Update parser

This commit is contained in:
qananasikq
2026-04-27 20:47:44 +03:00
parent d937aa2386
commit 9f215dc70a
10 changed files with 145 additions and 17 deletions

View File

@@ -5,22 +5,42 @@ x-app-env: &app-env
CELERY_BROKER_URL: ${CELERY_BROKER_URL:-redis://redis:6379/0}
CELERY_RESULT_BACKEND: ${CELERY_RESULT_BACKEND:-redis://redis:6379/0}
IAAI_DATABASE_POOL_RECYCLE_SECONDS: ${IAAI_DATABASE_POOL_RECYCLE_SECONDS:-1800}
CELERY_TASK_SOFT_TIME_LIMIT: ${CELERY_TASK_SOFT_TIME_LIMIT:-900}
CELERY_TASK_TIME_LIMIT: ${CELERY_TASK_TIME_LIMIT:-1200}
CELERY_BROKER_VISIBILITY_TIMEOUT: ${CELERY_BROKER_VISIBILITY_TIMEOUT:-2400}
# Для больших Search-выборок (десятки тысяч лотов) run должен жить дольше одного батча.
CELERY_TASK_SOFT_TIME_LIMIT: ${CELERY_TASK_SOFT_TIME_LIMIT:-7200}
CELERY_TASK_TIME_LIMIT: ${CELERY_TASK_TIME_LIMIT:-7500}
CELERY_BROKER_VISIBILITY_TIMEOUT: ${CELERY_BROKER_VISIBILITY_TIMEOUT:-10800}
# 0 = без лимита.
CELERY_BEAT_SYNC_LIMIT: ${CELERY_BEAT_SYNC_LIMIT:-0}
# Следующий плановый прогон — не раньше чем через час после предыдущего tick beat.
CELERY_BEAT_SYNC_INTERVAL_MINUTES: ${CELERY_BEAT_SYNC_INTERVAL_MINUTES:-60}
# Любой внутренний follow-up после неполного bootstrap тоже не стартует сразу.
IAAI_SYNC_FOLLOWUP_MIN_DELAY_SECONDS: ${IAAI_SYNC_FOLLOWUP_MIN_DELAY_SECONDS:-3600}
CELERY_WORKER_MAX_TASKS_PER_CHILD: ${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
CELERY_BATCH_SIZE: ${CELERY_BATCH_SIZE:-200}
CELERY_BATCH_SIZE: ${CELERY_BATCH_SIZE:-1500}
IAAI_INTER_BATCH_DELAY_SECONDS: ${IAAI_INTER_BATCH_DELAY_SECONDS:-0.3}
# Стабильный fast-профиль: не душим IAAI слишком большим числом HTTPS detail-соединений.
IAAI_SCRAPING_PROFILE: fast
IAAI_FETCH_CONCURRENCY: ${IAAI_FETCH_CONCURRENCY:-96}
IAAI_MAX_RETRIES: ${IAAI_MAX_RETRIES:-2}
IAAI_RETRY_DELAY_SECONDS: ${IAAI_RETRY_DELAY_SECONDS:-0.35}
IAAI_CHALLENGE_REFRESH_ATTEMPTS: ${IAAI_CHALLENGE_REFRESH_ATTEMPTS:-1}
CELERY_PARALLEL_SEGMENTS: ${CELERY_PARALLEL_SEGMENTS:-false}
CELERY_WORKER_CONCURRENCY: 4
CELERY_WORKER_POOL: ${CELERY_WORKER_POOL:-prefork}
IAAI_FAIL_RATE_THRESHOLD: ${IAAI_FAIL_RATE_THRESHOLD:-0.9}
IAAI_PARALLEL_TABS: ${IAAI_PARALLEL_TABS:-8}
IAAI_BLOCK_RESOURCES: ${IAAI_BLOCK_RESOURCES:-true}
IAAI_LISTING_SEGMENTS: ${IAAI_LISTING_SEGMENTS:-auto}
IAAI_FAST_PATH_TIMEOUT_MS: ${IAAI_FAST_PATH_TIMEOUT_MS:-10000}
IAAI_LISTING_SEGMENTS: ${IAAI_LISTING_SEGMENTS:-[]}
IAAI_DISCOVERY_MODE: ${IAAI_DISCOVERY_MODE:-listing}
IAAI_HOURLY_MODE: ${IAAI_HOURLY_MODE:-rolling_refresh}
IAAI_HOURLY_REFRESH_BATCH_SIZE: ${IAAI_HOURLY_REFRESH_BATCH_SIZE:-500}
IAAI_MAX_PAGES_PER_RUN: ${IAAI_MAX_PAGES_PER_RUN:-9999}
IAAI_MAX_VEHICLES_PER_RUN: ${IAAI_MAX_VEHICLES_PER_RUN:-50000}
IAAI_ALWAYS_FULL_SCAN: ${IAAI_ALWAYS_FULL_SCAN:-true}
IAAI_HUMAN_PACE_ENABLED: ${IAAI_HUMAN_PACE_ENABLED:-true}
IAAI_ALWAYS_FULL_SCAN: ${IAAI_ALWAYS_FULL_SCAN:-false}
IAAI_HUMAN_PACE_ENABLED: ${IAAI_HUMAN_PACE_ENABLED:-false}
# Первый sync после clean restart стартует сразу; следующий запуск — только через hourly guard/beat.
IAAI_STARTUP_SYNC_ENABLED: ${IAAI_STARTUP_SYNC_ENABLED:-true}
IAAI_TOKENS_FILE: ${IAAI_TOKENS_FILE:-/data/tokens.json}
IAAI_RUNTIME_CONFIG_FILE: ${IAAI_RUNTIME_CONFIG_FILE:-/app/runtime_config.json}
IAAI_BROWSER_ENGINE: chromium
@@ -149,9 +169,11 @@ services:
stop_grace_period: 60s
command: >
celery -A iaai_scraper.worker.celery_app worker
--loglevel=info --concurrency=1 --pool=solo
--loglevel=info
--concurrency=${CELERY_WORKER_CONCURRENCY:-4}
--pool=${CELERY_WORKER_POOL:-prefork}
--pidfile=/tmp/celery-worker.pid
-Q scraping --max-tasks-per-child=${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
-Q iaai_sync --max-tasks-per-child=${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
healthcheck:
# Проверка worker.
test: ["CMD-SHELL", "test -f /tmp/celery-worker.pid && kill -0 $(cat /tmp/celery-worker.pid) && pgrep -f 'iaai_scraper.worker.self_heal' >/dev/null"]