Compare commits

..

4 Commits

Author SHA1 Message Date
codex-bot c8db6118d4 Restore all MM 2026 program tracks with complete presentation evidence
Metadata integrity checks / metadata (push) Successful in 3m39s
2026-11-17 22:28:51 +08:00
Zayden-Jung 1c64971a52 Merge pull request '记录 RecSys 47 条关联验收与最新六个空白年度槽位' (#64) from codex/recsys-acceptance into main
Metadata integrity checks / metadata (push) Successful in 3m28s
Reviewed-on: #64
2026-11-17 14:22:45 +00:00
Zayden-Jung ac14df0625 Merge pull request '封装 ISMIR 2026 节目摘要与自愿公开评审来源' (#63) from codex/ismir2026 into main
Metadata integrity checks / metadata (push) Successful in 3m15s
Reviewed-on: #63
2026-11-17 14:22:40 +00:00
codex-bot 8c147f5e6e Collect frozen ISMIR 2026 program with abstracts and opt-in review provenance
Metadata integrity checks / metadata (push) Successful in 3m55s
2026-11-17 21:01:04 +08:00
11 changed files with 756 additions and 1 deletions
+35
View File
@@ -46,6 +46,26 @@ jobs:
subprocess.run([python,'-B','-m','pip','install','--no-deps',str(wheel)],check=True) subprocess.run([python,'-B','-m','pip','install','--no-deps',str(wheel)],check=True)
outside = root / 'outside-checkout' outside = root / 'outside-checkout'
outside.mkdir() outside.mkdir()
mm_fixture = json.loads(gzip.decompress((src/'tests/fixtures/mm2026-source.json.gz').read_bytes()))
mm_snapshots = root/'mm2026-snapshots'
for name, body in mm_fixture['files'].items():
path = mm_snapshots/name
path.parent.mkdir(parents=True,exist_ok=True)
path.write_bytes(body.encode('utf-8'))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','mm2026-program','--snapshots',str(mm_snapshots),'--out',str(root/'installed-mm2026')],cwd=outside,check=True)
mm_report = json.loads((root/'installed-mm2026/collection-report.json').read_text())
assert (mm_report['records'],mm_report['program_observations'],mm_report['extra_presentation_observations']) == (2143,4326,2183)
assert len(mm_report['files']) == 28 and mm_report['missing_abstract_count'] == 216
assert mm_report['session_count'] == 157 and not mm_report['database_written'] and mm_report['network_requests'] == 0 and mm_report['approved_for_import'] == 0
mm_rows = {r['paper_id']:r for f in mm_report['files'] for r in map(json.loads,pathlib.Path(f['file']).read_text().splitlines())}
import hashlib
assert set(mm_rows) == {g['paper_id'] for g in mm_fixture['independent_inventory']['inventory']}
for gold in mm_fixture['independent_inventory']['inventory']:
row = mm_rows[gold['paper_id']]
assert (row['title'],row['authors'],row['affiliations'],row['track']) == (gold['title'],gold['authors'],gold['affiliations'],gold['track_literal'])
assert hashlib.sha256(row['abstract'].encode()).hexdigest() == gold['abstract_sha256']
assert [c['source_pointer'] for c in row['contexts']] == gold['locations']
assert not row['approved_for_import'] and not row['final_decision_verified'] and 'emails' not in row
subprocess.run([str(venv/'bin/paper-atlas'),'--help'],cwd=outside,check=True) 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','cvf','--help'],cwd=outside,check=True)
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','rss2026-proceedings','--help'],cwd=outside,check=True) subprocess.run([python,'-I','-B','-m','paper_atlas','collect','rss2026-proceedings','--help'],cwd=outside,check=True)
@@ -166,6 +186,21 @@ jobs:
recsys_rows = [json.loads(line) for manifest in recsys_report['files'] for line in pathlib.Path(manifest['file']).read_text().splitlines()] 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() 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'] assert [{k:r['metadata']['program_fields'][k] for k in gold_keys} for r in recsys_rows] == recsys_fixture['independent_gold']
ismir_fixture = json.loads(gzip.decompress((src/'tests/fixtures/ismir2026-source.json.gz').read_bytes()))
ismir_snapshots = root/'ismir2026-snapshots'
ismir_snapshots.mkdir()
for name,encoded in ismir_fixture['files'].items():
target = ismir_snapshots/name
target.parent.mkdir(parents=True,exist_ok=True)
target.write_bytes(base64.b64decode(encoded))
subprocess.run([python,'-I','-B','-m','paper_atlas','collect','ismir2026-program','--snapshots',str(ismir_snapshots),'--out',str(root/'installed-ismir2026')],cwd=outside,check=True)
ismir_report = json.loads((root/'installed-ismir2026/collection-report.json').read_text())
assert (ismir_report['records'],ismir_report['missing_abstract_count'],ismir_report['review_optin_yes'],ismir_report['review_optin_no']) == (140,0,101,39)
assert ismir_report['literal_title_version_differences'] == 2 and ismir_report['author_sequence_version_differences'] == 3
assert not ismir_report['database_written'] and ismir_report['network_requests'] == 0 and ismir_report['approved_for_import'] == 0
ismir_rows = [json.loads(line) for line in pathlib.Path(ismir_report['files'][0]['file']).read_text().splitlines()]
assert [(r['external_id'],r['title'],r['authors'],r['abstract']) for r in ismir_rows] == [('2026/'+g['program_id'],g['title'],g['authors'],g['abstract']) for g in ismir_fixture['independent_gold']]
