Withdraw wikisource: one CC BY-SA label over mixed per-page licences (#60)

- Withdraw wikisource: one CC BY-SA label over mixed per-page licences (b81c1327875578d9a774491d1ce5e6e1c34830f9)

CHANGELOG.md

@@ -60,6 +60,17 @@ Removed from main:
   right. As with HPLT, the registry keeps its `release: "0.2.0"` entry (with a
   `legal_note`) and `make_docs.py` skips it; `plwikiquote` was dropped from
   `src/fetch_wikimedia.py`. The v0.2.0-v0.2.5 tags still contain it.
+- `wikisource` (632,005 documents / 801,947,427 tokens in v0.2.0-v0.2.5) was
+  removed on 2026-10-05. Every row carries one CC-BY-SA-3.0 label, but the
+  October rights audit found in-copyright post-war imprints among 12 checked
+  rows, each released by its rights holder under its own per-page licence (CC
+  BY 3.0 PL, CC BY-SA 3.0 PL, CC BY-SA 3.0, GFDL, CC BY 4.0) and needing
+  attribution the rows lack. The data came through SpeakLeash with an
+  undocumented selection and no title, author or page id, so it cannot be
+  screened. The registry keeps its `release: "0.2.0"` entry with a
+  `legal_note`, and its `traceable` text no longer claims every text is public
+  domain. A rebuild from the official dump with per-page licences is planned
+  for a later release. The v0.2.0-v0.2.5 tags still contain it.
 
 Changed on main:
 
@@ -74,9 +85,9 @@ Changed on main:
   in v0.2.5 → 1,818 / 207,181 on main.
 - `nkjp1m` is relabelled `CC-BY-4.0` per the NKJP Consortium, with an agreed
   `ATTRIBUTION.md` (Hub PR #51); documents and text unchanged.
-- The 18 v0.2.5 sources still on main total 2,725,031 documents /
-  7,078,287,856 tokens; all 33 sources on main total 4,010,443 documents /
-  8,335,077,871 tokens.
+- The 17 v0.2.5 sources still on main total 2,093,026 documents /
+  6,276,340,429 tokens; all 32 sources on main total 3,378,438 documents /
+  7,533,130,444 tokens.
 
 Tooling and process:
 

README.md

@@ -23,9 +23,9 @@ family (Enevoldsen et al., [arXiv:2508.02271](https://arxiv.org/abs/2508.02271))
 > (tiktoken proxy; canonical Llama-3 count at release) · 20 sources
 > Updated: **2026-08-14**
 
-> **main, not a release (2026-10-05)** · 33 sources · 4,010,443 documents ·
-> **8.34B tokens**: 18 of the 20 v0.2.5 sources (`european_hplt_v3_pl` and
-> `wikiquote` withdrawn)
+> **main, not a release (2026-10-05)** · 32 sources · 3,378,438 documents ·
+> **7.53B tokens**: 17 of the 20 v0.2.5 sources (`european_hplt_v3_pl`,
+> `wikiquote` and `wikisource` withdrawn)
 > plus the 15 sources [planned for v0.2.6](#v026-dev-planned-sources-on-main-not-yet-released). Cite
 > the `v0.2.5` tag, not main.
 
