wikipedia: drop the two copied pages and copyright-reason {{ek}} pages

COPIED_PAGES in src/fetch_wikipedia.py lists the pages the copyright sample found
copied, keyed on the wiki page id with a reason each: 833396 "Obnazyc prawde"
(Filmweb description) and 1126780 "Czarnka" (dpn.pl lake text). prepare drops
them and fails if a listed page is not reached.

{{ek}} notices (and their redirect aliases from the dump) are read with their
reason; a copyright reason (npa, plagiat, copypaste, kopiuj-wklej, copyvio,
"praw... autorsk...") drops the page. plwiki 20260901 has 0 live {{ek}} notices
on articles, so none are dropped; build.json records ek_tagged.

Rebuilt from the same dump: 1,468,656 -> 1,468,654 documents,
1,226,913,508 -> 1,226,912,191 tokens. Only the two listed rows differ.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

artifacts/source_findings.md

@@ -1537,3 +1537,24 @@ before release.
 source came from the article's own citation and history ("Czarnka": a 2025 forum hit, the
 park's 2012 text behind it). (3) The Terms of Use give the day of the licence switch, not
 the time, so a last-revision label is inexact for one day of edits.
+
+## wikipedia — update: copied pages and `{{ek}}` dropped
+
+Drops added to `src/fetch_wikipedia.py` (`prepare`), keyed on the wiki page id; the dump
+(plwiki 20260901) is unchanged. `COPIED_PAGES`: page 833396 "Obnażyć prawdę" (Filmweb's
+official description; excluded) and page 1126780 "Czarnka" (Drawieński Park Narodowy lake
+text; held for review, Art. 4 not assessed). The build fails if a listed page is not reached.
+
+`{{ek}}` (speedy deletion; aliases EK, Ekspresowe kasowanko, Usuń, Speedy, Delete from the
+dump's redirects) is read with its reason; a reason naming a copyright violation (npa,
+plagiat, copypaste, kopiuj-wklej, copyvio, "praw… autorsk…") drops the page. The dump has 0
+live `{{ek}}` notices on articles, so 0 dropped and no reasons to count; a looser text scan
+of all 7 shards found 3, all inside `<!-- -->` comments (redirect, content moved, no
+reason), none about copyright. The hand-check of 10 was not possible.
+
+Rebuilt from the same dump: 1,468,656 → 1,468,654 documents, 1,226,913,508 → 1,226,912,191
+tokens (3.0: 190,836 docs / 81,022,938 tokens; 4.0: 1,277,818 / 1,145,889,253). Diff against
+the previous parquet: only `wikipedia_833396` and `wikipedia_1126780` removed, no row added
+or changed, same in the attribution sidecar. Row order differs between builds
+(`build_dynaword.py` uses `imap_unordered`); content per id is identical. The copyright
+sample keeps its population (the previous build) and records the two dropped ids.

data/wikipedia/wikipedia.attribution.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:09470d9e87b26cb1b2a677c50004c0b6b0b9c0877eec116b227fcd47dca3ada1
-size 344192463
+oid sha256:046067a7992e0f98950ba6d7e0e8b7499e75e58bb5aac0a64e0779e6d69e0a2e
+size 344192023

data/wikipedia/wikipedia.build.json

@@ -1,6 +1,6 @@
 {
-  "attribution_rows": 1468656,
-  "attribution_sha256": "09470d9e87b26cb1b2a677c50004c0b6b0b9c0877eec116b227fcd47dca3ada1",
+  "attribution_rows": 1468654,
+  "attribution_sha256": "046067a7992e0f98950ba6d7e0e8b7499e75e58bb5aac0a64e0779e6d69e0a2e",
   "by_sa_4_cutoff": "2023-06-07T00:00:00+00:00",
   "copyvio_tagged": [
     {
@@ -147,6 +147,7 @@
   "discovered_at": "2026-10-05T22:27:57.741359+00:00",
   "dump_base": "https://dumps.wikimedia.org/plwiki/20260901",
   "dump_date": "20260901",
+  "ek_tagged": [],
   "extract": {
     "extracted": 1706319,
     "ns0": 2304425,
@@ -199,12 +200,27 @@
     }
   ],
   "prepare": {
+    "copied_pages": [
+      {
+        "pageid": "833396",
+        "reason": "Obnażyć prawdę: the 'Opis fabuły' paragraph closely paraphrases Filmweb's production description of the film; added in revision 15302011 (2008-12-26), which cites Filmweb",
+        "title": "Obnażyć prawdę"
+      },
+      {
+        "pageid": "1126780",
+        "reason": "Czarnka: the lake paragraph (revision 42979631, 2015-06-19) closely paraphrases the Drawieński Park Narodowy page dpn.pl/czarne, archived 2012-01-25; whether Art. 4 of the Polish copyright act (official materials) applies was not assessed",
+        "title": "Czarnka"
+      }
+    ],
     "counts": {
-      "CC-BY-SA-3.0": 284076,
-      "CC-BY-SA-4.0": 1421861,
+      "CC-BY-SA-3.0": 284075,
+      "CC-BY-SA-4.0": 1421860,
+      "copied_page": 2,
+      "ek_copyright": 0,
       "exact_duplicate": 382,
       "extracted": 1706319
     },
+    "ek_copyright": [],
     "exact_duplicate_pageids": [
       "28456",
       "28459",
@@ -595,6 +611,36 @@
     "url": "https://creativecommons.org/licenses/by-sa/4.0/deed.pl"
   },
   "sha1sums_url": "https://dumps.wikimedia.org/plwiki/20260901/plwiki-20260901-sha1sums.txt",
+  "tag_template_redirects": [
+    {
+      "target": "Szablon:Ek",
+      "title": "Szablon:EK"
+    },
+    {
+      "target": "Szablon:Ek",
+      "title": "Szablon:Ekspresowe kasowanko"
+    },
+    {
+      "target": "Szablon:Ek",
+      "title": "Szablon:Usuń"
+    },
+    {
+      "target": "Szablon:Ek",
+      "title": "Szablon:Speedy"
+    },
+    {
+      "target": "Szablon:Ek",
+      "title": "Szablon:Delete"
+    },
+    {
+      "target": "Szablon:NPA",
+      "title": "Szablon:Copyvio"
+    },
+    {
+      "target": "Szablon:NPA",
+      "title": "Szablon:Plagiat"
+    }
+  ],
   "terms_of_use": {
     "effective": "These Terms of Use went into effect on June 7, 2023",
     "sha256": "c96ef93b439ff9b8242dc3233f41ede57c0984ef9153fafa51aa5f8cf0678832",

