Real-estate parser pipeline

Monthly pipeline: monitor наш.дом.рф → new-object detection → run parsers (parsers/*.py per developer or add_rashet_from_pd.py from declarations) → fill_rashet → consolidated table → manual check → fill_baza.

Highlights

• API-first architecture — moved monitor_nash_dom + kpdgazstroi.ru from Playwright (5 min) to curl_cffi+API (12-30 sec) using __NEXT_DATA__ SSR-state extraction
• Idempotent fill_rashet — code API ≠ apartment numbers in Kristall buildings, separate matching by area+floor+rooms
• COM Excel write — Times New Roman 11, taskkill before COM, no openpyxl conflict
• Backup-before-write — main.py overwrites, daily backup auto + restore on demand
• 2 data sources unified — site parsers and government declarations both feed same fill_rashet

What you get

Industry-specific market data pipeline. Same shape works for any monthly aggregation: car listings, vacancy reports, FX rates — anything that has 10+ heterogeneous sources to homogenize.