assert all(not r['identifiers'] and not r['submission_identifiers'] and r['metadata']['paper_type'] is None for r in ismir_rows)
corl_fixture = json.loads(gzip.decompress((src/'tests/fixtures/corl2026-source.json.gz').read_bytes())) corl_fixture = json.loads(gzip.decompress((src/'tests/fixtures/corl2026-source.json.gz').read_bytes()))
corl_official = root/'corl2026-official' corl_official = root/'corl2026-official'
corl_archive = root/'corl2026-archive' corl_archive = root/'corl2026-archive'
+4
View File
@@ -131,6 +131,8 @@ ICAPS 2026 也提供独立的 `collect icaps2026-registry`,保留 Crossref 卷
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。 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 ismir2026-program` 核验 140 项 ISMIR 官方录用节目及原始摘要,保留两处标题、三处作者名单差异和 101 组自愿公开评审文本;不推断原始 OpenReview ID 或拒稿。见 [ISMIR 2026 使用与恢复说明](docs/ismir2026-program.md)。
`collect icassp2026-registry` 从冻结的 10 页 Crossref 元数据核验 4,582 条论文,明确排除 7 条非论文及 56 条其他会议记录;保留摘要缺失和单字段作者姓名,不自动融合节目表。见 [ICASSP 2026 采集与恢复说明](docs/icassp2026-registry.md)。 `collect icassp2026-registry` 从冻结的 10 页 Crossref 元数据核验 4,582 条论文,明确排除 7 条非论文及 56 条其他会议记录;保留摘要缺失和单字段作者姓名,不自动融合节目表。见 [ICASSP 2026 采集与恢复说明](docs/icassp2026-registry.md)。
`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 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)。
@@ -138,3 +140,5 @@ ECCV 2026 的[暂定节目采集器](docs/eccv2026-program.md)合并重复节目
`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 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)。 `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)。
`collect mm2026-program` 从已冻结官方快照恢复全部 28 个来源 track 的 2,143 个节目 ID,保留 4,326 次展示的来源位置与场次上下文,单列 216 个摘要缺失项;输出供贡献类型、出版及作品身份核查使用,不是可直接导入的论文/决定队列。见 [MM 2026 使用、来源和恢复说明](docs/mm2026-source-status.md)。
+50
View File
@@ -0,0 +1,50 @@
# ISMIR 2026 官方节目与摘要采集
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,不联网、不执行网页脚本、不下载论文 PDF、不连接数据库。输出仍须经过作品身份与投稿历史审查。
## 使用
```sh
paper-atlas collect ismir2026-program \
--snapshots /data/paper-index/collected \
--out /data/paper-index/collected/ismir2026-normalized-<new-run>
```
输入父目录需要三个冻结目录:
- `remaining2026-listings-research-v2`:清单、`ismir-accepted.html`、`ismir-call.html`。
- `ismir2026-script-research-v1`:清单及 `papers.js`,用于证明官方程序明确加载的数据文件和 poster 链接模板;只读取脚本原文。
- `ismir2026-json-research-v1`:清单及 `papers.json`,提供 140 项完整元数据和公开摘要。
固定清单和全部输入原始字节均校验 SHA256、HTTP 回执、官方 URL 及带时区的采集时间。已有输出目录不能覆盖;先检查原任务是否已结束,再选择新目录。失败保留 `failed.json`,只有全批成功才发布 `collection-report.json`。
## 来源与版本
[正式录用名单](https://ismir2026.ismir.net/accepted-papers)和[官方虚拟平台](https://ismir2026program.ismir.net/papers.html)各有 140 项,经来源内标题对应核查一致;其中 2 处标题有直引号/弯引号文字差异,3 处作者名单有差异,两个来源的原始写法分别保留并标记,不静默修正或拼接作者。
作者差异涉及节目 ID `60`、`126`、`433`,包含大小写和姓名完整度问题。正式名单的作者机构可能含括号和分号,采集器只在括号外拆分作者;作者原始字符串也保留。来源内名单对应不代表数据库作品身份已经获准关联。
- 原始 JSON 摘要全文保留,包括数学符号、换行和 Unicode,不作扁平化提取。
- 数字 `id`、`forum` 是 miniconf 节目定位符,不视为原始 OpenReview 投稿 ID;不推断 DOI 或 arXiv ID,投稿标识列表保持空。
- `accepted` 来自正式录用名单与节目平台,不是原始签名决定。`track=program_unclassified`,不从主题、session 或远程展示方式推断研究方法、workshop 或论文类型。
- `ismir_program:keyword` 保存来源非空关键词,规范空白并去除同条记录内完全重复值;原始关键词列表(含空项)保留在元数据。
- `ismir_program:presentation` 保存来源展示方式。Remote/In-person 是展示方式,不是研究类型。
- poster URL 严格依据冻结的官方 JS 模板生成;本命令不请求 poster 页面。
## 公开评审与拒稿边界
原始来源中有 **101 组明确选择公开的评审材料**;其余 39 项 `publish_reviews=No`,评审正文为空。采集器验证此偏好和正文的一致性,原文保存在来源元数据,不把评审意见或 meta-review 中的推荐转换成额外的最终决定。
这些记录来自录用节目,不能据此当作 101 条公开拒稿记录。公开拒稿导出是否可得仍记为 unknown,需要单独核查。[征稿与评审流程](https://ismir2026.ismir.net/authors/call-for-papers)中的评审规则也不等于具体投稿决定。
输出包含一份 140 条 JSONL、部分覆盖清单、七个原始来源文件及报告,批准导入数为零。采集成功不等于会议论文/track/公开负面决定覆盖完整,也不替代最后的全库去重审计。
## 恢复与验证
`papers.json` SHA256:`864d384cbc928554504e1b39849137bb380143e247a48f2d71563a698cb8e62d`。
原始 JSON 和 JS 恢复包:服务器 `/data/paper-index/validation/ismir2026-json-research-recovery-v1.tar.gz`,本地 `/Users/zzyang/Documents/Codex/2026-11-06/paper-index/` 同名文件;SHA256 为 `bba7b236cfe4d8b259d6c1230b3fe09a3e9071f5203884c92b016ca0718fb095`。
正式名单与政策恢复包:同目录 `remaining2026-listings-research-v2-recovery.tar.gz`,SHA256 为 `f5ef7c2e73d46712146e4d6711ad5ad562e16b8b55c0cd75d1041cd0b81142cb`。安全恢复三个源目录、核验文件后可重跑上述命令。
完整来源和独立 140 项清单在测试夹具中冻结。测试、构建、打包、安装及实际隔离目录 CLI 验证只在用户指定 `ces-control:host` runner 执行。生产规范化后还需独立核验全部记录,并用 PostgreSQL 只读预检检查候选作品和既往投稿;不自动合并来源差异或人物身份。
+17 -1
View File
@@ -73,4 +73,20 @@
服务器恢复包:`/data/paper-index/validation/mm2026-source-discovery-and-inventory-recovery-v1.tar.gz`;本地备份:`/Users/zzyang/Documents/Codex/2026-11-06/paper-index/` 同名文件。先安全解包并核验回执,再做规范化,不能重新抓取 403 入口作为恢复方式。 服务器恢复包:`/data/paper-index/validation/mm2026-source-discovery-and-inventory-recovery-v1.tar.gz`;本地备份:`/Users/zzyang/Documents/Codex/2026-11-06/paper-index/` 同名文件。先安全解包并核验回执,再做规范化,不能重新抓取 403 入口作为恢复方式。
本批尚未形成生产导入队列。下一步需要可复用采集器保留每个节目 ID 的全部展示上下文、明确贡献类型与来源边界,单列缺失字段,再做作品/出版身份及既往投稿审查。已消费的其他会议导入队列不可重放;最终全库去重和完整审计仍为独立必做阶段。 ## 可复用采集命令
在已配置的服务器环境中运行,`--snapshots` 是包含 `mm2026-program-and-public-probes-v1/` 的父目录;恢复时先安全解包上述已核验恢复包。命令只读快照,不请求网络、不连接数据库,不重新访问403入口,也不下载PDF:
