news-mcp/scripts/clear_all_but_feeds.py

#!/usr/bin/env python3
"""Clear all news data but keep feed_state rows intact."""

import os
import sqlite3
import sys
from pathlib import Path

DB_PATH = Path(os.getenv("NEWS_MCP_DB_PATH", str(Path(__file__).resolve().parent.parent / "data" / "news.sqlite")))

if len(sys.argv) > 1:
    DB_PATH = Path(sys.argv[1])

if not DB_PATH.exists():
    print(f"DB not found: {DB_PATH}", file=sys.stderr)
    sys.exit(1)

TABLES_TO_CLEAR = [
    "cluster_entities",
    "cluster_keywords",
    "entity_metadata",
    "clusters",
    "meta",
]

conn = sqlite3.connect(str(DB_PATH))
for table in TABLES_TO_CLEAR:
    conn.execute(f"DELETE FROM {table}")
conn.commit()

# Report remaining feed_state count
cur = conn.execute("SELECT COUNT(*) FROM feed_state")
feed_count = cur.fetchone()[0]
conn.close()

print(f"Done. Cleared: {', '.join(TABLES_TO_CLEAR)}. feed_state rows kept: {feed_count}")