data/wikipedia/wikipedia.license-evidence.json

@@ -67,15 +67,15 @@
  ],
  "attribution": "author = \"Wikipedia contributors, <page URL>\"; the page history names the contributors; Terms of Use section 7 accepts a link or URL to the page. wikipedia.attribution.jsonl: id, pageid, revid, title, url, last_revision_at, license per row",
  "intermediary_removed": "v0.2.0-v0.2.5 used the SpeakLeash plwiki snapshot (speakleash-ds-pub); the v0.2.6 rebuild reads the official dump directly",
- "finding": "Contributors' text is CC BY-SA; its version follows the page's last revision. Pages carrying the plwiki copyright-violation notice ({{NPA}}, {{Copyvio}}, {{Plagiat}}) are dropped.",
+ "finding": "Contributors' text is CC BY-SA; its version follows the page's last revision. Pages carrying the plwiki copyright-violation notice ({{NPA}}, {{Copyvio}}, {{Plagiat}}) are dropped, as are pages whose speedy-deletion notice {{ek}} gives a copyright reason (none in this dump) and the pages the copyright sample found copied (src/fetch_wikipedia.py COPIED_PAGES).",
  "release_scope": {
   "labels": {
-   "CC-BY-SA-4.0": 1277819,
-   "CC-BY-SA-3.0": 190837
+   "CC-BY-SA-4.0": 1277818,
+   "CC-BY-SA-3.0": 190836
   },
   "tokens_by_label": {
-   "CC-BY-SA-4.0": 1145890205,
-   "CC-BY-SA-3.0": 81023303
+   "CC-BY-SA-4.0": 1145889253,
+   "CC-BY-SA-3.0": 81022938
   },
   "pages_scanned": 2877278,
   "namespace_0": 2304425,
@@ -85,9 +85,12 @@
   "parser_failures": 0,
   "converted": 1706319,
   "exact_duplicates": 382,
+  "copied_pages": 2,
+  "ek_notices": 0,
+  "ek_copyright": 0,
   "under_200_chars": 231717,
   "non_polish": 5564,
-  "documents_released": 1468656,
+  "documents_released": 1468654,
   "last_revised_before_2009": 0
  },
  "copyvio_notice_recheck": {
@@ -113,7 +116,22 @@
    ]
   ]
  },
- "audit_trail": "wikipedia.build.json lists the dropped copyright-notice pages (copyvio_tagged) and the page ids dropped as exact duplicates (prepare.exact_duplicate_pageids)",
+ "ek_notice": {
+  "template": "Szablon:Ek",
+  "redirects_in_dump": [
+   "Delete",
+   "EK",
+   "Ekspresowe kasowanko",
+   "Speedy",
+   "Usuń"
+  ],
+  "rule": "src/fetch_wikipedia.py ek_tag reads the first parameter (or powód=) of a live {{ek}} call; ek_is_copyright (EK_COPYRIGHT_RE) drops it when the reason is npa, plagiat, copypaste, kopiuj-wklej, copyvio or names 'praw autorskich'; tested in src/test_wikipedia_contract.py",
+  "articles_tagged": 0,
+  "by_reason": {},
+  "copyright_dropped": 0,
+  "cross_check": "a looser scan of all 7 shards for {{ek, {{usuń, {{speedy, {{delete and {{ekspresowe in namespace-0 articles found no live notice; 3 pages have one inside an HTML comment (not shown by MediaWiki, not counted): Pożar Londynu (1666) (powód=redir), Tabela wszech czasów w konkursie Miss Supranational (content moved to the main article), Dangdang (no reason)"
+ },
+ "audit_trail": "wikipedia.build.json lists the dropped copyright-notice pages (copyvio_tagged), every {{ek}} notice (ek_tagged), the pages dropped as copied (prepare.copied_pages) or for a copyright {{ek}} reason (prepare.ek_copyright), the page ids dropped as exact duplicates (prepare.exact_duplicate_pageids), and the redirects to the two notice templates (tag_template_redirects)",
  "copyright_sample": {
   "file": "wikipedia.copyvio-sample.jsonl",
   "checked_at": "2026-10-05",
@@ -186,7 +204,22 @@
     "note": "Earwig's hit is a 2025-02-21 forum trip report (gorybezgranic.pl) that quotes the lake paragraph verbatim in quotation marks (645 of 2705 chars). The paragraph entered Wikipedia in revision 42979631 (2015-06-19, MOs810, summary 'rozb., źr., ilustracja'), cited to 'Tablica informacyjna in situ' (an on-site board, not read). It follows, fact by fact and in the same order, the Drawieński Park Narodowy page dpn.pl/czarne, archived 2012-01-25 with the same text: 10 tys. lat / bryła martwego lodu, średnio 7 m, 'beztlenowe bakterie Desulfovibrio desulfuricans', 'purpurowe bakterie siarkowe ... utleniające siarkowodór do wolnej siarki', mixing by wind 'wiosną i jesienią', mezotroficzny. Close paraphrase of an earlier park text; the board may hold the verbatim original. Whether a national park's text is protected (Art. 4 of the Polish copyright act excludes official materials) was not assessed."
    }
   ],
-  "found_copied_decision": "kept in this release; drop or keep is an open decision"
+  "found_copied_decision": "dropped from the release (src/fetch_wikipedia.py COPIED_PAGES, keyed on the page id); maintainer rule: keep a source clean or drop what cannot be cleared",
+  "population": {
+   "drawn_from": "Hub PR 68 commit 3393e05c178bb29cd9d0bda0444ef9fb2573fbf3, data/wikipedia/wikipedia.parquet sha256 fca10e80f9ee0c8db289518a001178e1b9580d744976841144fb08784d47256e",
+   "documents": 1468656,
+   "rows_dropped_since": [
+    [
+     "wikipedia_833396",
+     "CC-BY-SA-3.0"
+    ],
+    [
+     "wikipedia_1126780",
+     "CC-BY-SA-4.0"
+    ]
+   ],
+   "note": "the sample was drawn from this earlier build; test_copyright_sample_matches_the_release reproduces it from the released ids plus rows_dropped_since, each of which must be a copyright drop"
+  }
  },
  "not_checked": [
   "pastes from copyrighted sources beyond the sample above",
@@ -194,7 +227,7 @@
   "text imported from elsewhere under another licence or with extra attribution requirements (Terms of Use, Re-use); template notices are removed by the conversion",
   "text first contributed under GFDL before the 2009 move to CC BY-SA (0 pages last edited before 2009); GFDL dual licensing; only the CC BY-SA grant is relied on",
   "quotations inside articles (Art. 29 of the Polish copyright act)",
-  "pages tagged for speedy deletion ({{ek}}) are kept; how many are tagged for copyright reasons was not counted",
+  "deletion discussions and maintenance notices other than {{NPA}} and {{ek}}; their reasons are not read",
   "personal data (NERGAL scrub runs before release)"
  ]
 }

data/wikipedia/wikipedia.md

@@ -19,8 +19,8 @@ dump:
 
 | last revision (UTC) | `license` | documents | tokens | share of tokens |
 |---|---|---:|---:|---:|
-| before 2023-06-07 | `CC-BY-SA-3.0` | 190,837 | 81,023,303 | 6.6% |
-| 2023-06-07 or later | `CC-BY-SA-4.0` | 1,277,819 | 1,145,890,205 | 93.4% |
+| before 2023-06-07 | `CC-BY-SA-3.0` | 190,836 | 81,022,938 | 6.6% |
+| 2023-06-07 or later | `CC-BY-SA-4.0` | 1,277,818 | 1,145,889,253 | 93.4% |
 
