-parsing/trimming.ass`: Libass regression test token-parsing/trimming: an ASS subtitle script whose style name carries leading asterisks and trailing spaces, rendered to a 144x180 frame at 0 s.
22
  Format: ASS subtitle script (.ass) plus reference PNG
23
  Origin: regression test of the libass maintainers; trimming-0000.png is the accepted rendering of the script at time 0
24
- Reference result: reference rendering (PNG) compared with libass compare at pass level 2
25
- Reference source: regression/token-parsing/trimming-0000.png, the rendering the maintainers accepted (regenerated only with ART_REG_GEN=1)
26
  Check: render trimming.ass at 0 s with libass compare using only the fonts in the test directory, compute the maximum error against trimming-0000.png and pass when it classifies at level 2 or better (max error below 4); hide the PNG when posing the task
27
  Tools: regression/run-single.sh <compare> regression/token-parsing (libass compare tool, headless render and image compare)
28
  Companions (not included): regression/token-parsing/trimming-0000.png; regression/.fonts/DejaVuSans.ttf; README.md; regression/README.md; regression/run-single.sh
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** regression/run-single.sh <compare> regression/token-parsing (libass compare tool, headless render and image compare)
12
 
13
  **Licence (SPDX):** NOASSERTION AND Bitstream-Vera
 
19
  - `regression/token-parsing/trimming.ass`: Libass regression test token-parsing/trimming: an ASS subtitle script whose style name carries leading asterisks and trailing spaces, rendered to a 144x180 frame at 0 s.
20
  Format: ASS subtitle script (.ass) plus reference PNG
21
  Origin: regression test of the libass maintainers; trimming-0000.png is the accepted rendering of the script at time 0
 
 
22
  Check: render trimming.ass at 0 s with libass compare using only the fonts in the test directory, compute the maximum error against trimming-0000.png and pass when it classifies at level 2 or better (max error below 4); hide the PNG when posing the task
23
  Tools: regression/run-single.sh <compare> regression/token-parsing (libass compare tool, headless render and image compare)
24
  Companions (not included): regression/token-parsing/trimming-0000.png; regression/.fonts/DejaVuSans.ttf; README.md; regression/README.md; regression/run-single.sh
campaign_files/2026-10-md-media-and-documents/libgd-tests/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Reference grayscale conversion (640 by 480 PNG); reference decode: the 128 by 128 24-bit TGA saved as PNG. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Hide tests/gdimagegrayscale/basic_exp.png; given tests/gdimagegrayscale/basic.png (640 by 480 RGB colour bar chart), the agent converts the colour image to grayscale (luma) and writes a PNG; pass if no pixel differs from the reference by a YIQ distance above the threshold below. The round checked that the reference is grey and within 1 of round(0.299 R + 0.587 G + 0.114 B) of the input; hide tests/tga/conformance/ctc24.png; the agent decodes tests/tga/conformance/ctc24.tga (24-bit uncompressed TrueVision conformance image of colour stripes) to a 128 by 128 PNG; pass if every pixel equals the reference. The round decoded the TGA with ffmpeg 7.1 and matched the reference exactly.
14
 
