Compare commits

..

41 Commits

Author SHA1 Message Date
codex-bot 33bb378369 Merge main and preserve independent collector command registration
Metadata integrity checks / metadata (push) Successful in 3m20s
2026-11-17 20:27:11 +08:00
Zayden-Jung 7ed5966d3a Merge pull request '封装 ICDM2026 官方日程与明确轨道来源' (#60) from codex/icdm2026 into main
Metadata integrity checks / metadata (push) Successful in 3m10s
Reviewed-on: #60
2026-11-17 12:24:11 +00:00
Zayden-Jung 871b2aa81e Merge pull request '记录 SDM25 入库验收与第14步剩余来源缺口' (#59) from codex/sdm2026-acceptance into main
Metadata integrity checks / metadata (push) Successful in 3m14s
Reviewed-on: #59
2026-11-17 12:23:50 +00:00
codex-bot b14193063a Collect RecSys 2026 source observations with explicit duplicate DOI versions
Metadata integrity checks / metadata (push) Successful in 3m20s
2026-11-17 19:34:03 +08:00
codex-bot 2f1b5c959a feat: collect frozen ICDM2026 program with source track evidence
Metadata integrity checks / metadata (push) Successful in 3m11s
2026-11-17 18:11:42 +08:00
codex-bot f139c7e49d docs: record SDM acceptance and remaining 2026 source gaps
Metadata integrity checks / metadata (push) Successful in 3m14s
2026-11-17 17:48:24 +08:00
Zayden-Jung 6a89f5f6c2 Merge pull request '封装 ECCV2026 公开元数据与不可变断点缓存' (#56) from codex/eccv2026 into main
Metadata integrity checks / metadata (push) Successful in 5m55s
Reviewed-on: #56
2026-11-17 09:16:10 +00:00
Zayden-Jung cb3d97ed92 Merge pull request '记录 SDM2026 身份核查、历史保护与恢复入口' (#58) from codex/sdm2026-review into main
Metadata integrity checks / metadata (push) Successful in 3m22s
Reviewed-on: #58
2026-11-17 09:16:00 +00:00
codex-bot 06467835c7 docs: record SDM2026 identity evidence and recovery gates
Metadata integrity checks / metadata (push) Successful in 3m0s
2026-11-17 16:59:15 +08:00
codex-bot f2753bc741 Merge main and preserve both ECCV and SDM validation paths
Metadata integrity checks / metadata (push) Successful in 3m5s
2026-11-17 16:13:06 +08:00
Zayden-Jung 1ea17c810d Merge pull request '封装 SDM2026 注册论文与独立日程版本' (#57) from codex/sdm2026 into main
Metadata integrity checks / metadata (push) Successful in 2m53s
Reviewed-on: #57
2026-11-17 08:07:51 +00:00
codex-bot f800c6ecb4 fix: preserve already matching ECCV literal text before entity decoding
Metadata integrity checks / metadata (push) Successful in 3m8s
2026-11-17 16:07:06 +08:00
codex-bot 28255743ef Collect SDM 2026 registry records and separate program observations
Metadata integrity checks / metadata (push) Successful in 3m3s
2026-11-17 15:23:54 +08:00
codex-bot 279d6a2735 fix: preserve and validate encoded ECCV structured author text
Metadata integrity checks / metadata (push) Successful in 3m9s
2026-11-17 14:19:21 +08:00
codex-bot 8996a21e09 docs: record ECCV contribution types and rejected metadata access evidence
Metadata integrity checks / metadata (push) Successful in 3m19s
2026-11-17 13:49:32 +08:00
codex-bot e9bcbc3aef Preserve exact HTTP bytes when replaying ECCV fixtures
Metadata integrity checks / metadata (push) Successful in 3m7s
2026-11-17 05:20:10 +08:00
codex-bot c97ea94b49 Collect ECCV 2026 public metadata with immutable resumable cache
Metadata integrity checks / metadata (push) Failing after 2m44s
2026-11-17 05:12:31 +08:00
Zayden-Jung 59053fd4fb Merge pull request '封装CoRL2026官方名单及分字段档案来源' (#54) from codex/corl2026 into main
Metadata integrity checks / metadata (push) Successful in 2m54s
Reviewed-on: #54
2026-11-16 16:50:20 +00:00
codex-bot b50136d5bc Merge main into CoRL branch and preserve both conference guides
Metadata integrity checks / metadata (push) Successful in 2m52s
2026-11-17 00:46:21 +08:00
Zayden-Jung 87a0689fcc Merge pull request '封装ICAPS2026官方名单与独立注册元数据' (#55) from codex/icaps2026 into main
Metadata integrity checks / metadata (push) Successful in 2m49s
Reviewed-on: #55
2026-11-16 16:41:57 +00:00
codex-bot 9093a8ce25 Preserve separate ICAPS registry publication versions and abstracts
Metadata integrity checks / metadata (push) Successful in 2m51s
2026-11-17 00:30:14 +08:00
codex-bot e19cb83cd6 Align ICAPS collector entry point with shared CLI
Metadata integrity checks / metadata (push) Successful in 2m47s
2026-11-17 00:13:30 +08:00
codex-bot 4d2df2a826 Preserve ICAPS 2026 official acceptance and presentation boundaries
Metadata integrity checks / metadata (push) Failing after 2m30s
2026-11-17 00:08:49 +08:00
codex-bot 19d03ced54 Verify installed CoRL collector against all 687 source records
Metadata integrity checks / metadata (push) Successful in 2m52s
2026-11-16 22:59:19 +08:00
codex-bot c33f5422de Collect CoRL 2026 program with versioned public metadata provenance
Metadata integrity checks / metadata (push) Successful in 2m51s
2026-11-16 22:54:52 +08:00
Zayden-Jung 1cfd44d8b6 Merge pull request '为RSS发表记录保留跨来源的同次投稿身份' (#53) from codex/rss-publication-identity into main
Metadata integrity checks / metadata (push) Successful in 2m48s
Reviewed-on: #53
2026-11-16 08:13:30 +00:00
Zayden-Jung 5de7f6ff22 Merge pull request '封装RSS2026官方论文集采集与来源差异' (#52) from codex-bot/rss2026 into main
Metadata integrity checks / metadata (push) Successful in 3m1s
Reviewed-on: #52
2026-11-16 08:13:16 +00:00
codex-bot b4e249e675 Recognize accepted RSS publisher DOI as an edition-scoped publication identity
Metadata integrity checks / metadata (push) Successful in 2m55s
2026-11-15 23:07:11 +08:00
codex-bot 93f00d1a64 Preserve failed-run checkpoint when snapshot receipts are invalid
Metadata integrity checks / metadata (push) Successful in 2m44s
2026-11-15 22:47:27 +08:00
codex-bot 8b7bc6d871 Include frozen RSS 2026 source fixture and independent 210-record inventory
Metadata integrity checks / metadata (push) Successful in 2m54s
2026-11-15 22:43:21 +08:00
codex-bot 7588a4e904 Collect reviewed RSS 2026 publisher metadata with separate program variants 2026-11-15 22:43:05 +08:00
Zayden-Jung 891a63d982 Merge pull request '封装来源身份只读预检与投稿历史证据' (#51) from codex/source-identity-audit into main
Metadata integrity checks / metadata (push) Successful in 2m49s
Reviewed-on: #51
2026-11-10 06:55:25 +00:00
codex-bot fe3e2502a5 Add read-only source identity candidate audit with evidence shards
Metadata integrity checks / metadata (push) Successful in 2m42s
2026-11-10 02:55:27 +08:00
Zayden-Jung b675a38b97 Merge pull request '适配AAMAS2026官方目录并保留撤稿与撤回证据' (#49) from codex/aamas2026 into main
Metadata integrity checks / metadata (push) Successful in 2m45s
Reviewed-on: #49
2026-11-09 18:49:37 +00:00
Zayden-Jung cb875b8eef Merge pull request '封装NeurIPS2026节目采集,分离投稿与重复展示' (#50) from codex/neurips-program into main
Metadata integrity checks / metadata (push) Successful in 2m34s
Reviewed-on: #50
2026-11-09 18:49:29 +00:00
codex-bot 57a557c66b Merge current main and preserve scope and AAMAS documentation entries
Metadata integrity checks / metadata (push) Successful in 2m27s
2026-11-10 01:12:45 +08:00
codex-bot 5b5e606f0c Correct independent affiliation inventory and target section heading mutation
Metadata integrity checks / metadata (push) Successful in 2m22s
2026-11-10 01:09:25 +08:00
codex-bot 9f84fd7caf Separate italic title content from publisher status notices
Metadata integrity checks / metadata (push) Failing after 2m6s
2026-11-10 01:04:46 +08:00
codex-bot bf05d3cdc8 Keep day-only withdrawal evidence through native ingestion
Metadata integrity checks / metadata (push) Failing after 2m6s
2026-11-10 01:02:54 +08:00
codex-bot 930df2a4d5 Preserve explicit withdrawal and byte-exact AAMAS fixture evidence
Metadata integrity checks / metadata (push) Failing after 2m4s
2026-11-10 01:00:58 +08:00
codex-bot 37380e1d3b Collect reviewed AAMAS 2026 tracks and hold explicit retraction evidence
Metadata integrity checks / metadata (push) Failing after 2m10s
2026-11-10 00:57:09 +08:00
52 changed files with 4190 additions and 7 deletions
+127 -1
View File
@@ -12,7 +12,7 @@ jobs:
- name: Verify metadata storage and query commands
run: |
python3 -B - <<'PY'
import gzip, io, json, os, pathlib, re, subprocess, tarfile, tempfile, urllib.request
import base64, gzip, io, json, os, pathlib, re, subprocess, tarfile, tempfile, urllib.request
sha = os.environ['ARCHIVE_SHA']
if not re.fullmatch(r'[0-9a-fA-F]{40,64}', sha):
raise SystemExit('Invalid source commit')
@@ -48,6 +48,121 @@ jobs:
outside.mkdir()
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','collect','rss2026-proceedings','--help'],cwd=outside,check=True)
eccv_fixture = json.loads(gzip.decompress((src/'tests/fixtures/eccv2026-source.json.gz').read_bytes()))
eccv_official = root/'eccv2026-official'
eccv_papers = root/'eccv2026-paper-samples'
eccv_official.mkdir(); eccv_papers.mkdir()
for name, html in eccv_fixture['official_pages'].items():
(eccv_official/name).write_bytes(html.encode('utf-8'))
(eccv_official/'manifest.json').write_bytes(eccv_fixture['official_manifest'].encode('utf-8'))
for pid, html in eccv_fixture['paper_pages'].items():
(eccv_papers/(pid+'.html')).write_bytes(html.encode('utf-8'))
eccv_encoding = eccv_fixture['encoding_sample']
(eccv_papers/(eccv_encoding['paper_id']+'.html')).write_bytes(eccv_encoding['html'].encode('utf-8'))
eccv_fixture['paper_manifest']['records'].append({**eccv_encoding['receipt'], 'paper_id': eccv_encoding['paper_id'], 'file': eccv_encoding['paper_id']+'.html'})
eccv_literal = eccv_fixture['literal_title_sample']
(eccv_papers/(eccv_literal['paper_id']+'.html')).write_bytes(eccv_literal['html'].encode('utf-8'))
eccv_fixture['paper_manifest']['records'].append({**eccv_literal['receipt'], 'paper_id': eccv_literal['paper_id'], 'file': eccv_literal['paper_id']+'.html'})
(eccv_papers/'manifest.json').write_text(json.dumps(eccv_fixture['paper_manifest']))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','eccv2026-program','--official-snapshots',str(eccv_official),'--paper-snapshots',str(eccv_papers),'--out',str(root/'installed-eccv2026')],cwd=outside,check=True)
eccv_report = json.loads((root/'installed-eccv2026/collection-report.json').read_text())
assert eccv_report['records'] == 5 and eccv_report['expected_program_ids'] == 2834
assert eccv_report['duplicate_program_occurrences'] == 17 and len(eccv_report['uncollected_program_ids']) == 2829
assert not eccv_report['database_written'] and eccv_report['files'][0]['track'] == 'program_unclassified'
eccv_rows = [json.loads(line) for line in pathlib.Path(eccv_report['files'][0]['file']).read_text().splitlines()]
eccv_encoded_row = next(row for row in eccv_rows if row['external_id'] == '2026/4199')
assert eccv_encoded_row['authors'] == eccv_encoding['gold']['authors']
assert 'jsonld_display_text_normalized_requires_review' in eccv_encoded_row['metadata']['public_page_fields']['review_flags']
eccv_literal_row = next(row for row in eccv_rows if row['external_id'] == '2026/3844')
assert eccv_literal_row['title'] == eccv_literal['gold']['visible_title']
assert '&amp;' in eccv_literal_row['title']
sdm_fixture = json.loads(gzip.decompress((src/'tests/fixtures/sdm2026-source.json.gz').read_bytes()))
sdm_snapshots = root/'sdm2026-snapshots'
for name, text in sdm_fixture['files'].items():
target = sdm_snapshots/name
target.parent.mkdir(parents=True, exist_ok=True)
target.write_bytes(text.encode('utf-8'))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','sdm2026-proceedings','--snapshots',str(sdm_snapshots),'--out',str(root/'installed-sdm2026')],cwd=outside,check=True)
sdm_report = json.loads((root/'installed-sdm2026/collection-report.json').read_text())
assert sdm_report['records'] == 53 and sdm_report['missing_abstract_count'] == 53
assert not sdm_report['database_written'] and sdm_report['network_requests'] == 0
assert sdm_report['program_observations']['count'] == 53 and sdm_report['program_observations']['abstract_pages'] == 52
assert sdm_report['program_registry_author_sequence_differences'] == 20
sdm_rows = [json.loads(line) for line in pathlib.Path(sdm_report['files'][0]['file']).read_text().splitlines()]
assert len(sdm_rows) == 53 and all(row['abstract'] is None for row in sdm_rows)
sdm_observations = [json.loads(line) for line in pathlib.Path(sdm_report['program_observations']['file']).read_text().splitlines()]
assert sum(row['cancelled_presentation'] for row in sdm_observations) == 1
assert all('status' not in row and 'identifiers' not in row for row in sdm_observations)
icdm_fixture = json.loads(gzip.decompress((src/'tests/fixtures/icdm2026-source.json.gz').read_bytes()))
icdm_snapshots = root/'icdm2026-snapshots'
for name, text in icdm_fixture['files'].items():
target = icdm_snapshots/name
target.parent.mkdir(parents=True, exist_ok=True)
target.write_bytes(text.encode('utf-8'))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','icdm2026-program','--snapshots',str(icdm_snapshots),'--out',str(root/'installed-icdm2026')],cwd=outside,check=True)
icdm_report = json.loads((root/'installed-icdm2026/collection-report.json').read_text())
assert icdm_report['records'] == 324 and icdm_report['missing_abstract_count'] == 324
assert icdm_report['track_counts'] == {'research':241,'applied':83}
assert icdm_report['presentation_counts'] == {'long':168,'short':156}
assert not icdm_report['database_written'] and icdm_report['network_requests'] == 0
icdm_rows = [json.loads(line) for f in icdm_report['files'] for line in pathlib.Path(f['file']).read_text().splitlines()]
assert len(icdm_rows) == 324 and all(row['abstract'] is None and not row['identifiers'] and not row['submission_identifiers'] for row in icdm_rows)
assert all(row['metadata']['paper_type'] is None and not row['metadata']['original_decision_signature_verified'] for row in icdm_rows)
icaps_fixture = json.loads(gzip.decompress((src/'tests/fixtures/icaps2026-source.json.gz').read_bytes()))
icaps_snapshots = root/'icaps2026-snapshots'
icaps_snapshots.mkdir()
for name, html in icaps_fixture['pages'].items():
(icaps_snapshots/name).write_text(html)
(icaps_snapshots/'manifest.json').write_text(json.dumps(icaps_fixture['manifest']))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','icaps2026-program','--snapshots',str(icaps_snapshots),'--out',str(root/'installed-icaps2026')],cwd=outside,check=True)
icaps_report = json.loads((root/'installed-icaps2026/collection-report.json').read_text())
assert icaps_report['records'] == 90 and icaps_report['previously_published_presentations'] == 10
assert icaps_report['missing_abstract_count'] == 90 and not icaps_report['database_written']
assert icaps_report['files'][0]['track'] == 'main_tracks_unclassified'
registry_snapshots = root/'icaps2026-registry-snapshots'
registry_snapshots.mkdir()
(registry_snapshots/'crossref-filtered.json').write_bytes(gzip.decompress((src/'tests/fixtures/icaps2026-crossref.json.gz').read_bytes()))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','icaps2026-registry','--snapshots',str(registry_snapshots),'--out',str(root/'installed-icaps2026-registry')],cwd=outside,check=True)
registry_report = json.loads((root/'installed-icaps2026-registry/collection-report.json').read_text())
assert registry_report['records'] == 90 and registry_report['missing_abstract_count'] == 0
assert len(registry_report['excluded_volume35_dois']) == 1 and not registry_report['database_written']
rss_fixture = json.loads(gzip.decompress((src/'tests/fixtures/rss2026-source.json.gz').read_bytes()))
rss_snapshots = root/'rss2026-snapshots'
rss_snapshots.mkdir()
for name, html in rss_fixture['pages'].items():
(rss_snapshots/name).write_bytes(html.encode('utf-8'))
(rss_snapshots/'receipts.json').write_text(json.dumps(rss_fixture['receipts']))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','rss2026-proceedings','--snapshots',str(rss_snapshots),'--out',str(root/'installed-rss2026')],cwd=outside,check=True)
assert json.loads((root/'installed-rss2026/collection-report.json').read_text())['records'] == 210
recsys_fixture = json.loads(gzip.decompress((src/'tests/fixtures/recsys2026-source.json.gz').read_bytes()))
recsys_snapshots = root/'recsys2026-snapshots'
recsys_snapshots.mkdir()
for name,encoded in recsys_fixture['files'].items():
path = recsys_snapshots/name
path.parent.mkdir(parents=True,exist_ok=True)
path.write_bytes(base64.b64decode(encoded))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','recsys2026-program','--snapshots',str(recsys_snapshots),'--out',str(root/'installed-recsys2026')],cwd=outside,check=True)
recsys_report = json.loads((root/'installed-recsys2026/collection-report.json').read_text())
assert (recsys_report['records'],recsys_report['unique_dois'],recsys_report['missing_abstract_count']) == (263,260,0)
assert recsys_report['duplicate_doi_observations'] == 6 and len(recsys_report['repeated_dois']) == 3
assert not recsys_report['database_written'] and recsys_report['network_requests'] == 0 and recsys_report['approved_for_import'] == 0
recsys_rows = [json.loads(line) for manifest in recsys_report['files'] for line in pathlib.Path(manifest['file']).read_text().splitlines()]
gold_keys = recsys_fixture['independent_gold'][0].keys()
assert [{k:r['metadata']['program_fields'][k] for k in gold_keys} for r in recsys_rows] == recsys_fixture['independent_gold']
corl_fixture = json.loads(gzip.decompress((src/'tests/fixtures/corl2026-source.json.gz').read_bytes()))
corl_official = root/'corl2026-official'
corl_archive = root/'corl2026-archive'
corl_official.mkdir()
corl_archive.mkdir()
(corl_official/'accepted.html').write_text(corl_fixture['official_html'])
(corl_official/'research.json').write_text(json.dumps(corl_fixture['research']))
(corl_archive/'explorer-paper-metadata-filtered-v1.json').write_text(corl_fixture['filtered'])
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','corl2026-program','--official-snapshots',str(corl_official),'--archive-snapshots',str(corl_archive),'--out',str(root/'installed-corl2026')],cwd=outside,check=True)
corl_report = json.loads((root/'installed-corl2026/collection-report.json').read_text())
assert corl_report['records'] == 687 and corl_report['presentations'] == 780
assert len(corl_report['title_variants']) == 34 and corl_report['public_rejected_availability'] == 'unknown'
assert not corl_report['database_written'] and not corl_report['pdfs_downloaded']
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','www-archive','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','aamas-proceedings','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','neurips-program','--help'],cwd=outside,check=True)
@@ -62,6 +177,16 @@ jobs:
assert not neurips_report['database_written'] and not neurips_report['pdfs_downloaded']
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','dblp-family','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','aamas-proceedings','--year','2025','--snapshots',str(src/'tests/fixtures/aamas2025'),'--out',str(root/'installed-aamas2025')],cwd=outside,check=True)
aamas2026 = json.loads(gzip.decompress((src/'tests/fixtures/aamas2026-fixture.json.gz').read_bytes()))
aamas2026_snapshots = root/'aamas2026-snapshots'
aamas2026_snapshots.mkdir()
(aamas2026_snapshots/'contents.html').write_text(aamas2026['contents'])
(aamas2026_snapshots/'receipts.json').write_text(json.dumps(aamas2026['receipts']))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','aamas-proceedings','--year','2026','--snapshots',str(aamas2026_snapshots),'--out',str(root/'installed-aamas2026')],cwd=outside,check=True)
aamas2026_report = json.loads((root/'installed-aamas2026/aamas-proceedings-2026.json').read_text())
assert aamas2026_report['records'] == 639 and len(aamas2026_report['held_publications']) == 1
assert aamas2026_report['status_counts'] == {'accepted': 638, 'withdrawn': 1}
assert not aamas2026_report['database_written'] and not aamas2026_report['pdfs_downloaded']
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','iclr-archive','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','acl','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','springer-selected','--help'],cwd=outside,check=True)
@@ -86,6 +211,7 @@ jobs:
subprocess.run([python,'-I','-B','-m','paper_atlas','audit','public-openreview','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','coverage','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','audit','histories','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','audit','source-identities','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','curate','identifiers','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-c','from paper_atlas.postgres.writer import PostgresStore, WriteConnection'],cwd=outside,check=True)
smoke = "from paper_atlas.db.store import Store; from paper_atlas.config import DEFAULT_REGISTRY; import json; assert {'COLM','MM','MLSys'} <= {v['acronym'] for v in json.loads(DEFAULT_REGISTRY.read_text())['venues']}; s=Store('installed.sqlite'); s.conn.close()"
+22
View File
@@ -8,6 +8,8 @@ Paper Atlas 将 arXiv、AI 会议论文集及公开投稿决定整理到可追
当前主动补漏从 **2020 年**起;2000 年起已发现的名单、缺口与待办保存在[补漏恢复清单](docs/coverage/backfill-handoff-2026-11-10.md),下次从这里接续。已有旧数据保留,范围排除不算补齐。
AAMAS 2026 的官方 HTML 目录由 `collect aamas-proceedings --year 2026` 适配,区分 track 和发表形式,keynote 排除、撤稿证据单独保留;[采集与恢复说明](docs/aamas-proceedings.md)。采集不写数据库,仍需身份核验和增量导入。
历史 ICLR 2014–2016 官方录用页由 `collect iclr-archive` 适配,保留主会/workshop及重复展示证据;运行和身份核验边界见 [历史档案说明](docs/coverage/iclr-archive-2014-2016.md)。
`collect springer-selected` 核验限定的会后选集,分开保存会议年份、出版年份及选集事件身份。目前支持 AISTATS 1993 对应的 1994 年出版选集,原始会议决定仍须单独核查;配置、证据和恢复见 [会后选集说明](docs/springer-selected-volumes.md)。
@@ -114,3 +116,23 @@ CI 在已授权的远程 runner 上运行测试、构建 wheel,并从源码目
整合后的运行版本依赖、交互约束和生产切换顺序见 [运行版本整合](docs/research-runtime.md)。
已导入整卷的逐条轨道修正见 [edition-tracks 维护流程](docs/edition-tracks.md):按完整证据生成只读计划,保留原投稿和决定 ID,事务性修正 edition、标识与覆盖,并防止旧整卷刷新覆盖分类。
入库前可使用通用的[来源身份只读预检](docs/source-identity-audit.md),导出候选论文、稳定标识和既有投稿决定证据;候选结果需要审核,不会自动批准导入。
RSS 2026 已独立适配官方论文集与录用表;使用方法和与 2025 demo 分类的区别见 [RSS 2026 来源说明](docs/rss2026-proceedings.md)。
ICAPS 2026 的[官方名单采集器](docs/icaps2026-program.md)区分 90 篇主会轨道组录用与 10 篇既有论文展示;逐篇 track、摘要和稳定标识尚未解决,输出均待审核,不能直接视作已完成入库。
ICAPS 2026 也提供独立的 `collect icaps2026-registry`,保留 Crossref 卷 36 的 DOI、作者与摘要版本;不覆盖节目表文字,不把卷 35 或版本差异自动并入。配置与边界见同一份[ICAPS 来源说明](docs/icaps2026-program.md)。
### CoRL 2026 官方程序快照
`collect corl2026-program` 将官方 687 个论坛与固定公开档案按 ID 对照,保留 34 条标题版本差异及分字段来源。不会自动接受 arXiv 匹配或推断拒稿;详情见 [CoRL 2026 使用与恢复说明](docs/corl2026-program.md)。
ECCV 2026 的[暂定节目采集器](docs/eccv2026-program.md)合并重复节目 ID,并以不可变缓存断点采集公开元数据;保留摘要原文、来源关键词及未分类 track,始终待身份审核,不自动入库。
`collect sdm2026-proceedings` 核验 53 个 Crossref 论文章节,并单独保留 53 项日程观察及 52 页摘要。作者差异、取消展示和摘要缺失分别保留,不自动拼接来源或绑定作品;见 [SDM 2026 使用与恢复说明](docs/sdm2026-proceedings.md)。
`collect icdm2026-program` 离线核验 324 条 ICDM2026 官方线下日程,按明确的研究/应用标签输出 241/83 条;展示长短、缺失摘要及未核实出版身份分别保留,不自动入库。见 [ICDM2026 使用与恢复说明](docs/icdm2026-program.md)。
`collect recsys2026-program` 核验九个 track 的 263 项官方录用贡献观察(260 个不同 DOI),保留摘要、原始作者和重复 DOI 的来源版本。见 [RecSys 2026 使用与恢复说明](docs/recsys2026-program.md)。
+26 -1
View File
@@ -1,6 +1,31 @@
# AAMAS 官方论文集元数据
`paper_atlas.collectors.aamas` 是按需运行的采集适配器,通过统一入口 `collect aamas-proceedings` 使用。支持已独立核验的 **AAMAS 2012、2025** HTML 版式;其他年份需先核对官方目录和逐篇结构,再增加适配,不能仅替换年份。2025 解析单独封装在 `paper_atlas.collectors.aamas2025`。
`paper_atlas.collectors.aamas` 是按需运行的采集适配器,通过统一入口 `collect aamas-proceedings` 使用。支持已独立核验的 **AAMAS 2012、2025、2026** HTML 版式;其他年份需先核对官方目录和逐篇结构,再增加适配,不能仅替换年份。2025、2026 解析分别封装在 `paper_atlas.collectors.aamas2025`、`paper_atlas.collectors.aamas2026`。
## AAMAS 2026
2026 只读取[官方 HTML 目录](https://www.ifaamas.org/Proceedings/aamas2026/forms/contents.htm),不下载 PDF。冻结目录 SHA256 `ce2bc7fc5da340930343901506a9ce37c0334d4068f07546431c9deee3f13de9`,643 个条目:keynote 3、research 338、extended abstract 185、AAAI 19、Blue Sky 17、doctoral 34、demo 44、JAAMAS 3。
排除 3 个 keynote;research 的 p238 明确标注 `Retracted June 8, 2026`,完整题名、作者、栏目、日期、原文和来源回执保存在采集报告 `held_publications` 中,**不会输出 accepted 或 withdrawn 投稿记录**。当前决定模型没有独立 retracted 状态,后续须单独建模和核验;原始证据不能删除。任何新增或改变的撤稿标记都会停止采集,等待人工核查。
另一个 extended abstract p3178 明确标注 `Withdrawn June 8, 2026`,输出 `withdrawn`,在 metadata 保留日期 `2026-06-08` 和日精度;原网页没有时区/时刻,不能伪造决定 timestamp,不补造其此前的接受决定或 PDF 链接。因此 639 个输出条目是 **638 accepted、1 withdrawn**,按六个 track 分组:`proceedings_unspecified` 522、`aaai_transfer` 19、`blue_sky` 17、`doctoral` 34、`demo` 44、`jaamas` 3。全文/扩展摘要是发表形式,不是研究类型;research 栏目不能证明每篇原始 main/special 投稿轨道。JAAMAS 保留期刊展示语义,不制造另一次独立审稿录用。
[AAAI 转投详细政策](https://cyprusconferences.org/aamas2026/call-for-papers-aaai-track/) 要求基于 AAAI-26 rejected 稿件;[总览](https://cyprusconferences.org/aamas2026/calls/) 却写 AAAI-25,须保留冲突。采集器保留 AAAI 栏目归属,不据此生成原 AAAI 投稿 ID、年份或逐篇拒稿决定。目录不是原始决定 note,未来关联仍需逐篇证据。
```sh
paper-atlas collect aamas-proceedings --year 2026 \
--snapshots /data/paper-index/collected/aamas2026-primary-evidence-v1 \
