Repair official collection and retire exhausted material tasks

backend/src/alphapulse/crawler/repair_collection.py

@@ -0,0 +1,51 @@
+"""本次采集修复初始化：默认只读，--execute 幂等建立一次新预算。运行前停止 crawler。"""
+import argparse
+import asyncio
+import copy
+import json
+from datetime import UTC, datetime
+
+from sqlalchemy import select, text
+
+from alphapulse.core.db import SessionLocal, engine
+from alphapulse.models import Source
+
+REPAIR = 'collection-repair-20261105-v1'
+CODES = ('sec', 'doj', 'congress')
+
+
+async def run(execute=False):
+    output = []
+    async with SessionLocal() as session:
+        if not execute:
+            await session.execute(text('SET TRANSACTION READ ONLY'))
+        query = select(Source).where(Source.code.in_(CODES)).order_by(Source.code)
+        if execute:
+            query = query.with_for_update()
+        for source in await session.scalars(query):
+            cursor = copy.deepcopy(source.history_cursor or {})
+            if cursor.get('collection_repair') == REPAIR:
+                output.append({'source': source.code, 'already_initialized': True})
+                continue
+            affected = []
+            for key in ('document_retries', 'attachment_retries', 'bill_retries'):
+                for url, entry in cursor.get(key, {}).items():
+                    affected.append({'kind': key, 'url': url, 'previous_attempts': entry['attempts']})
+                    entry['repair_previous'] = {k: v for k, v in entry.items() if k != 'document'}
+                    entry['attempts'] = 0
+                    entry['next_at'] = datetime.now(UTC).timestamp()
+                    entry.pop('retired_at', None)
+            cursor['collection_repair'] = REPAIR
+            if execute:
+                source.history_cursor = cursor
+            output.append({'source': source.code, 'items': affected})
+        if execute:
+            await session.commit()
+    await engine.dispose()
+    return {'mode': 'execute' if execute else 'dry-run', 'sources': output}
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument('--execute', action='store_true')
+    print(json.dumps(asyncio.run(run(parser.parse_args().execute)), ensure_ascii=False))