15
  **Licence (SPDX):** GD AND NOASSERTION
@@ -21,16 +19,12 @@ Files:
21
  - `tests/gdimagegrayscale/basic.png`: Libgd gdimagegrayscale test: basic.png with the expected grayscale result.
22
  Format: PNG input (640 by 480 RGB, 2009 bytes) plus the expected PNG (2092 bytes)
23
  Origin: libgd unit test of gdImageGrayScale; tests/gdimagegrayscale/basic.c line 17: "gdImageGrayScale(im);" then compares with tests/gdimagegrayscale/basic_exp.png (line 20)
24
- Reference result: reference grayscale conversion (640 by 480 PNG)
25
- Reference source: tests/gdimagegrayscale/basic_exp.png: the expected image the upstream test tests/gdimagegrayscale/basic.c compares against (line 20)
26
  Check: hide tests/gdimagegrayscale/basic_exp.png; given tests/gdimagegrayscale/basic.png (640 by 480 RGB colour bar chart), the agent converts the colour image to grayscale (luma) and writes a PNG; pass if no pixel differs from the reference by a YIQ distance above the threshold below. The round checked that the reference is grey and within 1 of round(0.299 R + 0.587 G + 0.114 B) of the input
27
  Companions (not included): tests/gdimagegrayscale/basic_exp.png; tests/gdimagegrayscale/basic.c; tests/gdtest/gdtest.c; tests/gdtest/gdtest.h; src/gd_perceptual_diff.c; COPYING
28
  Dependencies: any image library or a short script (libgd, Pillow, ImageMagick).
29
  - `tests/tga/conformance/ctc24.tga`: Libgd TGA conformance test ctc24.tga with its PNG reference.
30
  Format: TGA (24-bit uncompressed, 128 by 128, 21047 bytes) plus the reference PNG (418 bytes)
31
  Origin: TrueVision TGA 2.0 conformance image kept in libgd tests with a PNG reference made by libgd; tests/tga/conformance/LICENSE line 1: "TrueVision TGA (a.k.a. Targa) 2.0 conformance suite"; tests/tga/tga_conformance.c line 22: assert_conformance_file("ctc24")
32
- Reference result: reference decode: the 128 by 128 24-bit TGA saved as PNG
33
- Reference source: tests/tga/conformance/ctc24.png: the expected image tga_conformance.c compares the decoded TGA against (line 56)
34
  Check: hide tests/tga/conformance/ctc24.png; the agent decodes tests/tga/conformance/ctc24.tga (24-bit uncompressed TrueVision conformance image of colour stripes) to a 128 by 128 PNG; pass if every pixel equals the reference. The round decoded the TGA with ffmpeg 7.1 and matched the reference exactly
35
  Companions (not included): tests/tga/conformance/ctc24.png; tests/tga/tga_conformance.c; tests/gdtest/gdtest.c; tests/gdtest/gdtest.h; tests/tga/conformance/LICENSE; COPYING
36
  Dependencies: any image library or a short script (libgd, Pillow, ImageMagick).
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Hide tests/gdimagegrayscale/basic_exp.png; given tests/gdimagegrayscale/basic.png (640 by 480 RGB colour bar chart), the agent converts the colour image to grayscale (luma) and writes a PNG; pass if no pixel differs from the reference by a YIQ distance above the threshold below. The round checked that the reference is grey and within 1 of round(0.299 R + 0.587 G + 0.114 B) of the input; hide tests/tga/conformance/ctc24.png; the agent decodes tests/tga/conformance/ctc24.tga (24-bit uncompressed TrueVision conformance image of colour stripes) to a 128 by 128 PNG; pass if every pixel equals the reference. The round decoded the TGA with ffmpeg 7.1 and matched the reference exactly.
12
 
13
  **Licence (SPDX):** GD AND NOASSERTION
 
19
  - `tests/gdimagegrayscale/basic.png`: Libgd gdimagegrayscale test: basic.png with the expected grayscale result.
20
  Format: PNG input (640 by 480 RGB, 2009 bytes) plus the expected PNG (2092 bytes)
21
  Origin: libgd unit test of gdImageGrayScale; tests/gdimagegrayscale/basic.c line 17: "gdImageGrayScale(im);" then compares with tests/gdimagegrayscale/basic_exp.png (line 20)
 
 
22
  Check: hide tests/gdimagegrayscale/basic_exp.png; given tests/gdimagegrayscale/basic.png (640 by 480 RGB colour bar chart), the agent converts the colour image to grayscale (luma) and writes a PNG; pass if no pixel differs from the reference by a YIQ distance above the threshold below. The round checked that the reference is grey and within 1 of round(0.299 R + 0.587 G + 0.114 B) of the input
23
  Companions (not included): tests/gdimagegrayscale/basic_exp.png; tests/gdimagegrayscale/basic.c; tests/gdtest/gdtest.c; tests/gdtest/gdtest.h; src/gd_perceptual_diff.c; COPYING
24
  Dependencies: any image library or a short script (libgd, Pillow, ImageMagick).
25
  - `tests/tga/conformance/ctc24.tga`: Libgd TGA conformance test ctc24.tga with its PNG reference.
26
  Format: TGA (24-bit uncompressed, 128 by 128, 21047 bytes) plus the reference PNG (418 bytes)
27
  Origin: TrueVision TGA 2.0 conformance image kept in libgd tests with a PNG reference made by libgd; tests/tga/conformance/LICENSE line 1: "TrueVision TGA (a.k.a. Targa) 2.0 conformance suite"; tests/tga/tga_conformance.c line 22: assert_conformance_file("ctc24")
 
 
28
  Check: hide tests/tga/conformance/ctc24.png; the agent decodes tests/tga/conformance/ctc24.tga (24-bit uncompressed TrueVision conformance image of colour stripes) to a 128 by 128 PNG; pass if every pixel equals the reference. The round decoded the TGA with ffmpeg 7.1 and matched the reference exactly
29
  Companions (not included): tests/tga/conformance/ctc24.png; tests/tga/tga_conformance.c; tests/gdtest/gdtest.c; tests/gdtest/gdtest.h; tests/tga/conformance/LICENSE; COPYING
30
  Dependencies: any image library or a short script (libgd, Pillow, ImageMagick).
campaign_files/2026-10-md-media-and-documents/libjpeg-turbo-testimages/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: MD5 of the expected transformed JPEG file. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: The agent performs a lossless crop of 120x90 at offset +20+50 with transpose and perfect mode on testorig.jpg (task text names the operation, not the tool); pass when the MD5 of the output file equals b4197f377e621c4e9b1d20471432610d, as md5cmp does in the build test.
14
 
