-- 062_clean_avito_addresses.sql -- Strip CSS noise from listings.address for Avito source (one-shot backfill). -- Idempotent: re-run is safe (regexp_replace on already-clean string is a no-op). BEGIN; UPDATE listings SET address = trim(both ' ,.' FROM regexp_replace( regexp_replace(address, '\.?css-[a-z0-9_-]+\s*\{[^}]*\}', '', 'gi'), '\s*(Площадь \d+(\.\d+)?\s?(м²|м2|кв\.?\s?м)|от \d+\s?мин\.|css-[a-z0-9_-]+).*$', '', 'i' )) WHERE source = 'avito' AND (address ~* 'css-[a-z0-9_-]+' OR address ~* 'Площадь \d+(\.\d+)?\s?(м²|м2|кв\.?\s?м)|от \d+\s?мин\.'); COMMIT;