 A page last edited on or after the cutoff can still hold 3.0 text from earlier edits;
 [CC BY-SA 3.0](https://creativecommons.org/licenses/by-sa/3.0/legalcode) §4(b)(ii) lets such an adaptation be licensed under "a later
@@ -39,8 +39,13 @@ fetcher drops the 35 pages that carry plwiki's copyright-violation
 notice `{{NPA}}` (or its redirects `{{Copyvio}}`, `{{Plagiat}}`) in the dump.
 On 2026-10-05 the 31 of them not edited since the dump were re-read on
 pl.wikipedia: all carry the notice (some mark one section only; the whole page is dropped).
-The 4 edited after the dump were not re-read. A paste without the notice cannot
-be detected by the build.
+The 4 edited after the dump were not re-read. Pages tagged for speedy deletion with
+`{{ek}}` (or its redirects `{{Delete}}`, `{{EK}}`, `{{Ekspresowe kasowanko}}`, `{{Speedy}}`, `{{Usuń}}`) are also
+dropped when the stated reason is a copyright violation (`EK_COPYRIGHT_RE`: the
+template's codes `npa`, `plagiat`, `copypaste`, `kopiuj-wklej`, or free text with "copyvio" or
+"praw autorskich"). No article in this dump carries a live `{{ek}}` notice; a looser
+scan found 3 inside HTML comments, which MediaWiki does not show and the build does not
+count. A paste without either notice cannot be detected by the build.
 
 To measure undetected pastes, 40 released pages per label (seed
 20261006, page-id order) were run through Earwig's Copyvio Detector (web search and
@@ -56,8 +61,11 @@ with the page history and the date of the matching text (see `wikipedia.copyvio-
 | text copied or closely paraphrased from a copyrighted source | 1 | 1 |
 
 1 of 40 (`CC-BY-SA-3.0`) and 1 of 40 (`CC-BY-SA-4.0`) pages contain text
-copied or closely paraphrased from a copyrighted source: `wikipedia_833396` "Obnażyć prawdę" (`CC-BY-SA-3.0`); `wikipedia_1126780` "Czarnka" (`CC-BY-SA-4.0`). They stay in
-this release (the sample file notes why each is classed so). The share of such pages is below 12.9% (`CC-BY-SA-3.0`) and
+copied or closely paraphrased from a copyrighted source: `wikipedia_833396` "Obnażyć prawdę" (`CC-BY-SA-3.0`); `wikipedia_1126780` "Czarnka" (`CC-BY-SA-4.0`) (the sample file notes
+why each is classed so). Both are dropped from this release: `COPIED_PAGES` in
+`src/fetch_wikipedia.py` lists them by page id with the reason, and the build stops if one is
+not found. The sample was drawn before that drop, from the 1,468,656 documents of Hub PR 68
+commit 3393e05; the test reproduces the draw from the release plus the dropped rows. The share of such pages is below 12.9% (`CC-BY-SA-3.0`) and
 12.9% (`CC-BY-SA-4.0`), 95% Wilson upper bounds; the labels hold
 6.6% and 93.4% of the tokens. The bounds cover only
 what the detector can find: print sources, translations and paraphrase scoring below the
@@ -83,7 +91,8 @@ Parsed from the official Wikimedia dump `plwiki-20260901-pages-articles` by
 (`speakleash-ds-pub`); that intermediary was removed in this rebuild.
 
 Articles (namespace 0) only. Redirects, empty pages and pages with a copyright-violation
-notice are skipped. Wikitext is converted with mwparserfromhell
+notice are skipped; pages listed in `COPIED_PAGES` or with a copyright `{{ek}}` reason are
+dropped after the duplicate check, so a copy of their text on another page goes with them. Wikitext is converted with mwparserfromhell
 `strip_code(normalize=True, collapse=True)`, as in the sibling Wikimedia builders, after
 removing `<ref>` bodies, file and category links and behaviour switches (`__NOTOC__`).
 Templates (infoboxes, navigation boxes), `<math>`, galleries and comments are dropped;
@@ -93,18 +102,19 @@ pages with the same text (sha1 of the stripped text), the lower page id is kept.
 2,877,278 pages scanned → 2,304,425 in namespace 0 → 598,071 redirects,
 0 empty pages and 35 with a copyright-violation notice skipped
 (0 parser failures) → 1,706,319 converted →
-382 exact duplicates dropped → 1,705,937 read by the build →
-1,468,656 released.
+382 exact duplicates, 2 copied pages (`COPIED_PAGES`) and 0 copyright `{{ek}}`
+pages dropped → 1,705,935 read by the build →
+1,468,654 released.
 
 Not cleaned: appendix headings ("Zobacz też", "Uwagi", "Przypisy", "Bibliografia", "Linki
-zewnętrzne") stay in 828,389 documents (56.405%), with what is left of those sections
+zewnętrzne") stay in 828,388 documents (56.405%), with what is left of those sections
 after the conversion (2.96% of all characters from the first such heading to the end);
 174,668 documents (11.893%) have whitespace-only lines.
 
 ## Statistics
 | documents | characters | tokens (tiktoken proxy) |
 |---:|---:|---:|
-| 1,468,656 | 3,140,841,847 | 1,226,913,508 |
+| 1,468,654 | 3,140,838,409 | 1,226,912,191 |
 
 v0.2.5 (SpeakLeash snapshot) had 1,171,897 documents, 1,844,707,761 characters and
 707,194,207 tokens. The SpeakLeash extraction is not documented. Its text drops the
@@ -115,17 +125,17 @@ in v0.2.5 and 59,133 here.
 ## Per-document license metadata
 | license | documents |
 |---|---:|
-| `CC-BY-SA-3.0` | 190,837 |
-| `CC-BY-SA-4.0` | 1,277,819 |
+| `CC-BY-SA-3.0` | 190,836 |
+| `CC-BY-SA-4.0` | 1,277,818 |
 
-Author metadata present for **1,468,656** documents.
+Author metadata present for **1,468,654** documents.
 