--out /data/paper-index/collected/aamas2026-normalized-v1
```
在线模式将 `--snapshots` 换成 `--cache /data/paper-index/cache/aamas2026`。冻结模式 `receipts.json` 的 `receipts` 至少包含官方目录 URL、相对 `file`、`bytes`、`sha256`、`status=200` 和带时区 `finished_at`。拒绝越界路径、越界符号链接、失败状态、哈希/大小不符的页面。这里只读取目录对应回执,其他已冻结政策网页继续保留。
输出六组 JSONL/manifest、原始 HTML 与最后发布的 `aamas-proceedings-2026.json`;统一入口沿用输出目录锁和暂存目录。manifest 全部为 `partial`,639 条 HTML 摘要缺失如实记录。发布编号使用年份和起始页,随机 PDF 文档 ID 单独保留,不能沿用 2025 的 pN.pdf 规则。
首次部署先核对来源哈希和生产状态,再在新输出目录运行采集、独立比对 643 条标题/作者/栏目分区,随后做 PostgreSQL 身份预检与增量导入。采集、远程 CI 通过或目录含某个年份均不代表生产已导入、摘要齐全或最终去重完成。较早年份不自动启动补采,当前主动补漏范围按用户要求从 2020 年起。
## AAMAS 2025
2025 版式使用[官方首页](https://www.ifaamas.org/Proceedings/aamas2025/)和[出版目录](https://www.ifaamas.org/Proceedings/aamas2025/forms/contents.htm)。原始目录共 479 条:3 keynote、260 research paper、148 extended abstract、12 Blue Sky、25 doctoral、22 Demo 栏目、9 JAAMAS。排除 keynote,输出 476 条;Demo 栏目中的 p3000 是明确标题的 COMP24 竞赛报告,发表年份仍为 2025,格式记 `competition_report`、投稿 track 留作 `proceedings_unspecified`,保留原 Demo 栏目证据。其余输出按 track 分组为 409 unspecified、12 blue_sky、25 doctoral、21 demo、9 jaamas。
+24
View File
@@ -0,0 +1,24 @@
# CoRL 2026:官方程序与公开档案
这是长期维护、按需运行的来源适配器。命令只读已冻结的元数据快照,生成待核验文件;不连接数据库,不抓取 PDF、评审或作者个人档案,不执行网页脚本,也不尝试通过 OpenReview 验证挑战。
```sh
paper-atlas collect corl2026-program \
--official-snapshots /data/paper-index/collected/corl2026-official-listing-research-v1 \
--archive-snapshots /data/paper-index/collected/corl2026-public-archive-inventory-v1 \
--out /data/paper-index/collected/corl2026-normalized-v1
```
官方快照包含 `accepted.html` 和 `research.json` 的 HTTP 回执。适配器解析 Google Sites 的 `data-code`,不把外层文本提取为空误作无论文。687 个论坛有 780 次展示:687 次海报、21 次口头、72 次 Spotlight。论坛 ID 确定投稿身份,展示形式不是新的投稿或 track。
作者及摘要来自 `gisbi-kim/conference-explorers` 的固定提交 `75b3f7a04b3a77c454540fe2506bfa615b8e038e`,经字段过滤后的 `explorer-paper-metadata-filtered-v1.json`。输入、上游原始数据 SHA-256 及 Git blob 均固定;改变来源需要另行研究、更新快照及检查,不能静默接受。
这是独立整理者的版本证据,并非本次重新核验的 OpenReview 原页。每条保留档案标题、作者、摘要、分字段来源 URL、官方标题及全部展示场次。34 个官方与档案标题差异会显式标记,不能视作勘误或自动作者别名。规范记录中的标题与作者摘要保持档案版本一致;官方标题单独保留,避免拼成没有来源的版本。
档案的 arXiv 链接和匹配标题只存为未批准候选,不进入作品标识;不沿用其推断机构、个人档案、作者 ID 或生成标签。只有 OpenReview 论坛 ID 进入作品和投稿标识。
所有输出均为 partial。官方政策只证实录用论文及评审公开,没有证实 rejected 集可见性。不能由此生成拒稿记录,也不能声称这 687 条覆盖了公开拒稿或全部论文关系。
输出目录必须全新;失败留下 `failed.json`,成功报告最后原子发布。恢复时先检查原目录,不覆盖失败现场或重跑已消费的导入队列。生成文件仍需元数据版本审阅、当前数据库身份候选检查、独立保护基线及导入后核验,最后另做全库去重。
测试和安装验证只在用户授权的 `ces-control:host` runner 上运行。
+2
View File
@@ -1,5 +1,7 @@
# 会议覆盖与公开决定核查
最新生产进展和剩余工作见 [2026-11-17 补漏状态](backfill-status-2026-11-17.md):固定 278 个会议年份槽位中,270 个有记录、8 个仍为空;这不代表完整覆盖,第 14 步仍未完成。
**当前工作范围:2020 年起主动补漏。** 2000 年起已发现资料保留,恢复工作从 [补漏范围与恢复清单(2026-11-10)](backfill-handoff-2026-11-10.md) 开始;下文旧日期快照不改写,较早年份排除不算补齐。
这是 2026-11-08 的首份核查基线,**完整逐年政策普查和最终去重仍未完成**。收录 43 个目标,不能把“有记录”当作“完整覆盖”。结构化清单见 [inventory-2026-11-08.json](inventory-2026-11-08.json)。清单中的 `policy_checks` 均限定所列年份和轨道;空数组表示尚未核实,绝不表示没有公开拒稿。
@@ -0,0 +1,47 @@
# 2020 起补漏状态:2026-11-17
**总体仍为 13/17,第 14 步未完成。** 本页补充新的生产结果,不回写 [2026-11-10 恢复清单](backfill-handoff-2026-11-10.md) 或更早快照。用户确定的范围不变:2020 年起主动补漏;43 个更早历史缺口延期而非补齐,2000–2019 已发现的名单和身份线索继续保留。
## 最新数据库盘点
只读快照 `2026-11-17 09:38:31.677743+00`,固定 43 个目标会议、278 个应有会议年份槽位,其中 270 个至少有一条投稿记录,8 个仍为空,均为 2026 年:**CIKM、ECCV、ICASSP、ICDM、IROS、ISMIR、MM、RecSys**。本次 SDM2026 新增记录使空槽从 9 减到 8。
这只说明有无记录,**不表示 270 个槽位已经逐篇、逐 track、摘要或公开拒稿覆盖完整**;不能据此宣布第 14 步完成。2020–2025 没有整年空槽,仍有逐来源完整性和公开负面决定缺口。SDM2026 仅导入 25/53 条,状态为 partial,同样不能称为整届补齐。
| 生产文件(相对 `/data/paper-index/`) | SHA-256 |
| --- | --- |
| collected/active-scope-census-20261117-v17/summary.json | bfb1db8b50619c3756abb91ad0f73fa58ff998c7db58a216661342f12047749f |
| collected/active-scope-census-20261117-v17/snapshot.json | b5389205f8b34f7ed57946416f49c253e0aed4a05e45eb818801f18fca6a8cb4 |
| validation/active-scope-fixed278-v1.json | a835bef5b9c3a0f76069d840cecdb9a6bcf65d3a38796d70dbfdfd4b49242264 |
## 已推进的两批来源
- **SDM2026**:53 个正式注册章节、53 个独立日程观察和 52 个公开摘要页已冻结并规范化。25 条发表记录已在 PostgreSQL 独立验收,新增论文实体为 0;28 条待身份/版本核查。完整旧历史和 5,653 篇受保护论文的资料、来源、标签和既有标识保留。见 [生产验收与恢复](../sdm2026-identity-review.md)。这批队列已经消费,禁止重放。
- **ECCV2026**:2,834 个唯一初步日程条目及公开摘要已全量采集并独立验收;新的只读身份预检和 1,997 个候选历史证据包已保存,尚未入库。初步日程不等于已核验出版社最终论文集;负面决定元数据仍受匿名访问限制。见 [来源状态](../eccv2026-source-status.md)。
SDM 新增事件使 SeBA 和 AdamG 同时具有较早 ICLR 拒稿与后来的 SDM 发表事实。旧的两次 ICLR 投稿及四条决定来源观察仍保留,没有被较晚 accepted 覆盖;多个来源观察不直接累加成多次拒稿。
## 其余七个会议的最新入口核查
已对下表七个会议的 14 个明确官方 URL 保存原始响应和回执:13 个 HTTP 200,IROS 官方论文索引 HTTP 403。所有请求均匿名,不绕过访问限制,没有下载论文 PDF;保存入口不是完成逐篇采集或负面决定普查。
| 会议 | 已保存的明确官方入口 | 尚未完成 |
| --- | --- | --- |
| ICDM | [详细日程](https://icdm2026.neu.edu.cn/DetailedSessions/list.htm)、[研究轨道政策](https://icdm2026.neu.edu.cn/CallforResearchTrack/list.htm)、[应用轨道政策](https://icdm2026.neu.edu.cn/CallforAppliedTrack/list.htm) | 官方页面称为线下展示列表;需要核对逐篇范围、研究/应用轨道及演讲长短标签,不能把展示长短自动等同于论文类型;摘要和正式出版身份仍待核查 |
| ICASSP | [录用标题清单](https://cmsworkshops.com/ICASSP2026/papers/accepted_papers.php)、[作者指南](https://2026.ieeeicassp.org/author-guidelines/) | 取得逐篇完整作者/摘要/出版记录,区别会议投稿与期刊展示,不能只导入标题或从缺席推断拒稿 |
| IROS | [官方论文索引](https://2026.ieee-iros.org/program/paper_index/) | 当前服务器返回 403,保留缺口并停止该入口;第三方浏览器页面不能替代独立核验的官方原文 |
| ISMIR | [官网](https://ismir2026.ismir.net/)、[官方虚拟平台](https://ismir2026program.ismir.net/)、[作者指南](https://ismir2026.ismir.net/authors/author-guidelines) | 核查平台公开静态数据与正式开放论文集的逐篇映射;首页可读取不代表论文目录已经取得 |
| CIKM | [录用页](https://cikm2026.diag.uniroma1.it/accepted-papers/)、[日程页](https://cikm2026.diag.uniroma1.it/program/) | 网页提取的录用页仍显示 TBD;需要检查原文引用的节目数据与独立正式出版目录,不能当成没有论文 |
| RecSys | [官网](https://recsys.acm.org/recsys26/)、[投稿说明](https://recsys.acm.org/recsys26/call/) | 继续取得可核验的逐篇正式记录,并分开主轨、工业、复现、demo、challenge/workshop |
| MM | [官网](https://2026.acmmm.org/) | 继续核对官方完整日程和出版目录,分开研究论文与 grand challenge、workshop 等;首页响应不证明覆盖完整 |
冻结目录:`/data/paper-index/collected/remaining2026-official-leads-v1`。manifest SHA-256:`0e13313d94e3bff17f071d3a1384f6ed83ece156239fe5bc4633cc7a34fb8e04`。恢复包:`/data/paper-index/validation/remaining2026-official-leads-v1-recovery.tar.gz`,SHA-256:`0be8c84bb5c591e5afaaa5f95dd5334bf13a9be1671cb942e46e998f502495bb`。HTTP 成功仅是入口可读性,不能作为可导入记录数量。
## 第 14 步仍需做什么
1. 取得并规范化其余可公开取得的完整逐篇来源,记录应有范围、排除项和版本;对缺失、受限或尚未公开的入口保存明确证据,不能伪造“已补齐”。优先推进当前可读的 ICDM 详细日程、ICASSP 清单及其他会议的正式目录。
2. 完成来源与年度、track、摘要完整性的对应核查;已经有若干记录的年份也要看目录差集,不能只消除空槽。
3. 继续近期和历史年度实际公开的 rejected、desk_rejected、withdrawn 来源核查,特别是 UAI2024–2026 的 opt-in 拒稿、EMNLP2023 的实际公开范围以及其他会议的逐年限定政策。公开评审和公开决定不同;没有找到不等于零条或未曾拒稿。
4. 通过身份核查的记录再执行新快照守护、恢复备份和增量导入,保留较早决定。深层作品版本、跨会历史及类型/标签协调继续列为第 15 步;不能为减少空槽创建重复论文,也不能只盯住少量身份疑难项而停止其他来源采集。
第 16 步仍是所有导入结束后的独立全库去重与完整审计,第 17 步是最终长期运行/恢复和交付整理。上述任务均未因单批验收或 CI 成功而完成;总体 goal 继续保持 active。
+6
View File
@@ -342,3 +342,9 @@ ICML 2024 起的一个 PMLR 卷可以同时包含研究主轨道与 Position Pap
2024 年的官方节目单把两轨都放在 `Conference`,但提供 PMLR 论文自链接。该年仅采用组织方在 [2025 Position CFP](https://icml.cc/Conferences/2025/CallForPositionPapers) 明确链接的 [2024 Position 论文标题筛选](https://icml.cc/virtual/2024/papers.html?filter=titles&search=position%3A) 规则;同时保存规则页面的哈希。这个规则不用于其他会议或年份。官方节目单将 `guyard24a` 链接误写为 `theo24a`:只对这条已核查的映射额外要求标题精确一致、至少两名公开作者重合,不进行全库模糊匹配。
导入与身份回填拒绝 ICML 2024 起 parser_version<3 或没有官方轨道证据的旧文件。已有主库中的错误轨道需要在独占维护锁内迁移原投稿事件、保留决定与来源关联;仅重新导入拆分文件会留下旧的主轨道事件,不能作为修复方式。完成迁移前,不把新文件直接导入现有库。
### RSS 发表身份
已核验的 `rss_proceedings` 录用记录与其他出版社来源一样,将 DOI 作为届次范围内的 `publication_doi` 投稿身份。这样同一 RSS 出版物先由官方论文集、后由 DBLP 导入(或反向)时可以共享一次投稿,并保留各自来源和决定观察。此规则不把节目显示编号当原始 OpenReview 投稿号,不把拒稿/撤稿升级为录用,也不跨届次合并历史。
这不证明任意 DOI 的科学作品归属:导入前仍需核查原始标题、完整作者、版本和已有标识冲突。变更只作用于后续导入;不会自动回填、合并或改写旧 RSS 记录,旧记录需要独立审查计划。
+35
View File
@@ -0,0 +1,35 @@
# ECCV 2026:暂定节目与公开元数据
这是长期维护、按需运行的来源适配器。它读取已核验的官方节目快照,采集公开论文页的标题、完整作者和摘要,输出待身份审核文件;不连接数据库,不下载 PDF、评审、个人档案或受保护的直播内容。
```sh
paper-atlas collect eccv2026-program \
--official-snapshots /data/paper-index/collected/eccv2026-official-research-v1 \
--cache /data/paper-index/cache/eccv2026-public-pages \
--max-new-pages 200 \
--out /data/paper-index/collected/eccv2026-pass01
```
官方目录包含 `accepted.html`、`policies.html`、`faq.html` 和冻结的 `manifest.json`。当前适配器核验固定版本:节目 2,851 行、2,834 个节目 ID;17 个 ID 各重复一次,标题、作者、关键词、展示信息均一致。输出按节目 ID 合并这些重复行,并保留全部行序号。它不是跨来源作品去重。
`--max-new-pages` 限制本轮新请求数;不指定时遍历整个冻结名单。再次运行时使用同一 `--cache` 和**全新** `--out`,复用已验证缓存并继续读取缺失页。缓存不按 24 小时过期,不自动刷新或覆盖已成功的原始响应。每次运行输出包含本轮能核验的全部缓存条目,并明确列出尚未采集的 ID;不能未经审核把多轮重叠输出全部重复导入。
公开请求按固定来源和路径限制,只接受 HTTP 200、原 URL、HTML 和限定大小的完整响应。请求失败、重定向、访问挑战、非 HTML 或异常响应留下阻断标记,不自动重试或更换账号、域名、代理绕过限制。记录的 `Retry-After` 供人工核查使用。网络超时等未产生成功缓存的情况可在确认进程终止后用新输出目录继续。缓存数据与回执须成对且哈希一致;遇到孤立文件或 `.part` 时保留现场,先核查再把本任务的残留文件移至恢复目录,不静默丢弃。写同一缓存时持有非阻塞单写锁。
离线重放可改用 `--paper-snapshots /path/to/papers`,该目录包含 `<节目ID>.html` 和 `manifest.json`(带时区的 `at`、`records`,各条含文件、URL、最终 URL、HTTP 状态、类型、字节数、SHA-256)。缺少的公开页会明确列为未采集;三篇样本只会生成三条 partial 记录,不会冒充全量成功。
## 来源边界
[官方名单](https://eccv.ecva.net/Conferences/2026/AcceptedPapers)明确注明 **preliminary pending publisher checks**。这里的 accepted 表示冻结时列入暂定官方名单,不代表已核验 Springer 最终发表或原始签名决定。所有输出始终 partial。节目 ID 是官方节目定位符,不冒充原始 OpenReview 投稿号;尚未核实跨平台标识时作品标识和投稿标识均为空。
公开页的 canonical URL、可见标题/作者和 JSON-LD 相互核对;页面版本与冻结节目文字有差异时分别保留并标记待审阅,不制造人名别名。JSON-LD 日期只是网站条目时间,不当作论文发表日期或原始决定日期。摘要保留公开页 DOM 文本(规范空白),包括源头的断词、缺空格;不自动修补。原始 HTML 单独留存;遇到公式、上下标或图片结构会标记待审阅,不能把扁平化文本当作无损数学表达。页面上的登录提示对应直播内容,不用于访问受保护内容。
第 4199 页的 JSON-LD 作者姓名存在双重 HTML 实体编码;第 3844 页的可见标题和 JSON-LD 则已经一致地包含字面 `&amp;`。解析器优先保留已经一致的文字,仅在字段不一致时尝试最多两层实体解码来核对可见文本,不把已经一致的标题再次解码。完整作者序列仍须精确一致,真实差异或更多编码层次仍停止。`jsonld_raw_text` 保留原始结构化标题/姓名,发生文本规范化时增加 `jsonld_display_text_normalized_requires_review`,不生成姓名别名。恢复必须使用校验过的修复版本、新输出目录和原缓存;保留旧失败目录,不重新请求已有成功页面。
官方关键词保存在独立 `eccv_program:keyword` 命名空间,并保留原始单元格和证据。关键词、展示形式、贡献/研究类型、自定义标签相互独立,投稿 track 暂用 `program_unclassified`,不按主题推断 main、findings、workshop 或 position。
[2026 投稿政策](https://eccv.ecva.net/Conferences/2026/SubmissionPolicies)明确评审和作者回复不公开;这不能代替对公开 rejected、desk rejected、withdrawn **决定元数据**的单独核查。当前拒稿决定可见性仍记为 unknown,不能从录用表缺席推断拒稿。
贡献类型的官方定义、OpenReview 状态入口与匿名访问核查见[来源核查记录](eccv2026-source-status.md)。这是独立于冻结采集输入的证据,不会改变已核验节目版本,也不能用类型定义替论文推断标签。
失败输出保留 `failed.json`,成功报告最后发布;恢复前核查既有进程、输出及缓存,不因日志安静重启。身份、来源版本和完整投稿历史审阅后,另行准备专用入库队列、保护基线和导入后验收;全部来源导入结束后仍须全库去重、覆盖、摘要、约束及查询审计。测试、构建及安装验证只在已授权 `ces-control:host` runner 执行。
+59
View File
@@ -0,0 +1,59 @@
# ECCV 2026 来源核查记录
核查时间:2026-11-17,以下状态是当时的公开可见性。长期补漏时先核实来源是否变化,保留旧快照,不覆盖旧结论或重复导入。
## 贡献类型与 track
[作者贡献类型说明](https://eccv.ecva.net/Conferences/2026/AuthorContributionTypes)通过官方导航确认。它要求作者选择一项主要贡献类型;官方 FAQ 明确贡献类型不是独立 track。公开 OpenReview group 的 `source_submissions_query_mapping` 也包含五项对应值:
| 官方 group 字段值 | 含义摘要 |
| --- | --- |
| Algorithms/General | 方法、模型或算法;也是边界情形的默认选择 |
| Theory/Foundational | 理论、形式分析或基础理解 |
| Applied/Systems | 实际系统、部署与工程约束 |
| Datasets/Benchmarks | 数据集、基准或评估协议 |
| Concept & Feasibility | 新概念及可行性证据 |
这些是类型定义和配置值,**尚未核验逐篇论文的作者所选类型**。不能根据标题、摘要、主题关键词或展示形式补写类型;公开节目采集器继续输出 `program_unclassified`,主题关键词继续使用 `eccv_program:keyword`。以后获得逐篇类型证据时,保留独立命名空间、原始值、来源和版本,不把它写成投稿 track。
## 公开负面决定元数据
[投稿指南](https://eccv.ecva.net/Conferences/2026/AuthorGuide)链接的会议 group 是 [`thecvf.com/ECCV/2026/Conference`](https://openreview.net/group?id=thecvf.com/ECCV/2026/Conference)。匿名 API 读取 group 成功(HTTP 200),返回 `readers: ["everyone"]`。这只说明 group 配置公开,不证明论文或决定公开。
group 的 `rejected_venue_id` 实际配置为 `thecvf.com/ECCV/2026/Conference/Rejected_Submission`。使用该值查询公开 notes(`content.venueid`,`limit=1`)返回 HTTP 403,随后停止;没有自动重试、切换账号、代理或其他绕过操作。其余 desk-rejected、withdrawn 和 submission 状态查询没有执行,不能报告为已核查或零条。
结论仍为 **拒稿决定元数据覆盖未核实,当前匿名查询受限**。403 的具体原因未确定;不能由此推断真实投稿数量、全局不存在公开拒稿数据,或某篇论文遭拒。[投稿政策](https://eccv.ecva.net/Conferences/2026/SubmissionPolicies)关于评审与作者回复不公开的规定也不能代替决定元数据核查。
## 可恢复证据与后续入口
服务器证据目录:`/data/paper-index/collected/eccv2026-policy-research-v2`。本机工作记录目录保留同名备份;三份成功响应均已逐文件校验。证据不含登录凭据。
| 文件 | SHA-256 |
| --- | --- |
| manifest.json | 5a04e52bf6ea34ab25fe1d91fa9a1df18c9be5abb4f554586ee28fe9aa566e6e |
| contribution-types.html | 6a5128b1f47a275411acacbb08ab91320a767bdcc48aaad09fe1be9e31dff00f |
| author-guide.html | 779196716f5b37fcd10ddc191e9cd0530d7e9d43e5bc66b16a5546cd152959c4 |
| openreview-group.json | 055bcf99d412d77f79a760890739d4de13a20ea099857154ee3e714c0647e763 |
| anonymous-decision-probes/manifest.json | a522756ed3ec2880e5d942a211d7ad13819b18c8acada551dd3a8718b5e4d63b |
这里的补充证据与采集器的三份冻结节目/政策输入分开保存,不能替换采集器固定版本的 manifest。下次补漏先核查官方是否提供公开决定导出或明确的公开元数据说明;访问仍受限时保留缺口。若发现允许公开访问的新入口,另存带时间与哈希的快照,先审阅读者权限、决定语义和逐篇身份,再准备入库队列。
## 2026-11-17 全量采集与只读预检
第三次采集 `paper-atlas-eccv2026-collect-pass03.service`(invocation `147c8779a3af4223a4d5d60dbdc88264`)已终态成功:2,834 个唯一日程条目全部取得公开摘要,新取 1,775 页,复用已验收缓存 1,059 页;2,851 次日程出现中的 17 处重复保留原始位置。独立原文/回执/字段审计通过,缺少条目和摘要均为 0,5 处 JSON-LD 显示文字规范化仍留待逐篇审阅。这里仍是官方初步日程版本,不改称出版社最终论文集。
| 服务器证据 | SHA-256 |
| --- | --- |
| validation/eccv2026-pass03-audit-v1.json | ad90199e08a234c0b2a01f44e1edfbecb2a9101b45de87e00a64eadfd5c92848 |
| collected/eccv2026-pass03/eccv-program-2026-unclassified.jsonl | d0a80745fc6f58163f8722af351bbba86803af97857ea30554a0f3062d8ab2ef |
| validation/eccv2026-pass03-recovery-v1.tar.gz | df55493ad5f8d6413dca86e9b9aca1d8b544ac1dc222b3849f95b04f94accba9 |
| collected/eccv2026-pass03-preflight-v1/postgresql-preflight.json | 97bf57828cc67d05de0221188a9d988eef35631c845b18372186a6249f64e5da |
| validation/eccv2026-pass03-preflight-packets-check-v1.json | 8a4f44f5d2c3a15d05b5a1df96fe465a96a2fe8914d7ed68958e6978c30985d4 |
| validation/eccv2026-pass03-preflight-recovery-v1.tar.gz | 69c84fd0f24024ca56bfa4f848c977c578267bfc53c713105987e030df3f92a2 |
全量原文恢复包为 85,430,941 字节,保留输出、全部缓存原文与回执、官方输入和固定源码/CI 凭证。失败的第一次打包任务在创建文件前因命令引用错误退出;改用文件脚本的新单元 `paper-atlas-eccv2026-pass03-recovery-v2.service` 成功,未重启或重跑采集器。
新的 PostgreSQL 只读预检覆盖全部 2,834 条:1,519 条精确标题和作者集合一致、468 条仅部分作者重合、840 条无精确匹配、4 条没有作者重合、3 条命中多个同标题实体。1,997 个候选实体包的哈希独立核对通过,来源、投稿、决定与来源关系未触发截断。该结果只生成候选,`database_written=false`、`approved_for_import=0`;没有批准合并、创建新论文或导入 ECCV 事件。无精确匹配不等于新作品,相同作者也不证明人物或作品身份。
后续先保留这批完整来源并推进其他会议来源补漏,再做逐篇内容与版本审阅、当前快照守护和保护基线。当前全局最新基线来自 SDM25 验收,保护 5,653 篇论文;不能回退到 ICAPS34 之前的计数或重放已消费批次。最终全库去重仍须另行执行。
+40
View File
@@ -0,0 +1,40 @@
# ICAPS 2026 官方名单
这是长期维护、按需运行的元数据采集器。它只读取已核验的官方页面快照,生成待身份审核的文件;不连接数据库,不下载 PDF 或评审,不执行网页脚本。
```sh
paper-atlas collect icaps2026-program \
--snapshots /data/paper-index/collected/icaps2026-official-research-v1 \
--out /data/paper-index/collected/icaps2026-normalized-v1
```
快照目录包含 `accepted.html`、`cfp.html`、`previously-published.html` 和 `manifest.json`。采集器固定这三份已审阅页面的 SHA-256,核对 URL、最终 URL、字节数及带时区的观测时间;来源变化需要重新研究、固定证据并更新检查,不能静默接受。输出目录必须全新,失败会保留 `failed.json`,成功报告最后发布。恢复前先检查旧目录,不能覆盖失败现场或重放已消费的导入队列。
[官方录用表](https://icaps26.icaps-conference.org/program/accepted/)含两组:
- **90 篇 Papers accepted to the Main Tracks**:生成 partial JSONL,状态为 accepted。[CFP](https://icaps26.icaps-conference.org/calls/cfp/)包含主会及四个特别轨道,但该表没有逐篇归属,因此保存为 `main_tracks_unclassified`,不能把它们全部重命名为 main,也不能按论文主题猜特别轨道。position 是研究类别,不能直接推成投稿 track。
- **10 篇 Previously Published Papers**:保存在单独的 `previously-published-presentations.json`,不生成新的研究论文录用记录。[该展示轨道的官方说明](https://icaps26.icaps-conference.org/calls/previously_published_research/)须随原文保存;后续关联原发表记录时不能重复制造论文或替换原会议。
名单没有逐篇链接、摘要、DOI 或原始投稿号。90 条摘要明确留空,作品及投稿标识均为空;完整作者顺序与原始作者文本一并保留。`external_id` 里的哈希仅定位这份名单的标题和作者版本,不是原始投稿号,也不是跨来源作品身份。作者文字变化不代表人物别名,标题相同不代表可以自动合并作品。
主会 rejected、desk rejected、withdrawn 的公开范围仍未知,不能从缺席录用表推断拒稿,也不能把某一 workshop 的私密评审政策扩展到整个会议。出版社 OJS 页面曾返回访问挑战;本采集器不访问或绕过该挑战。当前来源只证明官方名单,未证明完整出版物或公开负面决定覆盖。
生产导入之前仍需当前数据库候选核查、独立身份审核、来源及历史保护基线、导入后逐项验收。所有来源导入完成后还必须另做全库去重、覆盖、摘要和约束查询审计。正常使用不会自动执行这些后续写入。
测试及安装后 CLI 验证只在已授权的 `ces-control:host` runner 上运行。冻结 fixture 含完整 100 条官方条目与独立盘点结果,用于检查两组边界、作者顺序、缺页、重复、来源篡改和恢复行为。
## 独立注册元数据版本
```sh
paper-atlas collect icaps2026-registry \
--snapshots /data/paper-index/collected/icaps2026-crossref-research-v1 \
--out /data/paper-index/collected/icaps2026-registry-normalized-v1
```
该命令只读固定 `crossref-filtered.json`,与官方节目表输出分开。来源是 Crossref 的 ICAPS 期刊注册元数据(ISSN 2334-0843),查询出版日期为 2026 年的结果共 91 条,其中卷 36、期 1 有 90 条,另有卷 35 的一条必须排除。不能仅凭出版日期归属会议届次。来源 URL、抓取时间、原响应哈希和字段过滤快照均保留。
卷 36 的 90 条均带 DOI 和摘要,输出保持注册标题、作者顺序及摘要版本一致;保留原始 JATS 摘要,只移除已核验的单个段落包装并规范空白,遇到其他标记(尤其公式、上下标)拒绝无声扁平化。作者个人 ID、机构、引用与文件链接没有进入过滤快照。原始注册类型 `journal-article` 留在 metadata 中,不据此给论文添加研究类型标签。
这里的 accepted 表示注册源记录了正式发表,不代表采到原始签名录用决定;publication_doi 标识发表对应的投稿事件,不是原始投稿号。具体 track 保持待分类。与官方 90 篇名单的来源核查发现:88 个规范标题一致,其中 78 个完整作者名单一致;另有 2 个标题和 10 个作者文字差异待审核。两份输出不会自动建立作品同一性、复制人名别名或覆盖文字。它们也不应未经审核同时导入,避免重复作品或投稿。
此来源仍标记 partial,不包含公开拒稿。生产导入前需核对 DOI 归属、既有作品及跨会议历史;已知 DOI 也可能存在错误绑定,不能代替最终全库去重验收。
+46
View File
@@ -0,0 +1,46 @@
# ICDM2026 冻结日程适配器
这是长期维护、按需运行的来源模块。它读取已经冻结的官方页面,离线生成可审阅的标准记录,不连接数据库、不请求网络、不下载论文 PDF;后续入库仍需作品身份、发表版本和旧投稿历史核查。
```sh
paper-atlas collect icdm2026-program \
--snapshots /data/paper-index/collected \
--out /data/paper-index/collected/icdm2026-normalized-pass01
```
输入根目录下须有 `remaining2026-official-leads-v1/manifest.json`,及 manifest 所列的三个 ICDM HTML:`icdm-detailed-sessions.html`、`icdm-research-policy.html`、`icdm-applied-policy.html`。共享研究 manifest 也保留其他会议的入口回执,但本适配器只读取并验证 ICDM 的三个明确 URL,不要求复制其他会议文件。固定 manifest SHA-256 为 `0e13313d94e3bff17f071d3a1384f6ed83ece156239fe5bc4633cc7a34fb8e04`。
适配器核对每个回执的 URL、HTTP 200、HTML 内容类型、时区时间戳、原始字节数和哈希;保留 CRLF。来源更新必须另存版本并重新审阅,不能修改旧 manifest 来冒充旧快照。源码测试 fixture 同时保存原始页面、回执和使用另一解析方法得到的独立逐行盘点。
## 两个明确来源轨道
[官方详细日程](https://icdm2026.neu.edu.cn/DetailedSessions/list.htm)自称为线下展示清单,日期为 2026 年 11 月 13–15 日,34 个分会场、324 个不同节目编号。适配器分别输出:
| 输出文件 | 官方明确标签 | 记录数 |
| --- | --- | ---: |
| icdm-program-2026-research.jsonl | research / Research Track | 241 |
| icdm-program-2026-applied.jsonl | applied / Applied Track | 83 |
`data-track` 属性必须与可见标签一致;轨道来自这两份明确证据,不根据标题、会场名称或研究主题推测。`data-format` 和可见 Long/Short 标签、展示说明必须一致;168 个 long 和 156 个 short 仅表示演讲形式,**不自动转成论文长短、position、findings 或研究类型**。日程主题也不生成自定义人物或研究分类。
每条保留日程显示标题、原始作者文字、分号分隔的来源姓名顺序、节目编号、分会场、时间、地点、全局来源位置、来源回执和对应原文文件。作者仅标为日程报告的姓名,不声称为最终出版社完整作者版本,也不生成姓名别名。
`2026/DM...` 是 `icdm_program` 来源下的节目定位键;页面锚点只是来源定位,不能当 DOI、arXiv 或原始签名投稿/决定标识。`identifiers` 和 `submission_identifiers` 都为空。`status=accepted` 表示列入官方研究/应用日程的来源观察,`raw_decision` 明确其并非原始签名决定或已核验出版社记录;不根据该状态生成原始决定日期。每条 `original_decision_signature_verified=false`。
## 始终保留范围限制
324 条日程记录都没有摘要,输出 `abstract=null`,每个文件的 `missing_abstract_count` 如实填写。manifest 始终为 partial;`source_expected_count` 仅对应这个冻结线下日程分区,不能当作所有录用、正式出版或线上展示的应有总数。其他页面的 workshop、博士论坛、BlueSky、demo 等不在本次输入范围内,也不能假定不存在。
标签使用 `icdm_program:session_track` 与 `icdm_program:presentation` namespace;它们是平台来源证据,不与其他来源标签或用户自定义标签混为一谈。`paper_type` 保留为空。
[研究轨政策](https://icdm2026.neu.edu.cn/CallforResearchTrack/list.htm)描述 triple-blind 审稿,[应用轨政策](https://icdm2026.neu.edu.cn/CallforAppliedTrack/list.htm)描述 single-blind 审稿,并指向 CyberChair 投稿系统。这些是审稿期间政策,**不能据此断言最终拒稿元数据不公开**。报告保留 `public_rejected_metadata=unknown_no_public_export_verified`;未访问登录资料、未绕过限制,也未从日程缺席推断 rejected、desk_rejected 或 withdrawn。
## 验证与恢复
完整测试在已授权 `ces-control:host` 远程 runner 执行,包含全部 324 条与独立盘点逐字段对照、轨道/展示属性与可见文字冲突、重复或变化的节目编号、会场计数、空作者、原始字节/回执校验和失败输出保护。CI 另在安装 wheel 后、工作目录外运行真实 CLI,检查 241/83 两个输出及缺失摘要、未生成稳定标识等边界。构建和测试不在用户本机运行;实际结果见对应提交的远程 CI。
输出目录必须不存在。输入校验失败时不会创建输出;输出生成失败会保存 `failed.json`,成功报告最后原子发布为 `collection-report.json`。遇到已有目录,先核对原操作,不覆盖、不盲目重放。
原始来源恢复包为 `/data/paper-index/validation/remaining2026-official-leads-v1-recovery.tar.gz`,SHA-256 为 `0be8c84bb5c591e5afaaa5f95dd5334bf13a9be1671cb942e46e998f502495bb`。独立盘点为 `/data/paper-index/validation/icdm2026-independent-program-inventory-v1.json`,SHA-256 为 `e13eaea0a158dcaafdf5dfb4af7948e4793b0e98041ec87d42b52513acfdc5f6`;本机工作记录目录也保留同名备份。
下次继续先验证源码版本、原始证据和实际输出,再寻找可公开取得的摘要与正式出版记录,核查逐篇版本、稳定标识及旧投稿历史,准备新的只读预检和可审阅入库队列。完成日程规范化不意味着 ICDM2026 整届出版覆盖、公开负面决定普查或最终全库去重完成。
+53
View File
@@ -0,0 +1,53 @@
# RecSys 2026 公开录用贡献采集
这是长期维护项目中的离线采集器,按需处理冻结的官方元数据,不连接数据库、不访问论文链接或下载 PDF。输出须经过作品、DOI 和投稿历史审查才能导入。
## 使用
```sh
paper-atlas collect recsys2026-program \
--snapshots /data/paper-index/collected \
--out /data/paper-index/collected/recsys2026-normalized-<new-run>
```
输入父目录应包含 `remaining2026-listings-research-v2/manifest.json` 与 `recsys-contributions.html`。采集器检查固定清单 SHA256、官方 URL、HTTP 200、采集时区、原始字节哈希、九个 track 的菜单及完整条目数。现有输出目录不可覆盖;失败保留 `failed.json`,成功才发布 `collection-report.json`。再次运行前先检查原任务是否结束。
## 数据范围与重复项
来源是 [RecSys 官方录用贡献页面](https://recsys.acm.org/recsys26/contributions/),冻结时间为 2026-11-17。九个来源 track:
| track | 观察数 |
| --- | ---: |
| long | 65 |
| short | 29 |
| past_present_future | 13 |
| reproducibility | 4 |
| resource | 7 |
| industry | 83 |
| research_practice_notes | 34 |
| demo | 18 |
| doctoral_symposium | 10 |
**263 项来源观察对应 260 个不同 DOI,并非 263 篇不同论文。** 每条记录用官方页面中的 accordion locator 标识,分别保留来源标题、作者原文、摘要和位置,不假设 locator 是原始投稿 ID。
以下 DOI 各出现两次,六个观察均显式标记待审查:
- `10.1145/3773078.3831839`:Dart 两条来源记录的标题、作者和摘要都有差异;不得静默选一个覆盖另一个。
- `10.1145/3773078.3831819`:标题与作者相同、摘要版本不同。
- `10.1145/3773078.3831861`:重复展示的文字相同,保留两个来源位置。
所有观察都有公开摘要。正文只转换为 HTML 展示文字并规范空白,原始页面完整保留;遇到未审查的数学或图片结构会停止,不把上标扁平化成数字。作者按来源的逗号和最后一个 and 分隔,原始名单同时保留;这不是已验证的作者人物身份或别名。
`accepted` 仅指官方页面明示录用,不是原始签名评审决定。DOI 来自官方超链接,尚未独立核查实时出版商版本;归入 `publication_doi`,不复制其他会议的投稿 ID。track 明确来自页面区块;`recsys_program:contribution_type` 仅保存页面原始类型标签(Research、PastPresentFuture、Doctoral Symposium),各 track 的 RES/IND/RPN 等缩写同时保留在来源字段,不统一推断研究方法类型。
本页只含录用贡献,不能据此推断未出现的稿件被拒。公开拒稿导出仍为 unknown,征稿页的 desk rejection 政策不等于具体拒稿记录。不得将本批采集成功当成会议全部来源、负面决定或全库去重完成。
## 输出与恢复
输出九份 JSONL、对应部分覆盖清单、原始 HTML 和清单证据,以及记录观察数和不同 DOI 数的报告;批准导入数为零。
清单 SHA256:`fea7b710587636db06794ee195cd083634c41e9054a217859d9dd0ccdd1afab9`。
服务器恢复包 `/data/paper-index/validation/remaining2026-listings-research-v2-recovery.tar.gz`,本地备份 `/Users/zzyang/Documents/Codex/2026-11-06/paper-index/remaining2026-listings-research-v2-recovery.tar.gz`,SHA256 为 `f5ef7c2e73d46712146e4d6711ad5ad562e16b8b55c0cd75d1041cd0b81142cb`。安全恢复原始目录、核对哈希后重跑上述命令。
完整来源与独立 263 项基准已保存在测试夹具,所有测试、构建和安装验证交给指定的 `ces-control:host` runner。生产导入前还需独立核验规范化结果,并检查当前数据库已有 DOI 绑定及历次投稿。最终全库去重审计仍是独立必做阶段。
+21
View File
@@ -0,0 +1,21 @@
# RSS 2026 论文集来源
`collect rss2026-proceedings` 是独立的年度来源适配器,不依赖尚未合入主线的 RSS 2025 PR #43,也不改变其 demo 分类。使用包内 HTML、HTTP 工具和统一 CLI;不连接数据库,不下载 PDF。
```sh
paper-atlas collect rss2026-proceedings --cache /data/paper-index/cache/rss2026 \
--out /data/paper-index/collected/rss2026-native-v1
# 固定证据复现:包含 receipts.json 和对应 HTML 文件
paper-atlas collect rss2026-proceedings --snapshots /data/paper-index/collected/rss2026-primary-v1 \
--out /data/paper-index/collected/rss2026-native-replay-v1
```
核对的来源为 [第22卷论文集](https://www.roboticsproceedings.org/rss22/)、[录用表](https://roboticsconference.org/2026/program/papers/)、[CFP](https://roboticsconference.org/2026/information/cfp/) 和 [Sponsor Demos](https://roboticsconference.org/2026/program/demos/)。当前审查目录为210篇。CFP明确2026没有单独 demo paper 类别,Science/Systems论文放在 main;赞助商展示不作为论文,也不推断逐篇是否现场演示。投稿录用后才公开、评审不公开;此来源不能提供拒稿或评审全集。
每篇保留出版社原始标题、完整作者顺序、摘要、日期、BibTeX、原始 DOI,以及独立节目标题、完整作者和缩略显示、session、显示链接。节目显示编号不是原始 OpenReview 投稿编号,不写入 submission_identifiers。六个缩略作者预览按完整隐藏副本恢复核验,不假设固定字符长度;六个姓名差异和一处 RLinf/RLux 标题差异按已审核原文指纹识别并保留,不推断人名别名或官方勘误。未知差异会停止。
快照收据必须逐文件匹配固定URL、HTTP200、字节数、哈希和带时区的真实采集时间;拒绝路径穿越、重复收据和符号链接。缓存模式只请求已审核的固定公开HTML URL,访问限制由HTTP工具正常报错,不绕过。每份原始HTML按哈希保存到输出 evidence/。
输出目录必须不存在;失败目录留存供检查,换新目录重试。`collection-report.json` 最后原子写入,缺少此文件不能视为完成。清单始终 partial;已采集不代表主库完整、身份已核验或科学作品已去重。DOI也是待核对的候选标识,入库前仍需实时 PostgreSQL 身份审计、版本/作者关联复核和保护基线;入库后独立验收,所有导入结束后再做全库去重。
2026原始来源审计已冻结214份HTML且210篇均有摘要。回归 fixture 同时包含原始来源和独立生成的逐篇预期值,测试比较210篇完整标题/作者/摘要/DOI/日期;构建和测试仅在已授权 ces-control:host 远程 runner 进行。
+97
View File
@@ -0,0 +1,97 @@
# SDM 2026:生产规范化与身份核查记录
本页记录 2026-11-17 的来源验收、早期只读核查和随后的生产导入。**当前 53 条注册记录中,25 条已入库并独立验收,28 条仍待处理;没有新增论文实体。** 下文早期报告保持原始状态,不将它们改写成事后批准;最终结果见“25 条生产导入验收”。
## 已核实的生产输出
生产规范化使用通过远程 CI 1216 的源码版本 `f2753bc741d0fff06134cf64019dac38002fae37`,对应 494 项测试、wheel 构建及安装后 CLI 检查。该版本包含已合并的 SDM 适配器;这里没有替换服务器默认代码或运行环境。本页是文档更新,不能把此前 CI 结果当作本次提交的 CI 结果。
`/data/paper-index/collected/sdm2026-normalized-v1` 保留:
- 53 条 Crossref 论文章节记录,全部没有注册摘要;排除论文集和前言。
- 53 条独立日程观察,其中 52 条有公开摘要;20 处作者序列差异仍保留。
- 取消展示的条目与已注册发表记录分别保留,不生成撤回决定。
规范化服务 `paper-atlas-sdm2026-normalize-v1.service` 的 invocation 为 `3021248cae4044ac88eb80fb8b964a26`,终态成功。独立审计逐条核对冻结注册字段、完整作者顺序、原始日程和摘要证据,未以适配器自身统计替代验收。
| 证据文件 | SHA-256 |
| --- | --- |
| validation/sdm2026-normalized-v1-audit.json | 66d4e698ff8828731ccb657e273fb2f0d19cbce8fd91ade2c81ed68124c6d3a3 |
| collected/sdm2026-normalized-v1/sdm-crossref-registry-2026-unclassified.jsonl | 7edd78170ced9b79e4405eb16d63f8a9a38d38dfc906e892580f0ac4be165b2c |
| collected/sdm2026-normalized-v1/program-observations.jsonl | 421a7187369817fa90aa3484575e8a8a2f3e008b9d152bed6beca678b1e12204 |
表中服务器相对路径均位于 `/data/paper-index/`。本机工作记录根目录为 `/Users/zzyang/Documents/Codex/2026-11-06/paper-index/`,保存同名输出及审计备份;数据证据不作为项目源码提交。
## 当前数据库只读预检
`paper-atlas-sdm2026-preflight-v1.service`,invocation `9be7227930354633abf8f41cdbd0dd73`,终态成功。数据库为 `foundry_data.research`,快照时间 `2026-11-17 08:34:34.270409+00`。报告核实事务为 READ ONLY,`database_written=false`、`approved_for_import=0`。
| 候选类别 | 记录数 | 含义 |
| --- | ---: | --- |
