diff --git a/REPORT.md b/REPORT.md index a4cc4c4..7cfaf8c 100644 --- a/REPORT.md +++ b/REPORT.md @@ -1,7 +1,7 @@ # omni-parse · автоотчёт -_обновлено 2026-10-26T16:20:23+00:00_ +_обновлено 2026-10-26T16:21:06+00:00_ | метрика | значение | |---|---| @@ -30,5 +30,5 @@ _обновлено 2026-10-26T16:20:23+00:00_ | ab_altenens_is | 119 | | ab_afflift_com | 119 | -самоаудит: дубли url: 1 +самоаудит: OK diff --git a/tools/selfcheck.py b/tools/selfcheck.py index 2e54eca..2adc595 100644 --- a/tools/selfcheck.py +++ b/tools/selfcheck.py @@ -81,8 +81,9 @@ def collect(): st["by_source_top"] = [{"name": r[0], "n": r[1]} for r in q( con, "SELECT source, count(*) FROM items GROUP BY 1 ORDER BY 2 DESC LIMIT 12")] - dupes = q(con, "SELECT url, count(*) c FROM items GROUP BY url HAVING c > 1 LIMIT 5") - st["dupe_urls_sample"] = len(dupes) + st["dupe_urls_sample"] = one( + con, "SELECT count(*) FROM (SELECT url FROM items WHERE url != '' GROUP BY url HAVING count(*) > 1)") + st["no_url"] = one(con, "SELECT count(*) FROM items WHERE url IS NULL OR url = ''") st["empty_sources"] = st["sources"] - one(con, "SELECT count(DISTINCT source) FROM items") st["abuse_stored"] = one(con, "SELECT count(*) FROM items WHERE category = 'abuse'") \ if "category" in cols else None