 ## Filters applied (build_dynaword.py)
 - drop documents < 200 chars: **231,717**
 - drop non-Polish (diacritic ratio): **5,564**
 - exact cross-source dedup (sha1): **0** (the fetcher already dropped 382)
 - OCR alpha-ratio < 0.70 (OCR sources only): **0**
-- read 1,705,937 → kept 1,468,656
+- read 1,705,935 → kept 1,468,654
 
 Token counts are a fast tiktoken (cl100k) proxy (~1% off Llama-3); the canonical
 Llama-3 count is computed at release.
@@ -139,6 +149,6 @@ Llama-3 count is computed at release.
 - text first contributed under GFDL before the 2009 move to CC BY-SA, and GFDL dual
   licensing; only the CC BY-SA grant is relied on
 - quotations inside articles (Art. 29 of the Polish copyright act)
-- pages tagged for speedy deletion (`{{ek}}`) are kept; how many are tagged for copyright
-  reasons was not counted
+- deletion discussions and maintenance notices other than `{{NPA}}` and `{{ek}}`; their
+  reasons are not read
 - personal data (the NERGAL scrub runs before release)

data/wikipedia/wikipedia.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fca10e80f9ee0c8db289518a001178e1b9580d744976841144fb08784d47256e
-size 1462996349
+oid sha256:f4abfff711b207691d1278efc28d8af18551454143f3d8f2292ec3ddcefc314a
+size 1462846393

data/wikipedia/wikipedia.stats.json

@@ -1,17 +1,17 @@
 {
-  "read": 1705937,
-  "kept": 1468656,
+  "read": 1705935,
+  "kept": 1468654,
   "drop_short": 231717,
   "drop_lang": 5564,
   "drop_dup": 0,
   "drop_ocr": 0,
-  "chars": 3140841847,
-  "tokens": 1226913508,
+  "chars": 3140838409,
+  "tokens": 1226912191,
   "licenses": {
-    "CC-BY-SA-4.0": 1277819,
-    "CC-BY-SA-3.0": 190837
+    "CC-BY-SA-4.0": 1277818,
+    "CC-BY-SA-3.0": 190836
   },
-  "authors_with_value": 1468656,
-  "secs": 36.0,
+  "authors_with_value": 1468654,
+  "secs": 34.6,
   "license": "CC BY-SA (3.0 or 4.0 per record, by last-revision date)"
 }

src/fetch_wikipedia.py

@@ -8,8 +8,9 @@ checked against the sha1 in dumpstatus.json and in the dump's sha1sums.txt.
     B=<build dir, never committed>
     python3 src/fetch_wikipedia.py discover    --out $B  # shards, sha1s, licence evidence
     python3 src/fetch_wikipedia.py acquire     --out $B  # resumable download + sha1 check
-    python3 src/fetch_wikipedia.py extract     --out $B  # ns 0, no redirects, no copyvio tag
-    python3 src/fetch_wikipedia.py prepare     --out $B  # $B/wikipedia.jsonl.zst
+    python3 src/fetch_wikipedia.py extract     --out $B  # ns 0, no redirects, no copyvio tag; {{ek}} recorded
+    python3 src/fetch_wikipedia.py prepare     --out $B  # $B/wikipedia.jsonl.zst; drops duplicates,
+                                                         # COPIED_PAGES, copyright {{ek}} pages
     python3 src/build_dynaword.py --sources wikipedia --speakleash-dir $B --out .
     python3 src/fetch_wikipedia.py attribution --out $B  # data/wikipedia sidecars
 """
@@ -58,6 +59,29 @@ BY_SA_4_CUTOFF = datetime(2023, 6, 7, tzinfo=timezone.utc)
 # plwiki copyright-violation notice and its redirects (Szablon:NPA <- Copyvio, Plagiat).
 # Namespace prefix is case-insensitive, the first letter of the name too.
 COPYVIO_RE = re.compile(r"\{\{\s*(?:(?i:szablon|template)\s*:\s*)?(?:[Nn]PA|[Cc]opyvio|[Pp]lagiat)\s*(?:\||\}\})")
+# plwiki speedy-deletion template and its redirects (Szablon:Ek <- EK, Ekspresowe kasowanko,
+# Usuń, Speedy, Delete); its first parameter (or powód=) is the stated reason.
+EK_RE = re.compile(r"\{\{\s*(?:(?i:szablon|template)\s*:\s*)?"
+                   r"(?:[Ee][Kk]|[Ee]kspresowe[ _]+kasowanko|[Uu]suń|[Ss]peedy|[Dd]elete)\s*(?:\||\}\})")
+EK_NAMES = {"Ek", "EK", "Ekspresowe kasowanko", "Usuń", "Speedy", "Delete"}
+# The build records every template redirect to these, so the alias lists above can be checked against the dump.
+TAG_TEMPLATES = {"Szablon:NPA", "Szablon:Ek"}
+# Pages the copyright sample (data/wikipedia/wikipedia.copyvio-sample.jsonl) found to hold text
+# copied or closely paraphrased from a copyrighted source. Keyed on the wiki page id, which is
+# stable across dumps (the dataset id is derived from it). prepare fails if one is not reached.
+COPIED_PAGES = {
+    833396: "Obnażyć prawdę: the 'Opis fabuły' paragraph closely paraphrases Filmweb's production "
+            "description of the film; added in revision 15302011 (2008-12-26), which cites Filmweb",
+    1126780: "Czarnka: the lake paragraph (revision 42979631, 2015-06-19) closely paraphrases the "
+             "Drawieński Park Narodowy page dpn.pl/czarne, archived 2012-01-25; whether Art. 4 of the "
+             "Polish copyright act (official materials) applies was not assessed",
+}
+# {{ek}} reasons that state a copyright violation: the template's own codes that render as
+# "naruszenie praw autorskich" (npa, plagiat, copypaste, kopiuj-wklej) and free text naming one
+# (copyvio, "praw autorskich"). "kopia"/"skopiowane" alone are left out: on plwiki they also
+# mark a duplicate of another article, which is not a copyright reason.
+EK_COPYRIGHT_RE = re.compile(
+    r"\bnpa\b|copyvio|plagiat|copy[ -]?paste|kopiuj[ -]?wklej|praw\w* autorsk", re.IGNORECASE)
 # Links that render as an image or a category, not as running text ([[:Kategoria:X]] is a text link).
 NON_PROSE_NS = {"plik", "file", "image", "grafika", "media", "kategoria", "category"}
 SWITCH_RE = re.compile(r"__[A-ZĄĆĘŁŃÓŚŹŻ]+__")  # __NOTOC__, __BEZSPISU__
@@ -91,6 +115,39 @@ def strip_article(wikitext: str) -> str | None:
     return SWITCH_RE.sub("", text).strip()
 