| 精确标题与作者集合一致 | 26 | 需要进一步核对作品、版本和已有标识 |
| 精确标题但仅部分作者重合 | 2 | 作者版本冲突尚未解决 |
| 没有精确匹配 | 25 | 不是新作品证明,不能直接创建论文 |
报告位于 `collected/sdm2026-preflight-v1/postgresql-preflight.json`,SHA-256 为 `eb0cc45fa5bf01159133af0536ba6621a2403914ccd9591a5d1a6a230bc3791e`。28 个候选论文压缩证据包的哈希均核对通过;来源记录、投稿、决定及其来源关系均未触发报告的截断标志。这里的完整性仅限这些候选包,不代表全库审计完成。
随后逐项阅读来源摘要、既有摘要、注册作者和已有标识,为 25 个候选记录了支持同一作品的具体方法、数据集、实验结果或摘要一致性证据,仍需最新数据库保护基线和导入前检查。相似度只用于安排审阅顺序,不作为自动批准阈值。以下三条保留作者版本问题:
| 章节 DOI 尾号 | 既有 paper_id | 尚存问题 |
| --- | ---: | --- |
| .10 | 2656770 | 注册姓名 Tiejun Chen 与既有及日程姓名 Tiejin Chen 不同;不生成别名 |
| .15 | 2745754 | 注册 Sarang Patil 与既有 Sarang Rajendra Patil 不同;日程另有作者顺序与姓名缩写变化 |
| .49 | 2506693 | 注册和既有完整作者集合相同,但日程列出不同名单;ScaffAug 内容一致性仍不能消除来源矛盾 |
这里的尾号均属于 `10.1137/1.9781611979329`。逐条内容审阅文件为本机工作记录根目录的 `sdm2026-candidate-evidence-review-v1.json`,SHA-256 为 `6ff791f87eafeb50fa853766c632e91e3f0377ab3f619365117d557f6d56100d`。其中 25 条为 `same_work_supported_pending_fresh_import_guards`,3 条为 `hold_author_version_conflict`;所有条目的 `approved_for_import` 仍为 false。不能据此直接重放来源、绑定 DOI 或拼接日程摘要。
## 需要保护的较早拒稿历史
| SDM 候选作品 | paper_id | 已有较早决定 |
| --- | ---: | --- |
| SeBA | 2672776 | ICLR 2026 rejected |
| Towards Reliability of Parameter-Free Optimization | 3282961 | ICLR 2025 rejected |
| Scaffold-Aware Generative Augmentation and Reranking for Enhanced Virtual Screening | 2506693 | ICLR 2026 desk_rejected |
前两条在后续生产验收中各自新增独立 SDM 发表事件;第三条仍保留作者版本问题,未导入。旧投稿、旧决定及各自来源均完整保留。两个来源对同一拒稿的观察不能直接统计成两次独立投稿或拒稿。不能覆盖旧状态,也不能因这批录用候选就宣称完整投稿历史已补齐。
## 未精确匹配的补充检索
对另外 25 条运行了有时限、数量上限和只读快照的补充检索,使用现有索引,不新增数据库索引或改变全局配置。
- 全文检索 `paper-atlas-sdm2026-title-variants-v1.service`,invocation `29c5d2a288804ce7a9c3e2f7db4dca6f`,终态成功。每项使用完整标题及明确记录的锚点词,8 条至少有一次查询达到候选数量上限,不能据此排除重复。
- 标题前缀范围检索 `paper-atlas-sdm2026-title-prefix-v2.service`,invocation `3087f1c0f252427485a661a63c50b27f`,终态成功。25 条均使用既有标题索引,没有查询错误或数量截断。18 字符前缀只是补充方法,不是全库模糊匹配验收,也不会消除前一轮查询的限制。
全文候选中的既有论文 `2025464` 与 SDM .9 共享作者,但前者讨论 CLOUD、离散/混合/连续变量,后者讨论 SLICE、类别数据与 HCR 模型。标题前缀候选 `938886` 与 SDM .28 只共享一名作者,既有版本讨论随机游走相似度、有效电阻和社交网络推断。不能凭研究主题或作者重合将这些论文合并。25 条仍保留为待进一步查证,没有获准创建新作品。
| 服务器恢复文件 | SHA-256 |
| --- | --- |
| validation/sdm2026-title-variants-v1-recovery.tar.gz | 1f387f2ff4d5684aa72abd419bddeb7e972624f6a56f6e9a943168aa944b9137 |
| validation/sdm2026-title-prefix-v2-recovery.tar.gz | 4941343bd6152c5046778f7c82588e3c6ad41c9bb938e0107f8b2b426ac10772 |
两个原报告的 SHA-256 分别为 `b6ab2e9228b8e8e9548b27df70834faf25bfda551844979f8410a74b6e6aadd0` 和 `c1f902b4e4f33997eb182b10dc78332e182ceb6051f1a7c14b21465c8878263d`。前者保留 1 个进一步审阅的完整候选包,后者保留全部 113 个前缀命中的候选包;这些数量不是论文批准数。
## 25 条生产导入验收
`paper-atlas-sdm2026-import25-v1.service`,invocation `003b490a2f2c449da078a2f8625b31b8`,终态成功。执行前核验 81 个源码文件,并确认 24 个核心写入文件与此前已验收的 canonical writer 相同;复用原生 `paper-atlas import records`,未修改默认运行目录。
- 新增 25 条来源、25 次会议事件、25 条发表决定观察及 25 个 `publication_doi` 投稿标识,新增论文实体为 0。
- 24 篇使用已核实的 arXiv 标识补注;AdamG 没有 arXiv 标识,采用单独审阅并实时守护的唯一精确标题、四名完整作者和具体方法证据。旧 OpenReview ID 未复制为 SDM 投稿标识。
- 28 条待处理记录全部排除;25 个注册来源摘要仍为空,52 份日程摘要保持独立来源版本。覆盖状态为 `partial`,应有 53 条、当前导入 25 条,track 保留 `proceedings_unclassified`。
- 全局旧原始投稿与决定历史、5,653 篇受保护论文的原资料/来源/标签/既有标识均通过导入后比较。SeBA、AdamG 的两次较早 ICLR 投稿及四条 rejected 来源观察保留,没有将四条观察说成四次投稿。
- 约束有效、实际全文查询与 GIN 计划通过;foundry-core `/readyz` 返回 200,受保护 SQLite 快照保持只读。
| 生产证据 | SHA-256 |
| --- | --- |
| collected/sdm2026-import25-v1/postgresql-postcheck-v1.json | 0c861722f0d455462a2686a42d634e0b963d075cf58f90fa9f48e1ae1fc82c27 |
| collected/sdm2026-import25-baseline-v1/baseline.json | 42ffbaf6f843a91ebf64f0bfefa4af36f6ac72f044b0e8587e65c5e6648c0931 |
| collected/sdm2026-prepared25-v1/independent-preparation-check.json | 8ea8eb59f4b1601a484d1bc0535b289cb950b83747ed5f82b0eac55e527c4a7c |
| validation/sdm2026-reviewed25-recovery-v1.tar.gz | 13447944ec34e0e2a6eee07a1d9820bfb8066756aa11016673e416c0d6f8b2b2 |
恢复包包含 263 个文件;本机已独立核对队列、70 份内容寻址证据和 28 个早期候选证据包。**25 条队列已消费,禁止重放。** 28 条剩余记录需要新的导入后快照;先前候选包不能当作当前数据库状态。后续批次须沿用最新 5,653 篇保护基线,并加上各自新目标。
## 下次恢复
1. 先核对实际服务终态、来源及报告哈希,再读取上述证据;终态成功的研究任务无需重启。
2. 保留已验收的 25 条,不重放;继续解决 3 条作者版本问题和 25 条未精确匹配记录;不把缺少候选当作新作品证明,不把源端 403 当作零条公开拒稿。
3. 对后续新获内容支持的候选另做新快照检查:作品与 DOI 绑定、来源是否已导入、已有会议事件及全部历史保护。建立新的明确入库队列和可恢复备份后再执行写入,不能直接使用本次候选报告当导入计划。
4. 导入后逐项核验新事件、旧历史、来源关系和精确数量变化;注册摘要缺失仍如实保留。完整来源范围结束后另做全库去重、标识冲突、近似标题/作者、覆盖、摘要、约束和查询验收。
+46
View File
@@ -0,0 +1,46 @@
# SDM 2026:论文集与日程版本
这是长期维护、按需运行的离线来源适配器。它复用固定的公开元数据快照,不请求网络、不下载论文 PDF、不连接数据库,也不会把候选版本关联当成已通过的作品去重。
```sh
paper-atlas collect sdm2026-proceedings \
--snapshots /data/paper-index/collected \
--out /data/paper-index/collected/sdm2026-normalized-pass01
```
`--snapshots` 下需保留四个目录:`sdm2026-official-research-v1`、`sdm2026-crossref-research-v1`、`sdm2026-program-research-v1`、`sdm2026-public-abstract-research-v1`。适配器固定核验各自 manifest 和全部成功响应的字节数、哈希、URL 与状态。测试 fixture 保留原始响应字节及独立盘点结果。使用全新输出目录恢复,保留失败目录中的 `failed.json`;成功报告最后以原子替换发布。来源更新时另存版本并重新审阅,不能修改旧 manifest 后冒充旧快照。
## 两种输出
[论文集](https://epubs.siam.org/doi/book/10.1137/1.9781611979329)的 ISBN 为 `9781611979329`。[Crossref 精确 ISBN 查询](https://api.crossref.org/works?filter=isbn%3A9781611979329&rows=1000)的冻结响应有 55 项:53 个论文章节、1 本论文集、1 份前言。适配器只将 53 个章节放入 `report.files` 的待身份审阅 JSONL,排除论文集和前言;核对完整章节 DOI 集合、父 DOI、容器标题、ISBN、作者顺序证据及出版年份。
正式记录保留 Crossref 作者版本与原始字段,使用 `proceedings_unclassified`,不按主题或分会场推断 track。注册发表记录表达已收入论文集的事实,不冒充原始签名录用决定。印刷与在线出版日期分别保留,不能用它们替代会议举办日期或决定日期。两个标题含 `<sup>*</sup>`:显示标题保留脚注星号并标记待审阅,原始标题完整保留;去掉这一特定脚注的文字仅用于列出待审阅候选,不用于自动改名或绑定作品。
**53 个 Crossref 章节都没有摘要字段。**主记录的 `abstract` 因此为空,报告如实记为 53 个摘要缺失,不能因另一来源有同标题摘要而直接填入。
[官方在线日程](https://meetings.siam.org/program.cfm?CONFCODE=sdm26)有 9 个研究论文分会场、53 项日程条目和 52 个公开摘要页。`program-observations.jsonl` 单独保留日程版本、公开页标题、来源报告的作者姓名、摘要、会场引用、回执与哈希;它不是入库 manifest,也不在 `report.files` 中。页面中作者行的单位、国家及联系邮箱不规范化成人物字段,原始页面留在证据文件中。
52 个公开页中,32 个作者序列与 Crossref 完全一致,20 个不同,包括姓名拼写、顺序和作者缺失。例如 *Scaffold-Aware Generative Augmentation and Reranking for Enhanced Virtual Screening* 的日程页列出 Xin Wang、Qinwen Ge,Crossref 列出 Xin Wang、Yu Wang、Yunchao Liu、Jens Meiler、Tyler Derr。它们必须保留为独立来源版本,不能仅凭标题相同拼接摘要与作者,或生成姓名别名。候选 DOI 字段明确表示待审阅关系,不是逐篇身份批准;作者列表相同也不能代替全库作品身份核查。
*A Trajectory View of Statistical Policy Comparison* 的日程展示标为取消,且没有公开摘要链接;它仍出现在冻结的 Crossref 论文章节集合中。适配器保留这两项事实,**不把取消展示转换成论文撤回、拒稿或删除发表记录**。邀请报告、博士论坛和其他非研究论文活动也不计入这里的 53 个章节。
## 公开拒稿与访问限制
[官方投稿说明](https://www.siam.org/conferences-events/siam-conferences/sdm26/submissions/)链接 OpenReview group `SIAM.org/2026/Conference_on_Data_Mining`。匿名读取 group 配置成功;按实际 `rejected_venue_id` 查询 notes 返回 HTTP 403,随后停止,没有绕过或继续查询其他负面状态。因此公开拒稿、desk-rejected 和 withdrawn 决定覆盖仍未核实,不能报告为零条或从录用名单缺席推断拒稿。
来源冻结时,服务器访问 SIAM 主站的两份说明页及出版社目录也返回 403;回执如实保留。可读取的是公开在线日程及独立公开的 Crossref 注册元数据。Crossref 记录不是受限出版社页面的重建,也不被称作已经核验的出版社摘要版本。
## 来源恢复与后续核查
| 冻结目录 | manifest SHA-256 |
| --- | --- |
| sdm2026-official-research-v1 | a39a6ee6a12b69051ba2dad7c9cbc5b0fe215af85100c7aa2f5166d44c22fe61 |
| sdm2026-crossref-research-v1 | 51d337a153a1046e4a00ebc031707faf72e733577e4fc243a31672ee93277da5 |
| sdm2026-program-research-v1 | 689a2aaaa15be532957c4c88e03e1e2e87909334b6f529be16905d8b9cc1fb77 |
| sdm2026-public-abstract-research-v1 | 8e1c29611e145cbfa54ec4afaa2efe152557b4e04e339901af5cd33abefc29a5 |
服务器恢复包:`/data/paper-index/validation/sdm2026-sources-recovery-v1.tar.gz`,SHA-256 为 `d6d14c0e4bb2c26e16a4888e3b150e44f982d689351638b1b9a28644736c4387`;本机工作记录目录有同名备份。包内保留原始来源、独立盘点及逐页版本比较记录,含义是可复现来源证据,不是已完成导入或身份批准。
下次继续时,先核对来源版本和访问状态,再做当前数据库的 DOI 绑定、标题/作者近似候选与完整投稿历史核查,逐条解决作者及摘要版本问题。批准后另建入库队列、保护基线和导入后验收;所有来源结束后仍需全库去重、覆盖、摘要、约束和实际查询验收。测试、构建和安装验证只在已授权 `ces-control:host` 远程 runner 执行。
生产规范化验收、当前只读候选数量、已有拒稿历史保护及下次恢复入口见 [SDM2026 身份核查记录](sdm2026-identity-review.md)。该记录不代表已批准或完成导入。
+21
View File
@@ -0,0 +1,21 @@
# 来源身份只读预检
`audit source-identities` 在 PostgreSQL 的同一 `REPEATABLE READ READ ONLY` 快照中,检查规范化 JSONL 的来源绑定、稳定标识及精确标题候选,并导出候选论文的原始来源和投稿决定历史。用于入库前评审,不执行导入、合并或标识修改。
```sh
PGSERVICEFILE=/etc/paper-atlas/pg_service.conf paper-atlas audit source-identities \
--pg-service foundry-research --pg-schema research \
--files /data/paper-index/collected/example/main.jsonl \
/data/paper-index/collected/example/position.jsonl \
--out /data/paper-index/collected/example-identity-preflight-v1
```
每个 JSONL 必须有同名 `.manifest.json`,包含 `source`、`venue`、`year`、`track`、`count` 和 `file_sha256`。逐行校验届次、显式决定状态、标识类型和跨文件来源主键唯一性;文件哈希和数量必须相符。只读取调用者指定的文件,不跟随 manifest 中的绝对 `file` 路径;原始行哈希包含换行符。原始 JSONL 和 manifest 应与报告一起保留。
输出目录必须不存在,防止复用旧报告或覆盖证据。失败会保留 `failed.json` 及已完成分片,不能把部分输出当成功;排查后使用新的输出目录重新核查。成功报告 `postgresql-preflight.json` 最后原子写入,记录输入哈希、数据库快照和每条候选。实体详情存为 `entities/<分片>/<paper_id>.json.gz`,报告中列出压缩文件的 SHA-256;包含完整 paper、标识和限量原始来源、投稿、决定及出处。每篇最多保留 100 条来源、200 次投稿,每次投稿最多 200 条决定,每组决定或投稿出处最多 100 条,所有截断均显式标记。精确标题最多取 21 个候选;达到上限即标记,不能据此推算总数。这里没有计算全表精确数量。
候选分类用于安排审核:既有来源记录需要核对并避免重放,多个绑定冲突必须暂缓;单个 DOI、arXiv 或 OpenReview 绑定也可能错误,必须核对原文、标题、作者和历史。作者比较保留完整规范化姓名的重复次数,但姓名相同不能证明是同一个人。未找到精确标题或标识不能证明是新作品,仍需近似标题、版本和跨会历史核查。
所有条目固定为 `not_approved_for_import`,批准数量为 0。不要直接把此报告当成 curation 计划或导入队列。后续获批导入前应重新检查实时来源绑定;来源摘要缺失不得覆盖已有摘要,旧拒稿与新录用应作为不同投稿经历保留。本命令不能代替所有导入结束后的全库去重、覆盖和约束验收。
构建与测试仅在已确认的 `ces-control:host` 远程 runner 执行。集成测试使用已有 PostgreSQL 服务做真实只读候选查询,不往生产库添加测试记录。
+252
View File
@@ -0,0 +1,252 @@
"""Read-only, evidence-preserving identity candidates for normalized source files.
A candidate is never permission to import or merge. Absence of an exact match is
never proof of a new work. Run again against a fresh snapshot before curation.
"""
import argparse
from collections import Counter
import gzip
import hashlib
import json
from pathlib import Path
from paper_atlas.db.store import author_keys, normalized
from paper_atlas.db.submissions import timestamp
from paper_atlas.postgres.client import ReadConnection
STATUSES = {'accepted', 'rejected', 'desk_rejected', 'withdrawn', 'under_review', 'unknown'}
def digest(data):
return hashlib.sha256(data).hexdigest()
def authors_multiset(authors):
# author_keys intentionally removes duplicates; identity review must retain them.
return Counter(key for author in authors for key in author_keys([author]))
def load_inputs(files):
records, inputs, seen = [], [], set()
for filename in files:
path = Path(filename).resolve()
manifest_path = path.with_suffix('.manifest.json')
raw_manifest = manifest_path.read_bytes()
manifest = json.loads(raw_manifest)
hasher, count = hashlib.sha256(), 0
with path.open('rb') as stream:
for line_number, line in enumerate(stream, 1):
hasher.update(line)
row = json.loads(line)
for key in ('source', 'external_id', 'title', 'venue', 'track'):
if not isinstance(row.get(key), str) or not row[key].strip():
raise ValueError(f'Invalid {key}: {path.name}:{line_number}')
if type(row.get('year')) is not int or row.get('status') not in STATUSES:
raise ValueError('Invalid year or explicit source status')
for key in ('source', 'venue', 'year', 'track'):
if row[key] != manifest.get(key):
raise ValueError(f'Manifest scope mismatch: {key}')
authors = row.get('authors')
if not isinstance(authors, list) or any(not isinstance(a, (str, list, dict)) for a in authors):
raise ValueError('Invalid authors')
identifiers = row.get('identifiers', {})
if not isinstance(identifiers, dict) or any(
not isinstance(k, str) or not k.strip() or not isinstance(v, str) or not v.strip()
for k, v in identifiers.items()):
raise ValueError('Identifiers must contain singular nonempty strings')
key = (row['source'], row['external_id'])
if key in seen:
raise ValueError('Duplicate source/external_id across inputs')
seen.add(key)
record = {k: row[k] for k in ('source', 'external_id', 'title', 'authors',
'venue', 'year', 'track', 'status')}
record.update(identifiers=identifiers, file=str(path), line=line_number,
raw_line_sha256=digest(line), source_record_hits=[],
identifier_hits=[], title_candidate_ids=[])
records.append(record)
count += 1
if not count or count != manifest.get('count') or hasher.hexdigest() != manifest.get('file_sha256'):
raise ValueError('Input count/hash mismatch or empty input')
inputs.append({'file': str(path), 'file_sha256': hasher.hexdigest(), 'count': count,
'manifest_sha256': digest(raw_manifest), 'manifest': manifest})
if not records:
raise ValueError('At least one nonempty input is required')
return records, inputs
def _lookup(connection, values, columns, lateral):
"""Bound scalar VALUES allow the existing text codec to encode each key."""
width = len(columns)
for start in range(0, len(values), 100):
batch = values[start:start + 100]
sql = 'WITH keys(' + ','.join(columns) + ') AS (VALUES ' + ','.join(
'(' + ','.join(['?'] * width) + ')' for _ in batch) + ') SELECT keys.n,hit.* FROM keys JOIN LATERAL (' + lateral + ') hit ON true'
for row in connection.execute(sql, tuple(value for item in batch for value in item)):
yield dict(row)
def find_candidates(connection, records):
source_keys = [(i, row['source'], row['external_id']) for i, row in enumerate(records)]
for hit in _lookup(connection, source_keys, ('n', 'source', 'external_id'), '''
SELECT r.id,r.paper_id FROM sources s JOIN source_records r ON r.source_id=s.id
WHERE s.name=keys.source AND r.external_id=keys.external_id'''):
records[hit.pop('n')]['source_record_hits'].append(hit)
identifiers = []
for i, row in enumerate(records):
pairs = {(row['source'], row['external_id']), *row['identifiers'].items()}
identifiers.extend((i, scheme, value) for scheme, value in sorted(pairs))
for hit in _lookup(connection, identifiers, ('n', 'scheme', 'value'), '''
SELECT scheme,value,paper_id FROM identifiers
WHERE scheme=keys.scheme AND value=keys.value'''):
records[hit.pop('n')]['identifier_hits'].append(hit)
title_keys = [(i, normalized(row['title'])) for i, row in enumerate(records)]
for hit in _lookup(connection, title_keys, ('n', 'title_key'), '''
SELECT id AS paper_id FROM papers WHERE title_key=keys.title_key ORDER BY id LIMIT 21'''):
records[hit['n']]['title_candidate_ids'].append(hit['paper_id'])
def entity_packet(connection, paper_id):
paper = connection.execute('SELECT * FROM papers WHERE id=?', (paper_id,)).fetchone()
if paper is None:
raise RuntimeError('Candidate disappeared from read-only snapshot')
sources = [dict(r) for r in connection.execute('''SELECT r.*,s.name AS source
FROM source_records r JOIN sources s ON s.id=r.source_id
WHERE r.paper_id=? ORDER BY r.id LIMIT 101''', (paper_id,))]
attempts = [dict(r) for r in connection.execute('''SELECT s.*,v.acronym AS venue,e.year,e.track,e.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 s.id LIMIT 201''', (paper_id,))]
packet = {'paper': dict(paper), 'identifiers': [dict(r) for r in connection.execute(
'SELECT * FROM identifiers WHERE paper_id=? ORDER BY scheme,value', (paper_id,))],
'source_records': sources[:100], 'source_records_truncated': len(sources) > 100,
'attempts': attempts[:200], 'attempts_truncated': len(attempts) > 200}
for attempt in packet['attempts']:
observations = [dict(r) for r in connection.execute(
'SELECT * FROM decisions WHERE submission_id=? ORDER BY id LIMIT 201', (attempt['id'],))]
attempt['decisions'], attempt['decisions_truncated'] = observations[:200], len(observations) > 200
provenance = [dict(r) for r in connection.execute('''SELECT ss.*,r.external_id,s.name AS source
FROM submission_sources ss JOIN source_records r ON r.id=ss.source_record_id
JOIN sources s ON s.id=r.source_id WHERE ss.submission_id=?
ORDER BY ss.source_record_id,ss.source_submission_id LIMIT 101''', (attempt['id'],))]
attempt['sources'], attempt['sources_truncated'] = provenance[:100], len(provenance) > 100
for decision in attempt['decisions']:
provenance = [dict(r) for r in connection.execute('''SELECT ds.*,r.external_id,s.name AS source
FROM decision_sources ds JOIN source_records r ON r.id=ds.source_record_id
JOIN sources s ON s.id=r.source_id WHERE ds.decision_id=?
ORDER BY ds.source_record_id,ds.source_decision_id LIMIT 101''', (decision['id'],))]
decision['sources'], decision['sources_truncated'] = provenance[:100], len(provenance) > 100
return packet
def classify(record, entities):
anchors = {r['paper_id'] for r in record['source_record_hits'] + record['identifier_hits']}
titles = record['title_candidate_ids']
flags = []
own_authors = authors_multiset(record['authors'])
if not own_authors:
flags.append('missing_source_authors')
if any(n > 1 for n in own_authors.values()):
flags.append('duplicate_normalized_author_names')
if len(normalized(record['title'])) < 25:
flags.append('short_title')
comparisons = []
for pid in sorted(anchors | set(titles)):
entity = entities[pid]
other = authors_multiset(entity['authors'])
comparisons.append({'paper_id': pid,
'normalized_title_equal': normalized(record['title']) == normalized(entity['title']),
'nonempty_author_multiset_equal': bool(own_authors) and own_authors == other,
'author_name_overlap': sorted(own_authors.keys() & other.keys())})
if len(anchors) > 1:
kind = 'conflicting_existing_bindings_hold'
elif record['source_record_hits']:
kind = 'existing_source_record_reconcile_no_replay'
elif anchors:
kind = 'existing_identifier_binding_requires_review'
elif len(titles) >= 21:
kind = 'title_candidate_limit_reached_hold'
elif not titles:
kind = 'no_exact_match_not_proof_of_novelty'
elif len(titles) > 1:
kind = 'multiple_exact_title_entities_hold'
elif comparisons[0]['nonempty_author_multiset_equal']:
kind = 'one_exact_title_author_multiset_requires_review'
elif comparisons[0]['author_name_overlap']:
kind = 'partial_author_overlap_requires_review'
else:
kind = 'no_author_agreement_hold'
return {'kind': kind, 'flags': flags, 'comparisons': comparisons,
'title_candidates_truncated': len(titles) >= 21,
'decision': 'not_approved_for_import'}
def _atomic_json(path, value):
temporary = path.with_suffix(path.suffix + '.part')
with temporary.open('w', encoding='utf-8') as stream:
json.dump(value, stream, ensure_ascii=False, indent=2)
stream.write('\n')
temporary.replace(path)
def audit(files, out, *, pg_service, pg_schema):
# Reject input corruption before opening a database connection or reserving output.
records, inputs = load_inputs(files)
out = Path(out)
out.mkdir(parents=True, exist_ok=False)
(out / 'audit.lock').write_text(timestamp() + '\n')
connection = None
try:
connection = ReadConnection(pg_service, pg_schema)
connection.execute("SET LOCAL statement_timeout='60s'")
if connection.execute('SHOW transaction_read_only').fetchone()[0] != 'on':
raise RuntimeError('A read-only transaction is required')
snapshot = dict(connection.execute('''SELECT current_database() AS database,current_schema() AS schema,
transaction_timestamp()::text AS transaction_at,pg_current_snapshot()::text AS snapshot''').fetchone())
find_candidates(connection, records)
ids = sorted({hit['paper_id'] for row in records for hit in row['source_record_hits'] + row['identifier_hits']} |
{pid for row in records for pid in row['title_candidate_ids']})
entities, index = {}, {}
for number, pid in enumerate(ids, 1):
packet = entity_packet(connection, pid)
relative = Path('entities') / str(pid // 10000) / f'{pid}.json.gz'
target = out / relative
target.parent.mkdir(parents=True, exist_ok=True)
raw = gzip.compress(json.dumps(packet, ensure_ascii=False).encode(), mtime=0)
target.write_bytes(raw)
index[str(pid)] = {'file': relative.as_posix(), 'sha256': digest(raw)}
entities[pid] = {'title': packet['paper']['title'],
'authors': json.loads(packet['paper']['authors_json'])}
if number % 250 == 0 or number == len(ids):
print(json.dumps({'phase': 'entity_evidence', 'completed': number, 'total': len(ids)}), flush=True)
kinds = Counter()
for row in records:
row.update(classify(row, entities))
kinds[row['kind']] += 1
report = {'status': 'completed_read_only_candidates_not_import_approval',
'started_snapshot': snapshot, 'completed_at': timestamp(), 'transaction_read_only': True,
'database_written': False, 'approved_for_import': 0, 'records': len(records),
'kind_counts': dict(kinds), 'inputs': inputs, 'entities': index, 'entries': records,
'limitations': ['Exact keys only; no match does not establish a new work.',
'Name agreement does not prove person or scientific-work identity.',
'Bound identifiers can be incorrect; review titles, authors and original evidence.',
'Source/history limits are explicit in compressed entity packets.',
'Not a final whole-database deduplication or coverage audit.']}
_atomic_json(out / 'postgresql-preflight.json', report)
return report
except Exception as error:
_atomic_json(out / 'failed.json', {'status': 'failed_no_import_approval',
'error_type': type(error).__name__, 'database_written': False})
raise
finally:
if connection is not None:
connection.close()
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--files', nargs='+', required=True, type=Path)
parser.add_argument('--pg-service', required=True)
parser.add_argument('--pg-schema', required=True)
parser.add_argument('--out', required=True, type=Path)
args = parser.parse_args(argv)
report = audit(args.files, args.out, pg_service=args.pg_service, pg_schema=args.pg_schema)
print(json.dumps({k: report[k] for k in ('status', 'records', 'kind_counts', 'approved_for_import')}, ensure_ascii=False))
+9
View File
@@ -23,9 +23,17 @@ COMMANDS = {
('collect','colm'): ('paper_atlas.collectors.proceedings', {'source':'colm'}, 'COLM 官方论文集'),
('collect','neurips'): ('paper_atlas.collectors.proceedings', {'source':'neurips'}, 'NeurIPS 官方论文集'),
('collect','mlsys'): ('paper_atlas.collectors.proceedings', {'source':'mlsys'}, 'MLSys 官方历年标题与摘要'),
('collect','icaps2026-registry'): ('paper_atlas.collectors.icaps_registry2026', {}, '核验 ICAPS 2026 Crossref 注册元数据并保留独立版本'),
('collect','eccv2026-program'): ('paper_atlas.collectors.eccv2026', {}, '核验ECCV2026暂定节目并断点采集公开元数据'),
('collect','icdm2026-program'): ('paper_atlas.collectors.icdm2026', {}, '核验 ICDM2026 冻结日程及明确的研究/应用轨道'),
('collect','sdm2026-proceedings'): ('paper_atlas.collectors.sdm2026', {}, '核验 SDM 2026 注册论文并单独保留公开日程版本'),
('collect','icaps2026-program'): ('paper_atlas.collectors.icaps2026', {}, '核验 ICAPS 2026 官方录用名单并分开既有论文展示'),
('collect','rss2026-proceedings'): ('paper_atlas.collectors.rss2026', {}, '核验 RSS 2026 官方论文集与独立录用表'),
('collect','recsys2026-program'): ('paper_atlas.collectors.recsys2026', {}, '核验 RecSys 2026 公开录用贡献并保留 DOI 来源版本'),
('collect','cvf'): ('paper_atlas.collectors.vision', {'source':'cvf'}, 'CVPR / ICCV 官方开放目录'),
('collect','ecva'): ('paper_atlas.collectors.vision', {'source':'ecva'}, 'ECCV 官方开放目录'),
('collect','aamas-proceedings'): ('paper_atlas.collectors.aamas', {}, '按官方目录区分 AAMAS 发表形式和摘要'),
('collect','corl2026-program'): ('paper_atlas.collectors.corl2026', {}, '核验 CoRL 2026 官方论坛名单与分字段公开档案'),
('collect','neurips-program'): ('paper_atlas.collectors.neurips_program', {}, '按公开节目来源区分 NeurIPS 2026 投稿与重复展示'),
('collect','www-archive'): ('paper_atlas.collectors.www_archive', {}, '核验早期WWW官方档案并分开待核查条目'),
('collect','dblp'): ('paper_atlas.collectors.dblp_records', {}, '核对并整理 DBLP 历史发表记录'),
@@ -46,6 +54,7 @@ COMMANDS = {
('audit','arxiv'): ('paper_atlas.audits.arxiv', {}, '核验 arXiv 原始编号、重复行与 ZIP 完整性'),
('audit','arxiv-tags'): ('paper_atlas.audits.arxiv_tags', {}, '逐记录核验 arXiv 类别及出处完整性'),
('audit','public-openreview'): ('paper_atlas.audits.public_openreview', {}, '核对公开负面决定与已有投稿历史'),
('audit','source-identities'): ('paper_atlas.audits.source_identity', {}, '只读核对来源文件的身份候选与既有投稿历史'),
('audit','histories'): ('paper_atlas.audits.histories', {}, '只读核对跨会议投稿历史及决定来源关系'),
}
+6 -3
View File
@@ -136,6 +136,9 @@ def read_pages(cache=None, snapshots=None):
def _collect(year, out, *, cache=None, snapshots=None):
if year == 2026:
from paper_atlas.collectors.aamas2026 import collect as collect_2026
return collect_2026(out, cache=cache, snapshots=snapshots)
if year == 2025:
from paper_atlas.collectors.aamas2025 import collect as collect_2025
return collect_2025(out, cache=cache, snapshots=snapshots)
@@ -193,8 +196,8 @@ def _collect(year, out, *, cache=None, snapshots=None):
def collect(year, out, *, cache=None, snapshots=None):
if year not in (2012, 2025):
raise ValueError('Only independently reviewed AAMAS 2012/2025 layouts are supported')
if year not in (2012, 2025, 2026):
raise ValueError('Only independently reviewed AAMAS 2012/2025/2026 layouts are supported')
out = Path(out)
out.mkdir(parents=True, exist_ok=True)
with (out/'aamas.collect.lock').open('a') as lock:
@@ -204,7 +207,7 @@ def collect(year, out, *, cache=None, snapshots=None):
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog)
parser.add_argument('--year', type=int, choices=[2012, 2025], required=True)
parser.add_argument('--year', type=int, choices=[2012, 2025, 2026], required=True)
parser.add_argument('--out', required=True)
source = parser.add_mutually_exclusive_group(required=True)
source.add_argument('--cache')
+219
View File
@@ -0,0 +1,219 @@
"""Reviewed AAMAS 2026 HTML metadata; retractions stay outside accepted queues."""
