Keep legacy source refreshes in their verified joint editions

.gitea/workflows/check.yml

@@ -49,7 +49,7 @@ jobs:
               subprocess.run([str(venv/'bin/paper-atlas'),'--help'],cwd=outside,check=True)
               subprocess.run([python,'-I','-B','-m','paper_atlas','collect','cvf','--help'],cwd=outside,check=True)
               subprocess.run([python,'-I','-B','-m','paper_atlas','maintain','postgres','--help'],cwd=outside,check=True)
-              subprocess.run([python,'-I','-B','-m','paper_atlas','maintain','event-identity','--help'],cwd=outside,check=True)
+              subprocess.run([python,'-I','-B','-m','paper_atlas','maintain','joint-editions','--help'],cwd=outside,check=True)
               subprocess.run([python,'-I','-B','-m','paper_atlas','import','records','--help'],cwd=outside,check=True)
               subprocess.run([python,'-I','-B','-m','paper_atlas','import','arxiv','--help'],cwd=outside,check=True)
               subprocess.run([python,'-I','-B','-m','paper_atlas','maintain','backfill','--help'],cwd=outside,check=True)

README.md

@@ -87,4 +87,4 @@ CI 在已授权的远程 runner 上运行测试、构建 wheel，并从源码目
 
 开发通过工作分支和 PR 交付，由用户手动合并。
 
-同年多届的 `event_key`、显式迁移、旧文件刷新与查询边界见 [会议事件身份](docs/event-identity.md)。
+历史联合会议的既有届次归属可通过 [`maintain joint-editions`](docs/joint-editions.md) 先只读规划，再按计划哈希执行；保留论文与投稿决定来源，不重导入。

docs/event-identity.md

@@ -1,44 +0,0 @@
-# 同年多届会议的事件身份
-
-`venue`、`year`、`track` 描述系列、年份和投稿轨道；`event_key` 区分同系列、同年、同轨道的实际会议事件。例如 WWW 1995 的春季与秋季会议分别使用 `www-1995-spring` 和 `www-1995-fall`，两者仍是 `main`。相同论文可以在两届有不同投稿与决定；论文去重不会把这两次投稿合成一次。
-
-普通单届数据继续使用空键。新输入在同一组已有具名事件或已知同年多届时必须提供明确键；不能任选第一个 edition。已有来源刷新可依据自己的原投稿关系回到原事件，不能被显式改到另一事件。精确固定 DBLP 快照、WWW Fall 书目名、1995 年及 `www1995f.html` 卷共同支持旧秋季文件的归属；不按标题猜年份或季节。
-
-规范 JSONL 行及 manifest 都可提供 `event_key`，非空值限制为最多80个小写字母、数字及 `._:-`，首字符为字母或数字。manifest 提供键时，每行必须一致；一份文件仍仅对应一个事件及轨道。跨行冲突会回滚整份业务导入。身份回填同样按事件选择，不能借机把来源移到另一届。
-
-## 新样本与既有数据
-
-新建 SQLite 样本的 `editions` 唯一键为 `(venue_id, year, track, event_key)`。已有 SQLite/PostgreSQL 不会因普通读写命令自动升级。其普通单届导入继续可用；WWW1994/1995或带事件键的导入会要求先迁移。冻结的 SQLite 生产备份仍受 cutover 标记保护，不能写入或升级。
-
-先在 PostgreSQL 生成只读计划：
-
-```sh
-PGSERVICEFILE=/etc/paper-atlas/pg_service.conf paper-atlas maintain event-identity \
-  --pg-service foundry-research --pg-schema research --out /data/paper-index/event-plan.json
-```
-
-检查计划的数据库身份、原 edition IDs、业务表数量、目标来源与历史指纹。精确计数在同一快照中按主键分页，每页最多5000行，只读取主键；不使用估计行数。CLI向stderr持续报告计数进度，避免大库长时间无输出。已有未标记 WWW1995 edition 必须逐投稿都有固定 DBLP Fall 卷证据，才会只为原 edition 填入 `www-1995-fall`；缺证或混合卷会拒绝计划。本文命令不表示生产已经迁移。
-
-应用时传入同一文件及命令输出的文件 SHA256：
-
-```sh
-PGSERVICEFILE=/etc/paper-atlas/pg_service.conf paper-atlas maintain event-identity \
-  --pg-service foundry-research --pg-schema research \
-  --apply-plan /data/paper-index/event-plan.json --plan-sha256 PLAN_FILE_SHA256 \
-  --out /data/paper-index/event-applied.json
-```
-
-原生 PostgreSQL writer 使用现有 schema advisory lock，并在事务中锁定 editions，核对计划后添加列及替换唯一约束。SQLite 样本使用独占写事务，重建父表时保留 IDs，并在提交前检查外键；不级联删除子记录。额外列、未知唯一约束或 SQLite 额外索引会拒绝自动升级，须单独评估。未改变原版四列的排序内容哈希、任何业务表数量或目标投稿/决定/来源内容，才可提交；失败回滚。迁移标记与 DDL 同事务。原计划立即重试为 `already_applied`；若后续业务已经变化，应重新生成计划，不重放旧计划。
-
-ECML 与 PKDD 的联合举办前事件已有不同 venue/edition，保持原身份即可。这项迁移不会把它们合并，也不会创建新的论文、投稿或决定。WWW 春季及1994两届的逐篇采集、来源核验和导入仍须另做。
-
-## 查询与验收
-
-```sh
-paper-atlas query --pg-service foundry-research --pg-schema research \
-  --venue WWW --year 1995 --track main --event-key www-1995-fall
-```
-
-详细投稿历史及 `known_acceptances` 返回 `event_key`。`--event-key` 可与 `--status` 一起限定同一事件中的最新已知决定。未迁移的旧库仍可普通查询，其事件键为空；不能声称已经区分季节。`audit histories` 的新库事件输出也保留键。
-
-远程测试应覆盖同一作品的两届独立投稿、拒稿后另一届录用、原 IDs 与 Unicode/NUL 来源内容保留、旧文件刷新、歧义及错误键拒绝、整文件回滚、明确事件的身份回填、旧计划失效、重试与真实 PostgreSQL 迁移。生产应用后还须独立核验约束、来源和历史、全文索引及 foundry-core 健康；单元测试不替代生产或最终全库去重验收。

docs/joint-editions.md

@@ -0,0 +1,67 @@
+# Repairing joint conference membership
+
+EURAI's [official history](https://www.eurai.org/ecai/history) identifies ECAI 2018
+and 2022 as joint IJCAI/ECAI conferences. The pinned DBLP volumes
+`conf/ijcai/2018` and `conf/ijcai/2022` use `IJCAI` as their booktitle and title.
+Existing records can therefore already be present under IJCAI while an ECAI query
+shows a missing year. Adding ECAI as a global alias to the ordinary IJCAI venue
+would incorrectly include all other IJCAI years.
+
+The existing canonical `ECAI+IJCAI` venue has both member aliases. This maintenance
+command changes only the venue reference of the two existing `main` edition rows.
+Edition IDs, submission IDs, decisions, source text, authors, identifiers, tags,
+source links and coverage records stay intact. It does not merge or create papers
+or submissions, resolve per-paper tracks, or turn reviews into negative decisions.
+The DBLP collector applies the joint mapping only to these two exact main volume
+keys; original booktitle/title remain in source metadata with the membership evidence.
+The shared ingestion entry point also resolves these exact pinned source volumes before
+creating an edition, so refreshing an older normalized file cannot recreate a separate
+IJCAI edition. Other sources, years, workshop volumes and additional tracks retain their existing rules.
+
+## Review a read-only plan
+
+Use a runtime containing this repair command. The configured PostgreSQL writer lock
+is used only for the apply operation; do not replace a newer ingestion runtime with
+an older main-based feature branch.
+
+```sh
+PGSERVICEFILE=/etc/paper-atlas/pg_service.conf paper-atlas maintain joint-editions \
+  --pg-service foundry-research --pg-schema research --out joint-plan.json
+```
+
+The preflight requires the existing joint venue and exactly its two member aliases,
+one nonempty unambiguous edition per year, and matching pinned DBLP volume evidence
+for every existing submission. A colliding target edition, missing provenance or
+unexpected source scope fails closed. The plan binds current database/schema, IDs,
+counts and full fingerprints of affected papers, identifiers, source records, tags,
+submissions, decisions and their provenance, and coverage rows. Observation time is
+recorded separately from conference year. These are scoped checks, not a final global
+identity or conference completeness audit.
+
+## Apply the exact reviewed plan
+
+Compute the file's SHA256 after review and pass it explicitly:
+
+```sh
+PGSERVICEFILE=/etc/paper-atlas/pg_service.conf paper-atlas maintain joint-editions \
+  --pg-service foundry-research --pg-schema research --plan joint-plan.json \
+  --plan-sha256 "$PLAN_SHA256" --out joint-applied.json
+```
+
+The plan hash is checked before opening a writer. Current evidence is checked again
+under the dataset writer lock; both edition updates and an audit marker commit in one
+transaction. Protected row hashes are checked again before commit. Changed evidence
+requires a fresh reviewed plan. Existing output reports are preserved; choose a new
+path for another run. SQLite samples use `--db sample.sqlite`; the frozen production
+backup remains protected.
+
+If a response is lost after commit, rerun with the same plan and a new output path.
+When protected evidence is unchanged and both editions already point to the joint
+venue, the result is `already_joint` with zero updates. It does not replay imports or
+regenerate source records. Keep the plan, apply report and independent postcheck
+alongside collection evidence. The audit marker is `joint_ijcai_ecai_2018_2022_v1`.
+
+Publication counts can differ between the DBLP volume and the current publisher
+page. This membership repair preserves all existing rows rather than deleting entries
+to force an aggregate match; section-level main/invited/other classification and
+publication completeness require separate per-paper evidence.