 
+def _template_name(template) -> str:
+    """MediaWiki title of a template call: namespace prefix dropped, _ as space, first letter upper."""
+    name = " ".join(str(template.name).replace("_", " ").split())
+    prefix, sep, rest = name.partition(":")
+    if sep and prefix.strip().lower() in ("szablon", "template"):
+        name = rest.strip()
+    return name[:1].upper() + name[1:]
+
+
+def _param(template, *names) -> str:
+    for name in names:
+        if template.has(name):
+            return str(template.get(name).value).strip()
+    return ""
+
+
+def ek_tag(wikitext: str) -> dict | None:
+    """Stated reason and source of the page's first {{ek}} speedy-deletion notice; None without one
+    (a call inside a comment or <nowiki> is not a notice)."""
+    if not EK_RE.search(wikitext):
+        return None
+    import mwparserfromhell
+    for template in mwparserfromhell.parse(wikitext).filter_templates():
+        if _template_name(template) in EK_NAMES:
+            return {"reason": _param(template, "1", "powód"), "source": _param(template, "2", "źródło")}
+    return None
+
+
+def ek_is_copyright(ek: dict) -> bool:
+    """True when an {{ek}} notice's stated reason is a copyright violation."""
+    return bool(EK_COPYRIGHT_RE.search(ek["reason"]))
+
+
 def license_for_revision(timestamp: str | None) -> str:
     """SPDX label for a page whose last revision is at `timestamp` (ISO 8601 with zone)."""
     if not timestamp:
@@ -150,10 +207,11 @@ def acquire(out: Path) -> None:
 
 
 def pages(fh):
-    """Yield (pageid, revid, ns, title, timestamp, is_redirect, wikitext) per <page>."""
+    """Yield (pageid, revid, ns, title, timestamp, redirect_target, wikitext) per <page>;
+    redirect_target is "" unless the page is a redirect."""
     ids, title = [], None
     ns = timestamp = text = None
-    redirect = False
+    redirect = ""
     for _, el in iterparse(fh, events=("end",)):
         tag = _local(el.tag)
         if tag == "title":
@@ -164,39 +222,51 @@ def pages(fh):
             ids.append(el.text)
         elif tag == "timestamp":
             timestamp = el.text
-        elif tag == "redirect":
-            redirect = True
+        elif tag == "redirect":  # <redirect title="target" />
+            redirect = el.get("title") or "(no target)"
         elif tag == "text":
             text = el.text
         elif tag == "page":
             yield ids[0], ids[1], ns, title, timestamp, redirect, text or ""
-            ids, title, ns, timestamp, text, redirect = [], None, None, None, None, False
+            ids, title, ns, timestamp, text, redirect = [], None, None, None, None, ""
             el.clear()
 
 
 def extract_shard(src: Path, dst: Path) -> dict:
-    counts, copyvio = Counter(), []
+    counts, copyvio, ek_tagged, tag_redirects = Counter(), [], [], []
     tmp = dst.with_suffix(".tmp")
     with bz2.open(src, "rb") as fh, gzip.open(tmp, "wt", encoding="utf-8") as fo:
         for pageid, revid, ns, title, timestamp, redirect, wikitext in pages(fh):
             counts["pages"] += 1
+            if ns == "10" and redirect in TAG_TEMPLATES:
+                tag_redirects.append({"title": title, "target": redirect})
             if ns != "0":
                 continue
             counts["ns0"] += 1
             if redirect:
                 counts["ns0_redirect"] += 1
-            elif not wikitext.strip():
+                continue
+            if not wikitext.strip():
                 counts["ns0_empty"] += 1
-            elif COPYVIO_RE.search(wikitext):
+                continue
+            ek = ek_tag(wikitext)  # recorded on every article; prepare drops the copyright ones
+            if ek is not None:
+                counts["ns0_ek_tag"] += 1
+                ek_tagged.append({"pageid": pageid, "title": title, **ek})
+            if COPYVIO_RE.search(wikitext):
                 counts["ns0_copyvio_tag"] += 1
                 copyvio.append({"pageid": pageid, "title": title})
             elif (text := strip_article(wikitext)) is None:
                 counts["ns0_strip_error"] += 1
             else:
                 counts["extracted"] += 1
-                fo.write(json.dumps({"pageid": pageid, "revid": revid, "title": title, "timestamp": timestamp,
-                                     "wikitext_chars": len(wikitext), "text": text}, ensure_ascii=False) + "\n")
-    result = {"shard": src.name, "counts": dict(counts), "copyvio_tagged": copyvio}
+                row = {"pageid": pageid, "revid": revid, "title": title, "timestamp": timestamp,
+                       "wikitext_chars": len(wikitext), "text": text}
+                if ek is not None:
+                    row["ek"] = ek
+                fo.write(json.dumps(row, ensure_ascii=False) + "\n")
+    result = {"shard": src.name, "counts": dict(counts), "copyvio_tagged": copyvio,
+              "ek_tagged": ek_tagged, "tag_template_redirects": tag_redirects}
     save(dst.with_name(src.name + ".counts.json"), result)
     tmp.rename(dst)
     return result
@@ -217,10 +287,12 @@ def extract(out: Path, workers: int) -> None:
 
