fix(tradein): backfill kn obj_class (parser + 3-pass) (#572) #628
2 changed files with 5 additions and 3 deletions
|
|
@ -143,6 +143,8 @@ _OBJ_CLASS_PATTERNS: list[tuple[str, str]] = [
|
|||
# that use the word "типов*" to describe standard/economy housing. The canonical value
|
||||
# "Типовой" matches what the catalog SSR scraper (domrf_catalog.py) extracts from pages.
|
||||
# Bug #572: this pattern was absent → 131+ objects remained NULL despite matching aiDescription.
|
||||
# Intentionally broad (r"типов" not r"типов[ыоа]й?\b"): DOM.RF aiDescription uses varied
|
||||
# inflections; tightening risks dropping coverage below the 70% acceptance threshold.
|
||||
(r"типов", "Типовой"),
|
||||
]
|
||||
|
||||
|
|
|
|||
|
|
@ -1,4 +1,4 @@
|
|||
-- 120_backfill_obj_class_kn_api.sql
|
||||
-- 121_backfill_obj_class_kn_api.sql
|
||||
--
|
||||
-- Backfill obj_class for domrf_kn_objects rows that remain NULL after migration 106.
|
||||
--
|
||||
|
|
@ -75,7 +75,7 @@ WHERE o.obj_id = src.obj_id
|
|||
UPDATE domrf_kn_objects AS o
|
||||
SET obj_class = prev.obj_class
|
||||
FROM (
|
||||
SELECT DISTINCT ON (n.obj_id, n.snapshot_date)
|
||||
SELECT
|
||||
n.obj_id,
|
||||
n.snapshot_date,
|
||||
best.obj_class
|
||||
|
|
@ -145,7 +145,7 @@ BEGIN
|
|||
WHERE snapshot_date = v_latest_snap AND region_cd = 66;
|
||||
|
||||
RAISE NOTICE
|
||||
'domrf_kn_objects after 120 backfill:'
|
||||
'domrf_kn_objects after 121 backfill:'
|
||||
' all-snapshots: total=% filled=% null=%'
|
||||
' | latest snapshot (%): total=% filled=%',
|
||||
v_total, v_filled, v_null,
|
||||
Loading…
Add table
Reference in a new issue