@@ -191,7 +191,7 @@ files is not listed as a data contribution.
 | european_hplt_v3_pl *(withdrawn from main after v0.2.5, licence under review)* | HPLT v3.0 Polish (web, top WDS bins 10-5) | `CC0-1.0` | 1,603,974 | 2,555.5M |
 | [eurlex](data/eurlex/eurlex.md) | EUR-Lex (EU legal acts, Polish) | `CC-BY-4.0` | 243,060 | 2,378.1M |
 | [parliamentary](data/parliamentary/parliamentary.md) | Polish Parliamentary Corpus (Sejm/Senat) | `public-domain (official documents)` | 324,622 | 1,646.8M |
-| [wikisource](data/wikisource/wikisource.md) | Polish Wikisource | `CC-BY-SA-3.0` | 632,005 | 801.9M |
+| wikisource *(withdrawn from main after v0.2.5, mixed per-page licences; rebuild planned)* | Polish Wikisource | `CC-BY-SA-3.0` | 632,005 | 801.9M |
 | [wikipedia](data/wikipedia/wikipedia.md) | Polish Wikipedia | `CC-BY-SA-3.0` | 1,171,897 | 707.2M |
 | [biblioteka_nauki](data/biblioteka_nauki/biblioteka_nauki.md) | Biblioteka Nauki | `per-record upstream license` | 42,071 | 673.5M |
 | [dziennik_ustaw](data/dziennik_ustaw/dziennik_ustaw.md) | Dziennik Ustaw + Monitor Polski (Polish primary legislation) | `public-domain (official documents)` | 35,442 | 486.1M |
@@ -227,8 +227,11 @@ Main has changed since:
   agreed attribution file; its text is unchanged ([PR #51](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/51)).
 - `wikiquote` was removed from main: most of its pages quote works still in
   copyright, and its CC BY-SA label covers only the contributors' selection.
+- `wikisource` was removed from main: its single CC BY-SA 3.0 label also covers
+  in-copyright post-war imprints that carry their own per-page licences. A
+  rebuild from the official dump is planned for a later release.
 
-The 18 v0.2.5 sources still on main total 2,725,031 documents / 7,078.3M tokens.
+The 17 v0.2.5 sources still on main total 2,093,026 documents / 6,276.3M tokens.
 
 ## v0.2.6-dev: planned sources (on main, not yet released)
 The planned content of v0.2.6. Built and documented, but **not** part of v0.2.5 and excluded from every total
@@ -323,7 +326,7 @@ This corpus contains **only** text that its upstream sources already published
 under open licenses or as official public-domain record. It therefore includes
 names and statements of **public figures acting in a public capacity** — e.g.
 parliamentary speakers (from [PPC](https://clip.ipipan.waw.pl/PPC), the Polish Parliamentary Corpus, and the Sejm API), authorities named in legal acts (EUR-Lex), and
-people described in encyclopedic articles (Wikipedia/Wikisource). No private,
+people described in encyclopedic articles (Wikipedia). No private,
 non-public personal data was collected or added. If you are a data subject and
 want content concerning you removed, contact **k.wikiel@gmail.com** — it will be dropped
 from the next version (see retroactive-removal policy below).
@@ -349,7 +352,7 @@ from the next version (see retroactive-removal policy below).
 
 ## License & attribution
 Released under **CC-BY-SA-4.0** (copyleft inherited from CC-BY-SA sources such as
-Wikipedia/Wikisource/Wolne Lektury). Attribution due to each upstream (see datasheets)
+Wikipedia and Wolne Lektury). Attribution due to each upstream (see datasheets)
 and to **SpeakLeash** as the intermediate aggregator. Retroactive-removal policy: a
 source that raises an objection is dropped from subsequent versions, recorded in the
 CHANGELOG.

artifacts/source_findings.md

@@ -1228,3 +1228,35 @@ L11; `reports/citations/wikimedia.md` section 3 (Q1-Q14); sample in
 