 
 def prepare(out: Path) -> None:
-    """Label each page, drop exact duplicates keeping the lowest page id, write the build input."""
+    """Label each page, drop exact duplicates keeping the lowest page id, drop COPIED_PAGES and
+    pages whose {{ek}} notice gives a copyright reason, write the build input. A dropped page's
+    text stays in `seen`, so a later page with the same text is dropped as its duplicate."""
     files = load(out / "selection.json")["files"]
     seen, counts = set(), Counter()
-    duplicates = []
+    duplicates, copied, ek_copyright = [], [], []
     zst = subprocess.Popen(["zstd", "-q", "-f", "-3", "-o", str(out / f"{SOURCE}.jsonl.zst")],
                            stdin=subprocess.PIPE)
     with gzip.open(out / "manifest.jsonl.gz", "wt", encoding="utf-8") as manifest:
@@ -234,6 +306,13 @@ def prepare(out: Path) -> None:
                         duplicates.append(row["pageid"])
                         continue
                     seen.add(key)
+                    if int(row["pageid"]) in COPIED_PAGES:
+                        copied.append({"pageid": row["pageid"], "title": row["title"],
+                                       "reason": COPIED_PAGES[int(row["pageid"])]})
+                        continue
+                    if "ek" in row and ek_is_copyright(row["ek"]):
+                        ek_copyright.append({"pageid": row["pageid"], "title": row["title"], **row["ek"]})
+                        continue
                     label = license_for_revision(row["timestamp"])
                     counts[label] += 1
                     row_id, url = f"{SOURCE}_{row['pageid']}", page_url(row["title"])
@@ -249,8 +328,14 @@ def prepare(out: Path) -> None:
     zst.stdin.close()
     if zst.wait():
         raise RuntimeError("zstd failed")
+    missing = set(COPIED_PAGES) - {int(p["pageid"]) for p in copied}
+    if missing:
+        raise ValueError(f"COPIED_PAGES not reached (gone from the dump, tagged or a duplicate): {sorted(missing)}")
     counts["exact_duplicate"] = len(duplicates)
-    save(out / "prepare.json", {"counts": dict(counts), "exact_duplicate_pageids": duplicates})
+    counts["copied_page"] = len(copied)
+    counts["ek_copyright"] = len(ek_copyright)
+    save(out / "prepare.json", {"counts": dict(counts), "exact_duplicate_pageids": duplicates,
+                                "copied_pages": copied, "ek_copyright": ek_copyright})
     print(dict(counts))
 
 
@@ -276,6 +361,8 @@ def attribution(out: Path, data_dir: Path) -> None:
         "by_sa_4_cutoff": selection["by_sa_4_cutoff"], "discovered_at": selection["observed_at"],
         "extract": dict(sum((Counter(s["counts"]) for s in shards), Counter())),
         "copyvio_tagged": [p for s in shards for p in s["copyvio_tagged"]],
+        "ek_tagged": [p for s in shards for p in s["ek_tagged"]],
+        "tag_template_redirects": [r for s in shards for r in s["tag_template_redirects"]],
         "prepare": load(out / "prepare.json"),
         "attribution_rows": n, "attribution_sha256": digest((data_dir / f"{SOURCE}.attribution.jsonl").read_bytes()),
     })

src/sources.py

@@ -125,9 +125,12 @@ SOURCES = {
                       "v0.2.0-v0.2.5 used the SpeakLeash plwiki snapshot; that intermediary "
                       "was removed in the v0.2.6 rebuild.",
         "legal_note": "Pages carrying the plwiki copyright-violation notice {{NPA}} (or its "
-                      "redirects {{Copyvio}}, {{Plagiat}}) in the dump are dropped. Text "
-                      "pasted into an article without such a notice cannot be detected; a "
-                      "stratified web sample is recorded in wikipedia.license-evidence.json.",
+                      "redirects {{Copyvio}}, {{Plagiat}}) in the dump are dropped, as are "
+                      "pages whose speedy-deletion notice {{ek}} gives a copyright reason and "
+                      "the pages the copyright sample found copied (COPIED_PAGES in "
+                      "src/fetch_wikipedia.py). Other text pasted into an article cannot be "
+                      "detected; the stratified web sample is recorded in "
+                      "wikipedia.license-evidence.json.",
         "license": "CC BY-SA (3.0 or 4.0 per record, by last-revision date)",
         "license_spdx": "MIXED",
         "traceable": "Contributors license their text under CC BY-SA 4.0 since the Wikimedia "

src/test_wikipedia_contract.py

@@ -57,6 +57,36 @@ def test_copyvio_tag(wikitext, tagged):
     assert bool(fw.COPYVIO_RE.search(wikitext)) is tagged
 
 
+@pytest.mark.parametrize("wikitext, ek", [
+    ("{{ek|npa|http://x.pl}}\nTekst", {"reason": "npa", "source": "http://x.pl"}),
+    ("{{EK|powód=reklama}}", {"reason": "reklama", "source": ""}),
+    ("{{ek|1=naruszenie praw autorskich}}", {"reason": "naruszenie praw autorskich", "source": ""}),
+    ("{{ Szablon:Ekspresowe_kasowanko | test }}", {"reason": "test", "source": ""}),
+    ("{{usuń}}", {"reason": "", "source": ""}),      # redirect, first letter case-insensitive
+    ("{{Delete|copyvio}}", {"reason": "copyvio", "source": ""}),
+    ("{{ek}}", {"reason": "", "source": ""}),
+    ("<!-- {{ek|npa}} -->Tekst", None),              # commented out: not a notice
+    ("<nowiki>{{ek|npa}}</nowiki>", None),
+    ("{{Eko}}", None),
+    ("{{EKspresowe kasowanko|x}}", None),           # only the first letter is case-insensitive
+    ("Skrót EK w tekście", None),
+])
+def test_ek_tag(wikitext, ek):
+    assert fw.ek_tag(wikitext) == ek
+
+
+@pytest.mark.parametrize("reason, copyright", [
+    ("npa", True), ("NPA", True), ("plagiat", True), ("copypaste", True), ("kopiuj-wklej", True),
+    ("naruszenie praw autorskich (ze strony http://x.pl)", True), ("Prawa autorskie", True),
+    ("copyvio z http://x.pl", True), ("copy-paste ze strony gminy", True),
+    ("reklama", False), ("nieency", False), ("test", False), ("", False),
+    ("kopia artykułu Ala", False),   # duplicate of another article, not a copyright reason
+    ("Inpakt", False),               # npa only as a whole word
+])
+def test_ek_is_copyright(reason, copyright):
+    assert fw.ek_is_copyright({"reason": reason, "source": ""}) is copyright
+
+
 def test_strip_article_drops_markup_left_by_strip_code():
     wikitext = ("'''Ala''' ma kota<ref>Jan K., ''Koty'', 2001.</ref> i psa<ref name=\"x\" />.\n"
                 "[[Plik:Kot.jpg|thumb|250px|Kot [[Ala|Ali]]]]\n[[File:a.png|mały|x]]\n"
@@ -67,24 +97,32 @@ def test_strip_article_drops_markup_left_by_strip_code():
 
 # --- dump pipeline on a fixture ------------------------------------------------------------
 
-def _page(pageid, title, ns, timestamp, text, redirect=False):
+def _page(pageid, title, ns, timestamp, text, redirect=""):
     return (f"<page><title>{title}</title><ns>{ns}</ns><id>{pageid}</id>"
-            + ('<redirect title="Ala" />' if redirect else "")
+            + (f'<redirect title="{redirect}" />' if redirect else "")
             + f"<revision><id>{pageid * 100}</id><parentid>1</parentid><timestamp>{timestamp}</timestamp>"
             f"<contributor><username>U</username><id>7</id></contributor>"
             f"<text bytes=\"1\">{text}</text><sha1>x</sha1></revision></page>")
 