paper_atlas/audits/histories.py

@@ -124,11 +124,7 @@ def report(db=None, *, pg_service=None, pg_schema=None, examples=20):
                 sample_attempts.append(rows)
 
         current = []
-        from paper_atlas.db.events import has_event_key
-        attempts_sql = ATTEMPTS
-        if has_event_key(connection):
-            attempts_sql = attempts_sql.replace('e.year,e.track,', 'e.year,e.track,e.event_key,', 1).replace('e.year,e.track\n', 'e.year,e.track,e.event_key\n')
-        for raw in connection.execute(attempts_sql):
+        for raw in connection.execute(ATTEMPTS):
             row = dict(raw)
             if current and row['paper_id'] != current[0]['paper_id']:
                 consume(current)

paper_atlas/cli.py

@@ -25,7 +25,7 @@ COMMANDS = {
     ('catalogue','dblp'): ('paper_atlas.collectors.dblp', {}, '盘点 DBLP 历史会议目录'),
     ('import','arxiv'): ('paper_atlas.ingestion.arxiv', {}, '导入 arXiv 元数据 ZIP'),
     ('import','records'): ('paper_atlas.ingestion.records', {}, '导入核验过的会议文件'),
-    ('maintain','event-identity'): ('paper_atlas.ingestion.event_identity', {}, '按核验计划增加同年会议事件身份'),
+    ('maintain','joint-editions'): ('paper_atlas.ingestion.joint_editions', {}, '按已核验计划修正既有联合会议届次归属'),
     ('maintain','merge'): ('paper_atlas.ingestion.maintenance', {}, '协调单写入合并与恢复'),
     ('maintain','backfill'): ('paper_atlas.ingestion.backfill', {}, '回填既有投稿的来源身份'),
     ('maintain','arxiv-tags'): ('paper_atlas.ingestion.arxiv_tags', {}, '恢复逐来源记录的 arXiv 类别证据'),

paper_atlas/collectors/dblp_records.py

@@ -88,6 +88,8 @@ def normalize_row(row,volumes,targets,include_workshops=False,through_year=2026)
               'scope_rule':edition['rule'],'author_names_raw':fields.get('author',[]),'publtype':row.get('publtype'),
               'bibliography':{key:value for key,value in fields.items() if key in ('booktitle','year','ee','url','pages','volume','number','publisher','series','isbn')},
               'ambiguous_identifiers':ambiguous,'publication_evidence':'inproceedings bibliography; not an original decision note'}
+    if edition.get('joint_edition_evidence'):
+        metadata['joint_edition_evidence'] = edition['joint_edition_evidence']
     if edition.get('track_scope_note'):
         metadata['track_scope_note']=edition['track_scope_note']
     flagged=bool(re.search(r'\b(?:withdrawn|retracted)\b',row.get('publtype') or '',re.I))

paper_atlas/collectors/dblp_scope.py

@@ -96,8 +96,15 @@ def scope(row,volume,through_year=2026):
     elif suffix.casefold() in ('poster','posters') and venue not in ('ICLR','ACL','EMNLP','NAACL','EACL','COLING','CoNLL'):track='poster'
     if venue=='AAAI+IAAI':track='proceedings_unspecified'
     if venue=='AAMAS' and year<2002:return None,'predecessor_not_aamas'
+    joint_evidence = None
+    if (family == 'ijcai' and base == 'IJCAI' and track == 'main'
+            and year in (2018, 2022) and volume['key'] == 'conf/ijcai/' + str(year)):
+        venue, aliases = 'ECAI+IJCAI', ['ECAI', 'IJCAI']
+        joint_evidence = 'https://www.eurai.org/ecai/history'
     result = {'venue':venue,'venue_aliases':aliases,'year':year,'track':track,'bibliographic_year':int(bibliographic[0]),
               'rule':family+':'+base,'proceedings_title':title,'edition_year_basis':year_basis}
+    if joint_evidence:
+        result['joint_edition_evidence'] = joint_evidence
     if venue=='MM' and track=='main':
         # A main MM volume also includes doctoral, tutorials, companion papers
         # and other sections. DBLP XML omits the per-paper TOC section.

paper_atlas/db/events.py

@@ -1,71 +0,0 @@
-"""Stable event identity within a venue/year/track, independent of paper identity."""
-import json
-import re
-
-
-def has_event_key(c):
-    if getattr(c, 'postgres', False):
-        return bool(c.execute("SELECT 1 FROM information_schema.columns WHERE table_schema=current_schema() AND table_name='editions' AND column_name='event_key'").fetchone())
-    return any(row[1] == 'event_key' for row in c.execute('PRAGMA table_info(editions)'))
-
-
-def validate_event_key(value):
-    if value is not None and (not isinstance(value, str) or
-            (value != '' and not re.fullmatch(r'[a-z0-9][a-z0-9._:-]{0,79}', value))):
-        raise ValueError('Invalid conference event_key')
-    return value
-
-
-def pinned_www_fall(source, metadata):
-    from paper_atlas.collectors.dblp import XML_SHA
-    b = metadata.get('bibliography') or {}
-    urls = b.get('url') or []
-    return (source == 'dblp' and metadata.get('snapshot_sha256') == XML_SHA
-            and metadata.get('bibliographic_year') == 1995
-            and str(metadata.get('proceedings_key', '')).startswith('conf/www/')
-            and b.get('booktitle') == ['WWW (Fall)'] and len(urls) == 1
-            and urls[0].split('#')[0] == 'db/conf/www/www1995f.html')
-
-
-def find_edition(store, venue_id, venue, year, track, record, external_id,
-                 event_key=None, *, create=True):
-    c = store.conn
-    validate_event_key(event_key)
-    enabled = has_event_key(c) if getattr(store, 'event_identity', None) is None else store.event_identity
-    known_multi = venue.upper() == 'WWW' and int(year) in (1994, 1995)
-    source = c.execute('SELECT src.name,r.metadata_json FROM source_records r JOIN sources src ON src.id=r.source_id WHERE r.id=?', (record,)).fetchone()
-    inferred = ('www-1995-fall' if known_multi and int(year) == 1995 and source
-                and pinned_www_fall(source[0], json.loads(source[1])) else None)
-    if inferred:
-        if event_key is not None and event_key != inferred:
-            raise ValueError('event_key conflicts with pinned WWW Fall source evidence')
-        event_key = inferred
-    if not enabled:
-        if event_key or known_multi:
-            raise ValueError('Explicit event identity migration required before this conference import')
-        row = c.execute('SELECT id FROM editions WHERE venue_id=? AND year=? AND track=?', (venue_id, year, track)).fetchone()
-        if row or not create:
-            return row[0] if row else None
-        c.execute('INSERT INTO editions(venue_id,year,track) VALUES(?,?,?) ON CONFLICT DO NOTHING', (venue_id, year, track))
-        return c.execute('SELECT id FROM editions WHERE venue_id=? AND year=? AND track=?', (venue_id, year, track)).fetchone()[0]
-    anchored = c.execute('''SELECT DISTINCT e.id,e.event_key FROM editions e JOIN submissions s ON s.edition_id=e.id
-        JOIN submission_sources ss ON ss.submission_id=s.id WHERE e.venue_id=? AND e.year=? AND e.track=?
-        AND ss.source_record_id=? AND ss.source_submission_id=?''', (venue_id, year, track, record, external_id)).fetchall()
-    if len(anchored) > 1:
-        raise ValueError('Source record already has ambiguous conference events')
-    if anchored:
-        if event_key is not None and event_key != anchored[0][1]:
-            raise ValueError('Cannot move an existing source to another conference event')
-        event_key = anchored[0][1]
-    if event_key is None:
-        rows = c.execute('SELECT event_key FROM editions WHERE venue_id=? AND year=? AND track=?', (venue_id, year, track)).fetchall()
-        if len(rows) > 1 or any(row[0] for row in rows):
-            raise ValueError('Explicit event_key required; conference event is ambiguous')
-        event_key = ''
-    if known_multi and not event_key:
-        raise ValueError('Explicit event_key required for same-year WWW events')
-    row = c.execute('SELECT id FROM editions WHERE venue_id=? AND year=? AND track=? AND event_key=?', (venue_id, year, track, event_key)).fetchone()
-    if row or not create:
-        return row[0] if row else None
-    c.execute('INSERT INTO editions(venue_id,year,track,event_key) VALUES(?,?,?,?) ON CONFLICT DO NOTHING', (venue_id, year, track, event_key))
-    return c.execute('SELECT id FROM editions WHERE venue_id=? AND year=? AND track=? AND event_key=?', (venue_id, year, track, event_key)).fetchone()[0]

paper_atlas/db/schema.sql

@@ -29,8 +29,7 @@ CREATE TABLE IF NOT EXISTS venues (
 CREATE TABLE IF NOT EXISTS editions (
  id INTEGER PRIMARY KEY, venue_id INTEGER NOT NULL REFERENCES venues(id),
  year INTEGER NOT NULL, track TEXT NOT NULL DEFAULT 'main',
- event_key TEXT NOT NULL DEFAULT '',
- UNIQUE(venue_id,year,track,event_key)
+ UNIQUE(venue_id,year,track)
 );
 CREATE TABLE IF NOT EXISTS venue_aliases (
  venue_id INTEGER NOT NULL REFERENCES venues(id), alias TEXT NOT NULL,

paper_atlas/db/store.py

@@ -36,8 +36,6 @@ class Store:
         self.conn.execute('PRAGMA synchronous=NORMAL')
         self.conn.execute('PRAGMA cache_size=-32768')
         migrate_sources(self.conn)
-        from paper_atlas.db.events import has_event_key
-        self.event_identity = has_event_key(self.conn)
 
     def source(self,name,url):
         self.conn.execute('INSERT INTO sources(name,homepage) VALUES(?,?) ON CONFLICT DO NOTHING',(name,url))
@@ -144,15 +142,29 @@ class Store:
             tag=c.execute('SELECT id FROM tags WHERE namespace=? AND value=?',(namespace,value)).fetchone()[0]
             c.execute('INSERT INTO paper_tags VALUES(?,?,?,?) ON CONFLICT DO NOTHING',(paper,tag,record,evidence))
 
-    def decision(self,paper,record,venue,year,track,external_id,url,status,raw_value,aliases=None,identities=None,observed_at=None,decision_date=None,source_decision_id='',submitted_at=None,guard_status=False,event_key=None):
+    def decision(self,paper,record,venue,year,track,external_id,url,status,raw_value,aliases=None,identities=None,observed_at=None,decision_date=None,source_decision_id='',submitted_at=None,guard_status=False):
         c = self.conn
+        # An already collected file may still call these exact pinned volumes
+        # IJCAI. Resolve membership before finding/creating an edition so a
+        # later refresh cannot split the repaired joint event again.
+        if venue == 'IJCAI' and year in (2018,2022) and track == 'main':
+            from paper_atlas.collectors.dblp import XML_SHA
+            evidence=c.execute('SELECT src.name,r.metadata_json FROM source_records r JOIN sources src ON src.id=r.source_id WHERE r.id=?',(record,)).fetchone()
+            if evidence and evidence[0] == 'dblp':
+                metadata=json.loads(evidence[1])
+                if (metadata.get('snapshot_sha256') == XML_SHA
+                        and metadata.get('proceedings_key') == 'conf/ijcai/'+str(year)
+                        and metadata.get('bibliographic_year') == year
+                        and (metadata.get('bibliography') or {}).get('booktitle') in (['IJCAI'],['IJCAI/ECAI'])):
+                    venue='ECAI+IJCAI'
+                    aliases=sorted(set(aliases or []) | {'ECAI','IJCAI'})
         existing=c.execute('SELECT id FROM venues WHERE lower(acronym)=lower(?)',(venue,)).fetchone()
         if existing:venue_id=existing[0]
         else:venue_id=inserted_id(c,"INSERT INTO venues(acronym,name,field,venue_type) VALUES(?,?,?,'conference')",(venue,venue,'unclassified'))
         for alias in aliases or []:
             c.execute('INSERT INTO venue_aliases VALUES(?,?) ON CONFLICT DO NOTHING',(venue_id,alias))
-        from paper_atlas.db.events import find_edition
-        edition = find_edition(self,venue_id,venue,int(year),track,record,external_id,event_key)
+        c.execute('INSERT INTO editions(venue_id,year,track) VALUES(?,?,?) ON CONFLICT DO NOTHING',(venue_id,int(year),track))
+        edition = c.execute('SELECT id FROM editions WHERE venue_id=? AND year=? AND track=?',(venue_id,int(year),track)).fetchone()[0]
         # A previous event in this row may already have merged the input paper.
         paper=c.execute('SELECT paper_id FROM source_records WHERE id=?',(record,)).fetchone()[0]
         submission=attach_submission(self,edition,paper,record,external_id,identities or [])