 **Lesson** — a project licence covers what the project's contributors wrote. On a
 quotation site that is the framing, not the quoted text.
+
+## wikisource — withdrawn (2026-10-05)
+
+**Verdict** — `removed`, whole source, from main and from v0.2.6; a rebuild for a later
+release is tracked separately (polish-dynaword-v2 #68). Released in v0.2.0 and in every
+tag through v0.2.5, which stay as published. 632,005 documents / 801,947,427 tokens
+dropped. As with HPLT, the registry keeps its `release: "0.2.0"` entry with a
+`legal_note`. Its `traceable` text no longer claims every text is public domain, and
+its `upstream` is now pl.wikisource.org rather than the dumps it never came from.
+`count_vocab.py` and `sample_quality.py` no longer list it. There was no fetch script
+to remove: the data was a SpeakLeash pull.
+
+**Reason** — all rows carry one `CC-BY-SA-3.0` label, and the card said the underlying
+texts are public-domain works. Wikisource's own guidance says "mostly" (W1).
+- Among the 12 audited rows, several are in-copyright post-war imprints that their
+  rights holders released through consent programmes or OTRS, each page under its own
+  licence template: PIW 1971 and 1972, Czytelnik 1962 and 1966, WL 1974, a 1995 sailing
+  manual, 1990-91 regional press (W11-W21). The templates differ (CC BY 3.0 PL,
+  CC BY-SA 3.0 PL, CC BY-SA 3.0, GFDL, CC BY 4.0), so no single relabel fixes them (W17).
+- One row carries a translator's footnote for a translation still protected until 2045;
+  the row-to-page mapping is inferred (W8-W10).
+- The rows keep neither the consent, the licence version nor an author (W22). The
+  SpeakLeash selection is undocumented and rows have no title or page id, so the rows
+  cannot be screened afterwards. The audit's crude filter (372 rows, 1.7%) was not
+  reproduced; the in-copyright share is unknown (W23).
+
+References: `rights_audit_2026-10/reports/citations.md` L10;
+`reports/citations/wikimedia.md` section 2 (W1-W23); `REVIEW.html` (wikisource block).
+
+**Lesson** — a project-wide label on an aggregator pull hides per-page licences. Pull
+from the official dump with page ids and licence templates, so each page can be
+screened and labelled.

data/wikisource/wikisource.md

@@ -1,43 +0,0 @@
-# wikisource
-
-Polish Wikisource
-
-## Dataset description
-- **Source (upstream):** https://dumps.wikimedia.org/plwikisource/latest/
-- **Domain:** books/source-texts
-- **Language:** Polish (pl)
-- **License:** `CC-BY-SA-3.0`
-- **Created (range):** 2005-01-01, 2026-06-14
-- **Added:** 2026-07-02
-
-## Licensing — traceable basis
-Wikimedia dumps under CC-BY-SA 3.0; underlying texts are public-domain works transcribed by the community.
-
-## Provenance
-Pulled from SpeakLeash's public redistribution (`speakleash-ds-pub`, key `plwikisource`) of the upstream source above. SpeakLeash credited as intermediate aggregator; upstream license/attribution preserved.
-
-## Statistics
-| documents | characters | tokens (tiktoken proxy) |
-|---:|---:|---:|
-| 632,005 | 1,974,249,746 | 801,947,427 |
-
-
-## Per-document license metadata
-| license | documents |
-|---|---:|
-| `CC-BY-SA-3.0` | 632,005 |
-
-Author metadata present for **0** documents. Empty values mean the upstream record did not expose a machine-readable author field.
-
-Statistics were recomputed directly from the released parquet file.
-
-## Filters applied (build_dynaword.py)
-Minimal, per Dynaword guidelines (heavy filtering left to downstream use):
-- drop documents < 200 chars: **0**
-- drop non-Polish (diacritic ratio): **0**
-- exact cross-source dedup (sha1): **0**
-- OCR alpha-ratio < 0.70 (OCR sources only): **0**
-- read 632,005 → kept 632,005
-
-Token counts are a fast tiktoken (cl100k) proxy (~1% off Llama-3); the canonical
-Llama-3 count is computed at release.

data/wikisource/wikisource.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1d5a01e35a59a7388bd979c6744b5092fbf446cd0052faec9610d07e04228615
-size 983059726

data/wikisource/wikisource.stats.json

@@ -1,16 +0,0 @@
-{
-  "read": 632005,
-  "kept": 632005,
-  "drop_short": 0,
-  "drop_lang": 0,
-  "drop_dup": 0,
-  "drop_ocr": 0,
-  "chars": 1974249746,
-  "tokens": 801947427,
-  "licenses": {
-    "CC-BY-SA-3.0": 632005
-  },
-  "authors_with_value": 0,
-  "license": "CC-BY-SA-3.0",
-  "stats_recomputed_from_parquet": true
-}

src/count_vocab.py

@@ -6,7 +6,7 @@ ponytail: lowercased surface forms (Polish is heavily inflected, so this is
 import re
 import pyarrow.parquet as pq
 