 
 BODY = "Ala jest polskim imieniem żeńskim, noszonym przez wiele znanych osób w różnych epokach. " * 4
+MOUSE = "Mysz domowa jest gryzoniem z rodziny myszowatych, żyjącym zwykle w pobliżu człowieka. " * 4
+FROG = "Żaba trawna jest płazem bezogonowym z rodziny żabowatych, pospolitym w całej Polsce. " * 4
+MOLE = "Kret europejski jest owadożernym ssakiem, który większość życia spędza pod ziemią. " * 4
 OTHER = "Pies domowy jest udomowionym ssakiem z rodziny psowatych, towarzyszem człowieka od wieków. " * 4
 XML = ('<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/"><siteinfo><sitename>W</sitename></siteinfo>'
        + _page(10, "Ala", 0, "2023-06-07T00:00:00Z", BODY + "[[Kategoria:Imiona]]")
-       + _page(11, "Przekierowanie", 0, "2020-01-01T00:00:00Z", "#PATRZ [[Ala]]", redirect=True)
+       + _page(11, "Przekierowanie", 0, "2020-01-01T00:00:00Z", "#PATRZ [[Ala]]", redirect="Ala")
        + _page(12, "Dyskusja:Ala", 1, "2020-01-01T00:00:00Z", BODY)
        + _page(13, "Kot", 0, "2020-01-01T00:00:00Z", "{{NPA|http://x.pl}}" + OTHER)
        + _page(14, "Ala (imię)", 0, "2010-01-01T00:00:00Z", BODY)
        + _page(15, "Pusty", 0, "2010-01-01T00:00:00Z", "")
        + _page(16, "Pies", 0, "2023-06-06T23:59:59Z", OTHER)
+       + _page(17, "Mysz", 0, "2021-01-01T00:00:00Z", "{{ek|npa|http://y.pl}}" + MOUSE)
+       + _page(18, "Żaba", 0, "2021-01-01T00:00:00Z", "{{ek|reklama}}" + FROG)
+       + _page(19, "Kopia", 0, "2021-01-01T00:00:00Z", MOUSE)  # same text as 17
+       + _page(20, "Kret", 0, "2021-01-01T00:00:00Z", MOLE)
+       + _page(21, "Szablon:Usuń", 10, "2020-01-01T00:00:00Z", "#PATRZ [[Szablon:Ek]]", redirect="Szablon:Ek")
        + "</mediawiki>")
 SHARD = "plwiki-20260901-pages-articles1.xml-p1p20.bz2"
 
@@ -92,11 +130,12 @@ SHARD = "plwiki-20260901-pages-articles1.xml-p1p20.bz2"
 def test_pages_reads_page_and_revision_ids_not_contributor_id():
     rows = list(fw.pages(io.BytesIO(XML.encode())))
     assert [r[:3] for r in rows][:2] == [("10", "1000", "0"), ("11", "1100", "0")]
-    assert rows[1][5] is True and rows[0][5] is False
+    assert rows[1][5] == "Ala" and rows[0][5] == ""
 
 
 @pytest.mark.skipif(shutil.which("zstd") is None, reason="zstd CLI")
-def test_extract_prepare_attribution_end_to_end(tmp_path):
+def test_extract_prepare_attribution_end_to_end(tmp_path, monkeypatch):
+    monkeypatch.setattr(fw, "COPIED_PAGES", {20: "Kret: copied"})
     (tmp_path / "raw_dump").mkdir()
     (tmp_path / "extracted").mkdir()
     (tmp_path / "raw_dump" / SHARD).write_bytes(bz2.compress(XML.encode()))
@@ -105,21 +144,29 @@ def test_extract_prepare_attribution_end_to_end(tmp_path):
         "files": [{"name": SHARD}], "rightsinfo": {}, "terms_of_use": {}, "by_sa_4_cutoff": "c",
     })
     result = fw.extract_shard(tmp_path / "raw_dump" / SHARD, tmp_path / "extracted" / (SHARD + ".jsonl.gz"))
-    assert result["counts"] == {"pages": 7, "ns0": 6, "ns0_redirect": 1, "ns0_empty": 1,
-                                "ns0_copyvio_tag": 1, "extracted": 3}
+    assert result["counts"] == {"pages": 12, "ns0": 10, "ns0_redirect": 1, "ns0_empty": 1,
+                                "ns0_copyvio_tag": 1, "ns0_ek_tag": 2, "extracted": 7}
     assert result["copyvio_tagged"] == [{"pageid": "13", "title": "Kot"}]
+    assert result["ek_tagged"] == [{"pageid": "17", "title": "Mysz", "reason": "npa", "source": "http://y.pl"},
+                                   {"pageid": "18", "title": "Żaba", "reason": "reklama", "source": ""}]
+    assert result["tag_template_redirects"] == [{"title": "Szablon:Usuń", "target": "Szablon:Ek"}]
 
     fw.prepare(tmp_path)
     out = subprocess.run(["zstd", "-dc", str(tmp_path / "wikipedia.jsonl.zst")], capture_output=True, check=True)
     rows = [json.loads(line) for line in out.stdout.decode().splitlines()]
-    assert [r["id"] for r in rows] == ["wikipedia_10", "wikipedia_16"]  # 14 duplicates 10's text
-    assert [r["license"] for r in rows] == ["CC-BY-SA-4.0", "CC-BY-SA-3.0"]
-    assert [r["created"] for r in rows] == ["2023-06-07", "2023-06-06"]
+    # 14 duplicates 10's text; 17 (ek: npa) and 20 (COPIED_PAGES) are dropped, and 19, a copy
+    # of 17's text, goes with it as a duplicate; 18's ek reason is not copyright, so it stays
+    assert [r["id"] for r in rows] == ["wikipedia_10", "wikipedia_16", "wikipedia_18"]
+    assert [r["license"] for r in rows] == ["CC-BY-SA-4.0", "CC-BY-SA-3.0", "CC-BY-SA-3.0"]
+    assert [r["created"] for r in rows] == ["2023-06-07", "2023-06-06", "2021-01-01"]
     assert rows[0]["author"] == "Wikipedia contributors, https://pl.wikipedia.org/wiki/Ala"
-    assert "Kategoria" not in rows[0]["text"]
+    assert "Kategoria" not in rows[0]["text"] and "reklama" not in rows[2]["text"]
     assert fw.load(tmp_path / "prepare.json") == {
-        "counts": {"extracted": 3, "CC-BY-SA-4.0": 1, "CC-BY-SA-3.0": 1, "exact_duplicate": 1},
-        "exact_duplicate_pageids": ["14"]}
+        "counts": {"extracted": 7, "CC-BY-SA-4.0": 1, "CC-BY-SA-3.0": 2, "exact_duplicate": 2,
+                   "copied_page": 1, "ek_copyright": 1},
+        "exact_duplicate_pageids": ["14", "19"],
+        "copied_pages": [{"pageid": "20", "title": "Kret", "reason": "Kret: copied"}],
+        "ek_copyright": [{"pageid": "17", "title": "Mysz", "reason": "npa", "source": "http://y.pl"}]}
 