paper_atlas/ingestion/backfill.py

@@ -37,14 +37,9 @@ def backfill(db,files, *, pg_service=None, pg_schema=None):
                     seen.add(row['external_id'])
                     if any(row.get(key)!=manifest.get(key) for key in ('source','venue','year','track')):
                         raise ValueError('Evidence row differs from edition manifest')
-                    if 'event_key' in manifest and row.get('event_key')!=manifest['event_key']:
-                        raise ValueError('Event identity differs from manifest')
                     require_track_record(row, manifest)
                     record=c.execute('SELECT r.id,r.paper_id,r.metadata_json FROM source_records r JOIN sources s ON s.id=r.source_id WHERE s.name=? AND r.external_id=?',(row['source'],row['external_id'])).fetchone()
-                    from paper_atlas.db.events import find_edition
-                    venue=c.execute('SELECT id FROM venues WHERE lower(acronym)=lower(?)',(row['venue'],)).fetchone()
-                    eid=find_edition(store,venue[0],row['venue'],row['year'],row['track'],record[0],row['external_id'],row.get('event_key'),create=False) if record and venue else None
-                    edition=(eid,) if eid is not None else None
+                    edition=c.execute('SELECT e.id FROM editions e JOIN venues v ON v.id=e.venue_id WHERE lower(v.acronym)=lower(?) AND e.year=? AND e.track=?',(row['venue'],row['year'],row['track'])).fetchone()
                     if not record or not edition:raise ValueError('Evidence backfill requires an already imported record and edition')
                     if not c.execute('SELECT 1 FROM submission_sources ss JOIN submissions s ON s.id=ss.submission_id WHERE ss.source_record_id=? AND ss.source_submission_id=? AND s.edition_id=?',(record[0],row['external_id'],edition[0])).fetchone():
                         raise ValueError('No existing submission to backfill')

paper_atlas/ingestion/event_identity.py

@@ -1,217 +0,0 @@
-"""Explicit, plan-bound edition event-key upgrade; never reimport submissions."""
-import argparse
-from contextlib import closing
-import hashlib
-import json
-from pathlib import Path
-import sqlite3
-import sys
-
-from paper_atlas.db.backend import database_arguments, open_read_connection, require_sqlite_writable
-from paper_atlas.db.events import has_event_key, pinned_www_fall
-from paper_atlas.postgres.schema import TABLES, quote
-
-MARKER = 'edition_event_identity_v1:'
-
-
-def digest(value):
-    return hashlib.sha256(json.dumps(value, ensure_ascii=False, sort_keys=True, separators=(',', ':')).encode()).hexdigest()
-
-
-def values(rows):
-    return [[r[i] for i in range(len(r))] for r in rows]
-
-
-def exact_counts(c, *, page_size=5000, progress=None):
-    """Count every row in the same snapshot through bounded primary-key pages."""
-    if not isinstance(page_size, int) or page_size < 1:
-        raise ValueError('Positive count page size required')
-    result = {}
-    for table in TABLES:
-        if table == 'schema_migrations':
-            continue
-        if getattr(c, 'postgres', False):
-            columns = [r[0] for r in c.execute("SELECT a.attname FROM pg_constraint pc JOIN pg_attribute a ON a.attrelid=pc.conrelid AND a.attnum=ANY(pc.conkey) WHERE pc.conrelid=?::regclass AND pc.contype='p' ORDER BY array_position(pc.conkey,a.attnum)", (table,))]
-        else:
-            columns = [r[1] for r in sorted(c.execute('PRAGMA table_info('+quote(table)+')'), key=lambda r:r[5]) if r[5]]
-        if not columns:
-            raise ValueError('Primary key required for exact count: '+table)
-        fields = ','.join(quote(column) for column in columns)
-        last = None; count = 0; pages = 0
-        if progress:
-            progress({'table':table, 'rows_counted':0, 'complete':False})
-        while True:
-            where = (' WHERE ('+fields+') > ('+','.join('?' for _ in columns)+')') if last is not None else ''
-            batch = values(c.execute('SELECT '+fields+' FROM '+quote(table)+where+' ORDER BY '+fields+' LIMIT ?', (*last, page_size) if last is not None else (page_size,)).fetchall())
-            count += len(batch); pages += 1
-            complete = len(batch) < page_size
-            if progress and (complete or pages % 20 == 0):
-                progress({'table':table, 'rows_counted':count, 'complete':complete})
-            if complete:
-                break
-            last = batch[-1]
-        result[table] = count
-    return result
-
-
-def inspect(c, *, progress=None):
-    enabled = has_event_key(c)
-    if getattr(c, 'postgres', False):
-        columns = [r[0] for r in c.execute("SELECT column_name FROM information_schema.columns WHERE table_schema=current_schema() AND table_name='editions' ORDER BY ordinal_position")]
-        unique = [r[0] for r in c.execute("SELECT pg_get_constraintdef(oid) FROM pg_constraint WHERE conrelid='editions'::regclass AND contype='u'")]
-        expected_unique = ['UNIQUE (venue_id, year, track, event_key)' if enabled else 'UNIQUE (venue_id, year, track)']
-    else:
-        columns = [r[1] for r in c.execute('PRAGMA table_info(editions)')]
-        indexes = list(c.execute('PRAGMA index_list(editions)'))
-        unique = [tuple(r[2] for r in c.execute('PRAGMA index_info('+quote(index[1])+')')) for index in indexes if index[2]]
-        expected_unique = [('venue_id','year','track','event_key') if enabled else ('venue_id','year','track')]
-        if len(indexes) != 1:
-            raise ValueError('Unexpected edition indexes; require a reviewed migration')
-    if columns != ['id','venue_id','year','track'] + (['event_key'] if enabled else []) or unique != expected_unique:
-        raise ValueError('Unexpected edition schema; require a reviewed migration')
-    rows = values(c.execute('SELECT id,venue_id,year,track FROM editions ORDER BY id'))
-    keys = values(c.execute('SELECT id,event_key FROM editions ORDER BY id')) if enabled else [[r[0], ''] for r in rows]
-    counts = exact_counts(c, progress=progress)
-    targets = values(c.execute("SELECT e.id FROM editions e JOIN venues v ON v.id=e.venue_id WHERE v.acronym='WWW' AND e.year=1995 AND e.track='main' ORDER BY e.id"))
-    assignments = []
-    for (eid,) in targets:
-        current_key = dict(keys)[eid]
-        if current_key:
-            continue
-        evidence = values(c.execute('''SELECT s.id,src.name,r.id,r.external_id,r.metadata_json FROM submissions s
-            JOIN submission_sources ss ON ss.submission_id=s.id JOIN source_records r ON r.id=ss.source_record_id
-            JOIN sources src ON src.id=r.source_id WHERE s.edition_id=? ORDER BY s.id,r.id''', (eid,)))
-        count = c.execute('SELECT count(*) FROM submissions WHERE edition_id=?', (eid,)).fetchone()[0]
-        supported = {r[0] for r in evidence if pinned_www_fall(r[1], json.loads(r[4]))}
-        if not count or len(supported) != count:
-            raise ValueError('Existing unlabelled WWW1995 edition lacks complete pinned Fall evidence')
-        assignments.append({'edition_id':eid, 'event_key':'www-1995-fall', 'submissions':count,
-                            'source_evidence_sha256':digest(evidence)})
-    # Protect all rows in the affected attempt history, including non-DBLP evidence.
-    target_ids = [r[0] for r in targets]
-    protected = {}
-    for table,sql in {
-        'submissions': 'SELECT s.* FROM submissions s WHERE s.edition_id IN ({}) ORDER BY s.id',
-        'decisions': 'SELECT d.* FROM decisions d JOIN submissions s ON s.id=d.submission_id WHERE s.edition_id IN ({}) ORDER BY d.id',
-        'submission_sources': 'SELECT ss.* FROM submission_sources ss JOIN submissions s ON s.id=ss.submission_id WHERE s.edition_id IN ({}) ORDER BY ss.submission_id,ss.source_record_id,ss.source_submission_id',