from collections import Counter, defaultdict
import hashlib
import json
from pathlib import Path
import re
import tempfile
from paper_atlas.collectors.aamas2025 import one
from paper_atlas.db.submissions import timestamp
from paper_atlas.io.html import Document, Node, clean
from paper_atlas.io.http import fetch
ROOT = 'https://www.ifaamas.org/Proceedings/aamas2026/'
URL = ROOT+'forms/contents.htm'
TITLE = 'AAMAS 2026: Proceedings of the 2026 International Conference on Autonomous Agents and Multiagent Systems'
SECTIONS = {
'K': ('Keynote Talks', 'keynote', 'keynote', 3),
'R': ('Research Paper Track', 'full_paper', 'proceedings_unspecified', 338),
'1': ('Extended Abstracts', 'extended_abstract', 'proceedings_unspecified', 185),
'A': ('AAAI Track', 'full_paper', 'aaai_transfer', 19),
'B': ('Blue Sky Ideas Track', 'blue_sky_paper', 'blue_sky', 17),
'C': ('Doctoral Consortium', 'doctoral_abstract', 'doctoral', 34),
'D': ('Demonstration Track', 'demo', 'demo', 44),
'J': ('JAAMAS Track', 'journal_presentation', 'jaamas', 3),
}
PDF = re.compile(r'\.\./pdfs/([A-Z]{4}[0-9]{4})\.pdf')
RETRACTION_TITLE = 'Unifying Model-Free Efficiency and Model-Based Representations via Latent Dynamics'
WITHDRAWAL_TITLE = 'Cognitive Heterogeneity and Behavioral Biases in Multi-Stage Supply Chains: Evidence from LLM-Based Agent Simulations'
def parse_authors(paragraph):
font = one([n for n in paragraph.walk() if n.tag == 'font' and
n.attrs.get('color', '').lower() == '#222222'], 'ordered author block')
authors, affiliations = [], []
name, affiliation = '', ''
def flush():
nonlocal name, affiliation
if clean(name) or clean(affiliation):
if not clean(name):
raise ValueError('Affiliation without author')
authors.append(clean(name)); affiliations.append(clean(affiliation))
name, affiliation = '', ''
for child in font.children:
if isinstance(child, str):
name += child
elif child.tag == 'br':
flush()
elif child.tag == 'i':
affiliation += child.text()
else:
raise ValueError('Unreviewed author markup: '+child.tag)
flush()
if not authors:
raise ValueError('Empty author list')
return authors, affiliations
def parse_contents(html, evidence):
doc = Document(html)
if [clean(n.text()) for n in doc.root.walk() if n.tag == 'title'] != [TITLE]:
raise ValueError('Unexpected AAMAS proceedings title/year')
section = None
headers, records, excluded, held = [], [], [], []
counts, seen_pages, seen_pdf_ids = Counter(), set(), set()
for node in doc.root.walk():
if node.tag != 'p' or any(n is not node and n.tag == 'p' for n in node.walk()):
continue
anchors = [n.attrs['name'] for n in node.walk() if n.tag == 'a' and n.attrs.get('name') in SECTIONS]
if anchors:
section = one(anchors, 'section anchor')
if clean(node.text()) != SECTIONS[section][0]:
raise ValueError('Publisher section heading differs from reviewed layout')
headers.append(section)
page_fonts = [n for n in node.walk() if n.tag == 'font' and
n.attrs.get('color', '').lower() == '#444444']
if not page_fonts:
continue
page_text = clean(one(page_fonts, 'page marker').text())
page_match = re.fullmatch(r'\(Page ([1-9][0-9]*)\)', page_text)
if not page_match or section not in SECTIONS:
raise ValueError('Invalid page or paper outside reviewed sections')
page = int(page_match[1])
if page in seen_pages:
raise ValueError('Repeated publisher page ID')
seen_pages.add(page)
title_node = one([n for n in node.walk() if n.tag == 'strong'], 'paper title')
title = clean(title_node.text())
if not title:
raise ValueError('Empty paper title')
authors, affiliations = parse_authors(node)
link = one([n for n in node.walk() if n.tag == 'a' and 'href' in n.attrs], 'paper link')
href = link.attrs['href']
title_nodes = {id(n) for n in title_node.walk()}
retractions = [clean(n.text()) for n in link.walk() if n.tag == 'i' and id(n) not in title_nodes]
heading, publication_format, track, _ = SECTIONS[section]
counts[section] += 1
metadata = {'publisher_page': page, 'section_url': URL, 'section_anchor': section,
'section_heading': heading, 'publication_format': publication_format,
'author_affiliations': affiliations, 'source_evidence': evidence,
'decision_date_precision': 'not_supplied',
'missing_abstract_reason': 'Publisher HTML supplies no abstract; research PDF not downloaded.',
'publication_evidence': 'Published proceedings entry, not an original decision note.'}
withdrawn = retractions == ['Withdrawn June 8, 2026']
if withdrawn:
if (section, page, title, href) != ('1', 3178, WITHDRAWAL_TITLE, ''):
raise ValueError('Unreviewed withdrawal notice')
metadata.update({'publication_state': 'withdrawn',
'withdrawal_date': '2026-06-08',
'raw_publication_state': retractions[0],
'decision_date_precision': 'day',
'publication_evidence': 'Explicit dated publisher withdrawal notice, not an original review decision note.'})
elif retractions:
if (section, page, title, href, retractions) != ('R', 238, RETRACTION_TITLE, '', ['Retracted June 8, 2026']):
raise ValueError('Unreviewed retraction or publication status markup')
held.append({'source': 'aamas_proceedings', 'external_id': '2026/p238',
'title': title, 'authors': authors, 'year': 2026, 'track': track,
'url': URL+'#R', 'abstract': None, 'identifiers': {},
'observed_at': evidence['observed_at'], 'metadata': metadata,
'publication_state': 'retracted', 'retraction_date': '2026-06-08',
'raw_publication_state': retractions[0],
'hold_reason': 'Explicit retraction; not an accepted/withdrawn submission assertion. Dedicated decision modeling and identity review required.'})
continue
if not withdrawn:
match = PDF.fullmatch(href)
if not match or match[1] in seen_pdf_ids:
raise ValueError('Invalid or repeated publisher PDF ID')
seen_pdf_ids.add(match[1])
metadata.update({'publisher_document_id': match[1], 'paper_url': ROOT+'pdfs/'+match[1]+'.pdf'})
if section == 'K':
excluded.append({'publisher_page': page, 'title': title, 'authors': authors,
'metadata': metadata, 'reason': 'Publisher keynote, not a research submission.'})
continue
metadata['track_scope_note'] = ('Published format known; individual main/special submission track not established.'
if track == 'proceedings_unspecified' else 'Explicit publisher section.')
if section == 'A':
metadata['prior_submission_note'] = 'AAAI transfer section does not supply original submission IDs or individual prior decision notes; none inferred.'
if section == 'J':
metadata['track_scope_note'] = 'Journal presentation; not another independent journal review or research-track acceptance.'
records.append({'source': 'aamas_proceedings', 'external_id': '2026/p'+str(page),
'url': URL+'#'+section, 'title': title, 'abstract': None,
'authors': authors, 'identifiers': {}, 'venue': 'AAMAS', 'venue_aliases': [],
'year': 2026, 'track': track, 'status': 'withdrawn' if withdrawn else 'accepted',
'observed_at': evidence['observed_at'],
'raw_decision': retractions[0] if withdrawn else 'listed in official AAMAS 2026 published proceedings '+heading,
'metadata': metadata,
'tags': [{'namespace': 'aamas_proceedings:publication_format', 'value': publication_format}]})
if headers != list(SECTIONS) or dict(counts) != {k: v[3] for k, v in SECTIONS.items()}:
raise ValueError('Publisher section inventory differs from reviewed 2026 layout')
if (len(records), len(excluded), len(held)) != (639, 3, 1):
raise ValueError('Unexpected accepted/keynote/retraction partition')
if Counter(r['status'] for r in records) != {'accepted': 638, 'withdrawn': 1}:
raise ValueError('Unexpected publisher publication-state partition')
return records, excluded, held
def read_page(cache, snapshots):
if bool(cache) == bool(snapshots):
raise ValueError('Choose exactly one cache or frozen snapshot directory')
if snapshots:
base = Path(snapshots).resolve()
frozen = json.loads((base/'receipts.json').read_text())['receipts']
r = one([r for r in frozen if r['url'] == URL], 'frozen response receipt')
filename = Path(r['file'])
if (filename.is_absolute() or '..' in filename.parts or
not (base/filename).resolve().is_relative_to(base) or r.get('status') != 200):
raise ValueError('Unsafe or unsuccessful publisher response')
data = (base/filename).read_bytes()
if len(data) > 8*1024**2 or len(data) != r['bytes'] or hashlib.sha256(data).hexdigest() != r['sha256']:
raise ValueError('Frozen publisher response checksum/size mismatch')
observed = r['finished_at']
else:
data = fetch(URL, cache, 8*1024**2, delay=1)
r = json.loads((Path(cache)/hashlib.sha256(URL.encode()).hexdigest()).with_suffix('.json').read_text())
observed = r['retrieved_at']
timestamp(observed)
return data, {'url': URL, 'sha256': hashlib.sha256(data).hexdigest(), 'bytes': len(data),
'file': 'evidence/'+hashlib.sha256(data).hexdigest()+'.html',
'observed_at': observed, 'http_status': 200, 'response_receipt': r}
def collect(out, *, cache=None, snapshots=None):
data, evidence = read_page(cache, snapshots)
records, excluded, held = parse_contents(data.decode('utf-8'), evidence)
grouped = defaultdict(list)
for row in records:
grouped[row['track']].append(row)
out = Path(out); out.mkdir(parents=True, exist_ok=True); files = []
with tempfile.TemporaryDirectory(prefix='.aamas2026-', dir=out) as temporary:
stage = Path(temporary); (stage/'evidence').mkdir()
(stage/evidence['file']).write_bytes(data)
for track, group in sorted(grouped.items()):
p = stage/f'aamas-proceedings-2026-{track}.jsonl'
p.write_text(''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in group))
manifest = {'source': 'aamas_proceedings', 'parser_version': 3, 'venue': 'AAMAS', 'year': 2026,
'track': track, 'count': len(group), 'file': str(out/p.name),
'file_sha256': hashlib.sha256(p.read_bytes()).hexdigest(), 'list_url': URL,
'status': 'partial', 'missing_abstract_count': len(group),
'status_counts': dict(Counter(r['status'] for r in group)),
'publication_format_counts': dict(Counter(r['metadata']['publication_format'] for r in group)),
'note': 'Reviewed TOC partition excluding keynote/retracted entries; identity and full submission history remain unverified.'}
p.with_suffix('.manifest.json').write_text(json.dumps(manifest, ensure_ascii=False, indent=2)+'\n')
files.append(manifest)
result = {'source': 'aamas_proceedings', 'year': 2026, 'finished_at': timestamp(),
'status': 'collected_pending_identity_review', 'records': len(records), 'files': files,
'excluded_publications': excluded, 'held_publications': held,
'status_counts': dict(Counter(r['status'] for r in records)),
'source_receipts': [evidence], 'database_written': False, 'pdfs_downloaded': False,
'note': 'Retraction evidence held separately; publication is not an original decision note, and absence is not rejection.'}
report = 'aamas-proceedings-2026.json'
(stage/report).write_text(json.dumps(result, ensure_ascii=False, indent=2)+'\n')
(out/'evidence').mkdir(exist_ok=True)
for p in (stage/'evidence').iterdir():
p.replace(out/'evidence'/p.name)
for p in stage.iterdir():
if p.is_file() and p.name != report:
p.replace(out/p.name)
(stage/report).replace(out/report)
return result
+183
View File
@@ -0,0 +1,183 @@
"""Frozen CoRL 2026 official program joined to attributed public archive metadata."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
import re
import unicodedata
from urllib.parse import parse_qs, urlsplit
from paper_atlas.io.html import Document, clean
SOURCE = 'corl_program_archive'
OFFICIAL = 'https://2026.corl.org/program/accepted-papers'
REVISION = '75b3f7a04b3a77c454540fe2506bfa615b8e038e'
ARCHIVE = f'https://raw.githubusercontent.com/gisbi-kim/conference-explorers/{REVISION}/corl2026/data.json'
OFFICIAL_SHA = '3f026639c1a777b0b5fbd132e1935a27cdd519ca0a04e60ca3b650a3bc7eaef4'
FILTERED_SHA = '6e6d7ecd57ee139c61283188fec39ebe9dc7f5c30ec14c8c5839d1277b03a588'
RAW_SHA = '1e96ed2dffcd96c2584a7ba020201c831e49ae1a69eae3ef1553fd5a34802810'
BLOB = '322eeef71126b85e13a98fdce9c73fac81934d3f'
FIELDS = {'id', 'title', 'openreview', 'poster_session', 'poster_day', 'poster_date',
'poster_time', 'poster_venue', 'poster_number', 'index', 'presentation_type',
'arxiv', 'matched_title', 'abstract', 'metadata_source', 'author_source',
'abstract_source', 'identity_note', 'authors', 'openreview_updated', 'metadata_checked'}
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def comparable(text):
return ''.join(c for c in unicodedata.normalize('NFKD', text).casefold() if c.isalnum())
def forum_id(url):
if not isinstance(url, str):
raise ValueError('Forum URL must be text')
u = urlsplit(url); q = parse_qs(u.query, keep_blank_values=True)
if (u.scheme != 'https' or u.netloc != 'openreview.net' or u.path != '/forum'
or u.fragment or set(q) != {'id'} or len(q['id']) != 1
or not re.fullmatch(r'[A-Za-z0-9_-]{8,32}', q['id'][0])):
raise ValueError('Invalid original forum identity')
return q['id'][0]
def parse_program(raw):
embeds = [n.attrs['data-code'] for n in Document(raw).root.walk() if 'data-code' in n.attrs]
if len(embeds) != 1:
raise ValueError('Expected one official embedded program')
doc = Document(embeds[0]); nodes = list(doc.root.walk())
if [clean(n.text()) for n in nodes if n.tag == 'h1'] != ['CoRL 2026 Accepted Papers']:
raise ValueError('Unexpected program edition')
summary = [clean(n.text()) for n in nodes if n.tag == 'p' and n.attrs.get('class') == 'sub']
if len(summary) != 1 or not summary[0].startswith('687 accepted papers · 21 orals · 72 spotlights · 687 posters'):
raise ValueError('Reviewed program census changed')
papers = {}; sessions = {}; occurrences = []
for section in (n for n in nodes if n.tag == 'section' and 'session' in n.attrs.get('class', '').split()):
sid = section.attrs.get('data-sess', section.attrs.get('id'))
if not isinstance(sid, str) or not re.fullmatch(r'(?:O[1-7]|S[1-8]|post-p[1-3])', sid) or sid in sessions:
raise ValueError('Unexpected or repeated presentation session')
kind = 'oral' if sid.startswith('O') else 'spotlight' if sid.startswith('S') else 'poster'
headings = [clean(n.text()) for n in section.walk() if n.tag == 'h3']
if len(headings) != 1:
raise ValueError('Ambiguous session heading')
sessions[sid] = {'kind': kind, 'heading': headings[0]}
seen = set()
for link in (n for n in section.walk() if n.tag == 'a' and 'openreview.net' in n.attrs.get('href', '')):
fid = forum_id(link.attrs['href']); title = clean(link.text())
if fid in seen or not title:
raise ValueError('Repeated or empty paper in session')
seen.add(fid)
paper = papers.setdefault(fid, {'title': title, 'presentations': []})
if paper['title'] != title:
raise ValueError('Conflicting official titles share a forum')
event = {'session_id': sid, **sessions[sid]}
paper['presentations'].append(event); occurrences.append((fid, sid, kind))
counts = Counter(kind for _, _, kind in occurrences)
if len(papers) != 687 or len(sessions) != 18 or counts != {'poster': 687, 'oral': 21, 'spotlight': 72}:
raise ValueError('Incomplete official program')
for paper in papers.values():
kinds = Counter(p['kind'] for p in paper['presentations'])
if kinds['poster'] != 1 or len(paper['presentations']) not in (1, 2):
raise ValueError('Invalid poster/talk multiplicity')
return papers
def join_metadata(program, archive):
if (archive.get('repository') != 'gisbi-kim/conference-explorers'
or archive.get('revision') != REVISION or archive.get('source_url') != ARCHIVE
or archive.get('source_data_sha256') != RAW_SHA or archive.get('source_git_blob_sha1') != BLOB
or not isinstance(archive.get('papers'), list)):
raise ValueError('Unpinned public metadata provenance')
indexed = {}
for row in archive['papers']:
if not isinstance(row, dict) or set(row) - FIELDS:
raise ValueError('Unexpected metadata/profile fields')
fid = forum_id(row.get('openreview'))
if fid != row.get('id') or fid in indexed:
raise ValueError('Repeated or mismatched archive forum')
if any(forum_id(row.get(key)) != fid for key in ('metadata_source', 'author_source', 'abstract_source')):
raise ValueError('Field source belongs to another submission')
if (not isinstance(row.get('title'), str) or not row['title'].strip()
or not isinstance(row.get('abstract'), str) or not row['abstract'].strip()
or not isinstance(row.get('authors'), list) or not row['authors']
or any(not isinstance(a, str) or not a.strip() or '@' in a or 'anonymous' in a.casefold() for a in row['authors'])):
raise ValueError('Incomplete scholarly metadata')
indexed[fid] = row
if set(indexed) != set(program):
raise ValueError('Official and archive forum sets differ')
return indexed
def collect(out, *, official_snapshots, archive_snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
try:
official_root = Path(official_snapshots); archive_root = Path(archive_snapshots)
official = official_root/'accepted.html'; filtered = archive_root/'explorer-paper-metadata-filtered-v1.json'
if official.is_symlink() or filtered.is_symlink():
raise ValueError('Snapshot symlinks rejected')
raw = official.read_bytes(); data = filtered.read_bytes()
if len(raw) > 8*1024**2 or len(data) > 8*1024**2 or digest(raw) != OFFICIAL_SHA or digest(data) != FILTERED_SHA:
raise ValueError('Reviewed snapshot hash changed')
receipts = json.loads((official_root/'research.json').read_bytes())['receipts']
matches = [r for r in receipts if r.get('url') == OFFICIAL]
if len(matches) != 1:
raise ValueError('Missing official source receipt')
receipt = matches[0]; observed = receipt['at']
if (receipt.get('final_url') != OFFICIAL or receipt['http_status'] != 200
or receipt['sha256'] != digest(raw) or receipt['bytes'] != len(raw)
or datetime.fromisoformat(observed.replace('Z', '+00:00')).utcoffset() is None):
raise ValueError('Invalid official receipt')
program = parse_program(raw.decode('utf-8')); archive = json.loads(data); indexed = join_metadata(program, archive)
evidence = out/'evidence'; evidence.mkdir()
sources = {}
for body, suffix, url, name in [(raw, 'html', OFFICIAL, 'program'), (data, 'json', ARCHIVE, 'archive')]:
rel = f'evidence/{digest(body)}.{suffix}'; (out/rel).write_bytes(body)
sources[name] = {'file': rel, 'sha256': digest(body), 'url': url}
rows = []; variants = []
for fid in sorted(program):
p = program[fid]; a = indexed[fid]; variant = comparable(p['title']) != comparable(a['title'])
if variant: variants.append({'forum': fid, 'official_title': p['title'], 'archive_title': a['title']})
rows.append({'source': SOURCE, 'external_id': f'2026/{fid}', 'url': a['openreview'],
'title': a['title'], 'authors': a['authors'], 'abstract': a['abstract'],
'identifiers': {'openreview': fid}, 'submission_identifiers': [{'scheme': 'openreview', 'value': fid}],
'venue': 'CoRL', 'year': 2026, 'track': 'main', 'status': 'accepted',
'raw_decision': 'listed in official CoRL 2026 accepted program', 'evidence_url': OFFICIAL,
'observed_at': observed, 'tags': [], 'metadata': {
'official_program': p, 'archive_fields': a, 'title_variant_requires_review': variant,
'source_evidence': sources, 'archive_revision': REVISION,
'field_provenance': {k: a[k] for k in ('metadata_source', 'author_source', 'abstract_source')},
'original_decision_signature_verified': False,
'scope': 'Acceptance from official forum listing; title/authors/abstract are curator snapshot versions, not live original metadata. arXiv links remain unapproved candidates. No reject inference or person aliases.'}})
if len(variants) != 34:
raise ValueError('Reviewed title-variant census changed')
path = out/'corl-program-archive-2026-main.jsonl'
payload = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in rows).encode(); path.write_bytes(payload)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'CoRL', 'year': 2026, 'track': 'main',
'count': len(rows), 'file': str(path), 'file_sha256': digest(payload), 'list_url': OFFICIAL,
'status': 'partial', 'missing_abstract_count': 0}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n')
report = {'status': 'collected_pending_identity_review', 'records': len(rows), 'presentations': 780,
'files': [manifest], 'title_variants': variants, 'source_evidence': sources,
'database_written': False, 'pdfs_downloaded': False, 'public_rejected_availability': 'unknown',
'scope': 'Source collection only; requires identity and metadata-version review before import; no exhaustive deduplication or rejected-set claim.'}
part = out/'collection-report.json.part'; part.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n'); part.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--out', type=Path, required=True)
parser.add_argument('--official-snapshots', type=Path, required=True)
parser.add_argument('--archive-snapshots', type=Path, required=True)
args = parser.parse_args(argv)
report = collect(args.out, official_snapshots=args.official_snapshots, archive_snapshots=args.archive_snapshots)
print(json.dumps({'status': report['status'], 'records': report['records'], 'database_written': False}))
+254
View File
@@ -0,0 +1,254 @@
"""ECCV 2026 frozen preliminary program and resumable public metadata pages."""
import argparse
from collections import Counter
from datetime import datetime, timezone
import fcntl
from html import unescape
import json
from pathlib import Path
import re
from paper_atlas.io.html import Document, Node, clean
from paper_atlas.io.html_cache import PublicHTMLCache, checked_html, digest
SOURCE = 'eccv_program'
ORIGIN = 'https://eccv.ecva.net'
ACCEPTED = ORIGIN+'/Conferences/2026/AcceptedPapers'
TRACK = 'program_unclassified'
MANIFEST_SHA = '04f9734783bbbfa2e19f936d8f5a7aecfb524b0ceef33fc5d9fb6276c7b6678a'
PAGES = {'accepted.html': ACCEPTED, 'policies.html': ORIGIN+'/Conferences/2026/SubmissionPolicies',
'faq.html': ORIGIN+'/Conferences/2026/FAQs'}
def one(items):
if len(items) != 1:
raise ValueError('Expected one reviewed metadata field')
return items[0]
def has_class(node, value):
return value in node.attrs.get('class', '').split()
def parse_program(raw):
nodes = list(Document(raw).root.walk())
if len([n for n in nodes if n.tag == 'h2' and clean(n.text()) == 'ECCV 2026 Accepted Papers:']) != 1:
raise ValueError('Unexpected ECCV program edition')
if 'This paper list is preliminary pending publisher checks.' not in clean(Document(raw).root.text()):
raise ValueError('Reviewed provisional-publication notice changed')
table = one([n for n in nodes if n.tag == 'table' and n.attrs.get('id') == 'event-list-2026-poster-nodates-filter-vslinks-table'])
papers = {}; occurrences = 0
for tr in (n for n in table.walk() if n.tag == 'tr'):
cells = [n for n in tr.children if isinstance(n, Node) and n.tag == 'td']
if not cells:
continue
if len(cells) != 3:
raise ValueError('Unexpected program columns')
links = [n for n in cells[0].walk() if n.tag == 'a' and re.fullmatch(r'/virtual/2026/poster/[1-9][0-9]*', n.attrs.get('href', ''))]
link = one(links); pid = link.attrs['href'].rsplit('/', 1)[1]
authors = clean(one([n for n in cells[0].walk() if n.tag == 'i']).text())
row = {'title': clean(link.text()), 'authors_literal': authors,
'keywords_literal': clean(cells[1].text()), 'presentation_literal': clean(cells[2].text())}
if not row['title'] or not authors:
raise ValueError('Incomplete public program fields')
occurrences += 1
if pid in papers:
if papers[pid]['listing'] != row:
raise ValueError('Conflicting entries share a program ID')
papers[pid]['row_ordinals'].append(occurrences)
else:
papers[pid] = {'url': ORIGIN+link.attrs['href'], 'listing': row, 'row_ordinals': [occurrences]}
if occurrences != 2851 or len(papers) != 2834 or Counter(len(v['row_ordinals']) for v in papers.values()) != {1: 2817, 2: 17}:
raise ValueError('Incomplete reviewed program census')
return papers
def parse_paper(raw, url, program):
nodes = list(Document(raw).root.walk())
title = clean(one([n for n in nodes if n.tag == 'h1' and has_class(n, 'event-title')]).text())
authors_literal = clean(one([n for n in nodes if n.tag == 'div' and has_class(n, 'event-organizers')]).text())
abstract_node = one([n for n in nodes if n.tag == 'div' and has_class(n, 'abstract-text-inner')])
abstract = clean(abstract_node.text())
markup = sorted({n.tag for n in abstract_node.walk() if n.tag in ('math','sup','sub','svg','img')})
canonical = one([n.attrs.get('href') for n in nodes if n.tag == 'link' and n.attrs.get('rel') == 'canonical'])
scripts = [n for n in nodes if n.tag == 'script' and n.attrs.get('type') == 'application/ld+json']
ld = one([json.loads(''.join(x for x in n.children if isinstance(x, str))) for n in scripts])
authors = ld.get('author')
if (canonical != url or ld.get('@type') != 'CreativeWork' or ld.get('creditText') != 'ECCV 2026'
or not title or not isinstance(ld.get('name'), str) or not abstract
or not isinstance(authors, list) or not authors
or any(not isinstance(a, dict) or a.get('@type') != 'Person' or not isinstance(a.get('name'), str)
or not clean(a['name']) or '@' in a['name'] for a in authors)):
raise ValueError('Public visible/structured metadata identity or edition disagrees')
# Prefer source text that already agrees. Poster 3844 intentionally displays
# literal '&amp;', while 4199 double-encodes an author's apostrophe in JSON-LD.
# Decode only to test agreement with visible text, never to rewrite it.
def agrees_with_display(value, expected):
if clean(value) == expected:
return True
for _ in range(2):
value = unescape(value)
if clean(value) == expected:
return True
return False
raw_names = [a['name'] for a in authors]
names = [clean(name) for name in raw_names]
if not agrees_with_display(ld['name'], title):
raise ValueError('Public visible and structured titles disagree')
if ' ⋅ '.join(names) != authors_literal:
visible_names = authors_literal.split(' ⋅ ')
if len(visible_names) != len(raw_names) or any(
not agrees_with_display(raw_name, visible_name)
for raw_name, visible_name in zip(raw_names, visible_names)):
raise ValueError('Complete visible and structured authors disagree')
names = visible_names
if any('@' in name for name in names) or ' ⋅ '.join(names) != authors_literal:
raise ValueError('Complete visible and structured authors disagree')
flags = ['structured_abstract_markup_requires_review'] if markup else []
if raw_names != names or ld['name'] != title:
flags.append('jsonld_display_text_normalized_requires_review')
if title != program['listing']['title']:
flags.append('program_and_page_title_versions_differ_requires_review')
if authors_literal != program['listing']['authors_literal']:
flags.append('program_and_page_author_versions_differ_requires_review')
return {'title': title, 'authors': names, 'authors_literal': authors_literal, 'abstract': abstract,
'page_metadata_dates': {k: ld.get(k) for k in ('datePublished', 'dateModified')}, 'review_flags': flags, 'abstract_markup_tags': markup,
'jsonld_raw_text': {'name': ld['name'], 'author_names': raw_names}}
def official_snapshot(root):
root = Path(root); manifest = root/'manifest.json'
if root.is_symlink() or manifest.is_symlink() or digest(manifest.read_bytes()) != MANIFEST_SHA:
raise ValueError('Reviewed official snapshot manifest changed')
m = json.loads(manifest.read_bytes())
if datetime.fromisoformat(m['at'].replace('Z', '+00:00')).utcoffset() is None:
raise ValueError('Official snapshot requires a timezone')
if len(m['sources']) != 3 or {r['file'] for r in m['sources']} != set(PAGES):
raise ValueError('Official source set changed')
pages = {}; evidence = {}
for r in m['sources']:
name = r['file']; p = root/name
if p.is_symlink() or p.stat().st_size > 8*1024**2:
raise ValueError('Unsafe official snapshot')
raw = p.read_bytes()
if r['url'] != PAGES[name] or r['final_url'] != r['url'] or r['http_status'] != 200 or r['bytes'] != len(raw) or r['sha256'] != digest(raw):
raise ValueError('Invalid official source receipt')
pages[name] = raw; evidence[name] = {'url': r['url'], 'sha256': digest(raw)}
program = parse_program(pages['accepted.html'].decode('utf-8'))
return m, pages, evidence, program
def paper_snapshots(root, program):
root = Path(root); path = root/'manifest.json'
if root.is_symlink() or path.is_symlink():
raise ValueError('Paper snapshot symlinks rejected')
if path.stat().st_size > 4*1024**2:
raise ValueError('Oversized paper manifest')
m = json.loads(path.read_bytes()); result = {}
for r in m['records']:
pid = r['paper_id']
if pid not in program or pid in result or r['file'] != pid+'.html':
raise ValueError('Foreign or repeated paper snapshot')
p = root/r['file']
if p.is_symlink() or p.stat().st_size > 2*1024**2:
raise ValueError('Unsafe paper snapshot')
receipt = {**r, 'retrieved_at': r.get('retrieved_at', m['at'])}
result[pid] = checked_html(p.read_bytes(), receipt, program[pid]['url'], 2*1024**2)
return result
def collect(out, *, official_snapshots, paper_snapshots_dir=None, cache=None, max_new_pages=None):
if (paper_snapshots_dir is None) == (cache is None):
raise ValueError('Choose paper snapshots or public metadata cache')
if max_new_pages is not None and (type(max_new_pages) is not int or max_new_pages < 0 or cache is None):
raise ValueError('New-page budget requires cache mode and a nonnegative integer')
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
lock = None
try:
official, pages, source_evidence, program = official_snapshot(official_snapshots)
directory = out/'evidence'; directory.mkdir()
for name, raw in pages.items():
rel = 'evidence/'+digest(raw)+'.html'; (out/rel).write_bytes(raw); source_evidence[name]['file'] = rel
snapshots = paper_snapshots(paper_snapshots_dir, program) if paper_snapshots_dir is not None else None
client = None
if cache is not None:
cache = Path(cache); cache.mkdir(parents=True, exist_ok=True)
if cache.is_symlink() or (cache/'.collector.lock').is_symlink():
raise ValueError('Cache lock symlinks rejected')
lock = (cache/'.collector.lock').open('a+')
fcntl.flock(lock, fcntl.LOCK_EX | fcntl.LOCK_NB)
client = PublicHTMLCache(cache, ORIGIN)
rows = []; missing = []; fetched = 0; reused = 0
for pid in sorted(program, key=int):
p = program[pid]
if snapshots is not None:
response = snapshots.get(pid)
else:
response = client.cached(p['url'])
if response is not None:
reused += 1
elif max_new_pages is None or fetched < max_new_pages:
response = client.get(p['url']); fetched += 1
if response is None:
missing.append(pid); continue
raw, receipt = response; fields = parse_paper(raw.decode('utf-8'), p['url'], p)
rel = 'evidence/'+digest(raw)+'.html'; (out/rel).write_bytes(raw)
keyword = p['listing']['keywords_literal']
rows.append({'source': SOURCE, 'external_id': '2026/'+pid, 'url': p['url'],
'title': fields['title'], 'authors': fields['authors'], 'abstract': fields['abstract'],
'identifiers': {}, 'submission_identifiers': [], 'venue': 'ECCV', 'year': 2026,
'track': TRACK, 'status': 'accepted', 'observed_at': official['at'],
'raw_decision': 'listed in frozen preliminary ECCV 2026 accepted program, pending publisher checks',
'evidence_url': ACCEPTED,
'tags': [{'namespace': SOURCE+':keyword', 'value': keyword}] if keyword else [],
'metadata': {'official_program': p, 'public_page_fields': fields,
'public_page_receipt': receipt, 'public_page_evidence': {'file': rel, 'sha256': digest(raw), 'url': p['url']},
'program_source_evidence': source_evidence, 'publication_state': 'preliminary_pending_publisher_checks',
'original_decision_signature_verified': False,
'program_id_semantics': 'Official program locator, not a verified original submission number or cross-platform identifier.',
'abstract_quality': 'Public display literal retained, including broken hyphenation/missing spaces; no automatic text repair.',
'date_semantics': 'JSON-LD dates describe the website entry, not verified paper publication or decision dates.',
'label_semantics': 'Official keyword namespace independent of track, contribution/research type and custom labels.',
'scope': 'Public metadata only; no protected live content, reviews, person profiles or PDFs. Requires work identity, metadata version and complete history review before import.'}})
if len(rows) % 10 == 0:
print(json.dumps({'phase': 'public_metadata', 'records': len(rows), 'new_pages': fetched,
'cached_pages': reused, 'expected_program_ids': 2834, 'at': datetime.now(timezone.utc).isoformat()}), flush=True)
files = []
if rows:
path = out/'eccv-program-2026-unclassified.jsonl'; payload = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in rows).encode(); path.write_bytes(payload)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'ECCV', 'year': 2026, 'track': TRACK,
'count': len(rows), 'file': str(path), 'file_sha256': digest(payload), 'list_url': ACCEPTED,
'status': 'partial', 'missing_abstract_count': 0, 'source_expected_count': 2834}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n'); files.append(manifest)
report = {'status': 'collected_pending_identity_review' if rows else 'no_public_metadata_collected',
'records': len(rows), 'files': files, 'expected_program_ids': 2834, 'program_occurrences': 2851,
'duplicate_program_occurrences': 17, 'uncollected_program_ids': missing,
'new_pages': fetched, 'reused_pages': reused, 'database_written': False,
'public_review_visibility': 'not_public_per_2026_policy', 'public_rejected_decision_metadata': 'unknown',
'scope': 'Always partial: preliminary program, unresolved track/identity/version coverage; not final whole-database deduplication.'}
part = out/'collection-report.json.part'; part.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n'); part.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__,
'database_written': False, 'scope': 'Preserve this output and verified cache; resume with a new output directory after inspection.'})+'\n')
raise
finally:
if lock is not None:
lock.close()
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--official-snapshots', required=True)
group = parser.add_mutually_exclusive_group(required=True)
group.add_argument('--paper-snapshots')
group.add_argument('--cache')
parser.add_argument('--max-new-pages', type=int)
parser.add_argument('--out', required=True)
args = parser.parse_args(argv)
print(json.dumps(collect(args.out, official_snapshots=args.official_snapshots,
paper_snapshots_dir=args.paper_snapshots, cache=args.cache, max_new_pages=args.max_new_pages), ensure_ascii=False))
+155
View File
@@ -0,0 +1,155 @@
"""ICAPS 2026 frozen official listing, with unresolved track membership explicit."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
from paper_atlas.io.html import Document, Node, clean
SOURCE = 'icaps_official_program'
ROOT = 'https://icaps26.icaps-conference.org/'
PAGES = {
'accepted.html': (ROOT+'program/accepted/', '235543c3e5917ecb3e10a1429a4754f9cb7fae9844b54404076254584da86443'),
'cfp.html': (ROOT+'calls/cfp/', 'bdcf5a5b3b4144d6c6ecf3e324c856363f4f4e992e592c1f63988980434ecd5d'),
'previously-published.html': (ROOT+'calls/previously_published_research/', 'ecd60c8c056c9aa73bac16eb11f220c756ddff562511f0ec42ada41919fb1946'),
}
MAIN = 'Papers accepted to the Main Tracks'
PREVIOUS = 'Previously Published Papers'
TRACK = 'main_tracks_unclassified'
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def authors_from_literal(literal):