```sh
paper-atlas collect mm2026-program \
--snapshots /data/paper-index/collected \
--out /data/paper-index/collected/mm2026-native-v1
```
输出目录必须不存在;已存在、失败或部分输出均不能覆盖重跑,应先查看报告,另选新的目录。`collection-report.json` 仅在全量验证成功后生成;失败保留 `failed.json`,不冒充完成。
采集器核验固定回执、正文哈希、URL、抓取时间、完整逐轨数量和字段缺失盘点。28个逐轨 JSONL 保留原始标题、作者、机构、逐字摘要以及同一节目ID的全部展示位置、时间、poster/oral号和展示kind;157个场次及子场次/非论文活动另存 `sessions.jsonl`。外层空track不覆盖 `paper.track`,不同核心版本、未知字段或未处理嵌套需要重新审查,不能静默丢弃或取第一条。公开联系邮箱数组只留在原始证据中,不进入规范记录。`evidence/` 保留原始节目JSON、完整回执及四个既有403响应。
这些文件的格式为 `mm_program_observation_v1_not_ingestion_record`,所有项 `approved_for_import=false`。节目ID仅为 `mm_program:program_id` 来源定位符,track标签使用 `mm_program:track`;不建立科学作品、原始投稿ID或最终决定,不推断 DOI/arXiv/OpenReview 标识、论文类型、拒稿或撤稿。行业报告、艺术、workshop及其他来源贡献均保留,尚需逐轨核对贡献类型与正式出版依据。
本批尚未形成生产导入队列。后续做贡献类型、作品/出版身份及既往投稿审查,再生成经批准的增量队列。已消费的其他会议导入队列不可重放;最终全库去重和完整审计仍为独立必做阶段。
+2
View File
@@ -23,6 +23,7 @@ COMMANDS = {
('collect','colm'): ('paper_atlas.collectors.proceedings', {'source':'colm'}, 'COLM 官方论文集'), ('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','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','mlsys'): ('paper_atlas.collectors.proceedings', {'source':'mlsys'}, 'MLSys 官方历年标题与摘要'),
('collect','ismir2026-program'): ('paper_atlas.collectors.ismir2026', {}, '核验 ISMIR 2026 公开节目与录用名单并保留原始摘要、来源差异和自愿公开评审'),
('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','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','eccv2026-program'): ('paper_atlas.collectors.eccv2026', {}, '核验ECCV2026暂定节目并断点采集公开元数据'),
('collect','icassp2026-registry'): ('paper_atlas.collectors.icassp2026', {}, '核验 ICASSP 2026 冻结注册论文元数据'), ('collect','icassp2026-registry'): ('paper_atlas.collectors.icassp2026', {}, '核验 ICASSP 2026 冻结注册论文元数据'),
@@ -31,6 +32,7 @@ COMMANDS = {
('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','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','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','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','mm2026-program'): ('paper_atlas.collectors.mm2026', {}, '恢复 MM 2026 全部来源 track 与重复展示上下文,待贡献和身份核查'),
('collect','cvf'): ('paper_atlas.collectors.vision', {'source':'cvf'}, 'CVPR / ICCV 官方开放目录'), ('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','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','aamas-proceedings'): ('paper_atlas.collectors.aamas', {}, '按官方目录区分 AAMAS 发表形式和摘要'),
+208
View File
@@ -0,0 +1,208 @@
"""Frozen ISMIR 2026 accepted program, abstracts and opt-in review provenance."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
import re
import unicodedata
from paper_atlas.io.html import Document, Node, clean
SOURCE = 'ismir_program'
TRACK = 'program_unclassified'
PAPERS = 'https://ismir2026program.ismir.net/papers.html'
ACCEPTED = 'https://ismir2026.ismir.net/accepted-papers'
SNAPSHOTS = {
'remaining2026-listings-research-v2': ('fea7b710587636db06794ee195cd083634c41e9054a217859d9dd0ccdd1afab9',
'research_completed_not_coverage_or_import_approval', {
'ismir-accepted.html': ACCEPTED,
'ismir-call.html': 'https://ismir2026.ismir.net/authors/call-for-papers'}),
'ismir2026-script-research-v1': ('5885b4bcfc2fc65374e5dc7a0d74f4eb86a0a36b8689f9fe8edbd10d49a08794',
'public_script_research_only', {'papers.js': 'https://ismir2026program.ismir.net/static/js/papers.js'}),
'ismir2026-json-research-v1': ('8be3e23277c199720264d940917118572655d18c0f9ac6db1b9cd30a6561b34d',
'public_metadata_json_research_only', {'papers.json': 'https://ismir2026program.ismir.net/papers.json'}),
}
REVIEW_KEYS = {'meta_review','publish_reviews','review1','review2','review3','review4'}
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 title_key(text):
return ''.join(c for c in unicodedata.normalize('NFKD', text).casefold() if c.isalnum())
def load_snapshots(root):
root = Path(root)
if root.is_symlink():
raise ValueError('Unsafe snapshot root')
result = {}
for name, (expected, status, urls) in SNAPSHOTS.items():
directory = root/name; path = directory/'manifest.json'
if 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) != expected:
raise ValueError('Reviewed manifest hash changed')
manifest = json.loads(raw)
if manifest['status'] != status or manifest['database_written'] is not False:
raise ValueError('Unreviewed snapshot scope')
files = {'manifest.json': raw}; receipts = {}
for filename, url in urls.items():
receipt = one([r for r in manifest['entries'] if r.get('file') == filename])
path = directory/filename
if path.is_symlink() or path.stat().st_size > 8*1024**2:
raise ValueError('Unsafe snapshot source file')
body = path.read_bytes()
if (receipt['url'] != url or receipt['final_url'] != url or receipt['http_status'] != 200
or receipt['bytes'] != len(body) or receipt['sha256'] != digest(body)
or datetime.fromisoformat(receipt['at']).utcoffset() is None):
raise ValueError('Source receipt/body disagree')
files[filename] = body; receipts[filename] = receipt
result[name] = (files, receipts)
return result
def author_names(literal):