-        'decision_sources': 'SELECT ds.* FROM decision_sources ds JOIN decisions d ON d.id=ds.decision_id JOIN submissions s ON s.id=d.submission_id WHERE s.edition_id IN ({}) ORDER BY ds.decision_id,ds.source_record_id,ds.source_decision_id',
-        'source_records': 'SELECT DISTINCT r.* FROM source_records r JOIN submission_sources ss ON ss.source_record_id=r.id JOIN submissions s ON s.id=ss.submission_id WHERE s.edition_id IN ({}) ORDER BY r.id',
-    }.items():
-        records = values(c.execute(sql.format(','.join('?' for _ in target_ids)), target_ids)) if target_ids else []
-        protected[table] = {'count':len(records), 'sha256':digest(records)}
-    return {'event_column':enabled, 'editions_sha256':digest(rows), 'edition_ids':[r[0] for r in rows],
-            'event_keys':keys, 'counts':counts, 'assignments':assignments, 'protected':protected}
-
-
-def database_identity(c, db=None):
-    if getattr(c, 'postgres', False):
-        r = c.execute('SELECT current_database(),current_schema()').fetchone()
-        return {'engine':'postgres', 'database':r[0], 'schema':r[1]}
-    return {'engine':'sqlite', 'path':str(Path(db).resolve())}
-
-
-def plan(db=None, *, pg_service=None, pg_schema=None, progress=None):
-    with closing(open_read_connection(db, pg_service=pg_service, pg_schema=pg_schema)) as c:
-        return {'format':1, 'database':database_identity(c, db), 'before':inspect(c, progress=progress)}
-
-
-def apply_plan(c, proposal, *, progress=None):
-    before = proposal['before']
-    if proposal.get('format') != 1:
-        raise ValueError('Unsupported event identity plan')
-    if getattr(c, 'postgres', False):
-        c.execute('LOCK TABLE editions IN ACCESS EXCLUSIVE MODE')
-    current = inspect(c, progress=progress)
-    signature = digest(proposal)
-    marker = MARKER + signature
-    if c.execute('SELECT 1 FROM schema_migrations WHERE name=?', (marker,)).fetchone():
-        expected_keys = dict(before['event_keys'])
-        expected_keys.update({a['edition_id']:a['event_key'] for a in before['assignments']})
-        if (not current['event_column'] or current['editions_sha256'] != before['editions_sha256']
-                or current['protected'] != before['protected'] or current['counts'] != before['counts']
-                or dict(current['event_keys']) != expected_keys):
-            raise ValueError('Applied plan no longer matches current data; generate a new plan')
-        return {'status':'already_applied', 'updated_editions':0, 'after':current}
-    if current != before:
-        raise ValueError('Event identity plan is stale; generate a new plan')
-    if not current['event_column']:
-        if getattr(c, 'postgres', False):
-            constraints = c.execute("SELECT conname,pg_get_constraintdef(oid) FROM pg_constraint WHERE conrelid='editions'::regclass AND contype='u'").fetchall()
-            matches = [name for name,definition in constraints if definition == 'UNIQUE (venue_id, year, track)']
-            if len(matches) != 1 or len(constraints) != 1:
-                raise ValueError('Unexpected legacy edition uniqueness')
-            c.execute("ALTER TABLE editions ADD COLUMN event_key TEXT COLLATE \"C\" NOT NULL DEFAULT ''")
-            c.execute('ALTER TABLE editions DROP CONSTRAINT '+quote(matches[0]))
-            c.execute('ALTER TABLE editions ADD CONSTRAINT editions_event_identity_unique UNIQUE(venue_id,year,track,event_key)')
-        else:
-            unique = [tuple(r[2] for r in c.execute('PRAGMA index_info('+quote(index[1])+')'))
-                      for index in c.execute('PRAGMA index_list(editions)') if index[2]]
-            if unique != [('venue_id','year','track')]:
-                raise ValueError('Unexpected legacy edition uniqueness')
-            c.execute("CREATE TABLE editions_event_upgrade(id INTEGER PRIMARY KEY,venue_id INTEGER NOT NULL REFERENCES venues(id),year INTEGER NOT NULL,track TEXT NOT NULL DEFAULT 'main',event_key TEXT NOT NULL DEFAULT '',UNIQUE(venue_id,year,track,event_key))")
-            c.execute('INSERT INTO editions_event_upgrade(id,venue_id,year,track) SELECT id,venue_id,year,track FROM editions')
-            c.execute('DROP TABLE editions')
-            c.execute('ALTER TABLE editions_event_upgrade RENAME TO editions')
-    for item in before['assignments']:
-        result = c.execute("UPDATE editions SET event_key=? WHERE id=? AND event_key=''", (item['event_key'],item['edition_id']))
-        if result.rowcount != 1:
-            raise ValueError('Edition assignment changed during upgrade')
-    c.execute('INSERT INTO schema_migrations(name) VALUES(?)', (marker,))
-    after = inspect(c, progress=progress)
-    expected_keys = dict(before['event_keys'])
-    expected_keys.update({a['edition_id']:a['event_key'] for a in before['assignments']})
-    if (after['editions_sha256'] != before['editions_sha256'] or after['protected'] != before['protected']
-            or after['counts'] != before['counts'] or dict(after['event_keys']) != expected_keys):
-        raise ValueError('Event upgrade changed protected data')
-    if not getattr(c, 'postgres', False) and c.execute('PRAGMA foreign_key_check').fetchall():
-        raise ValueError('Foreign key violation after event upgrade')
-    return {'status':'applied', 'updated_editions':len(before['assignments']), 'after':after}
-
-
-def upgrade(proposal, db=None, *, pg_service=None, pg_schema=None, progress=None):
-    if pg_service:
-        from paper_atlas.postgres.writer import WriteConnection
-        with closing(WriteConnection(pg_service, pg_schema)) as c:
-            if proposal['database'] != database_identity(c):
-                raise ValueError('Plan targets a different database')
-            try:
-                result = apply_plan(c, proposal, progress=progress); c.commit(); return result
-            except BaseException:
-                c.rollback(); raise
-    require_sqlite_writable(db)
-    if not Path(db).is_file():
-        raise ValueError('Existing SQLite sample required')
-    with closing(sqlite3.connect(db, timeout=30)) as c:
-        if proposal['database'] != database_identity(c, db):
-            raise ValueError('Plan targets a different database')
-        c.execute('PRAGMA foreign_keys=OFF')  # Outside the transaction; no cascading deletes.
-        try:
-            c.execute('BEGIN IMMEDIATE')
-            result = apply_plan(c, proposal, progress=progress); c.commit(); return result
-        except BaseException:
-            c.rollback(); raise
-        finally:
-            c.execute('PRAGMA foreign_keys=ON')
-
-
-def main(argv=None, *, prog=None):
-    p = argparse.ArgumentParser(prog=prog); database_arguments(p)
-    p.add_argument('--out', required=True)
-    p.add_argument('--apply-plan'); p.add_argument('--plan-sha256')
-    args = p.parse_args(argv)
-    if bool(args.apply_plan) != bool(args.plan_sha256):
-        p.error('--apply-plan and --plan-sha256 must be supplied together')
-    target = Path(args.out)
-    if target.exists():
-        raise ValueError('Refusing to overwrite an existing plan/report')
-    progress = lambda event: print(json.dumps({'phase':'exact_row_count', **event}), file=sys.stderr, flush=True)
-    if args.apply_plan:
-        raw = Path(args.apply_plan).read_bytes()
-        if hashlib.sha256(raw).hexdigest() != args.plan_sha256:
-            raise ValueError('Plan file SHA256 differs from approved bytes')
-        result = upgrade(json.loads(raw), args.db, pg_service=args.pg_service, pg_schema=args.pg_schema, progress=progress)
-    else:
-        result = plan(args.db, pg_service=args.pg_service, pg_schema=args.pg_schema, progress=progress)
-    target.write_text(json.dumps(result, ensure_ascii=False, indent=2)+'\n')
-    print(json.dumps({'out':str(target), 'sha256':hashlib.sha256(target.read_bytes()).hexdigest(),
-                      'status':result.get('status','planned')}), flush=True)