# This frozen listing uses commas between authors and a final "and".
# Preserve the complete literal too; never infer person-name aliases.
if literal.count(' and ') != 1 or ', and ' in literal:
raise ValueError('Unreviewed author-list punctuation')
first, last = literal.rsplit(' and ', 1)
authors = first.split(', ') + [last]
if any(not a or a != clean(a) or ',' in a or '@' in a or 'anonymous' in a.casefold() for a in authors):
raise ValueError('Incomplete or ambiguous author list')
if ', '.join(authors[:-1])+' and '+authors[-1] != literal:
raise ValueError('Author literal cannot be reconstructed')
return authors
def parse_listing(raw):
nodes = list(Document(raw).root.walk())
headings = [clean(n.text()) for n in nodes if n.tag == 'h1']
if headings != ['List of Papers Accepted to ICAPS 2026']:
raise ValueError('Unexpected official conference edition')
section = None; sections = []; rows = []; seen = set()
for n in nodes:
if n.tag == 'h2':
section = clean(n.text()); sections.append(section)
if n.tag != 'div' or 'paper' not in n.attrs.get('class', '').split():
continue
if section not in (MAIN, PREVIOUS):
raise ValueError('Paper belongs to an unreviewed section')
# Require the reviewed title and complete-author blocks; fail on layout drift.
blocks = [b for b in n.children if isinstance(b, Node)]
if len(blocks) != 2 or any(b.tag != 'div' for b in blocks):
raise ValueError('Changed paper metadata layout')
titles = [b for b in blocks[0].walk() if b.tag == 'strong']
if len(titles) != 1 or any(b.tag == 'a' for b in n.walk()):
raise ValueError('Changed paper title/link layout')
title = clean(titles[0].text()); literal = clean(blocks[1].text())
if not title or clean(blocks[0].text()) != title or title.casefold() in seen:
raise ValueError('Empty or repeated paper title')
seen.add(title.casefold())
rows.append({'section': section, 'title': title, 'authors': authors_from_literal(literal),
'authors_literal': literal})
if sections != [MAIN, PREVIOUS] or Counter(r['section'] for r in rows) != {MAIN: 90, PREVIOUS: 10}:
raise ValueError('Incomplete reviewed official listing')
return rows
def collect(out, *, snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
try:
root = Path(snapshots)
manifest_path = root/'manifest.json'
if manifest_path.is_symlink():
raise ValueError('Snapshot symlinks rejected')
receipt = json.loads(manifest_path.read_bytes())
observed = receipt['at']
if (receipt.get('status') != 'collected_not_parsed_not_imported'
or receipt.get('database_written') is not False
or datetime.fromisoformat(observed.replace('Z', '+00:00')).utcoffset() is None):
raise ValueError('Invalid reviewed snapshot receipt')
receipts = receipt['sources']
if len(receipts) != 3 or {r['file'] for r in receipts} != set(PAGES):
raise ValueError('Unexpected snapshot source set')
bodies = {}; evidence = {}; directory = out/'evidence'; directory.mkdir()
for name, (url, expected) in PAGES.items():
matches = [r for r in receipts if r['file'] == name]
if len(matches) != 1:
raise ValueError('Repeated snapshot receipt')
r = matches[0]; p = root/name
if p.is_symlink() or p.stat().st_size > 2*1024**2:
raise ValueError('Unsafe or oversized snapshot')
raw = p.read_bytes()
if digest(raw) != expected or r['sha256'] != expected or r['bytes'] != len(raw):
raise ValueError('Reviewed snapshot hash changed')
if r['url'] != url or r['final_url'] != url:
raise ValueError('Unexpected snapshot URL or redirect')
rel = f'evidence/{expected}.html'; (out/rel).write_bytes(raw)
bodies[name] = raw.decode('utf-8')
evidence[name] = {'url': url, 'file': rel, 'sha256': expected}
entries = parse_listing(bodies['accepted.html']); rows = []; previous = []
for entry in entries:
entry_key = digest(json.dumps([entry['title'], entry['authors_literal']], ensure_ascii=False, separators=(',', ':')).encode())
if entry['section'] == PREVIOUS:
previous.append({**entry, 'entry_key': entry_key,
'presentation_kind': 'previously_published_research',
'evidence_url': PAGES['accepted.html'][0], 'observed_at': observed,
'scope': 'Official presentation listing, not evidence of new archival publication or a new research-paper acceptance.'})
continue
rows.append({'source': SOURCE, 'external_id': '2026/main-tracks/'+entry_key,
'url': PAGES['accepted.html'][0], 'title': entry['title'], 'authors': entry['authors'],
'abstract': None, 'identifiers': {}, 'submission_identifiers': [],
'venue': 'ICAPS', 'year': 2026, 'track': TRACK, 'status': 'accepted',
'raw_decision': 'listed in official ICAPS 2026 Papers accepted to the Main Tracks',
'evidence_url': PAGES['accepted.html'][0], 'observed_at': observed, 'tags': [],
'metadata': {'official_listing': entry, 'source_evidence': evidence,
'source_entry_key': {'algorithm': 'sha256-json-title-and-literal-authors', 'value': entry_key,
'scope': 'Snapshot entry locator only; not an original submission ID, DOI, or cross-source work identity.'},
'track_assignment': 'Unresolved among main and four special tracks; do not relabel all records as main.',
'abstract_availability': 'not_on_official_listing',
'original_decision_signature_verified': False,
'scope': 'Official listing evidence only; requires current identity review before import. No absence-based rejection or new-paper inference.'}})
path = out/'icaps-official-program-2026-main-tracks-unclassified.jsonl'
payload = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in rows).encode(); path.write_bytes(payload)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'ICAPS', 'year': 2026, 'track': TRACK,
'count': 90, 'file': str(path), 'file_sha256': digest(payload), 'list_url': PAGES['accepted.html'][0],
'status': 'partial', 'missing_abstract_count': 90}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n')
sidecar = out/'previously-published-presentations.json'
sidecar.write_text(json.dumps({'records': previous, 'count': 10, 'ingestion_ready': False,
'scope': 'Presentation evidence retained separately; not emitted as new accepted research-paper records.'}, ensure_ascii=False, indent=2)+'\n')
report = {'status': 'collected_pending_identity_review', 'records': 90, 'previously_published_presentations': 10,
'files': [manifest], 'previously_published_file': sidecar.name, 'previously_published_sha256': digest(sidecar.read_bytes()),
'source_evidence': evidence, 'database_written': False, 'pdfs_downloaded': False,
'public_rejected_availability': 'unknown', 'missing_abstract_count': 90,
'scope': 'Partial official list; individual tracks and stable identities unresolved; no all-paper, public-negative or final-dedup claim.'}
part = out/'collection-report.json.part'; part.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n'); part.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--snapshots', required=True)
parser.add_argument('--out', required=True)
args = parser.parse_args(argv)
print(json.dumps(collect(args.out, snapshots=args.snapshots), ensure_ascii=False))
@@ -0,0 +1,131 @@
"""Frozen Crossref ICAPS volume 36 metadata, kept separate from program versions."""
import argparse
from datetime import datetime, date
import hashlib
from html import unescape
import json
from pathlib import Path
import re
from paper_atlas.io.html import clean
SOURCE = 'crossref_icaps_registry'
SNAPSHOT_SHA = 'c7480a09f71680d457a16cc1eb35a479742b7510eaad5f236ce54c96e8a29a82'
RAW_RESPONSE_SHA = 'ac5268333b1ecdfa2e385680e2e02b5f6639939c3193738d213e66bb53e42a54'
URL = 'https://api.crossref.org/journals/2334-0843/works?filter=from-pub-date%3A2026-01-01%2Cuntil-pub-date%3A2026-12-31&rows=200'
CONTAINER = 'Proceedings of the International Conference on Automated Planning and Scheduling'
TRACK = 'main_tracks_unclassified'
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def abstract_text(raw):
if not isinstance(raw, str):
raise ValueError('Missing registry abstract')
match = re.fullmatch(r'<jats:p>(.*?)</jats:p>', raw, re.S)
if not match or re.search(r'<[^>]*>', match[1]):
raise ValueError('Unreviewed abstract markup; preserve rather than flatten math')
text = clean(unescape(match[1]))
if not text:
raise ValueError('Empty registry abstract')
return text
def registry_papers(snapshot):
observed = snapshot['at']
if (snapshot.get('status') != 'registry_snapshot_not_identity_approved_not_imported'
or snapshot.get('url') != URL or snapshot.get('raw_response_sha256') != RAW_RESPONSE_SHA
or snapshot.get('database_written') is not False
or datetime.fromisoformat(observed.replace('Z', '+00:00')).utcoffset() is None
or snapshot.get('total_results') != 91 or len(snapshot.get('records', [])) != 91):
raise ValueError('Unreviewed registry snapshot receipt')
papers = []; excluded = []; seen = set()
for row in snapshot['records']:
doi = row.get('DOI')
if not isinstance(doi, str) or doi in seen:
raise ValueError('Missing or repeated registry DOI')
seen.add(doi)
if row.get('volume') == '35':
excluded.append(doi)
continue
if (row.get('volume') != '36' or row.get('issue') != '1'
or row.get('container-title') != [CONTAINER] or row.get('type') != 'journal-article'
or not re.fullmatch(r'10\.1609/icaps\.v36i1\.[0-9]+', doi)
or row.get('URL') != 'https://doi.org/'+doi
or not isinstance(row.get('title'), list) or len(row['title']) != 1
or not isinstance(row['title'][0], str) or not clean(row['title'][0])):
raise ValueError('Unreviewed publication volume, DOI, type or title')
day, = row['published']['date-parts']
if len(day) != 3 or day[0] != 2026 or any(type(x) is not int for x in day):
raise ValueError('Unreviewed publication date')
publication_date = date(*day).isoformat()
authors = row.get('authors')
if (not isinstance(authors, list) or not authors
or any(not isinstance(a, dict) or set(a)-{'given','family','name','sequence'}
or not a.get('given') or not a.get('family') for a in authors)):
raise ValueError('Incomplete registry authors')
names = [clean(a['given']+' '+a['family']) for a in authors]
if (len(set(n.casefold() for n in names)) != len(names)
or any('@' in n or 'anonymous' in n.casefold() for n in names)
or [a.get('sequence') for a in authors] != ['first']+['additional']*(len(authors)-1)):
raise ValueError('Unreviewed registry author order or identity')
papers.append({'original': row, 'title': row['title'][0], 'authors': names,
'abstract': abstract_text(row.get('abstract')), 'publication_date': publication_date})
if len(papers) != 90 or len(excluded) != 1:
raise ValueError('Incomplete reviewed volume census')
return sorted(papers, key=lambda p: p['original']['DOI']), excluded
def collect(out, *, snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
try:
path = Path(snapshots)/'crossref-filtered.json'
if path.is_symlink() or path.stat().st_size > 4*1024**2:
raise ValueError('Unsafe or oversized registry snapshot')
raw = path.read_bytes()
if digest(raw) != SNAPSHOT_SHA:
raise ValueError('Reviewed registry snapshot hash changed')
snapshot = json.loads(raw); papers, excluded = registry_papers(snapshot)
directory = out/'evidence'; directory.mkdir(); rel = f'evidence/{digest(raw)}.json'; (out/rel).write_bytes(raw)
evidence = {'file': rel, 'sha256': digest(raw), 'url': URL}; rows = []
for paper in papers:
original = paper['original']; doi = original['DOI']
rows.append({'source': SOURCE, 'external_id': doi, 'url': original['URL'],
'title': paper['title'], 'authors': paper['authors'], 'abstract': paper['abstract'],
'identifiers': {'doi': doi}, 'submission_identifiers': [{'scheme': 'publication_doi', 'value': doi}],
'venue': 'ICAPS', 'year': 2026, 'track': TRACK, 'status': 'accepted',
'raw_decision': 'publication recorded in Crossref ICAPS volume 36 issue 1; not an original signed acceptance decision',
'evidence_url': URL, 'observed_at': snapshot['at'], 'tags': [],
'metadata': {'registry_fields': original, 'source_evidence': evidence,
'abstract_transform': 'Remove one reviewed jats:p wrapper, decode character entities and normalize whitespace; original JATS retained.',
'publication_date': paper['publication_date'], 'original_decision_signature_verified': False,
'track_assignment': 'Unresolved main/special-track membership; no per-paper track inferred from topic or publication order.',
'program_relationship': 'Separate registry version; not substituted into official-listing records. Title/author variants require review.',
'scope': 'Registered publication metadata, not original review decision or current live publisher page. Requires current work-identity and history review before import.'}})
path = out/'icaps-crossref-registry-2026-main-tracks-unclassified.jsonl'
payload = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in rows).encode(); path.write_bytes(payload)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'ICAPS', 'year': 2026, 'track': TRACK,
'count': 90, 'file': str(path), 'file_sha256': digest(payload), 'list_url': URL, 'status': 'partial', 'missing_abstract_count': 0}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n')
report = {'status': 'collected_pending_identity_review', 'records': 90, 'files': [manifest],
'excluded_volume35_dois': excluded, 'missing_abstract_count': 0, 'source_evidence': evidence,
'database_written': False, 'pdfs_downloaded': False, 'public_rejected_availability': 'unknown',
'scope': 'Crossref publication versions only; no automatic official-program linkage, person aliases, work merges or complete public-negative/final-dedup claim.'}
part = out/'collection-report.json.part'; part.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n'); part.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--snapshots', required=True)
parser.add_argument('--out', required=True)
args = parser.parse_args(argv)
print(json.dumps(collect(args.out, snapshots=args.snapshots), ensure_ascii=False))
+191
View File
@@ -0,0 +1,191 @@
"""Frozen ICDM2026 in-person program, with explicit source track evidence."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
import re
from paper_atlas.io.html import Document, Node, clean
SOURCE = 'icdm_program'
SNAPSHOT = 'remaining2026-official-leads-v1'
MANIFEST_SHA = '0e13313d94e3bff17f071d3a1384f6ed83ece156239fe5bc4633cc7a34fb8e04'
PROGRAM = 'https://icdm2026.neu.edu.cn/DetailedSessions/list.htm'
PAGES = {
'detailed-sessions': PROGRAM,
'research-policy': 'https://icdm2026.neu.edu.cn/CallforResearchTrack/list.htm',
'applied-policy': 'https://icdm2026.neu.edu.cn/CallforAppliedTrack/list.htm',
}
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def one(items):
if len(items) != 1:
raise ValueError('Expected exactly one reviewed field')
return items[0]
def has_class(node, name):
return name in node.attrs.get('class', '').split()
def field(node, name):
return clean(one([n for n in node.walk() if has_class(n, name)]).text())
def load_snapshot(root):
root = Path(root); folder = root/SNAPSHOT; manifest = folder/'manifest.json'
if root.is_symlink() or folder.is_symlink() or manifest.is_symlink() or manifest.stat().st_size > 256*1024:
raise ValueError('Unsafe source manifest')
raw = manifest.read_bytes()
if digest(raw) != MANIFEST_SHA:
raise ValueError('Frozen manifest hash changed')
data = json.loads(raw)
if data['status'] != 'research_completed_not_coverage_approval' or data['year'] != 2026:
raise ValueError('Wrong source edition or incomplete research run')
receipts = [r for r in data['entries'] if r['venue'] == 'ICDM']
if len(receipts) != 3 or {r['label'] for r in receipts} != PAGES.keys():
raise ValueError('Incomplete ICDM source set')
files = {'manifest.json': raw}; indexed = {}
for receipt in receipts:
filename = receipt['file']; path = folder/filename
if Path(filename).name != filename or path.is_symlink() or path.stat().st_size > 8*1024**2:
raise ValueError('Unsafe source file')
body = path.read_bytes()
if (receipt['url'] != PAGES[receipt['label']] or receipt['final_url'] != receipt['url']
or receipt['http_status'] != 200 or receipt['bytes'] != len(body)
or receipt['sha256'] != digest(body)
or not receipt['content_type'].lower().startswith('text/html')
or datetime.fromisoformat(receipt['at']).utcoffset() is None):
raise ValueError('Source receipt/body mismatch')
if filename in files:
raise ValueError('Repeated source filename')
files[filename] = body; indexed[receipt['label']] = receipt
return indexed, files
def parse_program(raw):
doc = Document(raw.decode('utf-8'))
main = one([n for n in doc.root.walk() if n.tag == 'main' and n.attrs.get('id') == 'session-details'])
text = clean(main.text())
if ('IEEE ICDM 2026' not in text or 'November 13–15, 2026' not in text
or 'In-person session presentation list.' not in text or '34 sessions · 324 papers' not in text):
raise ValueError('Wrong edition or unreviewed program scope')
sessions = [n for n in main.walk() if n.tag == 'details' and has_class(n, 'sd-session')]
expected = {f'L{i}' for i in range(1, 21)} | {f'S{i}' for i in range(1, 15)}
if len(sessions) != 34 or {n.attrs.get('data-session-id') for n in sessions} != expected:
raise ValueError('Incomplete or repeated session census')
records = []; seen = set()
for session in sessions:
sid = session.attrs['data-session-id']; track = session.attrs.get('data-track')
form = session.attrs.get('data-format')
if (session.attrs.get('id') != sid or track not in ('research', 'applied')
or form not in ('long', 'short')):
raise ValueError('Invalid session ID, track or presentation label')
label = 'Research Track' if track == 'research' else 'Applied Track'
if field(session, 'sd-track') != label:
raise ValueError('Visible track and source attribute disagree')
count = re.fullmatch(r'(Long|Short) · ([1-9][0-9]*) papers', field(session, 'sd-format'))
if count is None or count[1].lower() != form:
raise ValueError('Visible presentation label disagrees')
rows = [n for n in session.walk() if n.tag == 'tr' and 'data-paper-id' in n.attrs]
if len(rows) != int(count[2]):
raise ValueError('Session paper count changed')
note = field(session, 'sd-table-note')
if not note.startswith(f'{form.title()} Presentation ({len(rows)} papers) · '):
raise ValueError('Unreviewed presentation-note semantics')
context = {'id': sid, 'track_literal': track, 'presentation_literal': form,
'title': field(session, 'sd-session-title'),
'time_literal': field(session, 'sd-session-time'), 'room_literal': field(session, 'sd-room')}
for row in rows:
pid = row.attrs['data-paper-id']
cells = [n for n in row.children if isinstance(n, Node) and n.tag in ('th', 'td')]
if (not re.fullmatch(r'DM[1-9][0-9]*', pid) or pid in seen
or row.attrs.get('id') != 'paper-'+pid or len(cells) != 4
or [clean(n.text()) for n in cells] != [field(row, k) for k in
('sd-time', 'sd-paper-id', 'sd-paper-title', 'sd-authors')]
or field(row, 'sd-paper-id') != pid):
raise ValueError('Repeated/malformed program paper identity or cells')
seen.add(pid)
title = field(row, 'sd-paper-title'); authors_literal = field(row, 'sd-authors')
authors = [name.strip() for name in authors_literal.split(';')]
if not title or any(not name for name in authors):
raise ValueError('Missing program title or author')
records.append({'paper_id_literal': pid, 'title': title, 'authors_literal': authors_literal,
'source_reported_authors': authors, 'time_literal': field(row, 'sd-time'),
'session': context.copy(), 'source_position': len(records)+1})
if (len(records) != 324 or len([n for n in main.walk() if 'data-paper-id' in n.attrs]) != 324
or Counter(r['session']['track_literal'] for r in records) != {'research': 241, 'applied': 83}
or Counter(r['session']['presentation_literal'] for r in records) != {'long': 168, 'short': 156}):
raise ValueError('Incomplete reviewed paper/track/presentation census')
return records
def collect(out, *, snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
receipts, files = load_snapshot(snapshots)
program = receipts['detailed-sessions']; observations = parse_program(files[program['file']])
out.mkdir(parents=True)
try:
evidence = []; directory = out/'evidence'; directory.mkdir()
for name, raw in files.items():
rel = 'evidence/'+digest(raw)+Path(name).suffix; (out/rel).write_bytes(raw)
evidence.append({'file': name, 'output_file': rel, 'sha256': digest(raw)})
outputs = []
for track in ('research', 'applied'):
rows = []
for item in observations:
if item['session']['track_literal'] != track:
continue
pid = item['paper_id_literal']; url = PROGRAM+'#paper-'+pid
rows.append({'source': SOURCE, 'external_id': '2026/'+pid, 'url': url,
'title': item['title'], 'authors': item['source_reported_authors'], 'abstract': None,
'venue': 'ICDM', 'year': 2026, 'track': track, 'status': 'accepted',
'identifiers': {}, 'submission_identifiers': [], 'observed_at': program['at'],
'evidence_url': url,
'raw_decision': 'Listed in official ICDM2026 in-person '+track+' program; not an original signed decision or verified publisher record.',
'tags': [{'namespace': SOURCE+':session_track', 'value': track},
{'namespace': SOURCE+':presentation', 'value': item['session']['presentation_literal']}],
'metadata': {'program_fields': item, 'program_receipt': program,
'original_decision_signature_verified': False,
'program_id_scope': 'Source program locator only; not asserted original submission ID or publication DOI.',
'author_scope': 'Source-reported program names; no publisher completeness or person alias assertion.',
'abstract_availability': 'No abstracts in this frozen program table.',
'track_evidence': 'Explicit data-track attribute agrees with visible session label; not inferred from topic.',
'paper_type': None,
'scope': 'In-person program only; excludes other program sections. Long/short describe presentation format. Requires work/version and publication review before import.'}})
path = out/f'icdm-program-2026-{track}.jsonl'
payload = ''.join(json.dumps(row, ensure_ascii=False)+'\n' for row in rows).encode(); path.write_bytes(payload)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'ICDM', 'year': 2026,
'track': track, 'count': len(rows), 'file': str(path), 'file_sha256': digest(payload),
'list_url': PROGRAM, 'status': 'partial', 'missing_abstract_count': len(rows),
'source_expected_count': len(rows), 'scope': 'Frozen in-person program section, not all accepted/published papers.'}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n'); outputs.append(manifest)
report = {'status': 'collected_pending_identity_review', 'records': 324, 'sessions': 34,
'files': outputs, 'track_counts': {'research': 241, 'applied': 83},
'presentation_counts': {'long': 168, 'short': 156}, 'missing_abstract_count': 324,
'evidence': evidence, 'public_rejected_metadata': 'unknown_no_public_export_verified',
'negative_policy_scope': 'Research triple-blind and applied single-blind concern review-time visibility; neither proves final public decision coverage.',
'database_written': False, 'network_requests': 0, 'pdfs_downloaded': False,
'scope': 'In-person program observations only. No work merge, inferred author alias, publication identity, original signed decision or final whole-database audit.'}
temporary = out/'collection-report.json.part'; temporary.write_text(json.dumps(report, indent=2)+'\n')
temporary.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--snapshots', required=True, type=Path)
parser.add_argument('--out', required=True, type=Path)
args = parser.parse_args(argv)
print(json.dumps(collect(args.out, snapshots=args.snapshots), ensure_ascii=False))
+172
View File
@@ -0,0 +1,172 @@
"""Frozen RecSys 2026 accepted-contribution observations, preserving source variants."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
import re
from paper_atlas.io.html import Document, Node, clean
SOURCE = 'recsys_program'
SNAPSHOT = 'remaining2026-listings-research-v2'
MANIFEST_SHA = 'fea7b710587636db06794ee195cd083634c41e9054a217859d9dd0ccdd1afab9'
URL = 'https://recsys.acm.org/recsys26/contributions/'
TRACKS = [('long', 'Long Papers', 65), ('short', 'Short Papers', 29),
('past_present_future', 'PPF Papers', 13), ('reproducibility', 'Reproducibility', 4),
('resource', 'Resource', 7), ('industry', 'Industry', 83),
('research_practice_notes', 'R&P Notes', 34), ('demo', 'Demos', 18),
('doctoral_symposium', 'Doctoral', 10)]
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def one(items):
if len(items) != 1:
raise ValueError('Expected exactly one reviewed source field')
return items[0]
def load_snapshot(root):
root = Path(root); directory = root/SNAPSHOT; path = directory/'manifest.json'
if root.is_symlink() or directory.is_symlink() or path.is_symlink() or path.stat().st_size > 128*1024:
raise ValueError('Unsafe snapshot manifest')
raw = path.read_bytes()
if digest(raw) != MANIFEST_SHA:
raise ValueError('Reviewed manifest hash changed')
manifest = json.loads(raw)
if manifest['status'] != 'research_completed_not_coverage_or_import_approval' or manifest['database_written'] is not False:
raise ValueError('Unreviewed snapshot scope')
receipt = one([e for e in manifest['entries'] if e['venue'] == 'RecSys' and e['label'] == 'contributions'])
name = receipt['file']; path = directory/name
if Path(name).name != name or path.is_symlink() or path.stat().st_size > 4*1024**2:
raise ValueError('Unsafe snapshot body')
body = path.read_bytes()
if (receipt['url'] != URL or receipt['final_url'] != URL or receipt['http_status'] != 200
or not receipt['content_type'].startswith('text/html') or receipt['bytes'] != len(body)
or receipt['sha256'] != digest(body) or datetime.fromisoformat(receipt['at']).utcoffset() is None):
raise ValueError('Snapshot receipt/body disagree')
return receipt, {'manifest.json': raw, name: body}, body
def body_text(node, doi_url):
if not isinstance(node, Node):
return node
if node.tag in ('script', 'style') or (node.tag == 'a' and node.attrs.get('href') == doi_url):
return ''
text = ''.join(body_text(c, doi_url) for c in node.children)
return ' '+text+' ' if node.tag in ('p','div','li','br','section') else text
def parse_program(raw):
doc = Document(raw.decode('utf-8')); nodes = list(doc.root.walk())
if not any(n.tag == 'title' and 'RecSys 26 - Accepted Contributions' in n.text() for n in nodes):
raise ValueError('Wrong conference edition')
menu = one([n for n in nodes if n.tag == 'ul' and n.attrs.get('id') == 'tab-1'])
labels = [(n.attrs.get('href'), clean(n.text())) for n in menu.walk() if n.tag == 'a']
if labels != [(f'#content-tab-1-{i}', label) for i, (_,label,_) in enumerate(TRACKS)]:
raise ValueError('Unreviewed track menu')
observations = []; seen = set()
for tab, (track, label, count) in enumerate(TRACKS):
section = one([n for n in nodes if n.tag == 'div' and n.attrs.get('id') == f'content-tab-1-{tab}'])
listing = one([n for n in section.walk() if n.tag == 'ul' and n.attrs.get('id') == f'accordion-{tab+1}'])
rows = [n for n in listing.children if isinstance(n, Node) and n.tag == 'li']
if len(rows) != count:
raise ValueError('Incomplete reviewed track census')
for position, row in enumerate(rows, 1):
anchor = one([n for n in row.children if isinstance(n, Node) and n.tag == 'a'])
locator = anchor.attrs.get('rel', '').removeprefix('#')
if not re.fullmatch(f'accordion-{tab+1}-slide-[0-9]+', locator) or locator in seen:
raise ValueError('Missing or repeated program locator')
seen.add(locator)
types = [n for n in anchor.walk() if n.tag == 'span' and n.attrs.get('class') == 'paper-type']
type_node = one(types); type_literal = type_node.attrs.get('title'); abbreviation = clean(type_node.text())
expected_type = [('Research','RES'),('Research','RES'),('PastPresentFuture','PPF'),
('Research','REPR'),('Research','RSC'),('Research','IND'),('Research','RPN'),('Research','DEMO'),('Doctoral Symposium','DS')][tab]
if (type_literal,abbreviation) != expected_type:
raise ValueError('Unreviewed source contribution type')
title = clean(''.join(c.text() if isinstance(c, Node) else c for c in anchor.children
if not isinstance(c, Node) or c.tag not in ('span','em','br')))
authors_literal = clean(one([n for n in anchor.walk() if n.tag == 'em']).text())
if not authors_literal.startswith('by '):
raise ValueError('Missing program author prefix')
authors_literal = authors_literal[3:]
authors = [n.strip() for n in re.split(r',\s+|\s+and\s+', authors_literal)]
if not title or any(not n or '@' in n for n in authors):
raise ValueError('Missing title or program author')
body = one([n for n in row.children if isinstance(n, Node) and n.tag == 'div'])
doi_url = one([n.attrs.get('href') for n in body.walk() if n.tag == 'a'
and n.attrs.get('href', '').startswith('https://doi.org/')])
if not re.fullmatch(r'https://doi.org/10\.1145/3773078\.[0-9]+', doi_url):
raise ValueError('Wrong publication DOI family')
abstract = clean(body_text(body, doi_url))
if len(abstract) < 100 or any(n.tag in ('math','sup','sub','img','svg') for n in body.walk()):
raise ValueError('Missing abstract or unreviewed structured markup')
observations.append({'doi': doi_url.removeprefix('https://doi.org/'), 'tab': tab, 'position': position,
'locator': locator, 'title': title, 'authors': authors, 'authors_literal': authors_literal,
'abstract': abstract, 'type_labels': [type_literal], 'source_type_abbreviation': abbreviation,
'track': track, 'track_label': label})
counts = Counter(r['doi'] for r in observations)
if len(observations) != 263 or len(counts) != 260 or {d:n for d,n in counts.items() if n>1} != {
'10.1145/3773078.3831839':2,'10.1145/3773078.3831819':2,'10.1145/3773078.3831861':2}:
raise ValueError('Unreviewed DOI observation census')
for item in observations:
item['review_flags'] = ['repeated_program_doi_preserve_versions_requires_review'] if counts[item['doi']] > 1 else []
return observations
def collect(out, *, snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
try:
receipt, files, body = load_snapshot(snapshots); observations = parse_program(body)
evidence = out/'evidence'; evidence.mkdir()
for name, raw in files.items():
(evidence/name).write_bytes(raw)
outputs = []
for track,label,count in TRACKS:
rows = []
for item in observations:
if item['track'] != track:
continue
doi = item['doi']; url = URL+'#'+item['locator']
rows.append({'source': SOURCE, 'external_id': '2026/'+item['locator'], 'url': url,
'title': item['title'], 'authors': item['authors'], 'abstract': item['abstract'],
'identifiers': {'doi': doi}, 'submission_identifiers': [{'scheme':'publication_doi','value':doi}],
'venue':'RecSys', 'year':2026, 'track':track, 'status':'accepted', 'observed_at':receipt['at'],
'evidence_url':URL, 'raw_decision':'Listed as an accepted contribution in the official RecSys2026 '+label+' section; not a signed review decision.',
'tags':[{'namespace':SOURCE+':contribution_type','value':item['type_labels'][0]}],
'metadata':{'program_fields':item, 'program_receipt':receipt,
'source_evidence':{'file':'evidence/'+receipt['file'],'sha256':receipt['sha256'],'url':URL},
'source_manifest':{'file':'evidence/manifest.json','sha256':MANIFEST_SHA},
'original_decision_signature_verified':False,
'author_scope':'Source-reported comma/and-delimited names; original literal retained. Not verified publisher person identities or aliases.',
'doi_scope':'DOI asserted by official program hyperlink, not independently verified live publisher metadata.',
'abstract_transform':'HTML display text and whitespace normalization; source retained. No PDF or cross-source fusion.',
'scope':'Program observation keyed by source locator. Repeated DOI variants retained separately and flagged, not counted as distinct scientific works. Requires DOI/work/history review before import.'}})
path = out/f'recsys-program-2026-{track}.jsonl';raw=''.join(json.dumps(r,ensure_ascii=False)+'\n' for r in rows).encode();path.write_bytes(raw)
manifest={'source':SOURCE,'parser_version':1,'venue':'RecSys','year':2026,'track':track,'count':len(rows),
'file':str(path),'file_sha256':digest(raw),'list_url':URL,'status':'partial','missing_abstract_count':0,
'scope':'Official accepted-contribution observations; unique DOI counts differ from observation counts.'}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest,indent=2)+'\n');outputs.append(manifest)
report={'status':'collected_pending_identity_review','records':263,'unique_dois':260,'files':outputs,
'track_counts':{track:count for track,_,count in TRACKS},'missing_abstract_count':0,
'duplicate_doi_observations':6,'repeated_dois':{d:n for d,n in Counter(r['doi'] for r in observations).items() if n>1},
'source_manifest_sha256':MANIFEST_SHA,'database_written':False,'network_requests':0,'pdfs_downloaded':False,
'approved_for_import':0,'public_rejected_availability':'unknown_no_public_export_verified',
'scope':'263 program observations for260 DOIs, not263 unique papers or final work identities. Track/type labels are source scoped. Not complete public-negative coverage or a final whole-database audit.'}
part=out/'collection-report.json.part';part.write_text(json.dumps(report,ensure_ascii=False,indent=2)+'\n');part.replace(out/'collection-report.json');return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status':'failed_not_complete','error_type':type(error).__name__,'database_written':False})+'\n')
raise
def main(argv=None, *, prog=None):
parser=argparse.ArgumentParser(prog=prog,description=__doc__)
parser.add_argument('--snapshots',required=True);parser.add_argument('--out',required=True)
args=parser.parse_args(argv);print(json.dumps(collect(args.out,snapshots=args.snapshots),ensure_ascii=False))
+222
View File
@@ -0,0 +1,222 @@
"""Reviewed RSS 2026 publisher metadata, with separate program evidence."""