15
  **Licence (SPDX):** IJG AND BSD-3-Clause
@@ -21,8 +19,6 @@ Files:
21
  - `testimages/testorig.jpg`: Libjpeg-turbo jpegtran regression: lossless crop and transpose of testorig.jpg.
22
  Format: baseline JPEG photo (5770 bytes) plus the build file holding the expected MD5
23
  Origin: libjpeg-turbo regression test: lossless crop and transpose of the reference photo testorig.jpg with jpegtran; the build pins the MD5 of the expected output file
24
- Reference result: MD5 of the expected transformed JPEG file
25
- Reference source: CMakeLists.txt line 1552: set(MD5_JPEG_CROP b4197f377e621c4e9b1d20471432610d), used by the jpegtran crop test at lines 1869 to 1870
26
  Check: the agent performs a lossless crop of 120x90 at offset +20+50 with transpose and perfect mode on testorig.jpg (task text names the operation, not the tool); pass when the MD5 of the output file equals b4197f377e621c4e9b1d20471432610d, as md5cmp does in the build test
27
  Companions (not included): CMakeLists.txt
28
  Dependencies: libjpeg-turbo jpegtran, since the exact output bytes depend on the encoder version
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: The agent performs a lossless crop of 120x90 at offset +20+50 with transpose and perfect mode on testorig.jpg (task text names the operation, not the tool); pass when the MD5 of the output file equals b4197f377e621c4e9b1d20471432610d, as md5cmp does in the build test.
12
 
13
  **Licence (SPDX):** IJG AND BSD-3-Clause
 
19
  - `testimages/testorig.jpg`: Libjpeg-turbo jpegtran regression: lossless crop and transpose of testorig.jpg.
20
  Format: baseline JPEG photo (5770 bytes) plus the build file holding the expected MD5
21
  Origin: libjpeg-turbo regression test: lossless crop and transpose of the reference photo testorig.jpg with jpegtran; the build pins the MD5 of the expected output file
 
 
22
  Check: the agent performs a lossless crop of 120x90 at offset +20+50 with transpose and perfect mode on testorig.jpg (task text names the operation, not the tool); pass when the MD5 of the output file equals b4197f377e621c4e9b1d20471432610d, as md5cmp does in the build test
23
  Companions (not included): CMakeLists.txt
24
  Dependencies: libjpeg-turbo jpegtran, since the exact output bytes depend on the encoder version
campaign_files/2026-10-md-media-and-documents/libjxl-conformance/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Reference decoded image (float32 array) with error limits. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json).
14
 
15
  **Licence (SPDX):** BSD-3-Clause
@@ -21,16 +19,12 @@ Files:
21
  - `testcases/grayscale/input.jxl`: JPEG XL conformance case grayscale: decode and match the reference image.
22
  Format: JPEG XL codestream (1069 bytes) plus reference PNG, float32 .npy (1x200x200x1), ICC profiles and test.json
23
  Origin: JPEG XL conformance test case "grayscale": a codestream and the reference decoded image with per-frame RMS and peak error limits, from the libjxl conformance repository
24
- Reference result: reference decoded image (float32 array) with error limits
25
- Reference source: testcases/grayscale/reference_image.npy (sha256 listed in test.json) and the rms_error and peak_error limits in test.json
26
  Check: hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json)
27
  Companions (not included): testcases/grayscale/ref.png; testcases/grayscale/test.json; testcases/grayscale/reference_image.npy; testcases/grayscale/reference.icc; testcases/grayscale/original.icc; scripts/conformance.py; testcases/README.md; scripts/download_and_symlink.sh
28
  Dependencies: a JPEG XL decoder such as djxl; the judge needs numpy to read the .npy (plain array file, no pickle)
29
  - `testcases/lz77_flower/input.jxl`: JPEG XL conformance case lz77_flower: decode and match the reference image.
30
  Format: JPEG XL codestream (103540 bytes) plus reference PNG, float32 .npy (1x244x834x3), ICC profiles and test.json
31
  Origin: JPEG XL conformance test case "lz77_flower": a codestream and the reference decoded image with per-frame RMS and peak error limits, from the libjxl conformance repository
32
- Reference result: reference decoded image (float32 array) with error limits
33
- Reference source: testcases/lz77_flower/reference_image.npy (sha256 listed in test.json) and the rms_error and peak_error limits in test.json
34
  Check: hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json)
35
  Companions (not included): testcases/lz77_flower/ref.png; testcases/lz77_flower/test.json; testcases/lz77_flower/reference_image.npy; testcases/lz77_flower/reference.icc; testcases/lz77_flower/original.icc; scripts/conformance.py; testcases/README.md
36
  Dependencies: a JPEG XL decoder such as djxl; the judge needs numpy to read the .npy (plain array file, no pickle)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json).
12
 