paper_atlas/ingestion/joint_editions.py

@@ -0,0 +1,204 @@
+"""Evidence-bound repair of two existing IJCAI/ECAI edition memberships."""
+import argparse
+from contextlib import closing
+import hashlib
+import json
+from pathlib import Path
+
+from paper_atlas.db.backend import database_arguments, open_read_connection, open_store
+from paper_atlas.db.submissions import timestamp
+from paper_atlas.collectors.dblp import XML_SHA
+
+YEARS = (2018, 2022)
+TARGET = 'ECAI+IJCAI'
+MEMBERS = {'ECAI', 'IJCAI'}
+EVIDENCE = 'https://www.eurai.org/ecai/history'
+MARKER = 'joint_ijcai_ecai_2018_2022_v1'
+
+
+def _digest(c, sql, params=()):
+    sha = hashlib.sha256(); count = 0
+    for row in c.execute(sql, params):
+        sha.update((json.dumps([row[i] for i in range(len(row))], ensure_ascii=False,
+                              separators=(',', ':')) + '\n').encode())
+        count += 1
+        if count > 100000:
+            raise ValueError('Joint repair evidence scope exceeded bounded row limit')
+    return {'rows': count, 'sha256': sha.hexdigest()}
+
+
+def fingerprints(c, ids):
+    marks = ','.join('?' for _ in ids)
+    submissions = 'SELECT id FROM submissions WHERE edition_id IN (' + marks + ')'
+    papers = 'SELECT paper_id FROM submissions WHERE edition_id IN (' + marks + ')'
+    decisions = 'SELECT id FROM decisions WHERE submission_id IN (' + submissions + ')'
+    checks = {
+        'papers': ('id IN (' + papers + ')', 'id'),
+        'identifiers': ('paper_id IN (' + papers + ')', 'scheme,value'),
+        'source_records': ('paper_id IN (' + papers + ')', 'id'),
+        'paper_tags': ('paper_id IN (' + papers + ')', 'paper_id,tag_id,evidence_key'),
+        'submissions': ('edition_id IN (' + marks + ')', 'id'),
+        'decisions': ('submission_id IN (' + submissions + ')', 'id'),
+        'submission_sources': ('submission_id IN (' + submissions + ')', 'submission_id,source_record_id,source_submission_id'),
+        'submission_identifiers': ('edition_id IN (' + marks + ')', 'edition_id,scheme,value'),
+        'decision_sources': ('decision_id IN (' + decisions + ')', 'decision_id,source_record_id,source_decision_id'),
+        'edition_sources': ('edition_id IN (' + marks + ')', 'edition_id,source_id,list_url'),
+    }
+    return {table: _digest(c, 'SELECT * FROM ' + table + ' WHERE ' + where + ' ORDER BY ' + order, ids)
+            for table, (where, order) in checks.items()}
+
+
+def inspect(c):
+    target = c.execute('SELECT id FROM venues WHERE acronym=?', (TARGET,)).fetchall()
+    if len(target) != 1:
+        raise ValueError('Exactly one existing canonical joint venue required')
+    target = target[0][0]
+    aliases = {r[0] for r in c.execute('SELECT alias FROM venue_aliases WHERE venue_id=?', (target,))}
+    if aliases != MEMBERS:
+        raise ValueError('Joint target must already have exactly ECAI and IJCAI aliases')
+    editions = []
+    for year in YEARS:
+        rows = c.execute('''SELECT e.id,e.venue_id,v.acronym FROM editions e JOIN venues v ON v.id=e.venue_id
+            WHERE v.acronym IN ('IJCAI','ECAI+IJCAI') AND e.year=? AND e.track='main' ''', (year,)).fetchall()
+        if len(rows) != 1:
+            raise ValueError('Expected one existing edition per year; missing or colliding editions require separate review')
+        edition, venue, acronym = (rows[0][i] for i in range(3))
+        count = c.execute('SELECT count(*) FROM submissions WHERE edition_id=?', (edition,)).fetchone()[0]
+        if not 0 < count <= 5000:
+            raise ValueError('Existing edition must have a nonempty bounded submission scope')
+        # Every existing submission must be supported by its own pinned DBLP
+        # volume record. No title/author matching, new attempts or text writes.
+        evidence = c.execute('''SELECT s.id,r.external_id,r.metadata_json FROM submissions s
+            JOIN submission_sources ss ON ss.submission_id=s.id
+            JOIN source_records r ON r.id=ss.source_record_id JOIN sources src ON src.id=r.source_id
+            WHERE s.edition_id=? AND src.name='dblp' ORDER BY s.id,r.id''', (edition,)).fetchall()
+        seen = set()
+        for row in evidence:
+            sid, external, metadata = (row[i] for i in range(3))
+            meta = json.loads(metadata)
+            bibliography = meta.get('bibliography') or {}
+            if (meta.get('snapshot_sha256') != XML_SHA
+                    or meta.get('proceedings_key') != 'conf/ijcai/' + str(year)
+                    or meta.get('bibliographic_year') != year
+                    or bibliography.get('booktitle') not in (['IJCAI'], ['IJCAI/ECAI'])
+                    or not external.startswith('conf/ijcai/')):
+                raise ValueError('Source evidence outside the exact verified IJCAI/ECAI volume')
+            seen.add(sid)
+        if len(seen) != count:
+            raise ValueError('Every submission must have matching original volume evidence')
+        editions.append({'id': edition, 'year': year, 'track': 'main', 'from_venue_id': venue,
+                         'from_venue': acronym, 'target_venue_id': target, 'submissions': count,
+                         'evidence_url': EVIDENCE, 'proceedings_key': 'conf/ijcai/' + str(year)})
+    ids = tuple(e['id'] for e in editions)
+    return {'format_version': 1, 'kind': MARKER, 'checked_at': timestamp(), 'editions': editions,
+            'content_fingerprints': fingerprints(c, ids), 'target_aliases': sorted(aliases),
+            'status': 'already_joint' if all(e['from_venue_id'] == target for e in editions) else 'ready',
+            'scope': 'Only these two existing edition venue references; no source text, paper, submission, decision, track, identifier or tag change.'}
+
+
+def validate_plan(expected):
+    if not isinstance(expected, dict):
+        raise ValueError('Repair plan must be a JSON object')
+    if set(expected) != {'format_version','kind','checked_at','editions','content_fingerprints','target_aliases','status','scope','database'}:
+        raise ValueError('Unexpected repair plan fields')
+    if (type(expected['format_version']) is not int or expected['format_version'] != 1
+            or expected['kind'] != MARKER or expected['status'] not in ('ready','already_joint')):
+        raise ValueError('Unsupported repair plan')
+    if (not isinstance(expected['database'],dict) or not isinstance(expected['content_fingerprints'],dict)
+            or not isinstance(expected['editions'],list) or len(expected['editions']) != 2
+            or expected['target_aliases'] != sorted(MEMBERS)):
+        raise ValueError('Invalid repair plan structure')
+    for edition, year in zip(expected['editions'], YEARS):
+        if not isinstance(edition,dict) or edition.get('year') != year:
+            raise ValueError('Repair plan must name the two exact ordered years')
+        for key in ('id','year','from_venue_id','target_venue_id','submissions'):
+            if type(edition.get(key)) is not int or edition[key] <= 0:
+                raise ValueError('Repair plan IDs/counts must be positive integers')
+
+
+def apply_plan(c, expected):
+    validate_plan(expected)
+    actual = inspect(c)
+    if actual['content_fingerprints'] != expected['content_fingerprints'] or actual['target_aliases'] != expected['target_aliases']:
+        raise ValueError('Stale repair plan: protected content or aliases changed')
+    if len(expected['editions']) != 2:
+        raise ValueError('Exactly two editions required')
+    for old, now in zip(expected['editions'], actual['editions']):
+        if set(old) != set(now):
+            raise ValueError('Unexpected edition fields')
+        for key in ('id','year','track','target_venue_id','submissions','evidence_url','proceedings_key'):
+            if old[key] != now[key]:
+                raise ValueError('Stale repair plan: edition differs')
+        if now['from_venue_id'] != old['from_venue_id'] and now['from_venue_id'] != now['target_venue_id']:
+            raise ValueError('Edition membership differs from plan')
+    if actual['status'] == 'already_joint':
+        return {**actual, 'status':'already_joint', 'updated_editions':0}
+    for edition in actual['editions']:
+        if edition['from_venue_id'] == edition['target_venue_id']:
+            continue
+        changed = c.execute('''UPDATE editions SET venue_id=? WHERE id=? AND venue_id=? AND year=? AND track=?''',
+                            (edition['target_venue_id'],edition['id'],edition['from_venue_id'],edition['year'],edition['track']))
+        if changed.rowcount != 1:
+            raise ValueError('Concurrent edition change; entire repair must roll back')
+    after = inspect(c)
+    if after['status'] != 'already_joint' or after['content_fingerprints'] != actual['content_fingerprints']:
+        raise ValueError('Postcheck changed protected content; entire repair must roll back')
+    c.execute('INSERT INTO schema_migrations(name) VALUES(?) ON CONFLICT DO NOTHING', (MARKER,))
+    return {**after, 'status':'applied', 'updated_editions':sum(e['from_venue_id'] != e['target_venue_id'] for e in actual['editions'])}
+
+
+def database_identity(c, db):
+    if getattr(c, 'postgres', False):
+        row = c.execute('SELECT current_database(),current_schema()').fetchone()
+        return {'engine':'postgresql','database':row[0],'schema':row[1]}
+    return {'engine':'sqlite','path':str(Path(db).resolve())}
+
+
+def run(db, out, *, pg_service=None, pg_schema=None, plan=None, plan_sha256=None):
+    out = Path(out).resolve()
+    if not out.parent.is_dir():
+        raise ValueError('Report parent directory must already exist')
+    if out.exists():
+        raise ValueError('Preserve existing evidence report; choose a new output path')
+    expected = None
+    if plan:
+        raw = Path(plan).read_bytes()
+        if hashlib.sha256(raw).hexdigest() != plan_sha256:
+            raise ValueError('Repair plan hash differs before opening database')
+        expected = json.loads(raw)
+        validate_plan(expected)
+    elif plan_sha256:
+        raise ValueError('Plan hash requires an existing plan')
+    if db and out == Path(db).resolve():
+        raise ValueError('Report must not overwrite database')
+    writer = open_store(db, pg_service=pg_service, pg_schema=pg_schema) if expected is not None else None
+    c = writer.conn if writer else open_read_connection(db, pg_service=pg_service, pg_schema=pg_schema)
+    with closing(c):
+        try:
+            if expected is not None and not getattr(c, 'postgres', False):
+                c.commit(); c.execute('BEGIN IMMEDIATE')
+            identity = database_identity(c, db)
+            if expected is not None and expected['database'] != identity:
+                raise ValueError('Plan belongs to another database/schema')
+            report = apply_plan(c, expected) if expected is not None else inspect(c)
+            report['database'] = identity
+            if expected is not None:
+                report['plan_sha256'] = plan_sha256
+            if expected is not None:
+                c.commit()
+        except BaseException:
+            if expected is not None: c.rollback()
+            raise
+    out.write_text(json.dumps(report,ensure_ascii=False,indent=2)+'\n')
+    print(json.dumps({'status':report['status'],'editions':len(report['editions']),
+                      'updated_editions':report.get('updated_editions',0),'report':str(out)}))
+    return report
+
+
+def main(argv=None, *, prog=None):
+    parser = argparse.ArgumentParser(prog=prog); database_arguments(parser)
+    parser.add_argument('--out', required=True)
+    parser.add_argument('--plan', help='Apply an already reviewed plan; omitted means read-only preflight')
+    parser.add_argument('--plan-sha256', help='Required exact plan SHA256 when applying')
+    args = parser.parse_args(argv)
+    run(args.db,args.out,pg_service=args.pg_service,pg_schema=args.pg_schema,plan=args.plan,plan_sha256=args.plan_sha256)

paper_atlas/ingestion/records.py

@@ -42,11 +42,9 @@ def ingest_file(db,path, *, pg_service=None, pg_schema=None):
                 row = json.loads(line)
                 if row['source'] != source_name:
                     raise ValueError('Source differs from manifest')
-                if any(key in manifest and manifest[key]!=row.get(key) for key in ('venue','year','track','event_key')):
+                if any(key in manifest and manifest[key]!=row[key] for key in ('venue','year','track')):
                     raise ValueError('Conference edition differs from manifest')
                 require_track_record(row, manifest)
-                from paper_atlas.db.events import validate_event_key
-                validate_event_key(row.get('event_key'))
                 external = row['external_id']
                 if external in seen:
                     raise ValueError('Duplicate external ID in one collected file: '+external)
@@ -67,7 +65,7 @@ def ingest_file(db,path, *, pg_service=None, pg_schema=None):
                         status=event['status']
                         if events and status is None:raise ValueError('Explicit decision event requires a status')
                         if status is not None and status not in ('accepted','rejected','desk_rejected','withdrawn','under_review','unknown'):raise ValueError('Unsupported decision event')
-                        edition=store.decision(paper,record,row['venue'],row['year'],row['track'],external,event.get('evidence_url',row['url']),status,event.get('raw_value',''),row.get('venue_aliases'),identities=identities,observed_at=event.get('observed_at') or row_observed,decision_date=event.get('decision_date'),source_decision_id=event.get('source_decision_id',''),submitted_at=row.get('submitted_at'),guard_status=source_name in (PUBLIC_SOURCE,ARCHIVE_SOURCE),event_key=row.get('event_key'))
+                        edition=store.decision(paper,record,row['venue'],row['year'],row['track'],external,event.get('evidence_url',row['url']),status,event.get('raw_value',''),row.get('venue_aliases'),identities=identities,observed_at=event.get('observed_at') or row_observed,decision_date=event.get('decision_date'),source_decision_id=event.get('source_decision_id',''),submitted_at=row.get('submitted_at'),guard_status=source_name in (PUBLIC_SOURCE,ARCHIVE_SOURCE))
                     editions.add(edition)
                 else:
                     raise ValueError('This importer requires an explicit supported decision status')

paper_atlas/postgres/writer.py

@@ -85,5 +85,3 @@ class WriteConnection:
 class PostgresStore(Store):
     def __init__(self, service, schema=None):
         self.conn = WriteConnection(service, schema)
-        from paper_atlas.db.events import has_event_key
-        self.event_identity = has_event_key(self.conn)

paper_atlas/querying.py

@@ -14,9 +14,7 @@ ACCEPTED_PAPERS = "SELECT s.paper_id FROM submissions s JOIN decisions d ON d.su
 