REVIEWED_AUTHOR_VARIANTS = {16: '04de7e813b7c68b5c1764e90fe41bcd2e45259d46749bb2cc54bac5150c8f25b', 27: '975f6adc4d15045f9c8829689a759c5e3a44eecb3410a18ff99b37226d9c8b97', 78: '8208d0950468295c2b73dae28658566b1c726f65a4167f490039cf3449aa2db4', 132: '31eee6289de404f3d859dd0d9de4dd40e968f86b89312de41df87ae01c529711', 199: 'ced0bc9e38b28b4a490799fd94315f104ece0df28be428b3cc417db8ea797b9d', 204: '7c6959e6668bf44af09959388f6b9e2d883a76106bf7f191dc9949368c969ee5'}
import argparse
from datetime import datetime, date
import hashlib
import json
from pathlib import Path
import re
import unicodedata
from paper_atlas.io.html import Document, Node, clean
from paper_atlas.io.http import fetch
SOURCE = 'rss_proceedings'
ROOT = 'https://www.roboticsproceedings.org/rss22/'
HEADING = 'Robotics: Science and Systems XXII'
PAGES = {'index.html': ROOT,
'accepted.html': 'https://roboticsconference.org/2026/program/papers/',
'cfp.html': 'https://roboticsconference.org/2026/information/cfp/',
'sponsor-demos.html': 'https://roboticsconference.org/2026/program/demos/'}
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def one(values):
if len(values) != 1:
raise ValueError('Expected one reviewed HTML field')
return values[0]
def comparable(value):
return unicodedata.normalize('NFC', clean(value))
def parse_lists(index_html, program_html, policy_html, sponsors_html):
policy = clean(Document(policy_html).root.text())
for text in ('RSS 2026 will not have a separate category for demo papers.',
'papers will only be public after acceptance', 'reviews will not be publicly visible'):
if text not in policy:
raise ValueError('Reviewed 2026 publication/demo/privacy policy changed')
if [clean(n.text()) for n in Document(sponsors_html).root.walk() if n.tag == 'h1'] != ['Sponsor Demos']:
raise ValueError('Sponsor activity page changed; never classify these as papers')
nodes = list(Document(index_html).root.walk())
if [clean(n.text()) for n in nodes if n.tag == 'h1'] != [HEADING]:
raise ValueError('Unexpected publisher edition')
index = {}
for row in (n for n in nodes if n.tag == 'tr'):
links = [n for n in row.walk() if n.tag == 'a' and re.fullmatch(r'p[0-9]{3}\.html', n.attrs.get('href', ''))]
if not links:
continue
link = one(links); pid = int(link.attrs['href'][1:4])
if pid in index:
raise ValueError('Repeated publisher ID')
index[pid] = {'title': clean(link.text()), 'authors': clean(one([n for n in row.walk() if n.tag == 'i']).text())}
program = {}
if [clean(n.text()) for n in Document(program_html).root.walk() if n.tag == 'h1'] != ['Accepted Papers']:
raise ValueError('Unexpected accepted program')
for row in (n for n in Document(program_html).root.walk() if n.tag == 'tr'):
cells = [n for n in row.children if isinstance(n, Node) and n.tag == 'td']
if not cells:
continue
if len(cells) != 4 or not re.fullmatch('[1-9][0-9]*', clean(cells[0].text())):
raise ValueError('Unexpected program row')
pid = int(clean(cells[0].text()))
if pid in program:
raise ValueError('Repeated program ID')
preview = clean(' '.join(n for n in cells[3].children if isinstance(n, str)))
full = one([clean(n.text()) for n in cells[3].walk() if n.tag == 'div'
and 'display:none' in n.attrs.get('style', '').replace(' ', '')])
if not preview or not full.startswith(preview):
raise ValueError('Author preview is not a prefix of complete author evidence')
links = [n.attrs.get('href') for n in cells[2].walk() if n.tag == 'a']
if links != [f'/2026/program/papers/{pid}/']:
raise ValueError('Program display link changed')
program[pid] = {'display_id': pid, 'title': clean(cells[2].text()), 'authors_complete': full,
'authors_preview': preview, 'session': clean(cells[1].text()),
'program_url': 'https://roboticsconference.org'+links[0]}
if set(index) != set(program) or set(index) != set(range(1, 211)):
raise ValueError('Reviewed 210-entry source census changed')
return index, program
def parse_paper(html, pid, index, program):
nodes = list(Document(html).root.walk()); meta = {}
for n in nodes:
if n.tag == 'meta' and n.attrs.get('name', '').startswith('citation_'):
meta.setdefault(n.attrs['name'], []).append(n.attrs.get('content', ''))
title = one(meta.get('citation_title', [])); authors = meta.get('citation_author', [])
day = one(meta.get('citation_publication_date', []))
if not re.fullmatch(r'2026/[0-9]{2}/[0-9]{2}', day):
raise ValueError('Unexpected publication date')
date.fromisoformat(day.replace('/', '-'))
if (meta.get('citation_volume') != ['22'] or meta.get('citation_conference_title') != [HEADING]
or [clean(n.text()) for n in nodes if n.tag == 'h1'] != [HEADING]
or not authors or any(not clean(a) for a in authors)
or comparable(title) != comparable(index['title'])
or comparable(', '.join(authors)) != comparable(index['authors'])
or [comparable(n.text()) for n in nodes if n.tag == 'h3'] != [comparable(title)]):
raise ValueError('Publisher citation/index/visible metadata disagree')
pdf = one(meta.get('citation_pdf_url', []))
if pdf not in (ROOT+f'p{pid:03d}.pdf', ROOT.replace('https:', 'http:')+f'p{pid:03d}.pdf'):
raise ValueError('Citation PDF ID mismatch; PDFs are never fetched')
bib = one([n.text() for n in nodes if n.tag == 'pre'])
doi = one(re.findall(r'^\s*DOI\s*=\s*\{([^{}]+)\}', bib, re.I | re.M))
year = one(re.findall(r'^\s*YEAR\s*=\s*\{([^{}]+)\}', bib, re.I | re.M))
if doi != f'10.15607/RSS.2026.XXII.{pid:03d}' or year != '2026':
raise ValueError('BibTeX DOI/year mismatch')
variants = []
if comparable(', '.join(authors)) != comparable(program['authors_complete']):
key = digest(json.dumps([authors, program['authors_complete']], ensure_ascii=False, separators=(',', ':')).encode())
if REVIEWED_AUTHOR_VARIANTS.get(pid) != key:
raise ValueError('Unreviewed complete author variant')
variants.append('reviewed_program_author_name_variant_not_person_alias')
if comparable(title) != comparable(program['title']):
if (pid, title, program['title']) != (89,
'RLinf-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models',
'RLux-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models'):
raise ValueError('Unreviewed program title variant')
variants.append('reviewed_program_title_variant_not_erratum')
inside = False; parts = []; markers = []
for n in (n for n in nodes if n.tag == 'p'):
labels = [clean(b.text()) for b in n.walk() if b.tag == 'b']
if 'Abstract:' in labels:
inside = True; markers.append('start'); value = clean(n.text()).removeprefix('Abstract:').strip()
if value: parts.append(value)
elif 'Download:' in labels:
inside = False; markers.append('end')
elif inside:
parts.append(clean(n.text()))
if markers != ['start', 'end'] or not any(parts):
raise ValueError('Missing or changed publisher abstract layout')
return {'title': title, 'authors': authors, 'abstract': '\n\n'.join(v for v in parts if v),
'doi_raw': doi, 'publication_date': day, 'bibtex': bib, 'program_variants': variants}
def collect(out, *, snapshots=None, cache=None):
if (snapshots is None) == (cache is None):
raise ValueError('Specify snapshots or cache')
out = Path(out)
if out.exists():
raise ValueError('Output directory must not exist; inspect partial/completed runs first')
out.parent.mkdir(parents=True, exist_ok=True)
out.mkdir()
evidence = {}; receipts = {}
if snapshots is not None:
root = Path(snapshots)
try:
for r in json.loads((root/'receipts.json').read_bytes()):
name = r['file']
if not isinstance(name, str) or Path(name).name != name or name in receipts:
raise ValueError('Unsafe or repeated snapshot receipt')
receipts[name] = r
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def page(name, url):
if snapshots is not None:
path = root/name
if path.is_symlink(): raise ValueError('Snapshot symlink rejected')
raw = path.read_bytes(); r = receipts[name]
if (r['url'] != url or r.get('final_url', url) != url or r['http_status'] != 200
or r['bytes'] != len(raw) or r['sha256'] != digest(raw)):
raise ValueError('Snapshot hash/receipt mismatch')
at = r['at']
else:
raw = fetch(url, cache, limit=8*1024*1024)
sidecar = json.loads((Path(cache)/digest(url.encode())).with_suffix('.json').read_bytes())
if sidecar['sha256'] != digest(raw): raise ValueError('Cache receipt mismatch')
at = sidecar['retrieved_at']
if len(raw) > 8*1024*1024 or datetime.fromisoformat(at.replace('Z', '+00:00')).utcoffset() is None:
raise ValueError('Invalid source size/timezone')
relative = 'evidence/'+digest(raw)+'.html'; target = out/relative
target.parent.mkdir(exist_ok=True); target.write_bytes(raw)
evidence[name] = {'file': relative, 'sha256': digest(raw), 'url': url, 'observed_at': at}
return raw.decode('utf-8')
try:
pages = {name: page(name, url) for name, url in PAGES.items()}
index, program = parse_lists(pages['index.html'], pages['accepted.html'], pages['cfp.html'], pages['sponsor-demos.html'])
records = []
for pid in sorted(index):
name = f'p{pid:03d}.html'; fields = parse_paper(page(name, ROOT+name), pid, index[pid], program[pid])
row = {'source': SOURCE, 'external_id': f'2026/p{pid:03d}', 'url': ROOT+name,
'title': fields['title'], 'abstract': fields['abstract'], 'authors': fields['authors'],
'identifiers': {'doi': fields['doi_raw'].lower()}, 'venue': 'RSS', 'venue_aliases': [],
'year': 2026, 'track': 'main', 'status': 'accepted',
'raw_decision': 'listed in official RSS 2026 proceedings and accepted papers',
'observed_at': evidence[name]['observed_at'],
'metadata': {'publisher_date': fields['publication_date'], 'publisher_doi_raw': fields['doi_raw'],
'original_bibtex': fields['bibtex'], 'program': program[pid],
'program_variants': fields['program_variants'], 'source_evidence': evidence[name],
'index_evidence': evidence['index.html'], 'program_evidence': evidence['accepted.html'],
'policy_evidence': evidence['cfp.html'],
'scope': '2026 Science/Systems papers; no separate demo-paper category. Program display ID is not an original submission ID.'}}
records.append(row)
path = out/'rss-proceedings-2026-main.jsonl'
raw = b''.join((json.dumps(r, ensure_ascii=False)+'\n').encode() for r in records)
path.write_bytes(raw)
manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'RSS', 'year': 2026, 'track': 'main',
'count': 210, 'file': str(path), 'file_sha256': digest(raw), 'list_url': ROOT,
'status': 'partial', 'missing_abstract_count': 0}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest, indent=2)+'\n')
report = {'status': 'collected_pending_identity_review', 'source': SOURCE, 'records': 210,
'files': [manifest], 'source_evidence': evidence, 'database_written': False, 'pdfs_downloaded': False,
'limitations': 'Publisher metadata only; not database import, scientific-work deduplication, private reviews or a public rejection set.'}
temporary = out/'collection-report.json.part'; temporary.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n')
temporary.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
p = argparse.ArgumentParser(prog=prog, description=__doc__)
p.add_argument('--out', required=True, type=Path)
g = p.add_mutually_exclusive_group(required=True)
g.add_argument('--snapshots', type=Path); g.add_argument('--cache', type=Path)
args = p.parse_args(argv)
report = collect(args.out, snapshots=args.snapshots, cache=args.cache)
print(json.dumps({'status': report['status'], 'records': report['records'], 'database_written': False}))
+313
View File
@@ -0,0 +1,313 @@
"""Frozen SDM 2026 publication records and separate public program observations."""
import argparse
from collections import Counter
from datetime import date, datetime
import hashlib
from html.parser import HTMLParser
import json
from pathlib import Path
import re
from urllib.parse import parse_qs, urljoin, urlsplit
from paper_atlas.io.html import Document, clean
SOURCE = 'crossref_sdm_registry'
TRACK = 'proceedings_unclassified'
BOOK = '10.1137/1.9781611979329'
ISBN = '9781611979329'
TITLE = 'Proceedings of the 2026 SIAM International Conference on Data Mining (SDM)'
QUERY = 'https://api.crossref.org/works?filter=isbn%3A9781611979329&rows=1000'
PROGRAM = 'https://meetings.siam.org/program.cfm?CONFCODE=sdm26'
SNAPSHOTS = {
'sdm2026-official-research-v1': ('a39a6ee6a12b69051ba2dad7c9cbc5b0fe215af85100c7aa2f5166d44c22fe61', 'sources'),
'sdm2026-crossref-research-v1': ('51d337a153a1046e4a00ebc031707faf72e733577e4fc243a31672ee93277da5', 'sources'),
'sdm2026-program-research-v1': ('689a2aaaa15be532957c4c88e03e1e2e87909334b6f529be16905d8b9cc1fb77', 'sessions'),
'sdm2026-public-abstract-research-v1': ('8e1c29611e145cbfa54ec4afaa2efe152557b4e04e339901af5cd33abefc29a5', 'records'),
}
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def one(items):
if len(items) != 1:
raise ValueError('Expected exactly one reviewed source field')
return items[0]
def load_snapshot(root, name):
expected, key = SNAPSHOTS[name]
directory = root/name; path = directory/'manifest.json'
if root.is_symlink() or directory.is_symlink() or path.is_symlink() or path.stat().st_size > 256*1024:
raise ValueError('Unsafe snapshot manifest')
raw = path.read_bytes()
if digest(raw) != expected:
raise ValueError('Reviewed snapshot manifest hash changed')
manifest = json.loads(raw)
if datetime.fromisoformat(manifest['at']).utcoffset() is None:
raise ValueError('Snapshot timestamp requires timezone')
files = {'manifest.json': raw}
for receipt in manifest[key]:
if 'file' not in receipt:
# A recorded 403 is an access observation, not an empty data set.
continue
filename = receipt['file']; path = directory/filename
if Path(filename).name != filename or path.is_symlink() or path.stat().st_size > 24*1024**2:
raise ValueError('Unsafe snapshot source file')
body = path.read_bytes()
if (receipt['http_status'] != 200 or receipt['url'] != receipt['final_url']
or receipt['bytes'] != len(body) or receipt['sha256'] != digest(body)):
raise ValueError('Snapshot receipt/body disagree')
if filename in files:
raise ValueError('Repeated snapshot source file')
files[filename] = body
return manifest, files
def registry_papers(message):
items = message.get('items', [])
if message.get('total-results') != 55 or len(items) != 55:
raise ValueError('Incomplete exact-ISBN registry census')
if Counter(x.get('type') for x in items) != {'book-chapter': 53, 'book': 1, 'other': 1}:
raise ValueError('Unreviewed publication item types')
papers = []; excluded = []; seen = set()
for original in items:
doi = original.get('DOI')
if not isinstance(doi, str) or doi in seen or original.get('ISBN') != [ISBN]:
raise ValueError('Missing/repeated DOI or wrong ISBN')
seen.add(doi)
if doi in (BOOK, BOOK+'.fm'):
expected = ('book', [TITLE]) if doi == BOOK else ('other', ['Front Matter'])
if (original['type'], original.get('title')) != expected:
raise ValueError('Unreviewed excluded volume/front matter')
excluded.append({'doi': doi, 'type': original['type'], 'title': original['title']})
continue
if (original.get('type') != 'book-chapter' or original.get('container-title') != [TITLE]
or not re.fullmatch(re.escape(BOOK)+r'\.[1-9][0-9]*', doi)
or original.get('URL') != 'https://doi.org/'+doi
or original.get('publisher') != 'Society for Industrial and Applied Mathematics'
or original.get('relation', {}).get('is-part-of') != [
{'id-type': 'doi', 'id': 'https://doi.org/'+BOOK, 'asserted-by': 'object'}]):
raise ValueError('Wrong publication edition, publisher or parent DOI')
title = one(original.get('title', []))
if not isinstance(title, str) or not clean(title):
raise ValueError('Missing registry title')
flags = []
if '<' in title or '>' in title:
match = re.fullmatch(r'([^<>]+)<sup>\*</sup>\s*', title, re.S)
if not match:
raise ValueError('Unreviewed title markup')
comparison_title = clean(match[1])
flags.append('title_sup_asterisk_footnote_requires_review')
else:
comparison_title = clean(Document(title).root.text())
displayed_title = clean(Document(title).root.text())
authors = original.get('author')
if (not isinstance(authors, list) or not authors
or any(not isinstance(a.get('given'), str) or not isinstance(a.get('family'), str)
or not clean(a['given']) or not clean(a['family']) for a in authors)
or [a.get('sequence') for a in authors] != ['first']+['additional']*(len(authors)-1)):
raise ValueError('Incomplete registry authors or order evidence')
names = [clean(a['given']+' '+a['family']) for a in authors]
if any('@' in n or 'anonymous' in n.casefold() for n in names):
raise ValueError('Unreviewed registry author identity')
dates = {}
for field in ('published', 'published-print', 'published-online', 'issued'):
if field in original:
parts = one(original[field]['date-parts'])
if len(parts) != 3 or any(type(x) is not int for x in parts) or parts[0] != 2026:
raise ValueError('Unreviewed registry publication date')
dates[field] = date(*parts).isoformat()
if not dates or original.get('abstract'):
raise ValueError('Reviewed registry date/abstract availability changed')
papers.append({'original': original, 'title': displayed_title, 'comparison_title': comparison_title,
'authors': names, 'review_flags': flags, 'publication_dates': dates})
if {p['original']['DOI'] for p in papers} != {BOOK+'.'+str(n) for n in range(1, 54)} or len(excluded) != 2:
raise ValueError('Incomplete reviewed 53-chapter DOI set')
return sorted(papers, key=lambda p: int(p['original']['DOI'].rsplit('.', 1)[1])), excluded
class SessionParser(HTMLParser):
"""The source omits closing dt/dd tags; each next dt starts a new event."""
def __init__(self):
super().__init__(); self.rows = []; self.current = None; self.strong = False; self.author = False
def finish(self):
if self.current is not None:
match = re.fullmatch(r'(\d{1,2}:\d{2}-\d{1,2}:\d{2})\s+(.+)', clean(self.current.pop('timed_title')))
if not match or not clean(self.current['authors_literal']):
raise ValueError('Incomplete contributed program event')
self.current['time_literal'], self.current['title'] = match.groups()
self.current['authors_literal'] = clean(self.current['authors_literal'])
self.rows.append(self.current)
self.current = None; self.strong = False; self.author = False
def handle_starttag(self, tag, attrs):
attrs = dict(attrs)
if tag == 'dt':
self.finish()
self.current = {'timed_title': '', 'authors_literal': '', 'cancelled_presentation': False, 'abstract_links': []}
if self.current is not None:
if tag == 'strong': self.strong = True
elif tag == 'dd': self.author = True
elif tag == 's': self.current['cancelled_presentation'] = True
elif tag == 'a' and attrs.get('href', '').startswith('dsp_talk.cfm?p='):
self.current['abstract_links'].append(attrs['href'])
def handle_endtag(self, tag):
if tag == 'strong': self.strong = False
elif tag == 'dl': self.finish()
def handle_data(self, text):
if self.current is not None:
if self.strong: self.current['timed_title'] += text
elif self.author: self.current['authors_literal'] += text
elif clean(text) == 'Cancelled': self.current['cancelled_presentation'] = True
def talk_fields(raw, session_url):
nodes = list(Document(raw.decode('utf-8')).root.walk())
if one([clean(n.text()) for n in nodes if n.tag == 'h1']) != '2026 SIAM International Conference on Data Mining':
raise ValueError('Wrong program page edition')
title = clean(one([n for n in nodes if n.tag == 'div' and n.attrs.get('class') == 'ptitle']).text())
abstract_node = one([n for n in nodes if n.tag == 'div' and n.attrs.get('class') == 'abstract'])
abstract = clean(abstract_node.text())
if not title or not abstract.startswith('Abstract.') or not abstract[len('Abstract.'):].strip():
raise ValueError('Missing public program title/abstract')
reference = one([n for n in nodes if n.tag == 'div' and n.attrs.get('class') == 'sesref'])
link = one([n.attrs.get('href', '') for n in reference.walk() if n.tag == 'a'])
def session_key(url):
u = urlsplit(url)
query = {k.lower(): v for k, v in parse_qs(u.query).items()}
if u.scheme != 'https' or u.netloc != 'meetings.siam.org' or u.path != '/sess/dsp_programsess.cfm' or set(query) != {'sessioncode'}:
raise ValueError('Foreign program session link')
return query['sessioncode']
if session_key(link) != session_key(session_url):
raise ValueError('Program page belongs to a different session')
authors = []
for n in nodes:
if n.tag == 'li':
# Source rows contain name, affiliations, country and contact email.
# Only the independently checked leading name is a metadata field.
literal = clean(n.text()); name, separator, _ = literal.partition(',')
if not separator or not name or '@' in name:
raise ValueError('Unreviewed program author row')
authors.append(name)
if not authors:
raise ValueError('Missing source-reported program authors')
markup = sorted({n.tag for n in abstract_node.walk() if n.tag in ('math', 'sup', 'sub', 'svg', 'img')})
return {'title': title, 'source_reported_authors': authors, 'abstract': abstract[len('Abstract.'):].strip(),
'abstract_markup_tags': markup, 'review_flags': ['structured_abstract_markup_requires_review'] if markup else [],
'author_completeness': 'Program-reported names only; not asserted to be the complete or final publisher author sequence.'}
def program_observations(snapshots, papers):
official, official_files = snapshots['sdm2026-official-research-v1']
sessions, session_files = snapshots['sdm2026-program-research-v1']
pages, page_files = snapshots['sdm2026-public-abstract-research-v1']
main_nodes = list(Document(official_files['program.html'].decode()).root.walk())
links = {urljoin(PROGRAM, n.attrs.get('href', '')): clean(n.text()) for n in main_nodes
if n.tag == 'a' and re.match(r'CP[1-9]\s', clean(n.text()))}
if len(links) != 9 or {s['url']: s['title'] for s in sessions['sessions']} != links:
raise ValueError('Contributed session membership disagrees with main program')
receipts = {r['url']: r for r in pages['records']}
if len(receipts) != len(pages['records']) or len(receipts) != 52:
raise ValueError('Incomplete or repeated public talk receipts')
rows = []; used = set()
for session in sessions['sessions']:
parser = SessionParser(); parser.feed(session_files[session['file']].decode()); parser.finish()
for position, event in enumerate(parser.rows, 1):
urls = [urljoin(session['url'], link) for link in event.pop('abstract_links')]
event.update(record_kind='source_observation_not_import_record', session=session['title'],
session_url=session['url'], session_sha256=session['sha256'], session_position=position)
if not urls and event['cancelled_presentation']:
event.update(program_page=None, public_page_fields=None)
else:
url = one(urls)
if url not in receipts or url in used:
raise ValueError('Missing or repeated public program page')
used.add(url); receipt = receipts[url]
if receipt['session_url'] != session['url']:
raise ValueError('Public page receipt session disagrees')
fields = talk_fields(page_files[receipt['file']], session['url'])
if fields['title'] != event['title']:
fields['review_flags'].append('session_page_title_versions_differ_requires_review')
event.update(program_page=url, public_page_receipt=receipt, public_page_fields=fields)
candidates = []
for paper in papers:
if paper['comparison_title'].casefold() == event['title'].casefold():
fields = event['public_page_fields']
candidates.append({'registry_doi_candidate': paper['original']['DOI'],
'comparison': 'title_candidate_only_not_identity_approval',
'registry_title_review_flags': paper['review_flags'],
'author_sequence_exact': fields['source_reported_authors'] == paper['authors'] if fields else None})
event['registry_candidates'] = candidates
event['scope'] = 'Separate program version; cancellation is presentation status only. No original submission ID, decision, work identity or author alias inferred.'
rows.append(event)
if len(rows) != 53 or used != set(receipts) or sum(r['cancelled_presentation'] for r in rows) != 1:
raise ValueError('Incomplete reviewed program event census')
return rows
def collect(out, *, snapshots):
out = Path(out)
if out.exists():
raise ValueError('Output exists; inspect previous operation before retrying')
out.mkdir(parents=True)
try:
loaded = {name: load_snapshot(Path(snapshots), name) for name in SNAPSHOTS}
registry_manifest, registry_files = loaded['sdm2026-crossref-research-v1']
message = json.loads(registry_files['isbn-works.json'])['message']
papers, excluded = registry_papers(message)
observations = program_observations(loaded, papers)
evidence = []; directory = out/'evidence'; directory.mkdir()
for name, (manifest, files) in loaded.items():
for filename, raw in files.items():
rel = 'evidence/'+digest(raw)+Path(filename).suffix
(out/rel).write_bytes(raw)
evidence.append({'snapshot': name, 'file': filename, 'output_file': rel, 'sha256': digest(raw)})
receipt = one([r for r in registry_manifest['sources'] if r.get('file') == 'isbn-works.json'])
rows = []
for p in papers:
original = p['original']; doi = original['DOI']
rows.append({'source': SOURCE, 'external_id': doi, 'url': original['URL'], 'title': p['title'],
'authors': p['authors'], 'abstract': None, 'identifiers': {'doi': doi},
'submission_identifiers': [{'scheme': 'publication_doi', 'value': doi}],
'venue': 'SDM', 'year': 2026, 'track': TRACK, 'status': 'accepted', 'tags': [],
'observed_at': receipt['at'], 'evidence_url': QUERY,
'raw_decision': 'Registered research chapter in the exact SDM2026 proceedings ISBN; not an original signed review decision.',
'metadata': {'registry_fields': original, 'publication_dates': p['publication_dates'],
'registry_response_sha256': digest(registry_files['isbn-works.json']), 'review_flags': p['review_flags'],
'original_decision_signature_verified': False,
'abstract_availability': 'No abstract in frozen Crossref deposit; public program abstracts remain separate source versions.',
'title_transform': 'Display text with footnote asterisk retained; raw title preserved. Footnote-free text is used only to list review candidates.',
'scope': 'Requires fresh work/DOI/history review before import. No automatic program fusion, track, person alias or negative decision inference.'}})
path = out/'sdm-crossref-registry-2026-unclassified.jsonl'
payload = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in rows).encode(); path.write_bytes(payload)
file_manifest = {'source': SOURCE, 'parser_version': 1, 'venue': 'SDM', 'year': 2026, 'track': TRACK,
'count': 53, 'file': str(path), 'file_sha256': digest(payload), 'list_url': QUERY, 'status': 'partial', 'missing_abstract_count': 53}
path.with_suffix('.manifest.json').write_text(json.dumps(file_manifest, indent=2)+'\n')
sidecar = out/'program-observations.jsonl'
raw = ''.join(json.dumps(r, ensure_ascii=False)+'\n' for r in observations).encode(); sidecar.write_bytes(raw)
report = {'status': 'collected_pending_identity_review', 'records': 53, 'files': [file_manifest],
'missing_abstract_count': 53, 'excluded_registry_items': excluded,
'program_observations': {'file': str(sidecar), 'sha256': digest(raw), 'count': 53, 'abstract_pages': 52,
'cancelled_presentations': 1, 'not_an_import_manifest': True},
'program_registry_author_sequence_differences': sum(c['author_sequence_exact'] is False for o in observations for c in o['registry_candidates']),
'evidence': evidence, 'public_rejected_metadata': 'unknown_anonymous_query_http403',
'database_written': False, 'network_requests': 0, 'pdfs_downloaded': False,
'scope': 'Registry publication records and program source observations remain separate; no work merge, metadata fusion or final coverage/dedup claim.'}
part = out/'collection-report.json.part'; part.write_text(json.dumps(report, ensure_ascii=False, indent=2)+'\n'); part.replace(out/'collection-report.json')
return report
except Exception as error:
(out/'failed.json').write_text(json.dumps({'status': 'failed_not_complete', 'error_type': type(error).__name__, 'database_written': False})+'\n')
raise
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--snapshots', required=True)
parser.add_argument('--out', required=True)
args = parser.parse_args(argv)
print(json.dumps(collect(args.out, snapshots=args.snapshots), ensure_ascii=False))
+1 -1
View File
@@ -4,7 +4,7 @@ import re
from paper_atlas.io.publication_ids import mlsys_id
PUBLICATIONS={'pmlr','acl_anthology','colm','neurips','mlsys','cvf','ecva','dblp','aamas_proceedings'}
PUBLICATIONS={'pmlr','acl_anthology','colm','neurips','mlsys','cvf','ecva','dblp','aamas_proceedings','rss_proceedings'}
def timestamp(value=None):
+106
View File
@@ -0,0 +1,106 @@
"""Immutable, receipt-checked public HTML cache; no expiry or silent refresh."""
from datetime import datetime, timezone
import hashlib
import json
from pathlib import Path
import time
import urllib.request
import urllib.error
from urllib.parse import urlsplit
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def checked_html(raw, receipt, url, limit):
if (len(raw) > limit or receipt.get('url') != url or receipt.get('final_url') != url
or receipt.get('http_status') != 200 or receipt.get('bytes') != len(raw)
or receipt.get('sha256') != digest(raw)
or not str(receipt.get('content_type', '')).lower().startswith('text/html')
or datetime.fromisoformat(receipt['retrieved_at'].replace('Z', '+00:00')).utcoffset() is None):
raise ValueError('Invalid public HTML receipt, response or timestamp')
if any(v in raw.lower() for v in (b'<title>just a moment', b'cf-chl-', b'botstopper')):
raise ValueError('Access challenge; stop without bypass')
return raw, receipt
class NoRedirect(urllib.request.HTTPRedirectHandler):
def redirect_request(self, *args, **kwargs):
return None
class PublicHTMLCache:
"""Caller holds a single-writer lock and restricts the allowed URL paths."""
def __init__(self, directory, origin, *, limit=2*1024**2, delay=1):
self.root = Path(directory)
self.root.mkdir(parents=True, exist_ok=True)
if self.root.is_symlink():
raise ValueError('Cache symlinks rejected')
self.origin = origin
self.limit = limit
self.delay = delay
self.opener = urllib.request.build_opener(NoRedirect())
def paths(self, url):
u = urlsplit(url)
if u.scheme != 'https' or u.username or u.password or u.fragment or u.scheme+'://'+u.netloc != self.origin:
raise ValueError('Unapproved public metadata origin')
key = digest(url.encode())
return self.root/(key+'.html'), self.root/(key+'.json'), self.root/(key+'.failed.json')
def cached(self, url):
body, receipt, failure = self.paths(url)
parts = (body.with_suffix('.html.part'), receipt.with_suffix('.json.part'))
if any(p.is_symlink() for p in (body, receipt, failure, *parts)):
raise ValueError('Cache symlinks rejected')
if any(p.exists() for p in parts):
raise ValueError('Incomplete cache part; preserve and inspect orphan before recovery')
if failure.exists():
raise ValueError('Previous access failure requires inspection; no automatic retry')
if body.exists() != receipt.exists():
raise ValueError('Incomplete cache pair; preserve and inspect orphan before recovery')
if not body.exists():
return None
if body.stat().st_size > self.limit or receipt.stat().st_size > 16*1024:
raise ValueError('Oversized cache entry')
return checked_html(body.read_bytes(), json.loads(receipt.read_bytes()), url, self.limit)
def get(self, url):
cached = self.cached(url)
if cached is not None:
return cached
body, receipt, failure = self.paths(url)
try:
request = urllib.request.Request(url, headers={'User-Agent': 'PaperAtlas/0.1 (public scholarly metadata research)'})
with self.opener.open(request, timeout=30) as response:
rec = {'url': url, 'final_url': response.geturl(), 'http_status': response.status,
'content_type': response.headers.get('Content-Type', '')}
if rec['final_url'] != url or rec['http_status'] != 200 or not rec['content_type'].lower().startswith('text/html'):
raise ValueError('Unexpected metadata redirect, HTTP status or content type')
raw = response.read(self.limit+1)
if len(raw) > self.limit:
raise ValueError('Metadata response exceeds size limit')
expected = response.headers.get('Content-Length')
if expected is not None and int(expected) != len(raw):
raise ConnectionError('Incomplete metadata response')
rec.update(bytes=len(raw), sha256=digest(raw), retrieved_at=datetime.now(timezone.utc).isoformat())
checked_html(raw, rec, url, self.limit)
except (urllib.error.HTTPError, ValueError) as error:
if isinstance(error, urllib.error.HTTPError):
status = error.code
retry_after = error.headers.get('Retry-After') if error.headers else None
error.close()
else:
status = None
retry_after = None
# Access errors remain sticky until the operator verifies availability.
failure.write_text(json.dumps({'url': url, 'http_status': status, 'retry_after': retry_after, 'error_type': type(error).__name__,
'scope': 'Inspect access/redirect/response failure before moving this marker aside; no bypass or automatic retry.'})+'\n')
raise
part = body.with_suffix('.html.part')
part.write_bytes(raw); part.replace(body)
part = receipt.with_suffix('.json.part')
part.write_text(json.dumps(rec, indent=2)+'\n'); part.replace(receipt)
time.sleep(self.delay)
return raw, rec
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
+167
View File
@@ -0,0 +1,167 @@
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import sqlite3
import unittest
from unittest.mock import patch
from paper_atlas.collectors import aamas, aamas2026
from paper_atlas.ingestion.records import ingest_file
FIXTURE = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/aamas2026-fixture.json.gz').read_bytes()))
EVIDENCE = {'url': aamas2026.URL, 'observed_at': FIXTURE['receipts']['receipts'][0]['finished_at']}
class Aamas2026Tests(unittest.TestCase):
def parse(self, html=None):
return aamas2026.parse_contents(FIXTURE['contents'] if html is None else html, EVIDENCE)
def snapshots(self, folder):
folder.mkdir()
(folder/'contents.html').write_text(FIXTURE['contents'])
(folder/'receipts.json').write_text(json.dumps(FIXTURE['receipts']))
return folder
def test_every_publisher_title_author_affiliation_and_page_preserved(self):
rows, excluded, held = self.parse()
self.assertEqual((len(rows), len(excluded), len(held)), (639, 3, 1))
expected = {r['publisher_page']: r for r in FIXTURE['expected_entries']}
self.assertEqual(len(expected), 643)
all_rows = rows+excluded+held
self.assertEqual({r['metadata']['publisher_page'] for r in all_rows}, set(expected))
for r in all_rows:
e = expected[r['metadata']['publisher_page']]
self.assertEqual(r['title'], e['title'])
self.assertEqual(r['authors'], e['authors'])
self.assertEqual(r['metadata']['author_affiliations'], e['affiliations'])
self.assertEqual(r['metadata']['section_heading'], e['section'])
for r in rows:
self.assertIsNone(r['abstract'])
self.assertEqual(r['identifiers'], {})
if r['external_id'] != '2026/p3178':
self.assertEqual(r['status'], 'accepted')
self.assertNotIn('decision_date', r)
def test_explicit_withdrawal_preserves_status_date_and_no_pdf(self):
rows, _, _ = self.parse()
row, = [r for r in rows if r['external_id'] == '2026/p3178']
self.assertEqual(row['status'], 'withdrawn')
self.assertEqual(row['metadata']['withdrawal_date'], '2026-06-08')
self.assertNotIn('decision_date', row)
self.assertEqual(row['raw_decision'], 'Withdrawn June 8, 2026')
self.assertNotIn('paper_url', row['metadata'])
self.assertEqual(row['metadata']['publication_state'], 'withdrawn')
self.assertEqual(sum(r['status'] == 'accepted' for r in rows), 638)
def test_withdrawal_ingestion_retains_date_precision_without_invented_acceptance(self):
rows, _, _ = self.parse()
row, = [r for r in rows if r['external_id'] == '2026/p3178']
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); path = root/'withdrawn.jsonl'; db = root/'sample.sqlite'
path.write_text(json.dumps(row, ensure_ascii=False)+'\n')
path.with_suffix('.manifest.json').write_text(json.dumps({
'source': 'aamas_proceedings', 'list_url': aamas2026.URL,
'venue': 'AAMAS', 'year': 2026, 'track': row['track'],
'count': 1, 'status': 'partial', 'missing_abstract_count': 1,
'file_sha256': hashlib.sha256(path.read_bytes()).hexdigest()}))
ingest_file(db, path)
with sqlite3.connect(db) as c:
self.assertEqual(c.execute('SELECT status,raw_value,decision_date FROM decisions').fetchall(),
[('withdrawn', 'Withdrawn June 8, 2026', None)])
metadata = json.loads(c.execute('SELECT metadata_json FROM source_records').fetchone()[0])
self.assertEqual(metadata['withdrawal_date'], '2026-06-08')
self.assertEqual(metadata['decision_date_precision'], 'day')
self.assertEqual(c.execute('PRAGMA foreign_key_check').fetchall(), [])
def test_retraction_is_not_accepted_or_withdrawn_and_retains_date(self):
rows, _, held = self.parse()
self.assertNotIn('2026/p238', {r['external_id'] for r in rows})
h, = held
self.assertEqual(h['publication_state'], 'retracted')
self.assertEqual(h['retraction_date'], '2026-06-08')
self.assertEqual(h['raw_publication_state'], 'Retracted June 8, 2026')
self.assertNotIn('status', h)
self.assertNotIn('paper_url', h['metadata'])
self.assertEqual(h['metadata']['source_evidence'], EVIDENCE)
def test_tracks_formats_and_prior_decisions_are_not_conflated(self):
rows, _, _ = self.parse()
counts = {t: sum(r['track'] == t for r in rows) for t in {r['track'] for r in rows}}
self.assertEqual(counts, {'proceedings_unspecified': 522, 'aaai_transfer': 19,
'blue_sky': 17, 'doctoral': 34, 'demo': 44, 'jaamas': 3})
for row in rows:
self.assertEqual(row['tags'][0]['namespace'], 'aamas_proceedings:publication_format')
if row['track'] == 'aaai_transfer':
self.assertIn('none inferred', row['metadata']['prior_submission_note'])
self.assertNotIn('decision_events', row)
if row['track'] == 'jaamas':
self.assertEqual(row['metadata']['publication_format'], 'journal_presentation')
self.assertIn('not another independent', row['metadata']['track_scope_note'])
def test_refuses_changed_year_section_page_link_author_and_retraction(self):
html = FIXTURE['contents']
replacements = [('AAMAS 2026:', 'AAMAS 2027:'),
('name="R" id="R"></a><font color="#010101">Research Paper Track',
'name="R" id="R"></a><font color="#010101">Workshop Track'),
('(Page 4)', '(Page 1)'),
('../pdfs/HQQZ1937.pdf', 'https://other.example/paper.pdf'),
('Michael Wooldridge <i>', '<i>'),
('Retracted June 8, 2026', 'Withdrawn June 8, 2026')]
for old, new in replacements:
self.assertIn(old, html)
with self.subTest(change=old), self.assertRaises(ValueError):
self.parse(html.replace(old, new, 1))
def test_frozen_collection_is_partial_with_full_held_evidence(self):
with tempfile.TemporaryDirectory() as tmp, patch.object(aamas2026, 'fetch', side_effect=AssertionError('No HTTP')):
root = Path(tmp)
report = aamas.collect(2026, root/'out', snapshots=self.snapshots(root/'source'))
self.assertEqual(report['records'], 639)
self.assertEqual(report['status_counts'], {'accepted': 638, 'withdrawn': 1})
self.assertEqual(len(report['files']), 6)
self.assertEqual(len(report['held_publications']), 1)
self.assertFalse(report['database_written'])
self.assertFalse(report['pdfs_downloaded'])
for m in report['files']:
self.assertEqual(m['status'], 'partial')
self.assertEqual(m['missing_abstract_count'], m['count'])
self.assertEqual(hashlib.sha256(Path(m['file']).read_bytes()).hexdigest(), m['file_sha256'])
self.assertTrue((root/'out/aamas-proceedings-2026.json').is_file())
def test_frozen_response_rejects_tamper_unsafe_path_status_and_symlink(self):
for mode in ('tamper', 'path', 'status', 'symlink'):
with self.subTest(mode=mode), tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); source = self.snapshots(root/'source')
r = json.loads((source/'receipts.json').read_text())
if mode == 'tamper':
(source/'contents.html').write_text('bad')
elif mode == 'path':
r['receipts'][0]['file'] = '../outside.html'
elif mode == 'status':
r['receipts'][0]['status'] = 403
else:
(root/'outside.html').write_text(FIXTURE['contents'])
(source/'contents.html').unlink()
(source/'contents.html').symlink_to(root/'outside.html')
(source/'receipts.json').write_text(json.dumps(r))
with self.assertRaises(ValueError):
aamas.collect(2026, root/'out', snapshots=source)
self.assertFalse((root/'out/aamas-proceedings-2026.json').exists())
def test_live_mode_fetches_only_html_and_preserves_receipt(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); cache = root/'cache'; cache.mkdir()
receipt = {'retrieved_at': EVIDENCE['observed_at'], 'url': aamas2026.URL}
(cache/hashlib.sha256(aamas2026.URL.encode()).hexdigest()).with_suffix('.json').write_text(json.dumps(receipt))
with patch.object(aamas2026, 'fetch', return_value=FIXTURE['contents'].encode()) as fetch:
report = aamas.collect(2026, root/'out', cache=cache)
fetch.assert_called_once_with(aamas2026.URL, cache, 8*1024**2, delay=1)
self.assertEqual(report['source_receipts'][0]['response_receipt'], receipt)
def test_frozen_fixture_is_original_hashed_response(self):
r, = FIXTURE['receipts']['receipts']
self.assertEqual(hashlib.sha256(FIXTURE['contents'].encode()).hexdigest(), r['sha256'])
self.assertEqual(r['bytes'], len(FIXTURE['contents'].encode()))
+1 -1
View File
@@ -99,7 +99,7 @@ class AamasProceedingsTests(unittest.TestCase):
aamas.collect(2012, root/'out', snapshots=source)
self.assertFalse((root/'out/aamas-proceedings-2012.json').exists())
with self.assertRaises(ValueError):
aamas.collect(2026, root/'future', snapshots=source)
aamas.collect(2027, root/'future', snapshots=source)
self.assertFalse((root/'future').exists())
def test_ids_colliding_across_sections_do_not_produce_a_completed_collection(self):
+75
View File
@@ -0,0 +1,75 @@
import gzip
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import corl2026 as c
class CoRL2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.bundle = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/corl2026-source.json.gz').read_bytes()))
def test_official_census_and_presentation_dedup(self):
program = c.parse_program(self.bundle['official_html'])
self.assertEqual(len(program), 687)
self.assertEqual(sum(len(p['presentations']) for p in program.values()), 780)
self.assertEqual(sum(len(p['presentations']) == 2 for p in program.values()), 93)
def test_conflicting_official_title_is_not_silently_folded(self):
html = self.bundle['official_html']
title = 'Diagnosing Compositional Generalization in Sequential Robot Tasks'
self.assertIn(title, html)
with self.assertRaises(ValueError):
c.parse_program(html.replace(title, 'Different paper', 1))
def test_foreign_forum_source_is_rejected(self):
program = c.parse_program(self.bundle['official_html']); archive = json.loads(self.bundle['filtered'])
archive['papers'][0]['author_source'] = 'https://openreview.net/forum?id=WrongPaper'
with self.assertRaisesRegex(ValueError, 'another submission'):
c.join_metadata(program, archive)
def test_unknown_profile_fields_are_rejected(self):
program = c.parse_program(self.bundle['official_html']); archive = json.loads(self.bundle['filtered'])
archive['papers'][0]['authorids'] = ['profile']
with self.assertRaisesRegex(ValueError, 'profile'):
c.join_metadata(program, archive)
def test_source_sets_must_agree(self):
program = c.parse_program(self.bundle['official_html']); archive = json.loads(self.bundle['filtered'])
archive['papers'].pop()
with self.assertRaisesRegex(ValueError, 'forum sets'):
c.join_metadata(program, archive)
def test_complete_snapshot_and_explicit_version_provenance(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); official = root/'official'; archive = root/'archive'
official.mkdir(); archive.mkdir()
(official/'accepted.html').write_text(self.bundle['official_html'])
(official/'research.json').write_text(json.dumps(self.bundle['research']))
(archive/'explorer-paper-metadata-filtered-v1.json').write_text(self.bundle['filtered'])
out = root/'out'; report = c.collect(out, official_snapshots=official, archive_snapshots=archive)
self.assertEqual(report['records'], 687); self.assertEqual(len(report['title_variants']), 34)
self.assertEqual(report['public_rejected_availability'], 'unknown')
rows = [json.loads(l) for l in (out/'corl-program-archive-2026-main.jsonl').read_text().splitlines()]
self.assertEqual(len({r['external_id'] for r in rows}), 687)
self.assertTrue(all(set(r['identifiers']) == {'openreview'} for r in rows))
self.assertTrue(all(r['submission_identifiers'] == [{'scheme': 'openreview', 'value': r['identifiers']['openreview']}] for r in rows))
self.assertTrue(all(not r['tags'] and not r['metadata']['original_decision_signature_verified'] for r in rows))
self.assertTrue(all(r['title'] == r['metadata']['archive_fields']['title'] for r in rows))
self.assertTrue(all(r['authors'] and r['abstract'] for r in rows))
self.assertFalse(report['database_written']); self.assertFalse(report['pdfs_downloaded'])
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(out, official_snapshots=official, archive_snapshots=archive)
(official/'accepted.html').write_text('tampered')
with self.assertRaisesRegex(ValueError, 'hash changed'):
c.collect(root/'failed', official_snapshots=official, archive_snapshots=archive)
self.assertTrue((root/'failed/failed.json').exists())
self.assertFalse((root/'failed/collection-report.json').exists())
def test_forum_identity_validation(self):
for bad in ['https://openreview.net/challenge?redirect=forum', 'https://evil.test/forum?id=abcdefgh', 'https://openreview.net/forum?id=abcdefgh&other=1']:
with self.assertRaises(ValueError):
c.forum_id(bad)
+139
View File
@@ -0,0 +1,139 @@
import copy
import gzip
import json
from pathlib import Path
import tempfile
import unittest
from unittest.mock import patch
from paper_atlas.collectors import eccv2026 as c
from paper_atlas.io.html_cache import PublicHTMLCache
class ECCV2026Tests(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.bundle = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/eccv2026-source.json.gz').read_bytes()))
cls.program = c.parse_program(cls.bundle['official_pages']['accepted.html'])
def snapshots(self, root):
official = root/'official'; paper = root/'paper'; official.mkdir(); paper.mkdir()
for name, raw in self.bundle['official_pages'].items():
(official/name).write_bytes(raw.encode('utf-8'))
(official/'manifest.json').write_bytes(self.bundle['official_manifest'].encode('utf-8'))
for pid, raw in self.bundle['paper_pages'].items():
(paper/(pid+'.html')).write_bytes(raw.encode('utf-8'))
(paper/'manifest.json').write_text(json.dumps(self.bundle['paper_manifest']))
return official, paper
def test_complete_program_and_duplicate_occurrences_match_gold(self):
self.assertEqual(len(self.program), 2834)
self.assertEqual(sum(len(p['row_ordinals']) for p in self.program.values()), 2851)
for ordinal, gold in enumerate(self.bundle['inventory']['records'], 1):
item = self.program[gold['paper_id']]
self.assertEqual(item['listing'], {k: gold[k] for k in ('title','authors_literal','keywords_literal','presentation_literal')})
self.assertIn(ordinal, item['row_ordinals'])
def test_conflicting_duplicate_not_silently_folded(self):
title = 'Transferability Between Understanding and Generation in Unified Multimodal Models'
with self.assertRaisesRegex(ValueError, 'Conflicting'):
c.parse_program(self.bundle['official_pages']['accepted.html'].replace('href="/virtual/2026/poster/3137">'+title, 'href="/virtual/2026/poster/3137">Conflicting title', 1))
def test_missing_member_and_changed_provisional_status_rejected(self):
html = self.bundle['official_pages']['accepted.html']
for changed in [html.replace('/virtual/2026/poster/4048', '/virtual/2024/poster/4048', 1),
html.replace('This paper list is preliminary pending publisher checks.', 'Final archival papers.')]:
with self.assertRaises(ValueError):
c.parse_program(changed)
def test_public_page_fields_match_independent_sample_review(self):
for gold in self.bundle['paper_gold']['records']:
pid = gold['paper_id']; fields = c.parse_paper(self.bundle['paper_pages'][pid], gold['url'], self.program[pid])
self.assertEqual(fields['title'], gold['title']); self.assertEqual(fields['authors'], gold['authors'])
self.assertEqual(fields['abstract'], gold['abstract_literal']); self.assertFalse(fields['review_flags'])
self.assertIn('sig-nal', c.parse_paper(self.bundle['paper_pages']['4048'], self.program['4048']['url'], self.program['4048'])['abstract'])
def test_foreign_canonical_and_inconsistent_jsonld_author_rejected(self):
raw = self.bundle['paper_pages']['4048']; url = self.program['4048']['url']
for changed in [raw.replace(url, 'https://example.org/other'), raw.replace('Suhyeon Ha', 'Other Author', 1)]:
with self.assertRaises(ValueError):
c.parse_paper(changed, url, self.program['4048'])
def test_program_and_page_versions_flagged_without_aliases(self):
program = copy.deepcopy(self.program['4048']); program['listing']['title'] = 'Earlier source title'
fields = c.parse_paper(self.bundle['paper_pages']['4048'], program['url'], program)
self.assertEqual(fields['review_flags'], ['program_and_page_title_versions_differ_requires_review'])
self.assertEqual(fields['title'], self.program['4048']['listing']['title'])
def test_partial_snapshot_collection_has_no_inferred_submission_ids(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); official, paper = self.snapshots(root)
report = c.collect(root/'out', official_snapshots=official, paper_snapshots_dir=paper)
self.assertEqual(report['records'], 3); self.assertEqual(len(report['uncollected_program_ids']), 2831)
self.assertEqual(report['public_rejected_decision_metadata'], 'unknown'); self.assertFalse(report['database_written'])
manifest, = report['files']; self.assertEqual(manifest['status'], 'partial'); self.assertEqual(manifest['source_expected_count'], 2834)
rows = [json.loads(l) for l in Path(manifest['file']).read_text().splitlines()]
self.assertEqual({r['external_id'] for r in rows}, {'2026/4048','2026/3137','2026/3652'})
self.assertTrue(all(not r['identifiers'] and not r['submission_identifiers'] and r['track']=='program_unclassified' for r in rows))
self.assertTrue(all(r['metadata']['publication_state']=='preliminary_pending_publisher_checks' and not r['metadata']['original_decision_signature_verified'] for r in rows))
self.assertTrue(all(t['namespace']=='eccv_program:keyword' for r in rows for t in r['tags']))
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(root/'out', official_snapshots=official, paper_snapshots_dir=paper)
def test_zero_network_budget_reuses_immutable_cache_only(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); official, _ = self.snapshots(root); client = PublicHTMLCache(root/'cache', c.ORIGIN)
for r in self.bundle['paper_manifest']['records']:
body, receipt, _ = client.paths(r['url']); body.write_text(self.bundle['paper_pages'][r['paper_id']]); receipt.write_text(json.dumps({**r,'retrieved_at':self.bundle['paper_manifest']['at']}))
with patch.object(PublicHTMLCache, 'get', side_effect=AssertionError('network forbidden')):
report = c.collect(root/'out', official_snapshots=official, cache=root/'cache', max_new_pages=0)
self.assertEqual((report['records'],report['new_pages'],report['reused_pages']), (3,0,3))
def test_tampered_snapshot_preserves_failure_without_success(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); official, paper = self.snapshots(root); (paper/'4048.html').write_text('tampered')
with self.assertRaises(ValueError):
c.collect(root/'out', official_snapshots=official, paper_snapshots_dir=paper)
self.assertTrue((root/'out/failed.json').exists()); self.assertFalse((root/'out/collection-report.json').exists())
def test_structured_math_markup_requires_explicit_review(self):
raw = self.bundle['paper_pages']['4048']
self.assertIn('<p>Proactive tamper', raw)
fields = c.parse_paper(raw.replace('<p>Proactive tamper', '<p>10<sup>6</sup> Proactive tamper', 1), self.program['4048']['url'], self.program['4048'])
self.assertIn('structured_abstract_markup_requires_review', fields['review_flags'])
self.assertEqual(fields['abstract_markup_tags'], ['sup'])
def test_fixture_preserves_exact_http_bytes_including_crlf(self):
import hashlib
for receipt in json.loads(self.bundle['official_manifest'])['sources']:
raw = self.bundle['official_pages'][receipt['file']].encode('utf-8')
self.assertEqual(len(raw), receipt['bytes'])
self.assertEqual(hashlib.sha256(raw).hexdigest(), receipt['sha256'])
for receipt in self.bundle['paper_manifest']['records']:
raw = self.bundle['paper_pages'][receipt['paper_id']].encode('utf-8')
self.assertEqual(hashlib.sha256(raw).hexdigest(), receipt['sha256'])
def test_real_double_encoded_author_requires_exact_visible_agreement(self):
import hashlib
sample = self.bundle['encoding_sample']; pid = sample['paper_id']; raw = sample['html']
self.assertEqual(hashlib.sha256(raw.encode('utf-8')).hexdigest(), sample['receipt']['sha256'])
self.assertEqual(sample['receipt']['sha256'], 'b575318fa569e83eea5e96f937dd60f078fbed482877318013f797c16a02ff09')
fields = c.parse_paper(raw, sample['receipt']['url'], self.program[pid])
self.assertEqual(fields['authors'], sample['gold']['authors'])
self.assertEqual(fields['jsonld_raw_text']['author_names'][0], 'Zi&amp;#x27;ang Lu')
self.assertIn('jsonld_display_text_normalized_requires_review', fields['review_flags'])
# Real changes to the author or an extra encoding layer remain errors.
for replacement in ('Other Author', 'Zi&amp;amp;#x27;ang Lu'):
with self.assertRaisesRegex(ValueError, 'authors disagree'):
c.parse_paper(raw.replace('Zi&amp;#x27;ang Lu', replacement), sample['receipt']['url'], self.program[pid])
def test_real_already_agreeing_literal_title_is_not_decoded_again(self):
import hashlib
sample = self.bundle['literal_title_sample']; pid = sample['paper_id']; raw = sample['html']
self.assertEqual(hashlib.sha256(raw.encode('utf-8')).hexdigest(), 'c208525ad5f02b4ca274dff7e24a0392dabdc55bb3620497079a84df81a02a72')
fields = c.parse_paper(raw, sample['receipt']['url'], self.program[pid])
self.assertIn('&amp;', fields['title'])
self.assertEqual(fields['title'], sample['gold']['visible_title'])
self.assertEqual(fields['jsonld_raw_text']['name'], sample['gold']['jsonld_title_raw'])
self.assertNotEqual(fields['title'], sample['gold']['blind_double_decoded_title'])
self.assertNotIn('jsonld_display_text_normalized_requires_review', fields['review_flags'])
+90
View File
@@ -0,0 +1,90 @@
from datetime import datetime, timezone
import json
import os
from pathlib import Path
import tempfile
import unittest
from unittest.mock import MagicMock
import urllib.error
from paper_atlas.io.html_cache import PublicHTMLCache, digest, NoRedirect
URL = 'https://eccv.ecva.net/virtual/2026/poster/4048'
class PublicHTMLCacheTests(unittest.TestCase):
def receipt(self, raw):
return {'url':URL,'final_url':URL,'http_status':200,'content_type':'text/html; charset=utf-8',
'bytes':len(raw),'sha256':digest(raw),'retrieved_at':datetime.now(timezone.utc).isoformat()}
def test_old_cache_reused_without_network_or_expiry(self):
with tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp, 'https://eccv.ecva.net', delay=0); raw = b'<html>metadata</html>'
body, receipt, _ = c.paths(URL); body.write_bytes(raw); receipt.write_text(json.dumps(self.receipt(raw))); os.utime(body,(0,0))
c.opener = MagicMock(); self.assertEqual(c.get(URL)[0], raw); c.opener.open.assert_not_called()
def test_pair_tampering_and_orphan_parts_fail_without_refetch(self):
for kind in ('hash','orphan','part'):
with self.subTest(kind=kind), tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp, 'https://eccv.ecva.net', delay=0); body, receipt, _ = c.paths(URL); body.write_bytes(b'original')
if kind == 'hash': receipt.write_text(json.dumps(self.receipt(b'different')))
if kind == 'part': body.unlink(); body.with_suffix('.html.part').write_bytes(b'partial')
c.opener = MagicMock()
with self.assertRaises(ValueError): c.get(URL)
c.opener.open.assert_not_called()
def test_http_access_failure_sticky_without_automatic_retry(self):
with tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp, 'https://eccv.ecva.net', delay=0); c.opener = MagicMock()
c.opener.open.side_effect = urllib.error.HTTPError(URL,403,'Forbidden',{},None)
with self.assertRaises(urllib.error.HTTPError): c.get(URL)
with self.assertRaisesRegex(ValueError,'Previous access'): c.get(URL)
self.assertEqual(c.opener.open.call_count,1)
def test_redirect_nonhtml_challenge_and_oversize_rejected(self):
for kind in ('redirect','type','challenge','size'):
with self.subTest(kind=kind), tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp,'https://eccv.ecva.net',limit=100,delay=0); c.opener=MagicMock(); r=c.opener.open.return_value.__enter__.return_value
r.status=200; r.geturl.return_value=URL; r.headers={'Content-Type':'text/html'}; r.read.return_value=b'<html>metadata</html>'
if kind=='redirect':r.geturl.return_value='https://example.org/login'
if kind=='type':r.headers={'Content-Type':'application/pdf'}
if kind=='challenge':r.read.return_value=b'<title>Just a moment</title>'
if kind=='size':r.read.return_value=b'x'*101
with self.assertRaises(ValueError):c.get(URL)
self.assertFalse(c.paths(URL)[0].exists())
self.assertIsNone(NoRedirect().redirect_request(None,None,None,None,None,None))
def test_valid_response_has_atomic_body_and_receipt(self):
with tempfile.TemporaryDirectory() as tmp:
c=PublicHTMLCache(tmp,'https://eccv.ecva.net',delay=0);c.opener=MagicMock();r=c.opener.open.return_value.__enter__.return_value;r.status=200;r.geturl.return_value=URL;r.headers={'Content-Type':'text/html','Content-Length':'4'};r.read.return_value=b'data'
raw, receipt=c.get(URL);self.assertEqual(raw,b'data');self.assertEqual(receipt['sha256'],digest(raw));self.assertEqual(c.cached(URL),(raw,receipt));self.assertFalse(list(Path(tmp).glob('*.part')))
def test_foreign_origin_and_cache_symlink_rejected(self):
with tempfile.TemporaryDirectory() as tmp:
c=PublicHTMLCache(tmp,'https://eccv.ecva.net',delay=0)
with self.assertRaises(ValueError):c.get('https://example.org/other')
body,_,_=c.paths(URL);target=Path(tmp)/'outside';target.write_bytes(b'other');body.symlink_to(target)
with self.assertRaises(ValueError):c.get(URL)
def test_successful_fetch_publishes_receipt_and_reuses_without_refetch(self):
with tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp, 'https://eccv.ecva.net', delay=0)
c.opener = MagicMock(); response = c.opener.open.return_value.__enter__.return_value
response.status = 200; response.geturl.return_value = URL
raw = b'<html>public metadata</html>'
response.headers = {'Content-Type':'text/html','Content-Length':str(len(raw))}; response.read.return_value = raw
self.assertEqual(c.get(URL)[0], raw); self.assertEqual(c.get(URL)[0], raw)
self.assertEqual(c.opener.open.call_count, 1)
body, receipt, _ = c.paths(URL)
self.assertTrue(body.exists()); self.assertTrue(receipt.exists())
self.assertFalse(body.with_suffix('.html.part').exists())
self.assertFalse(receipt.with_suffix('.json.part').exists())
def test_changed_url_is_rejected_before_reading_body(self):
with tempfile.TemporaryDirectory() as tmp:
c = PublicHTMLCache(tmp, 'https://eccv.ecva.net', delay=0)
c.opener = MagicMock(); response = c.opener.open.return_value.__enter__.return_value
response.status = 200; response.geturl.return_value = 'https://example.org/login'
response.headers = {'Content-Type':'text/html'}
with self.assertRaises(ValueError): c.get(URL)
response.read.assert_not_called()
+95
View File
@@ -0,0 +1,95 @@
import gzip
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import icaps2026 as c
class ICAPS2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.bundle = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/icaps2026-source.json.gz').read_bytes()))
def snapshots(self, root):
root.mkdir()
for name, html in self.bundle['pages'].items():
(root/name).write_text(html)
(root/'manifest.json').write_text(json.dumps(self.bundle['manifest']))
return root
def test_full_listing_matches_independently_recorded_literals(self):
rows = c.parse_listing(self.bundle['pages']['accepted.html'])
self.assertEqual([{k: r[k] for k in ('section', 'title', 'authors_literal')} for r in rows],
[{k: r[k] for k in ('section', 'title', 'authors_literal')} for r in self.bundle['expected']])
self.assertEqual(sum(r['section'] == c.MAIN for r in rows), 90)
self.assertEqual(sum(r['section'] == c.PREVIOUS for r in rows), 10)
self.assertTrue(all(len(r['authors']) >= 2 for r in rows))
def test_incomplete_and_duplicate_listing_rejected(self):
html = self.bundle['pages']['accepted.html']
with self.assertRaises(ValueError):
c.parse_listing(html.replace('class=paper>', 'class=omitted>', 1))
with self.assertRaisesRegex(ValueError, 'repeated paper title'):
c.parse_listing(html.replace(self.bundle['expected'][1]['title'], self.bundle['expected'][0]['title']))
def test_changed_edition_or_section_rejected(self):
html = self.bundle['pages']['accepted.html']
for changed in [html.replace('ICAPS 2026</h1>', 'ICAPS 2027</h1>'),
html.replace(c.MAIN, 'Workshop Papers')]:
with self.assertRaises(ValueError):
c.parse_listing(changed)
def test_ambiguous_author_lists_not_guessed(self):
for literal in ['Anonymous', 'Alice, and Bob', 'Alice and Bob and Carol', 'Alice, Bob and name@example.org']:
with self.assertRaises(ValueError):
c.authors_from_literal(literal)
self.assertEqual(c.authors_from_literal('Augusto B. Corrêa, Simon Dold and Malte Helmert'),
['Augusto B. Corrêa', 'Simon Dold', 'Malte Helmert'])
def test_collection_separates_presentations_and_unresolved_tracks(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = self.snapshots(root/'src'); out = root/'out'
report = c.collect(out, snapshots=src)
self.assertEqual((report['records'], report['previously_published_presentations'], report['missing_abstract_count']), (90, 10, 90))
self.assertEqual(report['public_rejected_availability'], 'unknown')
rows = [json.loads(l) for l in Path(report['files'][0]['file']).read_text().splitlines()]
previous = json.loads((out/report['previously_published_file']).read_text())
self.assertFalse(previous['ingestion_ready']); self.assertEqual(previous['count'], 10)
self.assertFalse({r['title'] for r in rows} & {r['title'] for r in previous['records']})
self.assertTrue(all(r['track'] == 'main_tracks_unclassified' and r['status'] == 'accepted' for r in rows))
self.assertTrue(all(not r['identifiers'] and not r['submission_identifiers'] and r['abstract'] is None and not r['tags'] for r in rows))
self.assertEqual(len({r['external_id'] for r in rows}), 90)
self.assertTrue(all(r['metadata']['source_entry_key']['value'] in r['external_id'] for r in rows))
self.assertFalse(report['database_written']); self.assertFalse(report['pdfs_downloaded'])
self.assertEqual(report['files'][0]['status'], 'partial')
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(out, snapshots=src)
def test_tampering_retains_failure_without_success_receipt(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = self.snapshots(root/'src')
(src/'accepted.html').write_text('tampered')
with self.assertRaisesRegex(ValueError, 'hash changed'):
c.collect(root/'out', snapshots=src)
self.assertTrue((root/'out/failed.json').exists())
self.assertFalse((root/'out/collection-report.json').exists())
def test_receipt_redirect_naive_time_and_duplicate_source_rejected(self):
for kind in ('redirect', 'time', 'duplicate'):
with self.subTest(kind=kind), tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = self.snapshots(root/'src'); p = src/'manifest.json'; r = json.loads(p.read_bytes())
if kind == 'redirect': r['sources'][0]['final_url'] = 'https://example.org/challenge'
if kind == 'time': r['at'] = '2026-11-16T12:00:00'
if kind == 'duplicate': r['sources'][1] = r['sources'][0]
p.write_text(json.dumps(r))
with self.assertRaises(ValueError):
c.collect(root/'out', snapshots=src)
self.assertFalse((root/'out/collection-report.json').exists())
def test_snapshot_symlink_rejected(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = self.snapshots(root/'src'); p = src/'accepted.html'; raw = p.read_bytes(); p.unlink(); target = root/'external.html'; target.write_bytes(raw); p.symlink_to(target)
with self.assertRaisesRegex(ValueError, 'Unsafe'):
c.collect(root/'out', snapshots=src)
+96
View File
@@ -0,0 +1,96 @@
import copy
import gzip
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import icaps_registry2026 as c
class ICAPSRegistry2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.raw = gzip.decompress((Path(__file__).parent/'fixtures/icaps2026-crossref.json.gz').read_bytes())
cls.snapshot = json.loads(cls.raw)
def test_exact_edition_not_publication_year_alone(self):
papers, excluded = c.registry_papers(self.snapshot)
self.assertEqual(len(papers), 90); self.assertEqual(len(excluded), 1)
self.assertTrue(all(p['original']['volume'] == '36' for p in papers))
self.assertEqual(excluded, [r['DOI'] for r in self.snapshot['records'] if r['volume'] == '35'])
self.assertTrue(all(p['abstract'] and p['authors'] for p in papers))
def test_full_collection_keeps_registered_versions_and_publication_identity(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = root/'src'; src.mkdir(); (src/'crossref-filtered.json').write_bytes(self.raw)
report = c.collect(root/'out', snapshots=src)
self.assertEqual((report['records'], report['missing_abstract_count']), (90, 0))
self.assertFalse(report['database_written']); self.assertFalse(report['pdfs_downloaded'])
self.assertEqual(report['public_rejected_availability'], 'unknown')
manifest, = report['files']; self.assertEqual(manifest['status'], 'partial')
rows = [json.loads(l) for l in Path(manifest['file']).read_text().splitlines()]
original = {r['DOI']: r for r in self.snapshot['records'] if r['volume'] == '36'}
self.assertEqual({r['external_id'] for r in rows}, set(original))
for r in rows:
doi = r['external_id']; o = original[doi]
self.assertEqual(r['metadata']['registry_fields'], o)
self.assertEqual(r['title'], o['title'][0])
self.assertEqual(r['authors'], [a['given']+' '+a['family'] for a in o['authors']])
self.assertEqual(r['identifiers'], {'doi': doi})
self.assertEqual(r['submission_identifiers'], [{'scheme': 'publication_doi', 'value': doi}])
self.assertFalse(r['tags']); self.assertFalse(r['metadata']['original_decision_signature_verified'])
self.assertEqual(r['track'], 'main_tracks_unclassified')
self.assertEqual(r['metadata']['publication_date'], '2026-06-08')
changed = next(r for r in rows if r['external_id'] == '10.1609/icaps.v36i1.42860')
self.assertEqual(changed['title'], 'Analyzing Planner Design Trade-Offs for MAPF Under ADG-Based Realistic Execution')
self.assertIn('Stephen F. Smith', changed['authors'])
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(root/'out', snapshots=src)
def test_math_markup_is_not_flattened_as_plain_numbers(self):
self.assertEqual(c.abstract_text('<jats:p>A &lt; B and C &amp; D.\n Next line.</jats:p>'), 'A < B and C & D. Next line.')