     data = tmp_path / "data"
     data.mkdir()
@@ -131,6 +178,21 @@ def test_extract_prepare_attribution_end_to_end(tmp_path):
                         "last_revision_at": "2023-06-06T23:59:59Z", "license": "CC-BY-SA-3.0"}]
     build = fw.load(data / "wikipedia.build.json")
     assert build["attribution_rows"] == 1 and build["extract"]["ns0_copyvio_tag"] == 1
+    assert [p["pageid"] for p in build["ek_tagged"]] == ["17", "18"]
+    assert build["tag_template_redirects"] == [{"title": "Szablon:Usuń", "target": "Szablon:Ek"}]
+
+
+@pytest.mark.skipif(shutil.which("zstd") is None, reason="zstd CLI")
+def test_prepare_fails_when_a_listed_page_is_not_reached(tmp_path, monkeypatch):
+    (tmp_path / "extracted").mkdir()
+    fw.save(tmp_path / "selection.json", {"files": [{"name": SHARD}]})
+    import gzip
+    with gzip.open(tmp_path / "extracted" / (SHARD + ".jsonl.gz"), "wt", encoding="utf-8") as fo:
+        fo.write(json.dumps({"pageid": "10", "revid": "1", "title": "Ala", "timestamp": "2020-01-01T00:00:00Z",
+                             "text": BODY}) + "\n")
+    monkeypatch.setattr(fw, "COPIED_PAGES", {99: "gone"})
+    with pytest.raises(ValueError, match="99"):
+        fw.prepare(tmp_path)
 
 
 # --- released data -------------------------------------------------------------------------
@@ -157,18 +219,40 @@ def test_schema_ids_and_counts():
     assert schema.names == ["id", "text", "source", "added", "created", "token_count", "license", "author"]
     table, cols = _released()
     assert table.num_rows == stats["kept"] == build["attribution_rows"]
-    assert stats["read"] == build["prepare"]["counts"]["extracted"] - build["prepare"]["counts"]["exact_duplicate"]
+    counts = build["prepare"]["counts"]
+    assert stats["read"] == (counts["extracted"] - counts["exact_duplicate"]
+                             - counts["copied_page"] - counts["ek_copyright"])
     assert stats["drop_dup"] == 0  # the fetcher already dropped exact duplicates, lowest page id kept
     assert len(set(cols["id"])) == table.num_rows
     assert all(re.fullmatch(r"wikipedia_\d+", i) for i in cols["id"])
     assert set(cols["source"]) == {"wikipedia"}
     tagged = {f"wikipedia_{p['pageid']}" for p in build["copyvio_tagged"]}
     assert tagged and not tagged & set(cols["id"])
+    copied = {p["pageid"] for p in build["prepare"]["copied_pages"]}
+    assert copied == {str(i) for i in fw.COPIED_PAGES}
+    ek = build["prepare"]["ek_copyright"]
+    assert len(ek) == counts["ek_copyright"] and all(fw.ek_is_copyright(p) for p in ek)
+    dropped = {f"wikipedia_{i}" for i in copied | {p["pageid"] for p in ek}}
+    assert not dropped & set(cols["id"])
     assert [f["name"] for f in build["files"]] == sorted(
         (f["name"] for f in build["files"]), key=lambda n: int(re.search(r"-p(\d+)p", n).group(1)))
     assert all(re.fullmatch(r"[0-9a-f]{40}", f["sha1"]) for f in build["files"])
 
 
+def test_tag_template_aliases_match_the_dump():
+    """Every redirect to the notice templates in the dump is one the matchers know."""
+    build = json.loads((DATA / "wikipedia.build.json").read_text(encoding="utf-8"))
+    names = {}
+    for r in build["tag_template_redirects"]:
+        names.setdefault(r["target"], set()).add(r["title"].removeprefix("Szablon:"))
+    assert names["Szablon:Ek"] == fw.EK_NAMES - {"Ek"}
+    assert names["Szablon:NPA"] == {"Copyvio", "Plagiat"}
+    for name in names["Szablon:NPA"] | {"NPA"}:
+        assert fw.COPYVIO_RE.search("{{%s|http://x.pl}}" % name)
+    for name in fw.EK_NAMES:
+        assert fw.ek_tag("{{%s|npa}}" % name) == {"reason": "npa", "source": ""}
+
+
 def test_every_row_label_follows_its_last_revision():
     table, cols = _released()
     meta = {}
@@ -189,16 +273,24 @@ def test_every_row_label_follows_its_last_revision():
 
 
 def test_copyright_sample_matches_the_release():
-    """The sample must be redrawn if the released rows change."""
+    """The sample must be redrawn if the released rows change, except by the copyright drops.
+
+    It was drawn from an earlier build; the rows dropped since are listed with their label, so
+    the draw is reproduced from the release plus those rows, and each must be a copyright drop."""
     import random
 
     sample = json.loads((DATA / "wikipedia.license-evidence.json").read_text(encoding="utf-8"))["copyright_sample"]
     rows = [json.loads(line) for line in (DATA / sample["file"]).read_text(encoding="utf-8").splitlines()]
     _, cols = _released()
+    prepare = json.loads((DATA / "wikipedia.build.json").read_text(encoding="utf-8"))["prepare"]
+    copyright_drops = {f"wikipedia_{p['pageid']}" for p in prepare["copied_pages"] + prepare["ek_copyright"]}
+    dropped = dict(sample["population"]["rows_dropped_since"])
+    assert set(dropped) <= copyright_drops and not set(dropped) & set(cols["id"])
+    population = list(zip(cols["id"], cols["license"])) + list(dropped.items())
     random.seed(sample["seed"])
     expected = []
     for label in sorted(sample["per_stratum"]):
-        ids = sorted((i for i, lic in zip(cols["id"], cols["license"]) if lic == label),
+        ids = sorted((i for i, lic in population if lic == label),
                      key=lambda i: int(i.split("_")[1]))  # page-id order, not parquet row order
         expected += [(label, i) for i in random.sample(ids, sample["per_stratum"][label])]
     assert [(r["stratum"], r["id"]) for r in rows] == expected