 
 def known_acceptances(conn,paper):
-    from paper_atlas.db.events import has_event_key
-    event = 'e.event_key' if has_event_key(conn) else "''"
-    return [dict(row) for row in conn.execute('''SELECT s.id AS submission_id,v.acronym AS venue,e.year,e.track,'''+event+''' AS event_key,
+    return [dict(row) for row in conn.execute('''SELECT s.id AS submission_id,v.acronym AS venue,e.year,e.track,
         d.id AS decision_id,d.decision_date,d.observed_at,d.evidence_url
         FROM submissions s JOIN editions e ON e.id=s.edition_id JOIN venues v ON v.id=e.venue_id
         JOIN decisions d ON d.submission_id=s.id WHERE s.paper_id=? AND d.status='accepted' AND '''+LATEST_DECISION+
@@ -35,7 +33,6 @@ def main(argv=None, *, prog=None):
     parser.add_argument('--venue', help='Conference acronym')
     parser.add_argument('--year', type=int)
     parser.add_argument('--track',help='Exact track, e.g. main, findings, workshop')
-    parser.add_argument('--event-key',help='Exact event within a venue/year/track')
     parser.add_argument('--status', choices=['accepted','rejected','desk_rejected','withdrawn','under_review','unknown'])
     parser.add_argument('--outcome',choices=['accepted','no_known_acceptance'],help='跨会议的已知录用情况；未查到录用不等于从未录用')
     parser.add_argument('--limit', type=int, default=20)
@@ -50,11 +47,6 @@ def main(argv=None, *, prog=None):
         conn = sqlite3.connect(Path(args.db).resolve().as_uri()+'?mode=ro', uri=True)
         conn.row_factory = sqlite3.Row
         conn.execute('BEGIN')  # One consistent read snapshot while ingestion continues.
-    from paper_atlas.db.events import has_event_key
-    event = 'e.event_key' if has_event_key(conn) else "''"
-    if args.event_key is not None and event == "''":
-        conn.close()
-        raise ValueError('Event identity migration required for event-key queries')
     if args.stats:
         result = {name:conn.execute('SELECT count(*) FROM '+name).fetchone()[0] for name in ('papers','source_records','submissions','decisions','tags','match_candidates')}
         result['records_by_source'] = [dict(r) for r in conn.execute('SELECT s.name,count(r.id) AS imported_records FROM sources s LEFT JOIN source_records r ON r.source_id=s.id GROUP BY s.id ORDER BY s.name')]
@@ -81,15 +73,13 @@ def main(argv=None, *, prog=None):
             where.append('EXISTS(SELECT 1 FROM paper_tags pt JOIN tags t ON t.id=pt.tag_id WHERE pt.paper_id=p.id AND t.namespace=? AND t.value=?)'); params.extend([namespace,value])
         if args.outcome:
             where.append('p.id '+('IN' if args.outcome=='accepted' else 'NOT IN')+' ('+ACCEPTED_PAPERS+')')
-        if args.venue or args.year or args.status or args.track or args.event_key is not None:
+        if args.venue or args.year or args.status or args.track:
             clause = ['1=1']
             if args.venue:
                 if conn.execute("SELECT 1 FROM sqlite_master WHERE type='table' AND name='venue_aliases'").fetchone():
                     clause.append('(lower(v.acronym)=lower(?) OR EXISTS(SELECT 1 FROM venue_aliases va WHERE va.venue_id=v.id AND lower(va.alias)=lower(?)))'); params.extend([args.venue,args.venue])
                 else:
                     clause.append('lower(v.acronym)=lower(?)');params.append(args.venue)
-            if args.event_key is not None:
-                clause.append('e.event_key=?');params.append(args.event_key)
             if args.track:
                 clause.append('e.track=?');params.append(args.track)
             if args.year:
@@ -111,7 +101,7 @@ def main(argv=None, *, prog=None):
                 paper['tags'] = [dict(r) for r in conn.execute('SELECT t.namespace,t.value,pt.evidence_key FROM paper_tags pt JOIN tags t ON t.id=pt.tag_id WHERE pt.paper_id=?',(row['id'],))]
                 paper['sources'] = [dict(r) for r in conn.execute('SELECT s.name,r.external_id,r.url,r.title,r.abstract,r.retrieved_at FROM source_records r JOIN sources s ON s.id=r.source_id WHERE r.paper_id=?',(row['id'],))]
                 submissions = []
-                for sub in conn.execute('SELECT s.id,s.source_submission_id,s.submitted_at,v.acronym,e.year,e.track,'+event+' AS event_key FROM submissions s JOIN editions e ON e.id=s.edition_id JOIN venues v ON v.id=e.venue_id WHERE s.paper_id=? ORDER BY e.year,s.id',(row['id'],)):
+                for sub in conn.execute('SELECT s.id,s.source_submission_id,s.submitted_at,v.acronym,e.year,e.track FROM submissions s JOIN editions e ON e.id=s.edition_id JOIN venues v ON v.id=e.venue_id WHERE s.paper_id=? ORDER BY e.year,s.id',(row['id'],)):
                     submission = dict(sub)
                     submission['decisions'] = []
                     for d in conn.execute('SELECT id,status,raw_value,evidence_url,observed_at,decision_date FROM decisions WHERE submission_id=? ORDER BY observed_at,COALESCE(decision_date,observed_at),id',(sub['id'],)):