# Institution names can contain semicolons; split only outside parentheses.
parts = []; depth = 0; start = 0
for index, char in enumerate(literal):
if char == '(': depth += 1
elif char == ')':
depth -= 1
if depth < 0: raise ValueError('Unbalanced author affiliation')
elif char == ';' and depth == 0:
parts.append(literal[start:index]); start = index+1
if depth:
raise ValueError('Unbalanced author affiliation')
parts.append(literal[start:])
names = [clean(p.split('(', 1)[0]) for p in parts]
if any(not name for name in names):
raise ValueError('Missing accepted-list author')
return names
def accepted_listing(raw):
doc = Document(raw.decode('utf-8')); tables = []
for table in doc.root.walk():
if table.tag != 'table': continue
rows = []
for row in table.walk():
if row.tag == 'tr':
rows.append([clean(n.text()) for n in row.children if isinstance(n, Node) and n.tag in ('th','td')])
if rows and rows[0] == ['Paper Title', 'Authors']:
tables.append(rows)
rows = one(tables)[1:]; result = {}
if len(rows) != 140:
raise ValueError('Incomplete official accepted-list census')
for row in rows:
if len(row) != 2 or not row[0] or not row[1] or title_key(row[0]) in result:
raise ValueError('Missing or ambiguous accepted-list title')
result[title_key(row[0])] = {'title': row[0], 'authors_literal': row[1], 'author_names': author_names(row[1])}
return result
def program_observations(rows, listing):
if not isinstance(rows, list) or len(rows) != 140:
raise ValueError('Incomplete public program census')
observations = []; seen = set(); used = set()
for original in rows:
if not isinstance(original, dict) or not isinstance(original.get('content'), dict):
raise ValueError('Invalid public program record')
pid = original.get('id'); content = original.get('content', {})
if not isinstance(pid, str) or not re.fullmatch(r'[1-9][0-9]*', pid) or pid in seen or original.get('forum') != pid:
raise ValueError('Missing/repeated numeric program locator, not an OpenReview ID')
seen.add(pid); title = content.get('title'); authors = content.get('authors'); abstract = content.get('abstract')
if (not isinstance(title, str) or not clean(title) or not isinstance(abstract, str) or not abstract.strip()
or not isinstance(authors, list) or not authors or any(not isinstance(a, str) or not clean(a) for a in authors)):
raise ValueError('Missing program title, author sequence or abstract')
key = title_key(title)
if key not in listing or key in used:
raise ValueError('Program/accepted-list title membership mismatch')
used.add(key); accepted = listing[key]
keywords = content.get('keywords'); presentation = content.get('paper_presentation')
session = content.get('session'); reviews = original.get('reviews')
if (not isinstance(keywords, list) or any(not isinstance(k, str) for k in keywords)
or presentation not in ('Remote','In-person','Remote during oral, In-person during nominee')
or not isinstance(session, list) or len(session) != 1 or not isinstance(session[0], str) or session[0] != original.get('session')
or not re.fullmatch(r'(?:[1-9]|1[0-4])', session[0])):
raise ValueError('Unreviewed source keywords, presentation or session')
if (not isinstance(reviews, dict) or set(reviews) != REVIEW_KEYS
or any(not isinstance(v, str) for v in reviews.values())
or reviews['publish_reviews'] not in ('Yes','No')):
raise ValueError('Unreviewed opt-in review structure')
review_text = [reviews[k] for k in REVIEW_KEYS-{'publish_reviews'}]
if (reviews['publish_reviews'] == 'No' and any(review_text)) or (reviews['publish_reviews'] == 'Yes' and not any(review_text)):
raise ValueError('Review publication preference/body disagree')
flags = []
if title != accepted['title']: flags.append('program_and_accepted_title_literal_versions_differ')
if authors != accepted['author_names']: flags.append('program_and_accepted_author_sequence_versions_differ')
observations.append({'original': original, 'accepted_listing': accepted, 'review_flags': flags})
if (used != listing.keys() or sum('program_and_accepted_title_literal_versions_differ' in o['review_flags'] for o in observations) != 2
or {o['original']['id'] for o in observations if 'program_and_accepted_author_sequence_versions_differ' in o['review_flags']} != {'60','126','433'}
or Counter(o['original']['reviews']['publish_reviews'] for o in observations) != {'Yes':101,'No':39}):
raise ValueError('Unreviewed source-version or opt-in census')
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:
sources = load_snapshots(snapshots)
official, _ = sources['remaining2026-listings-research-v2']; script, _ = sources['ismir2026-script-research-v1']
program, receipts = sources['ismir2026-json-research-v1']; receipt = receipts['papers.json']
if b"d3.json('papers.json')" not in script['papers.js'] or b'href="poster_${openreview.id}.html"' not in script['papers.js']:
raise ValueError('Unreviewed public program data path or poster link template')
observations = program_observations(json.loads(program['papers.json']), accepted_listing(official['ismir-accepted.html']))
evidence = out/'evidence'; evidence.mkdir()
for name, (files, _) in sources.items():
directory = evidence/name; directory.mkdir()
for filename, body in files.items(): (directory/filename).write_bytes(body)
rows = []
for observation in observations:
original = observation['original']; content = original['content']; pid = original['id']
tags = [{'namespace':SOURCE+':presentation','value':content['paper_presentation']}]
tags += [{'namespace':SOURCE+':keyword','value':value} for value in dict.fromkeys(clean(k) for k in content['keywords'] if clean(k))]
rows.append({'source':SOURCE,'external_id':'2026/'+pid,'url':'https://ismir2026program.ismir.net/poster_'+pid+'.html',
'title':content['title'],'authors':content['authors'],'abstract':content['abstract'],
'identifiers':{},'submission_identifiers':[],'venue':'ISMIR','year':2026,'track':TRACK,'status':'accepted',
'raw_decision':'Present in official ISMIR2026 accepted list and public virtual program; not an original signed decision.',
'evidence_url':ACCEPTED,'observed_at':receipt['at'],'tags':tags,
'metadata':{'program_entry':original,'accepted_listing':observation['accepted_listing'],'review_flags':observation['review_flags'],
'program_receipt':receipt,'source_evidence':{'file':'evidence/ismir2026-json-research-v1/papers.json','sha256':receipt['sha256'],'url':receipt['url']},
'original_decision_signature_verified':False,'paper_type':None,
'id_scope':'Numeric miniconf id/forum fields are program locators, not original OpenReview submission IDs; no inferred DOI or arxiv ID.',
'track_scope':'Unclassified regular accepted-list program; no workshop or research-method class inferred from topic/session/remote presentation.',
'review_scope':'101 source-opt-in review sets retained as raw public text;39No contain no review text. Reviewer recommendations/meta-review prose are not separately inferred final negative decisions.',
'abstract_transform':'Original JSON abstract string retained verbatim, including math, newlines and Unicode; no markup flattening or PDF extraction.',
'program_listing_scope':'Source versions remain separate.2title literal differences and3author-list differences flagged, not silently fused or resolved.',
'scope':'Pending work/history review. Poster URL follows frozen official JS template; poster pages not downloaded. No work merge, aliases, negative decision or final coverage/dedup claim.'}})
path = out/'ismir-program-2026-unclassified.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':'ISMIR','year':2026,'track':TRACK,'count':140,'file':str(path),
'file_sha256':digest(raw),'list_url':ACCEPTED,'status':'partial','missing_abstract_count':0,'source_expected_count':140}
path.with_suffix('.manifest.json').write_text(json.dumps(manifest,indent=2)+'\n')
report={'status':'collected_pending_identity_review','records':140,'files':[manifest],'missing_abstract_count':0,
'literal_title_version_differences':2,'author_sequence_version_differences':3,'review_optin_yes':101,'review_optin_no':39,
'database_written':False,'network_requests':0,'pdfs_downloaded':False,'approved_for_import':0,
'public_rejected_availability':'unknown_accepted_program_reviews_are_not_a_rejected_export',
'scope':'140accepted observations;101opt-in review sets, not101rejected records. Source keyword/presentation tags are namespace-scoped; no original OpenReview ID assumption or source-version fusion.'}
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))
+202
View File
@@ -0,0 +1,202 @@
"""Restore frozen ACM MM 2026 program observations, not publication decisions."""
