Peach: Local-first video metadata scraper and media manager.
"""存量身份数据的两项修正,默认只列计划。
- `--downgrade`:标成 observed 却不出自代码判据(`source:`、`script:fc2-seller@`)的分类断言与
身份关联降为 candidate,等用户复核(ADR-0052 决策一)。来源、证据与检查时间不变。
- `--directories`:某账号在一个目录子树里的每件有效作品都已被逐个否定时,补一条目录级否定,
以后进入该目录的新文件不再归给这个账号。只留最上层目录,不顺带否定任何未被否定的作品。
两项都不给时两项都列。`--report` 把完整计划写成 JSON;`--apply` 必须同时给 `--backup`,
在一个事务里写完并做完整性与外键自检。
repair_identity_claims.py --report build/identity-claims.json
repair_identity_claims.py --downgrade --apply --backup <备份路径>
"""
from __future__ import annotations
import argparse
import json
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).resolve().parents[1] / 'src'))
from peach.entities import apply_directory_rejections, derived_directory_rejections # noqa: E402
from peach.entity_classification import downgrade_untrusted_observed, untrusted_observed # noqa: E402
from peach.field_owners import script_owner # noqa: E402
from peach.fsutil import atomic_write_text # noqa: E402
from peach.scripting import add_ledger_write_args, counts_of, open_for_write, verify_after_write # noqa: E402
REVIEWER = script_owner('repair_identity_claims')
EXTRA_COUNTS = {
'observed_claims': "SELECT count(*) FROM entity_classification WHERE status='observed'",
'candidate_claims': "SELECT count(*) FROM entity_classification WHERE status='candidate'",
'attribution_rejections': "SELECT count(*) FROM review_decision WHERE category='creator-attribution' AND status='rejected'",
}
def build_plan(connection, *, downgrade: bool, directories: bool) -> dict:
plan = {}
if downgrade:
plan['downgrade'] = untrusted_observed(connection)
if directories:
plan['directories'] = derived_directory_rejections(connection)
return plan
def summary(plan: dict) -> dict:
result = {}
if 'downgrade' in plan:
result['downgrade'] = {table: _by_source(rows) for table, rows in plan['downgrade'].items()}
if 'directories' in plan:
result['directories'] = [{key: row[key] for key in ('name', 'directory', 'subtree_assets', 'name_rejections')}
for row in plan['directories']]
return result
def _by_source(rows) -> dict[str, int]:
counts: dict[str, int] = {}
for row in rows:
counts[row['source']] = counts.get(row['source'], 0) + 1
return counts
def main(argv=None) -> int:
parser = add_ledger_write_args(argparse.ArgumentParser(description='存量身份断言降级与目录级否定'))
parser.add_argument('--downgrade', action='store_true', help='只处理非代码判据的 observed 降级')
parser.add_argument('--directories', action='store_true', help='只处理由作品级否定派生的目录级否定')
parser.add_argument('--report', type=Path, help='完整计划写成 JSON')
args = parser.parse_args(argv)
both = not (args.downgrade or args.directories)
selected = {'downgrade': args.downgrade or both, 'directories': args.directories or both}
connection = open_for_write(args)
try:
plan = build_plan(connection, **selected)
if args.report:
atomic_write_text(args.report, json.dumps(plan, ensure_ascii=False, indent=2))
output = {'plan': summary(plan)}
if args.apply:
connection.execute('BEGIN IMMEDIATE')
output['before'] = counts_of(connection, EXTRA_COUNTS)
_, known_violations = verify_after_write(connection)
if selected['downgrade']:
output['downgraded'] = downgrade_untrusted_observed(connection)
if selected['directories']:
output['directory_rejections'] = apply_directory_rejections(
connection, plan['directories'], reviewer=REVIEWER)
integrity, violations = verify_after_write(connection)
if integrity != 'ok' or violations > known_violations:
connection.rollback()
print(json.dumps({'error': '完整性或外键自检未通过', 'integrity': integrity,
'foreign_key_violations': violations}, ensure_ascii=False), file=sys.stderr)
return 1
connection.commit()
output['after'] = counts_of(connection, EXTRA_COUNTS)
else:
output['dry_run'] = '未写 ledger;加 --apply --backup <路径> 才写入'
print(json.dumps(output, ensure_ascii=False, indent=2))
finally:
connection.close()
return 0
if __name__ == '__main__':
raise SystemExit(main())