fix batch pool leak and add page cap

This commit is contained in:
qananasikq
2026-04-17 23:50:34 +03:00
parent 0c08fc7f63
commit 12b313ad52

View File

@@ -919,6 +919,41 @@ class EncarScraper:
excluded_brands: set[str] | None = None,
runtime_filters: FiltersConfig | None = None,
probe_all_photos: bool = False,
) -> dict[str, Any]:
"""Публичная обёртка: гарантирует закрытие HTTP pool даже при ошибках."""
try:
return self._sync_listing_impl(
limit=limit,
filters=filters,
lane=lane,
only_new=only_new,
batch_size=batch_size,
redis_client=redis_client,
allowed_brands=allowed_brands,
excluded_brands=excluded_brands,
runtime_filters=runtime_filters,
probe_all_photos=probe_all_photos,
)
finally:
if self._batch_pool is not None:
try:
self._batch_pool.close()
except Exception:
logger.debug("Failed to close batch pool", exc_info=True)
self._batch_pool = None
def _sync_listing_impl(
self,
limit: int | None = None,
filters: EncarFilters | None = None,
lane: str = "encar",
only_new: bool = False,
batch_size: int = 1000,
redis_client: Any | None = None,
allowed_brands: set[str] | None = None,
excluded_brands: set[str] | None = None,
runtime_filters: FiltersConfig | None = None,
probe_all_photos: bool = False,
) -> dict[str, Any]:
"""Полная синхронизация листинга Encar.
@@ -988,11 +1023,21 @@ class EncarScraper:
max_consecutive_errors = 5
consecutive_zero_new = 0
max_consecutive_zero_new = 5
# Защита от бесконечной пагинации: Encar API отдаёт максимум
# ~10k записей на query, шарды строятся ≤9500 → не более ~10
# страниц при page_size=1000. Жёсткий cap в 50 — паранойя.
max_pages_per_shard = 50
shard_query = shard_filters.build_query()
shard_collected = 0
while True:
if page >= max_pages_per_shard:
logger.warning(
"Shard %d hit hard page cap (%d), moving to next",
shard_idx, max_pages_per_shard,
)
break
offset = page * page_size
try:
response = self._fetch_listing_page(
@@ -1152,11 +1197,6 @@ class EncarScraper:
self._clear_checkpoint(redis_client)
# Закрываем pool batch API
if self._batch_pool is not None:
self._batch_pool.close()
self._batch_pool = None
logger.info(
"Full sync complete: %d shards, %d collected, %d synced, %d failed, %d marked sold",
len(shards), items_collected, synced, failed, marked_sold,