tests/test_event_identity.py

@@ -1,202 +0,0 @@
-from contextlib import closing, redirect_stdout, contextmanager
-import io
-import json
-import os
-from pathlib import Path
-import sqlite3
-import tempfile
-import unittest
-import uuid
-from importlib.resources import files
-
-from paper_atlas.db.store import Store
-from paper_atlas.db.backend import open_store
-from paper_atlas.db.events import has_event_key
-from paper_atlas.collectors.dblp import XML_SHA
-from paper_atlas.ingestion.event_identity import plan, upgrade, exact_counts
-from paper_atlas.ingestion.records import ingest_file
-from paper_atlas.ingestion.backfill import backfill
-from paper_atlas.postgres.client import connect
-from paper_atlas.postgres.schema import quote
-from paper_atlas.postgres.migration import migrate
-from paper_atlas import querying
-
-
-@contextmanager
-def opened_store(**kwargs):
-    store=open_store(**kwargs)
-    try:
-        yield store
-    finally:
-        store.conn.close()
-
-
-def legacy(db):
-    sql = files('paper_atlas.db').joinpath('schema.sql').read_text().replace(" event_key TEXT NOT NULL DEFAULT '',\n", '').replace('UNIQUE(venue_id,year,track,event_key)', 'UNIQUE(venue_id,year,track)')
-    c = sqlite3.connect(db); c.executescript(sql); c.close()
-    s = Store(db)
-    p,r = s.record('dblp','conf/www/Fall95','https://example.org/fall','Same paper in two events',
-                    'Abstract\x00with\x01text',['A Person'],{'arxiv':'2501.00001'})
-    s.record_tags(p,r,'dblp',[{'namespace':'dblp:booktitle','value':'WWW (Fall)'}])
-    s.decision(p,r,'Legacy',1995,'main','conf/www/Fall95','https://example.org/accept','accepted','Accept')
-    s.conn.execute("UPDATE venues SET acronym='WWW',name='WWW'")
-    meta={'snapshot_sha256':XML_SHA,'proceedings_key':'conf/www/1995f','bibliographic_year':1995,
-          'bibliography':{'booktitle':['WWW (Fall)'],'url':['db/conf/www/www1995f.html#Fall95']}}
-    s.conn.execute('UPDATE source_records SET metadata_json=?',(json.dumps(meta),));s.conn.commit();s.conn.close()
-    return p,r
-
-
-def add(store, external, event=None, status='accepted', venue='WWW', year=1995):
-    p,r = store.record('fixture',external,'https://example.org/'+external,'Same paper in two events',
-                       'Abstract\x00with\x01text',['A Person'],{'arxiv':'2501.00001'})
-    edition = store.decision(p,r,venue,year,'main',external,'https://example.org/'+status,status,status,event_key=event)
-    return p,r,edition
-
-
-def file(root, name, rows, event=None):
-    p=root/(name+'.jsonl');p.write_text(''.join(json.dumps(r)+'\n'for r in rows))
-    manifest={'source':'fixture','venue':'WWW','year':1995,'track':'main','count':len(rows),'list_url':'https://example.org/'+name}
-    if event is not None:manifest['event_key']=event
-    p.with_suffix('.manifest.json').write_text(json.dumps(manifest));return p
-
-
-class EventIdentityTests(unittest.TestCase):
-    def setUp(self):
-        t=tempfile.TemporaryDirectory();self.addCleanup(t.cleanup);self.root=Path(t.name);self.db=self.root/'sample.sqlite'
-
-    def test_same_paper_distinct_events_keep_both_attempts_and_decisions(self):
-        s=Store(self.db);self.addCleanup(s.conn.close)
-        paper,_,spring=add(s,'spring','www-1995-spring','rejected');_,_,fall=add(s,'fall','www-1995-fall')
-        s.conn.commit();self.assertNotEqual(spring,fall)
-        with redirect_stdout(io.StringIO())as out:querying.main(['--db',str(self.db),'--paper',str(paper)])
-        row=json.loads(out.getvalue())[0];self.assertEqual({a['event_key']for a in row['submissions']},{'www-1995-spring','www-1995-fall'})
-        self.assertEqual([a['event_key']for a in row['known_acceptances']],['www-1995-fall'])
-        self.assertEqual({a['latest_status']for a in row['submissions']},{'rejected','accepted'})
-        with redirect_stdout(io.StringIO())as out:querying.main(['--db',str(self.db),'--venue','WWW','--event-key','www-1995-spring','--status','rejected'])
-        self.assertEqual(len(json.loads(out.getvalue())),1)
-
-    def test_explicit_migration_preserves_ids_history_and_source_bytes(self):
-        paper,record=legacy(self.db);proposal=plan(self.db);self.assertFalse(proposal['before']['event_column'])
-        result=upgrade(proposal,self.db);self.assertEqual(result['updated_editions'],1)
-        self.assertEqual(result['after']['protected'],proposal['before']['protected'])
-        self.assertEqual(result['after']['editions_sha256'],proposal['before']['editions_sha256'])
-        self.assertEqual(upgrade(proposal,self.db)['status'],'already_applied')
-        s=Store(self.db);self.addCleanup(s.conn.close)
-        self.assertEqual(s.conn.execute('SELECT id,event_key FROM editions').fetchall(),[(1,'www-1995-fall')])
-        # An old file without event_key still has its original source anchor.
-        self.assertEqual(s.decision(paper,record,'WWW',1995,'main','conf/www/Fall95','https://example.org/accept','accepted','Accept'),1)
-        self.assertEqual(s.conn.execute('SELECT count(*) FROM submissions').fetchone()[0],1)
-        self.assertEqual(s.conn.execute('PRAGMA foreign_key_check').fetchall(),[])
-
-    def test_unverified_legacy_volume_is_not_automatically_labelled(self):
-        legacy(self.db)
-        with sqlite3.connect(self.db)as c:c.execute("UPDATE source_records SET metadata_json='{}'")
-        with self.assertRaisesRegex(ValueError,'complete pinned Fall'):plan(self.db)
-        with sqlite3.connect(self.db)as c:self.assertFalse(has_event_key(c))
-
-    def test_stale_plan_rolls_back_before_schema_change(self):
-        legacy(self.db);proposal=plan(self.db)
-        with sqlite3.connect(self.db)as c:c.execute("UPDATE editions SET track='unspecified'")
-        with self.assertRaisesRegex(ValueError,'stale'):upgrade(proposal,self.db)
-        with sqlite3.connect(self.db)as c:self.assertFalse(has_event_key(c));self.assertEqual(c.execute('SELECT track FROM editions').fetchone()[0],'unspecified')
-
-    def test_unscoped_source_and_wrong_refresh_event_fail(self):
-        s=Store(self.db);self.addCleanup(s.conn.close);_,r,e=add(s,'first','www-1995-fall');s.conn.commit()
-        p,r2=s.record('fixture','unscoped','https://example.org/u','Other work',None,[])
-        with self.assertRaisesRegex(ValueError,'event_key'):s.decision(p,r2,'WWW',1995,'main','unscoped','https://example.org/u','accepted','Accept')
-        with self.assertRaisesRegex(ValueError,'Cannot move'):s.decision(p,r,'WWW',1995,'main','first','https://example.org/u','accepted','Accept',event_key='www-1995-spring')
-        self.assertEqual(s.conn.execute('SELECT count(*) FROM editions').fetchone()[0],1)
-
-    def test_pinned_fall_cannot_claim_spring(self):
-        p,r=legacy(self.db);upgrade(plan(self.db),self.db);s=Store(self.db);self.addCleanup(s.conn.close)
-        with self.assertRaisesRegex(ValueError,'pinned WWW Fall'):s.decision(p,r,'WWW',1995,'main','conf/www/Fall95','https://example.org/u','accepted','Accept',event_key='www-1995-spring')
-
-    def test_manifest_event_mismatch_rolls_back_the_whole_file(self):
-        s=Store(self.db);s.conn.close()
-        base={'source':'fixture','venue':'WWW','year':1995,'track':'main','status':'accepted','title':'Input work','url':'https://example.org/p','authors':[]}
-        path=file(self.root,'bad',[dict(base,external_id='a',event_key='www-1995-spring'),dict(base,external_id='b',event_key='www-1995-fall')],'www-1995-spring')
-        with self.assertRaisesRegex(ValueError,'edition differs'):ingest_file(self.db,path)
-        with sqlite3.connect(self.db)as c:
-            self.assertEqual(c.execute('SELECT count(*) FROM papers').fetchone()[0],0)
-            self.assertEqual(c.execute('SELECT count(*) FROM editions').fetchone()[0],0)
-
-    def test_backfill_selects_the_explicit_event_and_preserves_attempts(self):
-        s=Store(self.db);add(s,'spring','www-1995-spring','rejected');add(s,'fall','www-1995-fall');s.conn.commit();s.conn.close()
-        row={'source':'fixture','external_id':'fall','venue':'WWW','year':1995,'track':'main','event_key':'www-1995-fall','status':'accepted','identifiers':{'openreview':'forumFall'}}
-        path=file(self.root,'backfill',[row],'www-1995-fall')
-        with redirect_stdout(io.StringIO()):result=backfill(self.db,[path])
-        self.assertEqual(result['submissions_after'],2)
-        with sqlite3.connect(self.db)as c:
-            self.assertEqual(c.execute("SELECT e.event_key FROM submission_identifiers si JOIN submissions s ON s.id=si.submission_id JOIN editions e ON e.id=s.edition_id WHERE si.scheme='openreview' AND si.value='forumFall'").fetchone()[0],'www-1995-fall')
-
-    def test_exact_paged_counts_cover_composite_key_boundaries(self):
-        s=Store(self.db);self.addCleanup(s.conn.close)
-        for i in range(7):add(s,'entry'+str(i),'www-1995-fall')
-        s.conn.commit();events=[];counts=exact_counts(s.conn,page_size=2,progress=events.append)
-        self.assertEqual(counts['source_records'],7)
-        self.assertEqual(counts['identifiers'],8)
-        self.assertEqual(counts['submission_sources'],7)
-        self.assertEqual([e['rows_counted']for e in events if e['table']=='identifiers'and e['complete']],[8])
-
-    def test_legacy_single_event_writes_and_reads_remain_available(self):
-        legacy(self.db);s=Store(self.db);self.addCleanup(s.conn.close)
-        add(s,'ordinary',venue='COLM',year=2025);s.conn.commit()
-        with redirect_stdout(io.StringIO())as out:querying.main(['--db',str(self.db),'--venue','COLM'])
-        self.assertEqual(json.loads(out.getvalue())[0]['known_acceptances'][-1]['event_key'],'')
-        with self.assertRaisesRegex(ValueError,'migration required'):
-            querying.main(['--db',str(self.db),'--event-key','www-1995-fall'])
-
-    def test_unexpected_extra_edition_column_requires_review(self):
-        legacy(self.db)
-        with sqlite3.connect(self.db)as c:c.execute('ALTER TABLE editions ADD COLUMN custom_data TEXT')
-        with self.assertRaisesRegex(ValueError,'Unexpected edition schema'):plan(self.db)
-
-    def test_file_spanning_two_events_without_manifest_key_rolls_back(self):
-        Store(self.db).conn.close()
-        base={'source':'fixture','venue':'WWW','year':1995,'track':'main','status':'accepted','title':'Input work','url':'https://example.org/p','authors':[]}
-        path=file(self.root,'two',[dict(base,external_id='a',event_key='www-1995-spring'),dict(base,external_id='b',event_key='www-1995-fall')])
-        with self.assertRaisesRegex(ValueError,'exactly one'):ingest_file(self.db,path)
-        with sqlite3.connect(self.db)as c:self.assertEqual(c.execute('SELECT count(*) FROM submissions').fetchone()[0],0)
-
-    def test_cutover_sqlite_backup_remains_protected(self):
-        legacy(self.db);proposal=plan(self.db);Path(str(self.db.resolve())+'.postgresql-cutover.json').write_text('{}')
-        with self.assertRaisesRegex(RuntimeError,'cut over'):upgrade(proposal,self.db)
-
-
-@unittest.skipUnless(os.environ.get('PAPER_ATLAS_TEST_PG_SERVICE'),'Requires approved remote PostgreSQL test service')
-class PostgreSQLEventIdentityTests(unittest.TestCase):
-    def setUp(self):
-        self.service=os.environ['PAPER_ATLAS_TEST_PG_SERVICE'];self.schema='pa_event_'+uuid.uuid4().hex[:12]
-        t=tempfile.TemporaryDirectory();self.addCleanup(t.cleanup);self.root=Path(t.name);self.db=self.root/'seed.sqlite';legacy(self.db)
-        self.addCleanup(self.cleanup)
-        with redirect_stdout(io.StringIO()):migrate(self.db,self.service,self.root/'migrate.json',schema=self.schema,min_free_gib=0)
-
-    def cleanup(self):
-        with connect(self.service)as c:
-            for (name,)in c.execute('SELECT nspname FROM pg_namespace WHERE starts_with(nspname,%s)',(self.schema,)).fetchall():c.execute('DROP SCHEMA '+quote(name)+' CASCADE')
-
-    def test_native_upgrade_preserves_ids_and_codec_then_accepts_separate_event(self):
-        proposal=plan(pg_service=self.service,pg_schema=self.schema)
-        result=upgrade(proposal,pg_service=self.service,pg_schema=self.schema)
-        self.assertEqual(result['after']['protected'],proposal['before']['protected'])
-        self.assertEqual(upgrade(proposal,pg_service=self.service,pg_schema=self.schema)['status'],'already_applied')
-        with opened_store(pg_service=self.service,pg_schema=self.schema)as s:
-            paper,_,spring=add(s,'spring','www-1995-spring','rejected');s.conn.commit()
-            self.assertNotEqual(spring,1)
-            self.assertEqual(s.conn.execute('SELECT abstract FROM papers WHERE id=?',(paper,)).fetchone()[0],'Abstract\x00with\x01text')
-        with redirect_stdout(io.StringIO())as out:querying.main(['--pg-service',self.service,'--pg-schema',self.schema,'--paper',str(paper)])
-        row=json.loads(out.getvalue())[0];self.assertEqual(len(row['submissions']),2);self.assertEqual(len(row['known_acceptances']),1)
-
-    def test_native_paged_counts_preserve_encoded_composite_key_order(self):
-        with opened_store(pg_service=self.service,pg_schema=self.schema)as s:
-            for value in ['a\x00b','a\x01b','a\x02b','z']:
-                s.conn.execute('INSERT INTO identifiers(scheme,value,paper_id) VALUES(?,?,?)',('custom',value,1))
-            s.conn.commit();counts=exact_counts(s.conn,page_size=2)
-            self.assertEqual(counts['identifiers'],6)
-            self.assertEqual(counts['submissions'],1)
-
-    def test_stale_native_plan_is_rejected_and_lock_released(self):
-        proposal=plan(pg_service=self.service,pg_schema=self.schema)
-        with opened_store(pg_service=self.service,pg_schema=self.schema)as s:s.conn.execute("UPDATE editions SET track='unspecified'");s.conn.commit()
-        with self.assertRaisesRegex(ValueError,'stale'):upgrade(proposal,pg_service=self.service,pg_schema=self.schema)
-        with opened_store(pg_service=self.service,pg_schema=self.schema)as s:self.assertFalse(has_event_key(s.conn))

