import json
import subprocess

with open("daily_crawl_config.json", "r", encoding="utf-8") as f:
    config = json.load(f)
scripts = config.get("crawlers", [])
print(f"配置共 {len(scripts)} 条")

# 查DB中有哪些script_name
result = subprocess.run(
    ["sqlite3", "search.db"],
    input="SELECT DISTINCT script_name FROM gov_raw WHERE script_name IS NOT NULL",
    capture_output=True, text=True, timeout=30
)
in_db = set()
if result.stdout.strip():
    in_db = set(result.stdout.strip().split("\n"))
print(f"DB中有数据的脚本数: {len(in_db)}")

no_data = []
for s in scripts:
    sn = s.get("script", "")
    if sn and sn not in in_db:
        no_data.append(s)

print(f"\n无数据的脚本: {len(no_data)} 个\n")
for i, s in enumerate(no_data, 1):
    sn = s.get("script","?")
    g = s.get("group","?")
    ind = s.get("industry","?")
    nm = s.get("name","?")
    print(f"{i:3d}. {sn:45s} | 组:{g:10s} | 行业:{ind:20s} | 配置名:{nm:20s}")