for raw in ['<jats:p>10<sup>6</sup></jats:p>', '<jats:p>x</jats:p><jats:p>y</jats:p>', '<jats:p> </jats:p>', None]:
with self.assertRaises(ValueError):
c.abstract_text(raw)
def test_invalid_or_repeated_doi_and_edition_rejected(self):
for kind in ('doi', 'repeat', 'volume', 'container', 'year'):
with self.subTest(kind=kind):
s = copy.deepcopy(self.snapshot); rows = [r for r in s['records'] if r['volume'] == '36']; r = rows[0]
if kind == 'doi': r['DOI'] = '10.9999/foreign'
if kind == 'repeat': rows[1]['DOI'] = r['DOI']
if kind == 'volume': r['volume'] = '37'
if kind == 'container': r['container-title'] = ['Other Conference']
if kind == 'year': r['published']['date-parts'] = [[2025, 6, 8]]
with self.assertRaises(ValueError):
c.registry_papers(s)
def test_missing_authors_and_changed_order_evidence_rejected(self):
for kind in ('missing', 'sequence', 'profile'):
s = copy.deepcopy(self.snapshot); r = next(r for r in s['records'] if r['volume'] == '36')
if kind == 'missing': r['authors'][0].pop('family')
if kind == 'sequence': r['authors'][0]['sequence'] = 'additional'
if kind == 'profile': r['authors'][0]['ORCID'] = 'unreviewed-profile'
with self.assertRaises(ValueError):
c.registry_papers(s)
def test_receipt_scope_and_timezone_rejected(self):
for key, value in [('at', '2026-11-17T00:00:00'), ('url', 'https://example.org/works'), ('raw_response_sha256', 'other')]:
s = copy.deepcopy(self.snapshot); s[key] = value
with self.assertRaises(ValueError):
c.registry_papers(s)
def test_tamper_failure_retained_and_success_not_published(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = root/'src'; src.mkdir(); (src/'crossref-filtered.json').write_bytes(self.raw+b' ')
with self.assertRaisesRegex(ValueError, 'hash changed'):
c.collect(root/'out', snapshots=src)
self.assertTrue((root/'out/failed.json').exists())
self.assertFalse((root/'out/collection-report.json').exists())
def test_symlink_rejected(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); src = root/'src'; src.mkdir(); (root/'data.json').write_bytes(self.raw); (src/'crossref-filtered.json').symlink_to(root/'data.json')
with self.assertRaisesRegex(ValueError, 'Unsafe'):
c.collect(root/'out', snapshots=src)
+113
View File
@@ -0,0 +1,113 @@
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import icdm2026 as c
class ICDM2026Tests(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/icdm2026-source.json.gz').read_bytes()))
cls.raw = cls.fixture['files'][c.SNAPSHOT+'/icdm-detailed-sessions.html'].encode()
cls.gold = cls.fixture['independent_inventory']
def snapshots(self, root):
for name, text in self.fixture['files'].items():
path = root/name; path.parent.mkdir(parents=True, exist_ok=True); path.write_bytes(text.encode())
def test_all_rows_match_separately_parsed_frozen_inventory(self):
result = c.parse_program(self.raw)
self.assertEqual(len(result), len(self.gold['entries']))
for position, (row, gold) in enumerate(zip(result, self.gold['entries']), 1):
for field in ('paper_id_literal', 'title', 'authors_literal', 'source_reported_authors', 'time_literal'):
self.assertEqual(row[field], gold[field])
self.assertEqual(row['source_position'], position)
for field, value in gold['session'].items():
self.assertEqual(row['session'][field], value)
def test_instantiable_collection_retains_scope_fields_and_source_bytes(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src')
report = c.collect(root/'out', snapshots=root/'src')
self.assertEqual(report['track_counts'], self.gold['track_counts_literal'])
self.assertEqual(report['presentation_counts'], self.gold['presentation_counts_literal'])
self.assertEqual((report['records'], report['missing_abstract_count']), (324, 324))
self.assertFalse(report['database_written']); self.assertEqual(report['network_requests'], 0)
self.assertEqual(report['public_rejected_metadata'], 'unknown_no_public_export_verified')
all_rows = []
for manifest in report['files']:
path = Path(manifest['file']); raw = path.read_bytes()
self.assertEqual(hashlib.sha256(raw).hexdigest(), manifest['file_sha256'])
self.assertEqual(manifest['status'], 'partial')
rows = [json.loads(line) for line in raw.splitlines()]
self.assertEqual(len(rows), manifest['missing_abstract_count'])
for row in rows:
self.assertEqual((row['venue'], row['year'], row['track']), ('ICDM', 2026, manifest['track']))
self.assertIsNone(row['abstract']); self.assertIsNone(row['metadata']['paper_type'])
self.assertEqual(row['identifiers'], {}); self.assertEqual(row['submission_identifiers'], [])
self.assertFalse(row['metadata']['original_decision_signature_verified'])
self.assertEqual(row['external_id'], '2026/'+row['metadata']['program_fields']['paper_id_literal'])
self.assertTrue(row['url'].startswith(c.PROGRAM+'#paper-DM'))
self.assertTrue(all(t['namespace'].startswith(c.SOURCE+':') for t in row['tags']))
self.assertNotIn('doi', row['metadata']); self.assertNotIn('decision_date', row)
all_rows.extend(rows)
self.assertEqual(len({r['external_id'] for r in all_rows}), 324)
for item in report['evidence']:
raw = (root/'out'/item['output_file']).read_bytes()
self.assertEqual(raw, self.fixture['files'][c.SNAPSHOT+'/'+item['file']].encode())
self.assertTrue((root/'out/collection-report.json').is_file())
self.assertFalse((root/'out/collection-report.json.part').exists())
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(root/'out', snapshots=root/'src')
def test_stale_or_different_conference_rejected(self):
for old, new in [(b'IEEE ICDM 2026', b'IEEE ICDM 2025'),
(b'In-person session presentation list.', b'Workshop presentation list.')]:
with self.subTest(old=old), self.assertRaises(ValueError):
c.parse_program(self.raw.replace(old, new))
def test_track_attribute_must_agree_with_visible_label(self):
changed = self.raw.replace(b'data-track="research"', b'data-track="applied"', 1)
with self.assertRaisesRegex(ValueError, 'track'):
c.parse_program(changed)
def test_presentation_attribute_is_not_inferred_from_session_id(self):
changed = self.raw.replace(b'data-format="long"', b'data-format="short"', 1)
with self.assertRaisesRegex(ValueError, 'presentation'):
c.parse_program(changed)
def test_repeated_paper_and_source_locator_disagreement_rejected(self):
for old, new in [(b'DM1278', b'DM827'), (b'id="paper-DM827"', b'id="paper-DM99999"')]:
with self.subTest(old=old), self.assertRaises(ValueError):
c.parse_program(self.raw.replace(old, new))
def test_lost_session_or_changed_session_counts_rejected(self):
for old, new in [(b'data-session-id="L1"', b'data-session-id="L2"'),
(b'Long \xc2\xb7 7 papers', b'Long \xc2\xb7 6 papers')]:
with self.subTest(old=old), self.assertRaises(ValueError):
c.parse_program(self.raw.replace(old, new, 1))
def test_empty_author_is_not_silently_dropped(self):
changed = self.raw.replace(b'Qishuang Fu; Andreas Deppeler;', b'Qishuang Fu;;', 1)
with self.assertRaisesRegex(ValueError, 'author'):
c.parse_program(changed)
def test_snapshot_manifest_and_body_corruption_rejected_before_output(self):
for name in ('manifest.json', 'icdm-detailed-sessions.html', 'icdm-research-policy.html', 'icdm-applied-policy.html'):
with self.subTest(name=name), tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src'); path = root/'src'/c.SNAPSHOT/name
path.write_bytes(path.read_bytes()+b' ')
with self.assertRaises(ValueError):
c.collect(root/'out', snapshots=root/'src')
self.assertFalse((root/'out').exists())
def test_snapshot_source_symlink_rejected(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src'); path = root/'src'/c.SNAPSHOT/'icdm-detailed-sessions.html'
real = root/'real.html'; path.rename(real); path.symlink_to(real)
with self.assertRaisesRegex(ValueError, 'Unsafe source'):
c.collect(root/'out', snapshots=root/'src')
+79
View File
@@ -0,0 +1,79 @@
import base64
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import recsys2026 as c
class RecSys2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture=json.loads(gzip.decompress((Path(__file__).parent/'fixtures/recsys2026-source.json.gz').read_bytes()))
cls.files={n:base64.b64decode(raw) for n,raw in cls.fixture['files'].items()}
cls.body=cls.files[c.SNAPSHOT+'/recsys-contributions.html']
def materialize(self,root):
for name,raw in self.files.items():
path=root/name;path.parent.mkdir(parents=True,exist_ok=True);path.write_bytes(raw)
def test_all263_observations_against_independent_gold(self):
items=c.parse_program(self.body)
keys=self.fixture['independent_gold'][0].keys()
self.assertEqual([{k:r[k] for k in keys} for r in items],self.fixture['independent_gold'])
self.assertEqual(len(items),263);self.assertEqual(len({r['doi'] for r in items}),260)
self.assertEqual(sum(bool(r['review_flags']) for r in items),6)
variants=[r for r in items if r['doi']=='10.1145/3773078.3831839']
self.assertNotEqual(variants[0]['title'],variants[1]['title']);self.assertNotEqual(variants[0]['authors'],variants[1]['authors'])
self.assertNotEqual(variants[0]['abstract'],variants[1]['abstract'])
def test_installed_format_tracks_evidence_and_original_fields(self):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);src=root/'src';self.materialize(src);out=root/'out'
report=c.collect(out,snapshots=src)
self.assertEqual((report['records'],report['unique_dois'],report['missing_abstract_count']),(263,260,0))
self.assertFalse(report['database_written']);self.assertFalse(report['pdfs_downloaded']);self.assertEqual(report['network_requests'],0);self.assertEqual(report['approved_for_import'],0)
self.assertEqual(report['track_counts'],{k:n for k,_,n in c.TRACKS});self.assertEqual(report['duplicate_doi_observations'],6)
gold={r['locator']:r for r in self.fixture['independent_gold']};seen=set()
for manifest in report['files']:
path=Path(manifest['file']);raw=path.read_bytes();self.assertEqual(hashlib.sha256(raw).hexdigest(),manifest['file_sha256'])
rows=[json.loads(l) for l in raw.splitlines()];self.assertEqual(len(rows),manifest['count']);self.assertEqual(manifest['status'],'partial')
for r in rows:
item=r['metadata']['program_fields'];locator=item['locator'];seen.add(locator);expected=gold[locator]
for key in expected:self.assertEqual(item[key],expected[key])
self.assertEqual(r['external_id'],'2026/'+locator);self.assertEqual(r['title'],expected['title']);self.assertEqual(r['authors'],expected['authors']);self.assertEqual(r['abstract'],expected['abstract'])
self.assertEqual(r['track'],manifest['track']);self.assertEqual(r['identifiers'],{'doi':expected['doi']});self.assertEqual(r['submission_identifiers'],[{'scheme':'publication_doi','value':expected['doi']}])
self.assertEqual(r['tags'],[{'namespace':'recsys_program:contribution_type','value':expected['type_labels'][0]}]);self.assertFalse(r['metadata']['original_decision_signature_verified'])
self.assertEqual(seen,set(gold));self.assertEqual((out/'evidence/recsys-contributions.html').read_bytes(),self.body)
with self.assertRaisesRegex(ValueError,'Output exists'):c.collect(out,snapshots=src)
def test_edition_and_track_label_drift_rejected(self):
for old,new in [(b'RecSys 26 - Accepted Contributions',b'RecSys 25 - Accepted Contributions'),(b'>Long Papers</a>',b'>Workshops</a>'),(b"title='Research'",b"title='Unreviewed'")]:
with self.subTest(old=old),self.assertRaises(ValueError):c.parse_program(self.body.replace(old,new,1))
def test_duplicate_locator_missing_author_and_foreign_doi_rejected(self):
for old,new in [(b'accordion-1-slide-2"',b'accordion-1-slide-1"'),(b'by Md Aminul Islam, Kathryn Vasilaky and Elena Zheleva',b'by '),(b'https://doi.org/10.1145/3773078.3831824',b'https://doi.org/10.1145/9999999.3831824')]:
with self.subTest(old=old),self.assertRaises(ValueError):c.parse_program(self.body.replace(old,new,1))
def test_math_markup_requires_review_and_is_not_flattened(self):
changed=self.body.replace(b'Learning-to-rank (LTR)',b'Learning-to-rank <sup>2</sup> (LTR)',1)
with self.assertRaisesRegex(ValueError,'markup'):c.parse_program(changed)
def test_corruption_and_missing_receipt_body_preserve_failure(self):
for kind in ('manifest','body','missing'):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);src=root/'src';self.materialize(src);path=src/c.SNAPSHOT/('manifest.json' if kind=='manifest' else 'recsys-contributions.html')
if kind=='missing':path.unlink()
else:path.write_bytes(path.read_bytes()+b' ')
with self.assertRaises((ValueError,FileNotFoundError)):c.collect(root/'out',snapshots=src)
self.assertTrue((root/'out/failed.json').exists());self.assertFalse((root/'out/collection-report.json').exists())
def test_symlink_source_and_root_rejected(self):
for kind in ('source','root'):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);src=root/'src';self.materialize(src);path=src/c.SNAPSHOT/'recsys-contributions.html' if kind=='source' else src
original=path.with_name(path.name+'-original');path.rename(original);path.symlink_to(original,target_is_directory=kind=='root')
with self.assertRaisesRegex(ValueError,'Unsafe'):c.collect(root/'out',snapshots=src)
+100
View File
@@ -0,0 +1,100 @@
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from unittest.mock import patch
from paper_atlas.collectors import rss2026 as rss
FIXTURE = Path(__file__).parent/'fixtures/rss2026-source.json.gz'
class Rss2026Tests(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture=json.loads(gzip.decompress(FIXTURE.read_bytes()))
def setUp(self):
self.tmp=tempfile.TemporaryDirectory();self.addCleanup(self.tmp.cleanup)
self.root=Path(self.tmp.name);self.snap=self.root/'snapshots';self.snap.mkdir()
for name,text in self.fixture['pages'].items():(self.snap/name).write_bytes(text.encode())
(self.snap/'receipts.json').write_text(json.dumps(self.fixture['receipts']))
self.pages=self.fixture['pages'];self.index,self.program=self.lists()
def lists(self,**changes):
pages={**self.pages,**changes}
return rss.parse_lists(*(pages[n] for n in rss.PAGES))
def test_all210_records_match_independent_primary_inventory_and_evidence(self):
report=rss.collect(self.root/'out',snapshots=self.snap)
self.assertEqual(report['records'],210);self.assertFalse(report['database_written'])
self.assertFalse(report['pdfs_downloaded']);self.assertEqual(len(report['source_evidence']),214)
manifest,=report['files'];self.assertEqual(manifest['status'],'partial')
rows=[json.loads(l) for l in Path(manifest['file']).read_text().splitlines()]
for row,gold in zip(rows,self.fixture['independent_gold']):
self.assertEqual(row['external_id'],f"2026/p{gold['paper_id']:03d}")
for key in ('title','authors','abstract'):self.assertEqual(row[key],gold[key])
self.assertEqual(row['identifiers'],{'doi':gold['doi_raw'].lower()})
self.assertEqual(row['metadata']['publisher_date'],gold['publication_date'])
self.assertNotIn('submission_identifiers',row);self.assertEqual(row['track'],'main')
e=row['metadata']['source_evidence'];self.assertEqual(hashlib.sha256((self.root/'out'/e['file']).read_bytes()).hexdigest(),e['sha256'])
self.assertEqual(manifest['file_sha256'],hashlib.sha256(Path(manifest['file']).read_bytes()).hexdigest())
self.assertEqual(manifest['missing_abstract_count'],0)
def test_full_hidden_authors_are_separate_from_six_short_previews(self):
shortened=[p for p in self.program.values() if p['authors_complete']!=p['authors_preview']]
self.assertEqual(len(shortened),6)
p=rss.parse_paper(self.pages['p087.html'],87,self.index[87],self.program[87])
self.assertEqual(len(p['authors']),55)
self.assertFalse(p['program_variants'])
def test_six_reviewed_name_variants_preserve_original_strings(self):
for pid in rss.REVIEWED_AUTHOR_VARIANTS:
p=rss.parse_paper(self.pages[f'p{pid:03d}.html'],pid,self.index[pid],self.program[pid])
self.assertIn('reviewed_program_author_name_variant_not_person_alias',p['program_variants'])
self.assertNotEqual(', '.join(p['authors']),self.program[pid]['authors_complete'])
def test_reviewed_title_variant_is_not_an_automatic_erratum(self):
p=rss.parse_paper(self.pages['p089.html'],89,self.index[89],self.program[89])
self.assertTrue(p['title'].startswith('RLinf-VLA'))
self.assertTrue(self.program[89]['title'].startswith('RLux-VLA'))
self.assertIn('reviewed_program_title_variant_not_erratum',p['program_variants'])
def test_unreviewed_author_or_title_variant_stops(self):
for changed in [{'authors_complete':'An Unrelated Person'},{'title':'A different paper'}]:
with self.assertRaisesRegex(ValueError,'Unreviewed'):
rss.parse_paper(self.pages['p001.html'],1,self.index[1],{**self.program[1],**changed})
def test_demo_policy_cannot_be_reused_from2025(self):
with self.assertRaisesRegex(ValueError,'policy'):
self.lists(**{'cfp.html':self.pages['cfp.html'].replace('will not have a separate category','will have a separate category')})
with self.assertRaises(ValueError):
self.lists(**{'sponsor-demos.html':self.pages['sponsor-demos.html'].replace('Sponsor Demos','Demo Papers')})
def test_wrong_doi_year_pdf_or_missing_abstract_is_rejected(self):
html=self.pages['p001.html']
for old,new in [('10.15607/RSS.2026.XXII.001','10.15607/RSS.2025.XXI.001'),
('2026/07/13','2025/07/13'),('rss22/p001.pdf','rss22/p002.pdf'),('Abstract:','Summary:')]:
self.assertIn(old,html)
with self.assertRaises(ValueError):rss.parse_paper(html.replace(old,new),1,self.index[1],self.program[1])
def test_index_census_and_preview_tampering_stop(self):
with self.assertRaises(ValueError):self.lists(**{'index.html':self.pages['index.html'].replace('p001.html','p999.html')})
bad=self.pages['accepted.html'].replace('Huayi Zhou, Kui Jia','Somebody Else',1)
with self.assertRaisesRegex(ValueError,'prefix'):self.lists(**{'accepted.html':bad})
def test_snapshot_tamper_has_no_completion_marker_and_never_fetches(self):
p=self.snap/'p001.html';p.write_bytes(p.read_bytes()+b'changed')
with patch.object(rss,'fetch',side_effect=AssertionError('network must not be used')):
with self.assertRaisesRegex(ValueError,'receipt'):rss.collect(self.root/'out',snapshots=self.snap)
self.assertFalse((self.root/'out/collection-report.json').exists())
self.assertTrue((self.root/'out/failed.json').exists())
def test_existing_output_and_unsafe_receipt_are_rejected(self):
out=self.root/'existing';out.mkdir()
with self.assertRaises(ValueError):rss.collect(out,snapshots=self.snap)
receipts=[*self.fixture['receipts'],{**self.fixture['receipts'][0],'file':'../outside'}]
(self.snap/'receipts.json').write_text(json.dumps(receipts))
with self.assertRaisesRegex(ValueError,'Unsafe'):rss.collect(self.root/'unsafe',snapshots=self.snap)
self.assertTrue((self.root/'unsafe/failed.json').exists())
self.assertFalse((self.root/'unsafe/collection-report.json').exists())
+59
View File
@@ -0,0 +1,59 @@
"""RSS publisher/DBLP observations share one event without losing old decisions."""
from pathlib import Path
import json
import sqlite3
import tempfile
import unittest
from paper_atlas.db.submissions import identities_for_row
from tests.test_aamas_publication_identity import row, ingest
DOI='10.15607/rss.2026.xxii.041'
def rss(source,external,**kwargs):
return row(source,external,venue='RSS',year=2026,track='main',
identifiers=kwargs.pop('identifiers',{'doi':DOI}),**kwargs)
class RssPublicationIdentityTests(unittest.TestCase):
def test_only_verified_accepted_source_adds_publication_identity(self):
r=rss('rss_proceedings','2026/p041',identifiers={'doi':'https://doi.org/'+DOI.upper()})
self.assertEqual(identities_for_row(r),[('publication_doi',DOI)])
for status in ('rejected','desk_rejected','withdrawn','under_review','unknown'):
self.assertEqual(identities_for_row({**r,'status':status}),[])
self.assertEqual(identities_for_row({**r,'source':'unverified_rss_copy'}),[])
self.assertEqual(identities_for_row({**r,'identifiers':{'doi':'not-a-doi'}}),[])
def test_publisher_and_bibliography_both_orders_preserve_two_observations(self):
a=rss('rss_proceedings','2026/p041');b=rss('dblp','conf/rss/Example26')
for pair in ((a,b),(b,a)):
with self.subTest(first=pair[0]['source']),tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);db=root/'papers.sqlite'
for r in pair:ingest(db,root,r)
ingest(db,root,a)
with sqlite3.connect(db) as c:
self.assertEqual(c.execute('SELECT count(*) FROM papers').fetchone()[0],1)
self.assertEqual(c.execute('SELECT count(*) FROM submissions').fetchone()[0],1)
self.assertEqual(c.execute('SELECT count(*) FROM source_records').fetchone()[0],2)
self.assertEqual(c.execute('SELECT count(*) FROM submission_sources').fetchone()[0],2)
self.assertEqual(c.execute('SELECT count(*) FROM decisions').fetchone()[0],2)
self.assertEqual(c.execute('SELECT count(*) FROM decision_sources').fetchone()[0],2)
self.assertEqual(c.execute('SELECT scheme,value FROM submission_identifiers').fetchall(),[('publication_doi',DOI)])
metadata=[json.loads(r[0]) for r in c.execute('SELECT metadata_json FROM source_records')]
self.assertEqual({x['original_source'] for x in metadata},{'rss_proceedings','dblp'})
self.assertEqual(c.execute('PRAGMA foreign_key_check').fetchall(),[])
def test_prior_rejection_remains_independent_and_raw_evidence_unchanged(self):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);db=root/'papers.sqlite'
old=row('openreview_snapshot','old-forum',venue='ICLR',year=2025,track='main',
status='rejected',identifiers={'arxiv':'2501.00001','openreview':'old-forum'})
ingest(db,root,old)
with sqlite3.connect(db) as c:
before=c.execute('SELECT * FROM decisions').fetchall()
new=rss('rss_proceedings','2026/p041',identifiers={'arxiv':'2501.00001','doi':DOI})
ingest(db,root,new)
with sqlite3.connect(db) as c:
self.assertEqual(c.execute('SELECT * FROM decisions WHERE status=\'rejected\'').fetchall(),before)
self.assertEqual(c.execute('SELECT count(*) FROM submissions').fetchone()[0],2)
self.assertEqual(c.execute('SELECT count(*) FROM papers').fetchone()[0],1)
self.assertEqual(c.execute('SELECT count(*) FROM decisions WHERE status=\'accepted\'').fetchone()[0],1)
self.assertEqual(c.execute('PRAGMA foreign_key_check').fetchall(),[])
+142
View File
@@ -0,0 +1,142 @@
import copy
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import sdm2026 as c
class SDM2026Tests(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/sdm2026-source.json.gz').read_bytes()))
cls.message = json.loads(cls.fixture['files']['sdm2026-crossref-research-v1/isbn-works.json'])['message']
def snapshots(self, root):
for name, text in self.fixture['files'].items():
path = root/name; path.parent.mkdir(parents=True, exist_ok=True); path.write_bytes(text.encode('utf-8'))
def test_publication_census_excludes_book_and_front_matter_not_cancelled_paper(self):
papers, excluded = c.registry_papers(self.message)
self.assertEqual(len(papers), 53)
self.assertEqual({x['doi'] for x in excluded}, {c.BOOK, c.BOOK+'.fm'})
self.assertIn(c.BOOK+'.7', {p['original']['DOI'] for p in papers})
marked = [p for p in papers if p['review_flags']]
self.assertEqual({p['original']['DOI'] for p in marked}, {c.BOOK+'.12', c.BOOK+'.35'})
for p in marked:
self.assertTrue(p['title'].endswith('*'))
self.assertIn('<sup>*</sup>', p['original']['title'][0])
def test_full_collection_preserves_separate_source_versions_and_unknown_negatives(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src')
report = c.collect(root/'out', snapshots=root/'src')
self.assertEqual((report['records'], report['missing_abstract_count']), (53, 53))
self.assertFalse(report['database_written']); self.assertEqual(report['network_requests'], 0)
self.assertEqual(report['public_rejected_metadata'], 'unknown_anonymous_query_http403')
self.assertEqual(report['program_registry_author_sequence_differences'], 20)
manifest, = report['files']; self.assertEqual(manifest['status'], 'partial')
rows = [json.loads(line) for line in Path(manifest['file']).read_text().splitlines()]
originals = {x['DOI']: x for x in self.message['items'] if x['type'] == 'book-chapter'}
for row in rows:
self.assertEqual(row['metadata']['registry_fields'], originals[row['external_id']])
self.assertIsNone(row['abstract']); self.assertFalse(row['tags'])
self.assertEqual(row['identifiers'], {'doi': row['external_id']})
self.assertEqual(row['submission_identifiers'], [{'scheme': 'publication_doi', 'value': row['external_id']}])
self.assertEqual((row['status'], row['track']), ('accepted', 'proceedings_unclassified'))
self.assertEqual(row['metadata']['publication_dates']['published-online'], '2026-11-12')
sidecar = report['program_observations']
self.assertTrue(sidecar['not_an_import_manifest'])
observations = [json.loads(line) for line in Path(sidecar['file']).read_text().splitlines()]
self.assertEqual(len(observations), 53)
self.assertEqual(sum(x['public_page_fields'] is not None for x in observations), 52)
for observation in observations:
self.assertNotIn('status', observation); self.assertNotIn('identifiers', observation)
self.assertEqual(observation['record_kind'], 'source_observation_not_import_record')
cancelled, = [x for x in observations if x['cancelled_presentation']]
self.assertEqual(cancelled['title'], 'A Trajectory View of Statistical Policy Comparison')
self.assertIsNone(cancelled['public_page_fields'])
# The real source has conflicting author sets under the same title.
conflict = next(x for x in observations if x['title'].startswith('Scaffold-Aware Generative'))
self.assertEqual(conflict['public_page_fields']['source_reported_authors'], ['Xin Wang', 'Qinwen Ge'])
candidate, = conflict['registry_candidates']; self.assertFalse(candidate['author_sequence_exact'])
registry_row = next(x for x in rows if x['external_id'] == candidate['registry_doi_candidate'])
self.assertEqual(registry_row['authors'], ['Xin Wang', 'Yu Wang', 'Yunchao Liu', 'Jens Meiler', 'Tyler Derr'])
# Every copied source and manifest remains byte-identical, including CRLF.
for evidence in report['evidence']:
body = (root/'out'/evidence['output_file']).read_bytes()
self.assertEqual(body, self.fixture['files'][evidence['snapshot']+'/'+evidence['file']].encode('utf-8'))
self.assertEqual(hashlib.sha256(body).hexdigest(), evidence['sha256'])
with self.assertRaisesRegex(ValueError, 'Output exists'):
c.collect(root/'out', snapshots=root/'src')
def test_all_program_events_match_independent_inventory(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root)
loaded = {name: c.load_snapshot(root, name) for name in c.SNAPSHOTS}
papers, _ = c.registry_papers(self.message)
observations = c.program_observations(loaded, papers)
gold = {(x['session_url'], x['title']): x for x in self.fixture['program_gold']}
self.assertEqual({(x['session_url'], x['title']) for x in observations}, set(gold))
for item in observations:
original = gold[(item['session_url'], item['title'])]
for key in ('authors_literal', 'time_literal', 'cancelled_presentation'):
self.assertEqual(item[key], original[key])
def test_all_public_abstracts_and_source_reported_names_match_independent_gold(self):
files = self.fixture['files']
manifest = json.loads(files['sdm2026-public-abstract-research-v1/manifest.json'])
receipts = {r['url']: r for r in manifest['records']}
self.assertEqual(len(self.fixture['page_gold']), 52)
for gold in self.fixture['page_gold']:
receipt = receipts[gold['program_url']]
raw = files['sdm2026-public-abstract-research-v1/'+receipt['file']].encode('utf-8')
self.assertEqual(hashlib.sha256(raw).hexdigest(), gold['page_sha256'])
fields = c.talk_fields(raw, receipt['session_url'])
self.assertEqual(fields['title'], gold['title'])
self.assertEqual(fields['source_reported_authors'], gold['authors'])
self.assertEqual(fields['abstract'], gold['abstract'])
self.assertTrue(all('@' not in name for name in fields['source_reported_authors']))
def test_wrong_isbn_parent_doi_and_incomplete_census_rejected(self):
for kind in ('isbn', 'parent', 'repeat', 'type', 'year', 'author', 'markup', 'abstract', 'missing'):
with self.subTest(kind=kind):
message = copy.deepcopy(self.message)
row = next(x for x in message['items'] if x['type'] == 'book-chapter')
if kind == 'isbn': row['ISBN'] = ['9780000000000']
if kind == 'parent': row['relation']['is-part-of'][0]['id'] = 'https://doi.org/10.9999/other'
if kind == 'repeat': row['DOI'] = c.BOOK
if kind == 'type': row['type'] = 'other'
if kind == 'year': row['published']['date-parts'] = [[2025, 10, 31]]
if kind == 'author': row['author'][0]['sequence'] = 'additional'
if kind == 'markup': row['title'] = ['Changed <sup>2</sup> title']
if kind == 'abstract': row['abstract'] = 'Unreviewed new version'
if kind == 'missing': message['items'].pop()
with self.assertRaises(ValueError): c.registry_papers(message)
def test_foreign_session_and_changed_edition_rejected(self):
raw = self.fixture['files']['sdm2026-public-abstract-research-v1/162766.html'].encode()
url = 'https://meetings.siam.org/sess/dsp_programsess.cfm?SESSIONCODE=91511'
for changed in [raw.replace(b'sessioncode=91511', b'sessioncode=91507'),
raw.replace(b'2026 SIAM International Conference', b'2025 SIAM International Conference')]:
with self.assertRaises(ValueError): c.talk_fields(changed, url)
def test_corruption_preserves_failure_without_success_report(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src')
(root/'src/sdm2026-public-abstract-research-v1/162766.html').write_bytes(b'corrupt')
with self.assertRaisesRegex(ValueError, 'receipt/body'):
c.collect(root/'out', snapshots=root/'src')
self.assertTrue((root/'out/failed.json').exists())
self.assertFalse((root/'out/collection-report.json').exists())
def test_source_symlink_rejected(self):
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp); self.snapshots(root/'src')
p = root/'src/sdm2026-crossref-research-v1/isbn-works.json'
original = p.read_bytes(); p.unlink(); (root/'data.json').write_bytes(original); p.symlink_to(root/'data.json')
with self.assertRaisesRegex(ValueError, 'Unsafe'):
c.collect(root/'out', snapshots=root/'src')
+148
View File
@@ -0,0 +1,148 @@
import gzip
import hashlib
import json
import os
from pathlib import Path
import tempfile
import unittest
from unittest.mock import patch
from paper_atlas.audits.source_identity import audit, classify, load_inputs
from paper_atlas.postgres.client import ReadConnection
def record(**updates):
row = dict(source='reviewed_program', external_id='2026/forum-a',
title='A sufficiently specific scientific paper title', authors=['Jane Doe', 'John Smith'],
venue='NeurIPS', year=2026, track='main', status='accepted', identifiers={},
source_record_hits=[], identifier_hits=[], title_candidate_ids=[])
row.update(updates)
return row
def write_input(root, rows):
path = root / 'source.jsonl'
raw = b''.join((json.dumps(row) + '\n').encode() for row in rows)
path.write_bytes(raw)
manifest = {k: rows[0][k] for k in ('source', 'venue', 'year', 'track')}
manifest.update(count=len(rows), file_sha256=hashlib.sha256(raw).hexdigest(),
file='/ignored/untrusted/location.jsonl', status='partial')
path.with_suffix('.manifest.json').write_text(json.dumps(manifest))
return path
class CandidateTests(unittest.TestCase):
def test_duplicate_names_do_not_equal_distinct_author_multiset(self):
row = record(authors=['Jane Doe', 'Jane Doe'], title_candidate_ids=[7])
result = classify(row, {7: {'title': row['title'], 'authors': ['Jane Doe']}})
self.assertEqual(result['kind'], 'partial_author_overlap_requires_review')
self.assertIn('duplicate_normalized_author_names', result['flags'])
self.assertFalse(result['comparisons'][0]['nonempty_author_multiset_equal'])
def test_arxiv_author_representation_and_order_match_but_never_approve(self):
row = record(title_candidate_ids=[7])
result = classify(row, {7: {'title': row['title'], 'authors': [['Smith', 'John', ''], ['Doe', 'Jane', '']]}})
self.assertEqual(result['kind'], 'one_exact_title_author_multiset_requires_review')
self.assertEqual(result['decision'], 'not_approved_for_import')
def test_binding_conflict_overrides_existing_source(self):
row = record(source_record_hits=[{'paper_id': 7}], identifier_hits=[{'paper_id': 8}])
entities = {pid: {'title': row['title'], 'authors': row['authors']} for pid in (7, 8)}
self.assertEqual(classify(row, entities)['kind'], 'conflicting_existing_bindings_hold')
def test_wrong_doi_binding_retains_author_title_disagreement(self):
row = record(identifier_hits=[{'paper_id': 7}])
result = classify(row, {7: {'title': 'Completely different work', 'authors': ['Somebody Else']}})
self.assertEqual(result['kind'], 'existing_identifier_binding_requires_review')
self.assertFalse(result['comparisons'][0]['normalized_title_equal'])
self.assertEqual(result['comparisons'][0]['author_name_overlap'], [])
def test_no_hit_is_not_novelty_and_missing_authors_never_agree(self):
self.assertEqual(classify(record(), {})['kind'], 'no_exact_match_not_proof_of_novelty')
row = record(authors=[], title_candidate_ids=[7])
result = classify(row, {7: {'title': row['title'], 'authors': []}})
self.assertEqual(result['kind'], 'no_author_agreement_hold')
def test_title_candidate_cap_is_explicit(self):
row = record(title_candidate_ids=list(range(21)))
entities = {pid: {'title': row['title'], 'authors': row['authors']} for pid in range(21)}
self.assertTrue(classify(row, entities)['title_candidates_truncated'])
self.assertEqual(classify(row, entities)['kind'], 'title_candidate_limit_reached_hold')
class InputTests(unittest.TestCase):
def setUp(self):
self.temp = tempfile.TemporaryDirectory()
self.addCleanup(self.temp.cleanup)
self.root = Path(self.temp.name)
def test_raw_line_hash_and_manifest_path_not_followed(self):
path = write_input(self.root, [record()])
rows, inputs = load_inputs([path])
self.assertEqual(rows[0]['raw_line_sha256'], hashlib.sha256(path.read_bytes()).hexdigest())
self.assertEqual(inputs[0]['count'], 1)
def test_hash_tamper_rejected_before_database_connection(self):
path = write_input(self.root, [record()])
path.write_bytes(path.read_bytes().replace(b'Jane Doe', b'Jane Roe'))
with patch('paper_atlas.audits.source_identity.ReadConnection') as connection:
with self.assertRaisesRegex(ValueError, 'hash mismatch'):
audit([path], self.root / 'out', pg_service='test', pg_schema='research')
connection.assert_not_called()
def test_manifest_scope_and_count_tamper_and_duplicate_sources(self):
path = write_input(self.root, [record()])
manifest_path = path.with_suffix('.manifest.json')
manifest = json.loads(manifest_path.read_text())
for key, value in [('year', 2025), ('count', 2)]:
altered = dict(manifest, **{key: value})
manifest_path.write_text(json.dumps(altered))
with self.assertRaises(ValueError):
load_inputs([path])
manifest_path.write_text(json.dumps(manifest))
with self.assertRaisesRegex(ValueError, 'Duplicate source'):
load_inputs([path, path])
def test_existing_output_never_replayed_and_failure_not_completion(self):
path = write_input(self.root, [record()])
out = self.root / 'out'
with patch('paper_atlas.audits.source_identity.ReadConnection', side_effect=RuntimeError('unavailable')):
with self.assertRaises(RuntimeError):
audit([path], out, pg_service='test', pg_schema='research')
self.assertTrue((out / 'failed.json').exists())
self.assertFalse((out / 'postgresql-preflight.json').exists())
with self.assertRaises(FileExistsError):
audit([path], out, pg_service='test', pg_schema='research')
@unittest.skipUnless(os.environ.get('PAPER_ATLAS_TEST_PG_SERVICE'), 'Requires the authorized remote PostgreSQL service')
class ReadOnlyIntegrationTests(unittest.TestCase):
def test_existing_binding_full_evidence_and_no_import_approval(self):
service = os.environ['PAPER_ATLAS_TEST_PG_SERVICE']
connection = ReadConnection(service, 'research')
try:
self.assertEqual(connection.execute('SHOW transaction_read_only').fetchone()[0], 'on')
hit = connection.execute("SELECT value,paper_id FROM identifiers WHERE scheme='arxiv' ORDER BY value LIMIT 1").fetchone()
self.assertIsNotNone(hit)
paper = connection.execute('SELECT title,authors_json FROM papers WHERE id=?', (hit['paper_id'],)).fetchone()
finally:
connection.close()
with tempfile.TemporaryDirectory() as tmp:
root = Path(tmp)
path = write_input(root, [record(title=paper['title'], authors=json.loads(paper['authors_json']),
identifiers={'arxiv': hit['value']})])
report = audit([path], root / 'out', pg_service=service, pg_schema='research')
self.assertTrue(report['transaction_read_only'])
self.assertFalse(report['database_written'])
self.assertEqual(report['approved_for_import'], 0)
row, = report['entries']
self.assertIn(hit['paper_id'], [r['paper_id'] for r in row['identifier_hits']])
shard = report['entities'][str(hit['paper_id'])]
raw = (root / 'out' / shard['file']).read_bytes()
self.assertEqual(hashlib.sha256(raw).hexdigest(), shard['sha256'])
packet = json.loads(gzip.decompress(raw))
self.assertEqual(packet['paper']['id'], hit['paper_id'])
self.assertEqual(packet['paper']['title'], paper['title'])
self.assertEqual(packet['paper']['authors_json'], paper['authors_json'])
self.assertIn('attempts', packet)
self.assertTrue((root / 'out/postgresql-preflight.json').is_file())