tests/test_joint_editions.py

@@ -0,0 +1,187 @@
+from contextlib import closing, redirect_stdout
+import hashlib
+import io
+import json
+import os
+from pathlib import Path
+import tempfile
+import unittest
+from unittest.mock import patch
+import uuid
+
+from paper_atlas.collectors.dblp import XML_SHA
+from paper_atlas.collectors.dblp_scope import scope
+from paper_atlas.collectors.dblp_records import normalize_row
+from paper_atlas.db.store import Store
+from paper_atlas.db.backend import open_store
+from paper_atlas.ingestion.joint_editions import (TARGET, MARKER, inspect, apply_plan, run)
+from paper_atlas.postgres.client import connect
+from paper_atlas.postgres.migration import migrate
+from paper_atlas.postgres.schema import quote
+from paper_atlas import querying
+
+
+def seed(store):
+    c = store.conn
+    c.execute("INSERT INTO venues(acronym,name,field,venue_type) VALUES(?,?,'reasoning','conference')", (TARGET,TARGET))
+    target = c.execute('SELECT id FROM venues WHERE acronym=?',(TARGET,)).fetchone()[0]
+    for alias in ['ECAI','IJCAI']:
+        c.execute('INSERT INTO venue_aliases VALUES(?,?)',(target,alias))
+    for year in [2018,2019,2022]:
+        metadata = {'snapshot_sha256':XML_SHA,'proceedings_key':'conf/ijcai/'+str(year),
+                    'bibliographic_year':year,'bibliography':{'booktitle':['IJCAI']}}
+        p,r=store.record('dblp','conf/ijcai/fixture'+str(year),'https://dblp.org/rec/fixture',
+                         'Original publication '+str(year),'Abstract\x00with\x01data',[],
+                         {'doi':'10.24963/ijcai.'+str(year)+'/1'}, {})
+        store.record_tags(p,r,'dblp',[{'namespace':'dblp:booktitle','value':'IJCAI'}])
+        store.decision(p,r,'IJCAI',year,'main','fixture'+str(year),'https://example.org/decision','accepted','listed')
+        stored=json.loads(c.execute('SELECT metadata_json FROM source_records WHERE id=?',(r,)).fetchone()[0]);stored.update(metadata)
+        c.execute('UPDATE source_records SET metadata_json=? WHERE id=?',(json.dumps(stored),r))
+    c.commit()
+
+
+def plan(c):
+    return {**inspect(c),'database':{'engine':'sqlite','path':'fixture'}}
+
+
+class JointEditionTests(unittest.TestCase):
+    def setUp(self):
+        self.tmp=tempfile.TemporaryDirectory(); self.addCleanup(self.tmp.cleanup)
+        self.db=Path(self.tmp.name)/'sample.sqlite'
+        self.store=Store(self.db); self.addCleanup(self.store.conn.close)
+        seed(self.store); self.c=self.store.conn
+
+    def test_apply_preserves_every_protected_row_and_edition_id(self):
+        before=plan(self.c); ids=[r['id']for r in before['editions']]
+        result=apply_plan(self.c,before);self.c.commit()
+        self.assertEqual(result['updated_editions'],2)
+        self.assertEqual(result['content_fingerprints'],before['content_fingerprints'])
+        self.assertEqual([r['id']for r in result['editions']],ids)
+        untouched=self.c.execute('SELECT v.acronym FROM editions e JOIN venues v ON v.id=e.venue_id WHERE e.year=2019').fetchone()[0]
+        self.assertEqual(untouched,'IJCAI')
+        self.assertEqual(self.c.execute('PRAGMA foreign_key_check').fetchall(),[])
+        self.assertEqual(self.c.execute('SELECT count(*) FROM schema_migrations WHERE name=?',(MARKER,)).fetchone()[0],1)
+        with redirect_stdout(io.StringIO())as out:
+            querying.main(['--db',str(self.db),'--venue','ECAI'])
+        self.assertEqual(len(json.loads(out.getvalue())),2)
+        self.assertEqual(apply_plan(self.c,before)['updated_editions'],0)
+
+    def test_refresh_of_old_normalized_source_does_not_recreate_ordinary_edition(self):
+        before=plan(self.c);apply_plan(self.c,before)
+        for year in [2018,2022]:
+            row=self.c.execute('''SELECT r.paper_id,r.id,d.observed_at FROM source_records r
+                JOIN submissions s ON s.source_record_id=r.id JOIN decisions d ON d.submission_id=s.id
+                WHERE r.external_id=?''',('conf/ijcai/fixture'+str(year),)).fetchone()
+            self.store.decision(row[0],row[1],'IJCAI',year,'main','fixture'+str(year),
+                                'https://example.org/decision','accepted','listed',observed_at=row[2])
+        self.assertEqual(inspect(self.c)['content_fingerprints'],before['content_fingerprints'])
+        self.assertEqual(self.c.execute('SELECT count(*) FROM editions').fetchone()[0],3)
+
+    def test_any_changed_source_text_rejects_stale_plan_before_membership_updates(self):
+        before=plan(self.c)
+        self.c.execute("UPDATE source_records SET title='New evidence' WHERE external_id='conf/ijcai/fixture2018'")
+        with self.assertRaisesRegex(ValueError,'Stale repair plan'):
+            apply_plan(self.c,before)
+        self.assertEqual(inspect(self.c)['status'],'ready')
+
+    def test_conflicting_target_edition_is_never_merged_or_deleted(self):
+        target=self.c.execute('SELECT id FROM venues WHERE acronym=?',(TARGET,)).fetchone()[0]
+        self.c.execute("INSERT INTO editions(venue_id,year,track) VALUES(?,2018,'main')",(target,))
+        with self.assertRaisesRegex(ValueError,'colliding'):
+            inspect(self.c)
+        self.assertEqual(self.c.execute('SELECT count(*) FROM editions WHERE year=2018').fetchone()[0],2)
+
+    def test_wrong_source_volume_and_snapshot_fail_closed(self):
+        original=self.c.execute("SELECT metadata_json FROM source_records WHERE external_id='conf/ijcai/fixture2018'").fetchone()[0]
+        for key,value in [('proceedings_key','conf/ijcai/2018w'),('snapshot_sha256','wrong'),('bibliographic_year',2017),('bibliography',{'booktitle':['Other']})]:
+            m=json.loads(original);m[key]=value
+            self.c.execute("UPDATE source_records SET metadata_json=? WHERE external_id='conf/ijcai/fixture2018'",(json.dumps(m),))
+            with self.subTest(key=key),self.assertRaisesRegex(ValueError,'Source evidence'):
+                inspect(self.c)
+
+    def test_missing_source_support_and_wrong_aliases_are_not_guessed(self):
+        self.c.execute("DELETE FROM venue_aliases WHERE alias='ECAI'")
+        with self.assertRaisesRegex(ValueError,'aliases'):
+            inspect(self.c)
+        self.c.rollback()
+        self.c.execute('DELETE FROM submission_sources WHERE source_record_id=(SELECT id FROM source_records WHERE external_id=?)',('conf/ijcai/fixture2022',))
+        with self.assertRaisesRegex(ValueError,'Every submission'):
+            inspect(self.c)
+
+    def test_cli_plan_readonly_hash_binding_identity_and_repeat_apply(self):
+        out=Path(self.tmp.name)/'plan.json'
+        with redirect_stdout(io.StringIO()):run(self.db,out)
+        self.assertEqual(inspect(self.c)['status'],'ready')
+        raw=out.read_bytes();sha=hashlib.sha256(raw).hexdigest()
+        with patch('paper_atlas.ingestion.joint_editions.open_store')as writer:
+            with self.assertRaisesRegex(ValueError,'hash differs'):
+                run(self.db,Path(self.tmp.name)/'bad.json',plan=out,plan_sha256='0'*64)
+            writer.assert_not_called()
+        with redirect_stdout(io.StringIO()):
+            result=run(self.db,Path(self.tmp.name)/'applied.json',plan=out,plan_sha256=sha)
+            again=run(self.db,Path(self.tmp.name)/'retry.json',plan=out,plan_sha256=sha)
+        self.assertEqual(result['updated_editions'],2);self.assertEqual(again['updated_editions'],0)
+        self.assertEqual(result['plan_sha256'],sha);self.assertEqual(again['plan_sha256'],sha)
+        altered=json.loads(raw);altered['database']['path']='another.sqlite'
+        bad=Path(self.tmp.name)/'wrong-db.json';bad.write_text(json.dumps(altered))
+        with self.assertRaisesRegex(ValueError,'another database'):
+            run(self.db,Path(self.tmp.name)/'wrong-result.json',plan=bad,plan_sha256=hashlib.sha256(bad.read_bytes()).hexdigest())
+
+    def test_invalid_empty_plan_fails_before_opening_writer(self):
+        f=Path(self.tmp.name)/'empty.json';f.write_text('{}')
+        with patch('paper_atlas.ingestion.joint_editions.open_store')as writer:
+            with self.assertRaises(ValueError):
+                run(self.db,Path(self.tmp.name)/'empty-result.json',plan=f,plan_sha256=hashlib.sha256(f.read_bytes()).hexdigest())
+            writer.assert_not_called()
+
+    def test_collector_only_changes_the_two_exact_main_volume_keys(self):
+        for year in [2018,2019,2022]:
+            volume={'key':'conf/ijcai/'+str(year),'fields':{'title':['IJCAI '+str(year)],'year':[str(year)]}}
+            row={'record_type':'inproceedings','key':'conf/ijcai/One'+str(year),'fields':{'booktitle':['IJCAI'],
+                 'year':[str(year)],'crossref':[volume['key']],'title':['Original research title'],
+                 'author':['Jane Doe'],'ee':['https://doi.org/10.24963/ijcai.'+str(year)+'/1']}}
+            result,error=normalize_row(row,{volume['key']:volume},{'IJCAI','ECAI'})
+            self.assertIsNone(error)
+            self.assertEqual(result['venue'],TARGET if year in(2018,2022)else'IJCAI')
+            self.assertEqual(result['metadata']['bibliography']['booktitle'],['IJCAI'])
+            self.assertEqual(result['title'],'Original research title')
+            if year in(2018,2022):
+                self.assertIn('joint_edition_evidence',result['metadata'])
+                row['fields']['booktitle']=['IJCAI (Workshop)']
+                scoped,_=scope(row,volume)
+                self.assertEqual(scoped['venue'],'IJCAI')
+                self.assertEqual(scoped['track'],'workshop')
+
+
+@unittest.skipUnless(os.environ.get('PAPER_ATLAS_TEST_PG_SERVICE'),'Requires approved remote PostgreSQL service')
+class PostgreSQLJointTests(unittest.TestCase):
+    def setUp(self):
+        self.service=os.environ['PAPER_ATLAS_TEST_PG_SERVICE'];self.target='pa_joint_'+uuid.uuid4().hex[:12]
+        self.tmp=tempfile.TemporaryDirectory();self.addCleanup(self.tmp.cleanup)
+        self.db=Path(self.tmp.name)/'sample.sqlite';s=Store(self.db);seed(s);s.conn.close()
+        self.addCleanup(self.cleanup)
+        with redirect_stdout(io.StringIO()):migrate(self.db,self.service,Path(self.tmp.name)/'migration.json',schema=self.target,min_free_gib=0)
+
+    def cleanup(self):
+        with connect(self.service)as c:
+            for name, in c.execute('SELECT nspname FROM pg_namespace WHERE starts_with(nspname,%s)',(self.target,)).fetchall():
+                c.execute('DROP SCHEMA '+quote(name)+' CASCADE')
+
+    def test_native_repair_is_idempotent_keeps_codec_and_provenance(self):
+        p=Path(self.tmp.name)/'plan.json'
+        with redirect_stdout(io.StringIO()):
+            before=run(None,p,pg_service=self.service,pg_schema=self.target)
+            after=run(None,Path(self.tmp.name)/'applied.json',pg_service=self.service,pg_schema=self.target,plan=p,plan_sha256=hashlib.sha256(p.read_bytes()).hexdigest())
+        self.assertEqual(after['content_fingerprints'],before['content_fingerprints'])
+        with closing(open_store(pg_service=self.service,pg_schema=self.target).conn)as c:
+            self.assertEqual(apply_plan(c,before)['updated_editions'],0)
+            self.assertEqual(c.execute('SELECT abstract FROM papers ORDER BY id LIMIT 1').fetchone()[0],'Abstract\x00with\x01data')
+
+    def test_native_stale_source_plan_preserves_both_original_memberships(self):
+        with closing(open_store(pg_service=self.service,pg_schema=self.target).conn)as c:
+            before=plan(c)
+            c.execute("UPDATE source_records SET title='Changed' WHERE external_id=?",('conf/ijcai/fixture2022',));c.commit()
+            with self.assertRaisesRegex(ValueError,'Stale repair plan'):
+                apply_plan(c,before)
+            c.rollback()
+            self.assertEqual(inspect(c)['status'],'ready')

tests/test_submission_evidence.py

@@ -133,7 +133,7 @@ class SubmissionTests(unittest.TestCase):
         c.execute("INSERT INTO papers(id,title,title_key) VALUES(1,'Paper','paper')")
         c.execute("INSERT INTO source_records VALUES(1,1,'paper',1,'https://example.org/p','Paper','Abstract','[]','{}',CURRENT_TIMESTAMP)")
         c.execute("INSERT INTO venues VALUES(1,'ICLR','ICLR','machine_learning','conference')")
-        c.execute("INSERT INTO editions(id,venue_id,year,track) VALUES(1,1,2026,'main')")
+        c.execute("INSERT INTO editions VALUES(1,1,2026,'main')")
         c.execute("INSERT INTO submissions VALUES(1,1,1,1,'paper',NULL)")
         c.execute("INSERT INTO decisions VALUES(1,1,'accepted','Accept','https://example.org/decision','2026-01-01T00:00:00Z',NULL)")
         c.commit();c.close()