13
  **Licence (SPDX):** BSD-3-Clause
 
19
  - `testcases/grayscale/input.jxl`: JPEG XL conformance case grayscale: decode and match the reference image.
20
  Format: JPEG XL codestream (1069 bytes) plus reference PNG, float32 .npy (1x200x200x1), ICC profiles and test.json
21
  Origin: JPEG XL conformance test case "grayscale": a codestream and the reference decoded image with per-frame RMS and peak error limits, from the libjxl conformance repository
 
 
22
  Check: hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json)
23
  Companions (not included): testcases/grayscale/ref.png; testcases/grayscale/test.json; testcases/grayscale/reference_image.npy; testcases/grayscale/reference.icc; testcases/grayscale/original.icc; scripts/conformance.py; testcases/README.md; scripts/download_and_symlink.sh
24
  Dependencies: a JPEG XL decoder such as djxl; the judge needs numpy to read the .npy (plain array file, no pickle)
25
  - `testcases/lz77_flower/input.jxl`: JPEG XL conformance case lz77_flower: decode and match the reference image.
26
  Format: JPEG XL codestream (103540 bytes) plus reference PNG, float32 .npy (1x244x834x3), ICC profiles and test.json
27
  Origin: JPEG XL conformance test case "lz77_flower": a codestream and the reference decoded image with per-frame RMS and peak error limits, from the libjxl conformance repository
 
 
28
  Check: hide ref.png and reference_image.npy; the agent decodes input.jxl to pixels; compare with the reference array as CompareNPY in scripts/conformance.py does (same shape, RMS error and peak absolute error at or below the limits in test.json)
29
  Companions (not included): testcases/lz77_flower/ref.png; testcases/lz77_flower/test.json; testcases/lz77_flower/reference_image.npy; testcases/lz77_flower/reference.icc; testcases/lz77_flower/original.icc; scripts/conformance.py; testcases/README.md
30
  Dependencies: a JPEG XL decoder such as djxl; the judge needs numpy to read the .npy (plain array file, no pickle)
campaign_files/2026-10-md-media-and-documents/librivox-ozymandias-of-egypt/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Orthographic transcript of a poem reading. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Give the agent the mp3; it transcribes the poem; strip wiki markup ({{sc|...}}, {{ppoem}}, section tags) from the reference and the spoken preamble and closing from the hypothesis; score WER on the 14 poem lines.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `md053-12/ozymandias_of_egypt_shelley_db_64kb.mp3`: LibriVox reading of Shelley, "Ozymandias of Egypt", with the proofread Wikisource text of the poem.
22
  Format: MP3 64 kbit/s, one short poem reading with the LibriVox spoken preamble
23
  Origin: volunteer audiobook reading of a public-domain poem; text from a proofread scan transcription
24
- Reference result: orthographic transcript of a poem reading
25
- Reference source: wikisource_Page_360.txt lines 26 to 33 and wikisource_Page_361.txt lines 2 to 7 (14 lines of the sonnet, page quality level 4, validated); the LibriVox preamble and closing ("This is a LibriVox recording...") are not in the text and are excluded
26
  Check: give the agent the mp3; it transcribes the poem; strip wiki markup ({{sc|...}}, {{ppoem}}, section tags) from the reference and the spoken preamble and closing from the hypothesis; score WER on the 14 poem lines
27
  Companions (not included): wikisource_Page_360.txt; wikisource_Page_361.txt
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Give the agent the mp3; it transcribes the poem; strip wiki markup ({{sc|...}}, {{ppoem}}, section tags) from the reference and the spoken preamble and closing from the hypothesis; score WER on the 14 poem lines.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `md053-12/ozymandias_of_egypt_shelley_db_64kb.mp3`: LibriVox reading of Shelley, "Ozymandias of Egypt", with the proofread Wikisource text of the poem.
20
  Format: MP3 64 kbit/s, one short poem reading with the LibriVox spoken preamble
21
  Origin: volunteer audiobook reading of a public-domain poem; text from a proofread scan transcription
 
 
22
  Check: give the agent the mp3; it transcribes the poem; strip wiki markup ({{sc|...}}, {{ppoem}}, section tags) from the reference and the spoken preamble and closing from the hypothesis; score WER on the 14 poem lines
23
  Companions (not included): wikisource_Page_360.txt; wikisource_Page_361.txt