import argparse
from collections import Counter
from datetime import datetime
import hashlib
import json
from pathlib import Path
import re
SOURCE = 'mm_program'
SNAPSHOT = 'mm2026-program-and-public-probes-v1'
URL = 'https://2026.acmmm.org/site/program-data.json'
BODY_SHA = '08e2976ecba2d0d535000b0935f41c85076ec674e518a1df2c5843d9bde4d083'
MANIFEST_SHA = 'e43ac6cdfc86bac61305553a044dd6148800b5aa41a5d1bf06c486e90d0ed7b8'
TRACK_COUNTS = {'MRAC':35,'SUMAC':9,'NeuroMM':10,'IMPACT':8,'RichMediaGAI':10,
'MMSports':16,'MUWS':5,'EReL-MIR':7,'GreenMM':10,'DFF':12,'Main Track':1703,
'Grand Challenge':97,'Doctoral Symposium':4,'Datasets':50,'Interactive Art':23,
'Reproducibility':1,'Demo and Video':27,'Open Source':14,'Industry Talk':7,
'BNI':25,'QoEVMA':4,'HUMA':14,'WIPAI':8,'AMI':11,'Mulsemedia':9,
'UAVM':11,'MCHM':7,'JAV-CG':6}
CORE = ('paper_id','title','authors','affiliations','abstract','track')
PRESENTATION = ('kind','start_time','end_time','poster_id','oral_id')
def digest(raw):
return hashlib.sha256(raw).hexdigest()
def safe_read(path, limit):
if any(p.is_symlink() for p in (path, *path.parents)) or not path.is_file() or path.stat().st_size > limit:
raise ValueError('Unsafe snapshot path or size')
return path.read_bytes()
def load_snapshot(root):
directory = Path(root)/SNAPSHOT
raw = safe_read(directory/'manifest.json', 128*1024)
if digest(raw) != MANIFEST_SHA:
raise ValueError('Reviewed manifest hash changed')
manifest = json.loads(raw)
if manifest.get('database_written') is not False or len(manifest['entries']) != 5:
raise ValueError('Unreviewed manifest scope')
evidence = {'manifest.json':raw}
for receipt in manifest['entries']:
name = receipt['file']
if Path(name).name != name or name in evidence:
raise ValueError('Unsafe or duplicate receipt name')
body = safe_read(directory/name, 12*1024**2)
if (receipt['bytes'] != len(body) or receipt['sha256'] != digest(body)
or receipt['url'] != receipt['final_url']
or not receipt['content_type'].startswith('application/json')
or datetime.fromisoformat(receipt['at']).utcoffset() is None):
raise ValueError('Snapshot receipt/body disagree')
if name == 'program-data.json':
if receipt['url'] != URL or receipt['http_status'] != 200 or digest(body) != BODY_SHA:
raise ValueError('Wrong program source')
elif receipt['http_status'] != 403:
raise ValueError('Unreviewed public-access result')
evidence[name] = body
if 'program-data.json' not in evidence:
raise ValueError('Missing program source')
return evidence
def without_emails(value):
"""Keep public session structure; contact arrays remain in raw evidence only."""
if isinstance(value, dict):
return {k:without_emails(v) for k,v in value.items() if k != 'emails'}
if isinstance(value, list):
return [without_emails(v) for v in value]
return value
def program_locations(node, pointer=''):
"""Independent structural guard against silently skipping a new nesting form."""