-SOURCES = ["eurlex", "parliamentary", "wikisource", "wikipedia",
+SOURCES = ["eurlex", "parliamentary", "wikipedia",
            "dziennik_ustaw", "wolne_lektury"]
 WORD = re.compile(r"[^\W\d_]+", re.UNICODE)  # unicode letters, no digits/punct
 

src/make_docs.py

@@ -26,7 +26,7 @@ This corpus contains **only** text that its upstream sources already published
 under open licenses or as official public-domain record. It therefore includes
 names and statements of **public figures acting in a public capacity** — e.g.
 parliamentary speakers (from [PPC](https://clip.ipipan.waw.pl/PPC), the Polish Parliamentary Corpus, and the Sejm API), authorities named in legal acts (EUR-Lex), and
-people described in encyclopedic articles (Wikipedia/Wikisource). No private,
+people described in encyclopedic articles (Wikipedia). No private,
 non-public personal data was collected or added. If you are a data subject and
 want content concerning you removed, contact **{CONTACT}** — it will be dropped
 from the next version (see retroactive-removal policy below).
@@ -465,7 +465,7 @@ Sources we reviewed and **deliberately left out** — part of the curation:
 {DISCLAIMER}
 ## License & attribution
 Released under **CC-BY-SA-4.0** (copyleft inherited from CC-BY-SA sources such as
-Wikipedia/Wikisource/Wolne Lektury). Attribution due to each upstream (see datasheets)
+Wikipedia and Wolne Lektury). Attribution due to each upstream (see datasheets)
 and to **SpeakLeash** as the intermediate aggregator. Retroactive-removal policy: a
 source that raises an objection is dropped from subsequent versions, recorded in the
 CHANGELOG.

src/sample_quality.py

@@ -6,7 +6,7 @@ metrics look ambiguous and you need semantic quality, not just surface garble.
 import sys, re, random
 import pyarrow.parquet as pq
 
-SOURCES = ["eurlex", "parliamentary", "wikisource", "wikipedia",
+SOURCES = ["eurlex", "parliamentary", "wikipedia",
            "dziennik_ustaw", "wolne_lektury"]
 N = int(sys.argv[1]) if len(sys.argv) > 1 else 200  # docs sampled per source
 SHOW = 2  # raw samples printed per source

src/sources.py

@@ -132,12 +132,24 @@ SOURCES = {
         "pretty": "Polish Wikisource",
         "license": "CC-BY-SA-3.0",
         "license_spdx": "CC-BY-SA-3.0",
-        "traceable": "Wikimedia dumps under CC-BY-SA 3.0; underlying texts are "
-                     "public-domain works transcribed by the community.",
-        "upstream": "https://dumps.wikimedia.org/plwikisource/latest/",
+        "traceable": "Pulled through SpeakLeash (plwikisource), not from the Wikimedia "
+                     "dumps; the selection is undocumented and rows have no title, "
+                     "author or page id. Wikisource texts are mostly, not all, public "
+                     "domain: consented post-war imprints carry their own per-page "
+                     "licences (CC BY 3.0 PL, CC BY-SA 3.0 PL, CC BY-SA 3.0, GFDL, "
+                     "CC BY 4.0).",
+        "upstream": "https://pl.wikisource.org/",
         "domain": "books/source-texts",
         "created": "2005-01-01, 2026-06-14",
         "is_ocr": False,
+        "legal_note": "Withdrawn from main on 2026-10-05; not in v0.2.6. Every row carries "
+                      "one CC-BY-SA-3.0 label, but the October rights audit found "
+                      "in-copyright post-war imprints among 12 checked rows, released by "
+                      "their rights holders under per-page licences that differ from the "
+                      "label and need attribution the rows lack. The rows cannot be "
+                      "screened. A rebuild from the official dump with per-page licences "
+                      "is planned for a later release. The v0.2.0-v0.2.5 tags still "
+                      "contain it.",
     },
     "wiktionary_examples": {
         "added": "2026-07-19",