if isinstance(node, dict):
if 'paper_id' in node:
yield pointer
for key,value in node.items():
escaped = key.replace('~','~0').replace('/','~1')
yield from program_locations(value, pointer+'/'+escaped)
elif isinstance(node, list):
for i,value in enumerate(node):
yield from program_locations(value, pointer+'/'+str(i))
def parse_program(raw):
data = json.loads(raw)
if (data.get('generated_at') != '2026-11-14T18:36:26.860677+00:00'
or data.get('days') != [1,2,3,4,5] or len(data.get('rooms',[])) != 20
or len(data.get('entries',[])) != 157):
raise ValueError('Unreviewed program edition or census')
records = {}; sessions = []; seen = []
def observe(paper, pointer, session_pointer, subsession_pointer=None):
if not set(CORE) <= paper.keys() or set(paper)-set(CORE)-set(PRESENTATION)-{'emails'}:
raise ValueError('Unreviewed program fields')
for key in ('paper_id','title','abstract','track'):
if not isinstance(paper[key],str):
raise ValueError('Wrong program field type')
for key in ('authors','affiliations'):
if not isinstance(paper[key],list) or any(not isinstance(v,str) for v in paper[key]):
raise ValueError('Wrong author or affiliation list')
if not paper['paper_id'].strip() or not paper['title'].strip() or paper['track'] not in TRACK_COUNTS:
raise ValueError('Missing identity/title or unreviewed source track')
core = {k:paper[k] for k in CORE}
pid = paper['paper_id']; seen.append(pointer)
if pid not in records:
missing = [k for k in ('authors','abstract') if not paper[k]]
records[pid] = {**core,'source':SOURCE,'year':2026,'venue':'MM',
'source_locator':{'namespace':'mm_program:program_id','value':pid},
'observed_status':'listed_in_official_program',
'publication_identity_verified':False,'final_decision_verified':False,
'approved_for_import':False,'missing_fields':missing,
'tags':[{'namespace':'mm_program:track','value':paper['track']}],
'contexts':[]}
elif {k:records[pid][k] for k in CORE} != core:
raise ValueError('Conflicting core versions for one program ID; review required')
records[pid]['contexts'].append({'source_pointer':pointer,'session_pointer':session_pointer,
'subsession_pointer':subsession_pointer,
'presentation':{k:paper[k] for k in PRESENTATION if k in paper}})
for i,entry in enumerate(data['entries']):
pointer = '/entries/'+str(i)
if entry['track'] != '':
raise ValueError('Unreviewed outer track; do not override paper.track')
session = without_emails({k:v for k,v in entry.items() if k not in ('papers','subsessions')})
session.update(source_pointer=pointer,subsessions=[])
for j,paper in enumerate(entry.get('papers') or []):
observe(paper,pointer+'/papers/'+str(j),pointer)
for j,sub in enumerate(entry.get('subsessions') or []):
sp = pointer+'/subsessions/'+str(j)
saved = without_emails({k:v for k,v in sub.items() if k != 'items'})
saved.update(source_pointer=sp,nonpaper_items=[])
for k,item in enumerate(sub.get('items') or []):
ip = sp+'/items/'+str(k)
if 'paper_id' in item:
observe(item,ip,pointer,sp)
else:
saved['nonpaper_items'].append({'source_pointer':ip,'fields':without_emails(item)})
session['subsessions'].append(saved)
sessions.append(session)
if set(seen) != set(program_locations(data)) or len(set(seen)) != len(seen):
raise ValueError('Unhandled or duplicated program location')
items = list(records.values())
if (len(seen) != 4326 or len(items) != 2143
or Counter(r['track'] for r in items) != TRACK_COUNTS
or sum(not r['abstract'] for r in items) != 216
or {r['paper_id'] for r in items if not r['authors']} != {'mfp4971','mfp7801'}):
raise ValueError('Program count or missing-field census changed')
return items,sessions
def collect(out, *, snapshots):
out = Path(out)
if out.exists() or out.is_symlink():
raise ValueError('Output exists; use a fresh directory')
if any(p.is_symlink() for p in out.parents):
raise ValueError('Unsafe output parent')
out.mkdir(parents=True)
try:
evidence = load_snapshot(snapshots)
rows,sessions = parse_program(evidence['program-data.json'])
(out/'evidence').mkdir()
for name,body in evidence.items():
(out/'evidence'/name).write_bytes(body)
outputs = []
for track,count in TRACK_COUNTS.items():
slug = re.sub('[^a-z0-9]+','_',track.lower()).strip('_')
path = out/('mm-program-2026-'+slug+'.jsonl')
raw = ''.join(json.dumps(r,ensure_ascii=False)+'\n' for r in rows if r['track']==track).encode()
path.write_bytes(raw)
outputs.append({'file':str(path),'file_sha256':digest(raw),'count':count,
'track_literal':track,'format':'mm_program_observation_v1_not_ingestion_record',
'status':'pending_contribution_and_identity_review','approved_for_import':0})
session_raw = ''.join(json.dumps(r,ensure_ascii=False)+'\n' for r in sessions).encode()
(out/'sessions.jsonl').write_bytes(session_raw)
report = {'status':'collected_pending_contribution_and_identity_review','source':SOURCE,
'records':2143,'program_observations':4326,'extra_presentation_observations':2183,
'track_counts':TRACK_COUNTS,'files':outputs,'session_count':157,
'sessions_file_sha256':digest(session_raw),'missing_abstract_count':216,
'missing_author_program_ids':['mfp4971','mfp7801'],
'source_body_sha256':BODY_SHA,'source_manifest_sha256':MANIFEST_SHA,
'database_written':False,'pdfs_downloaded':False,'network_requests':0,'approved_for_import':0,
'public_negative_access':{'anonymous_notes_probes':4,'http_statuses':[403]*4,
'negative_decisions_inferred':0,'scope':'Observed endpoints only; not proof of zero public negative decisions.'},
'scope':'Program IDs are source locators, not scientific-work or original submission IDs. All28source tracks retained; presentation times/kinds/locators and session context kept separately. No publication status, paper type, DOI, arxiv/OpenReview ID, rejection or withdrawal inferred. Blank fields are not negative decisions. Contribution/publication/work/history review and final whole-database audit remain.'}
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))
Binary file not shown.
Binary file not shown.
+92
View File
@@ -0,0 +1,92 @@
import base64
import copy
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import ismir2026 as c
class ISMIR2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture=json.loads(gzip.decompress((Path(__file__).parent/'fixtures/ismir2026-source.json.gz').read_bytes()))
cls.files={name:base64.b64decode(raw) for name,raw in cls.fixture['files'].items()}
cls.program=json.loads(cls.files['ismir2026-json-research-v1/papers.json'])
cls.accepted=cls.files['remaining2026-listings-research-v2/ismir-accepted.html']
def materialize(self,root):
for name,raw in self.files.items():
p=root/name;p.parent.mkdir(parents=True,exist_ok=True);p.write_bytes(raw)
def test140_source_versions_and_review_optins_against_independent_gold(self):
listing=c.accepted_listing(self.accepted);observations=c.program_observations(self.program,listing)
self.assertEqual(len(observations),140)
for observation,gold in zip(observations,self.fixture['independent_gold']):
r=observation['original'];v=r['content'];official=observation['accepted_listing']
self.assertEqual((r['id'],r['forum'],v['title'],v['authors'],v['abstract']),(gold['program_id'],gold['forum_literal'],gold['title'],gold['authors'],gold['abstract']))
self.assertEqual(official,{'title':gold['official_title'],'authors_literal':gold['official_authors_literal'],'author_names':gold['official_author_names']})
self.assertEqual(v['session'],gold['session_literal']);self.assertEqual(v['paper_presentation'],gold['presentation_literal']);self.assertEqual(v['keywords'],gold['keywords'])
self.assertEqual(sum('program_and_accepted_title_literal_versions_differ' in r['review_flags'] for r in observations),2)
self.assertEqual({r['original']['id'] for r in observations if 'program_and_accepted_author_sequence_versions_differ' in r['review_flags']},{'60','126','433'})
self.assertEqual(sum(r['original']['reviews']['publish_reviews']=='Yes' for r in observations),101)
def test_full_collection_keeps_abstract_bytes_review_text_tags_and_no_inferred_ids(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['missing_abstract_count'],report['review_optin_yes'],report['review_optin_no']),(140,0,101,39))
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['public_rejected_availability'],'unknown_accepted_program_reviews_are_not_a_rejected_export')
manifest,=report['files'];raw=Path(manifest['file']).read_bytes();self.assertEqual(hashlib.sha256(raw).hexdigest(),manifest['file_sha256']);self.assertEqual(manifest['status'],'partial')
rows=[json.loads(l) for l in raw.splitlines()];self.assertEqual(len(rows),140)
for row,original,gold in zip(rows,self.program,self.fixture['independent_gold']):
self.assertEqual(row['metadata']['program_entry'],original);self.assertEqual(row['abstract'],gold['abstract']);self.assertEqual(row['title'],gold['title']);self.assertEqual(row['authors'],gold['authors'])
self.assertEqual(row['external_id'],'2026/'+gold['program_id']);self.assertEqual(row['identifiers'],{});self.assertEqual(row['submission_identifiers'],[])
self.assertEqual(row['status'],'accepted');self.assertEqual(row['track'],'program_unclassified');self.assertIsNone(row['metadata']['paper_type']);self.assertFalse(row['metadata']['original_decision_signature_verified'])
tags=[{'namespace':'ismir_program:presentation','value':gold['presentation_literal']}]+[{'namespace':'ismir_program:keyword','value':v} for v in dict.fromkeys(' '.join(k.split()) for k in gold['keywords'] if k.strip())]
self.assertEqual(row['tags'],tags);self.assertTrue(all(t['value'] for t in row['tags']))
if original['reviews']['publish_reviews']=='No':self.assertFalse(any(v for k,v in row['metadata']['program_entry']['reviews'].items() if k!='publish_reviews'))
for name,body in self.files.items():self.assertEqual((out/'evidence'/name).read_bytes(),body)
with self.assertRaisesRegex(ValueError,'Output exists'):c.collect(out,snapshots=src)
def test_duplicate_and_malformed_locator_or_incomplete_census_rejected(self):
listing=c.accepted_listing(self.accepted)
for kind in ('repeat','forum','non_numeric','missing_row'):
rows=copy.deepcopy(self.program)
if kind=='repeat':rows[1]['id']=rows[0]['id'];rows[1]['forum']=rows[0]['forum']
if kind=='forum':rows[0]['forum']='OriginalOpenReviewID'
if kind=='non_numeric':rows[0]['id']='abc';rows[0]['forum']='abc'
if kind=='missing_row':rows.pop()
with self.subTest(kind=kind),self.assertRaises(ValueError):c.program_observations(rows,listing)
def test_empty_fields_foreign_title_and_session_disagree_rejected(self):
listing=c.accepted_listing(self.accepted)
for field,value in [('authors',[]),('abstract',''),('title','A foreign paper'),('session',['99']),('keywords',[42])]:
rows=copy.deepcopy(self.program);rows[0]['content'][field]=value
with self.subTest(field=field),self.assertRaises(ValueError):c.program_observations(rows,listing)
def test_source_optin_no_cannot_publish_review_text(self):
rows=copy.deepcopy(self.program);r=next(r for r in rows if r['reviews']['publish_reviews']=='No');r['reviews']['meta_review']='Do not infer a negative decision'
with self.assertRaisesRegex(ValueError,'preference'):c.program_observations(rows,c.accepted_listing(self.accepted))
def test_parenthesized_affiliation_semicolons_do_not_split_people(self):
self.assertEqual(c.author_names('A Person (University A; Lab (X)); B Person (Institute B)'),['A Person','B Person'])
for raw in ['A Person (Broken','; A Person (Lab)','A Person )']:
with self.assertRaises(ValueError):c.author_names(raw)
def test_ambiguous_official_title_is_not_fused(self):
source=self.accepted.replace(b'FusID: Modality-Fused Semantic IDs for Generative Music Recommendation',b'Off-Manifold Robustness in Synthesizer Inversion with Joint Distribution Flow Matching',1)
with self.assertRaisesRegex(ValueError,'ambiguous'):c.accepted_listing(source)
def test_tamper_missing_file_and_symlink_fail_without_success_report(self):
for kind in ('manifest','body','missing','symlink'):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);src=root/'src';self.materialize(src);path=src/'ismir2026-json-research-v1'/('manifest.json' if kind=='manifest' else 'papers.json')
if kind=='missing':path.unlink()
elif kind=='symlink':other=root/'actual.json';path.rename(other);path.symlink_to(other)
else:path.write_bytes(path.read_bytes()+b' ')
with self.subTest(kind=kind),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())
+146
View File
@@ -0,0 +1,146 @@
import copy
import gzip
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from paper_atlas.collectors import mm2026 as c
class MM2026Test(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.fixture = json.loads(gzip.decompress((Path(__file__).parent/'fixtures/mm2026-source.json.gz').read_bytes()))
cls.files = {name:body.encode('utf-8') for name,body in cls.fixture['files'].items()}
cls.raw = cls.files[c.SNAPSHOT+'/program-data.json']
cls.data = json.loads(cls.raw)
cls.gold = cls.fixture['independent_inventory']
def materialize(self, root):
for name,body in self.files.items():
path = root/name; path.parent.mkdir(parents=True,exist_ok=True); path.write_bytes(body)
def parse_changed(self, data):
return c.parse_program(json.dumps(data,ensure_ascii=False).encode())
def test_all2143_ids_and4326_locations_against_independent_inventory(self):
rows,sessions = c.parse_program(self.raw)
self.assertEqual(len(rows),2143); self.assertEqual(len(sessions),157)
self.assertEqual(sum(len(r['contexts']) for r in rows),4326)
actual = {r['paper_id']:r for r in rows}
self.assertEqual(set(actual),{g['paper_id'] for g in self.gold['inventory']})
for gold in self.gold['inventory']:
r = actual[gold['paper_id']]
self.assertEqual((r['title'],r['authors'],r['track'],r['affiliations']),
(gold['title'],gold['authors'],gold['track_literal'],gold['affiliations']))
self.assertEqual(hashlib.sha256(r['abstract'].encode()).hexdigest(),gold['abstract_sha256'])
self.assertEqual([x['source_pointer'] for x in r['contexts']],gold['locations'])
def test_every_presentation_and_session_context_is_preserved(self):
rows,sessions = c.parse_program(self.raw)
for r in rows:
for context in r['contexts']:
node = self.data
for key in context['source_pointer'].split('/')[1:]:
node = node[int(key)] if isinstance(node,list) else node[key]
self.assertEqual(context['presentation'],{k:node[k] for k in c.PRESENTATION if k in node})
self.assertEqual(r['paper_id'],node['paper_id'])
for i,session in enumerate(sessions):
original = self.data['entries'][i]
self.assertEqual(session['source_pointer'],'/entries/'+str(i))
self.assertEqual(session['track'],'')
for key in original.keys()-{'papers','subsessions'}:
self.assertEqual(session[key],original[key])
for j,sub in enumerate(original.get('subsessions') or []):
saved = session['subsessions'][j]
self.assertEqual(saved['title'],sub['title'])
expected = [x for x in sub.get('items') or [] if 'paper_id' not in x]
self.assertEqual([x['fields'] for x in saved['nonpaper_items']],expected)
def test_all_tracks_retained_without_publication_or_negative_inference(self):
rows,_ = c.parse_program(self.raw)
self.assertEqual({r['track'] for r in rows},set(self.gold['unique_id_track_counts']))
for r in rows:
self.assertFalse(r['approved_for_import']); self.assertFalse(r['final_decision_verified'])
self.assertFalse(r['publication_identity_verified'])
self.assertEqual(r['source_locator'],{'namespace':'mm_program:program_id','value':r['paper_id']})
self.assertEqual(r['tags'],[{'namespace':'mm_program:track','value':r['track']}])
self.assertEqual(r['observed_status'],'listed_in_official_program')
self.assertTrue({'status','identifiers','submission_identifiers','paper_type','emails'}.isdisjoint(r))
missing = [r for r in rows if not r['authors']]
self.assertEqual({r['paper_id'] for r in missing},{'mfp4971','mfp7801'})
self.assertTrue(all(r['missing_fields']==['authors','abstract'] for r in missing))
self.assertEqual(sum(not r['abstract'] for r in rows),216)
def test_collect_evidence_hashes_and_no_contact_arrays_in_normalized_output(self):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);self.materialize(root/'src')
report=c.collect(root/'out',snapshots=root/'src')
self.assertEqual((report['records'],report['program_observations'],report['extra_presentation_observations']),(2143,4326,2183))
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['public_negative_access']['http_statuses'],[403]*4)
self.assertEqual(report['track_counts'],self.gold['unique_id_track_counts'])
for name,body in self.files.items():
self.assertEqual((root/'out/evidence'/Path(name).name).read_bytes(),body)
n=0
for item in report['files']:
raw=Path(item['file']).read_bytes();self.assertEqual(hashlib.sha256(raw).hexdigest(),item['file_sha256'])
self.assertNotIn(b'"emails"',raw);self.assertEqual(len(raw.splitlines()),item['count']);n+=item['count']
self.assertEqual(item['format'],'mm_program_observation_v1_not_ingestion_record')
self.assertEqual(n,2143)
raw=(root/'out/sessions.jsonl').read_bytes()
self.assertEqual(hashlib.sha256(raw).hexdigest(),report['sessions_file_sha256']);self.assertNotIn(b'"emails"',raw)
with self.assertRaisesRegex(ValueError,'Output exists'):c.collect(root/'out',snapshots=root/'src')
def test_conflicting_core_version_is_not_silently_merged(self):
data=copy.deepcopy(self.data)
first=data['entries'][0]['papers'][0]
data['entries'][0]['papers'].append({**first,'abstract':'conflicting abstract'})
with self.assertRaisesRegex(ValueError,'Conflicting core'):self.parse_changed(data)
def test_unknown_nesting_cannot_silently_lose_observations(self):
data=copy.deepcopy(self.data)
moved=data['entries'][0]['papers'].pop()
data['entries'][0]['unreviewed_bucket']=[moved]
with self.assertRaisesRegex(ValueError,'Unhandled'):self.parse_changed(data)
def test_track_drift_unknown_fields_or_missing_title_require_review(self):
for kind in ('outer_track','paper_track','field','title','authors'):
data=copy.deepcopy(self.data);paper=data['entries'][0]['papers'][0]
if kind=='outer_track':data['entries'][0]['track']='Main Track'
elif kind=='paper_track':paper['track']='New Track'
elif kind=='field':paper['doi']='unreviewed'
elif kind=='title':paper['title']=''
else:paper['authors']='Unparsed author string'
with self.subTest(kind=kind),self.assertRaises(ValueError):self.parse_changed(data)
def test_edition_or_census_change_requires_review(self):
for kind in ('edition','count'):
data=copy.deepcopy(self.data)
if kind=='edition':data['generated_at']='2025-11-14T18:36:26.860677+00:00'
else:data['entries'][0]['papers'].pop()
with self.subTest(kind=kind),self.assertRaises(ValueError):self.parse_changed(data)
def test_corruption_missing_evidence_and_symlink_fail_without_completed_report(self):
for kind in ('manifest','program','probe','missing','symlink'):
with tempfile.TemporaryDirectory() as tmp:
root=Path(tmp);self.materialize(root/'src');directory=root/'src'/c.SNAPSHOT
name='manifest.json' if kind=='manifest' else 'openreview-rejected-probe.json' if kind=='probe' else 'program-data.json'
path=directory/name
if kind=='missing':path.unlink()
elif kind=='symlink':
original=directory/'original.json';path.rename(original);path.symlink_to(original)
else:path.write_bytes(path.read_bytes()+b' ')
with self.subTest(kind=kind),self.assertRaises((ValueError,FileNotFoundError)):c.collect(root/'out',snapshots=root/'src')
self.assertTrue((root/'out/failed.json').exists());self.assertFalse((root/'out/collection-report.json').exists())
def test_literal_abstract_math_and_nonpaper_session_events_retained(self):
rows,sessions=c.parse_program(self.raw)
self.assertTrue(any('\\' in r['abstract'] for r in rows))
self.assertTrue(any('\n' in r['abstract'] for r in rows))
self.assertTrue(any(s['type']=='coffee_break' for s in sessions))
self.assertTrue(any(s['type']=='industry_talk' for s in sessions))
self.assertEqual(c.without_emails({'emails':['contact'],'items':[{'emails':['contact'],'title':'talk'}]}),{'items':[{'title':'talk'}]})