Prepare mobile de parser release

This commit is contained in:
qananasikq
2026-04-27 21:10:17 +03:00
parent 31f87a9bdf
commit b86bcd04b8
21 changed files with 2835 additions and 423 deletions

View File

@@ -1,6 +1,7 @@
x-app-env: &app-env
# Всегда используем Postgres.
IAAI_DATABASE_URL: postgresql+psycopg2://iaai:iaai@postgres:5432/iaai_scraper
# Legacy env name is still used by Settings; values are mobile.de defaults.
IAAI_DATABASE_URL: ${MOBILEDE_DATABASE_URL:-postgresql+psycopg2://mobilede:mobilede@postgres:5432/mobilede_scraper}
MOBILEDE_DATABASE_URL: ${MOBILEDE_DATABASE_URL:-postgresql+psycopg2://mobilede:mobilede@postgres:5432/mobilede_scraper}
IAAI_REDIS_URL: ${IAAI_REDIS_URL:-redis://redis:6379/0}
CELERY_BROKER_URL: ${CELERY_BROKER_URL:-redis://redis:6379/0}
CELERY_RESULT_BACKEND: ${CELERY_RESULT_BACKEND:-redis://redis:6379/0}
@@ -20,12 +21,32 @@ x-app-env: &app-env
IAAI_REQUEST_JITTER_MAX_S: ${IAAI_REQUEST_JITTER_MAX_S:-0}
IAAI_DETAIL_RETRIES: ${IAAI_DETAIL_RETRIES:-1}
IAAI_LISTING_RETRIES: ${IAAI_LISTING_RETRIES:-1}
# 0 = без лимита.
# 0 = без лимита.
CELERY_BEAT_SYNC_LIMIT: ${CELERY_BEAT_SYNC_LIMIT:-0}
CELERY_WORKER_CONCURRENCY: ${CELERY_WORKER_CONCURRENCY:-4}
CELERY_WORKER_POOL: ${CELERY_WORKER_POOL:-prefork}
CELERY_WORKER_MAX_TASKS_PER_CHILD: ${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
CELERY_BATCH_SIZE: ${CELERY_BATCH_SIZE:-1000}
MOBILEDE_STARTUP_MAX_PAGES: ${MOBILEDE_STARTUP_MAX_PAGES:-100}
MOBILEDE_BEAT_MAX_PAGES: ${MOBILEDE_BEAT_MAX_PAGES:-100}
MOBILEDE_REQUEST_DELAY_SECONDS: ${MOBILEDE_REQUEST_DELAY_SECONDS:-0}
MOBILEDE_CONCURRENT_PAGES: ${MOBILEDE_CONCURRENT_PAGES:-2}
MOBILEDE_CURSOR_ENABLED: ${MOBILEDE_CURSOR_ENABLED:-true}
MOBILEDE_CONTINUOUS_SYNC_ENABLED: ${MOBILEDE_CONTINUOUS_SYNC_ENABLED:-true}
MOBILEDE_CONTINUOUS_SYNC_DELAY_SECONDS: ${MOBILEDE_CONTINUOUS_SYNC_DELAY_SECONDS:-0}
MOBILEDE_PROGRESS_LOG_EVERY_PAGES: ${MOBILEDE_PROGRESS_LOG_EVERY_PAGES:-100}
MOBILEDE_SKIP_EMPTY_WINDOW: ${MOBILEDE_SKIP_EMPTY_WINDOW:-true}
MOBILEDE_ROTATE_RUNTIME_SEGMENTS: ${MOBILEDE_ROTATE_RUNTIME_SEGMENTS:-true}
MOBILEDE_RUNTIME_INITIAL_TASKS: ${MOBILEDE_RUNTIME_INITIAL_TASKS:-2}
MOBILEDE_SEGMENT_PAGE_WINDOW: ${MOBILEDE_SEGMENT_PAGE_WINDOW:-10}
MOBILEDE_SEGMENT_TARGET_RESULTS: ${MOBILEDE_SEGMENT_TARGET_RESULTS:-1800}
MOBILEDE_COMPACT_SEGMENTS: ${MOBILEDE_COMPACT_SEGMENTS:-true}
MOBILEDE_DYNAMIC_SEGMENT_PROBES: ${MOBILEDE_DYNAMIC_SEGMENT_PROBES:-false}
MOBILEDE_SKIP_EMPTY_DYNAMIC_SEGMENTS: ${MOBILEDE_SKIP_EMPTY_DYNAMIC_SEGMENTS:-true}
MOBILEDE_BOOTSTRAP_FULL_SCAN_ENABLED: ${MOBILEDE_BOOTSTRAP_FULL_SCAN_ENABLED:-true}
MOBILEDE_INCREMENTAL_AFTER_BOOTSTRAP: ${MOBILEDE_INCREMENTAL_AFTER_BOOTSTRAP:-true}
MOBILEDE_INCREMENTAL_PAGE_WINDOW: ${MOBILEDE_INCREMENTAL_PAGE_WINDOW:-1}
IAAI_STARTUP_SYNC_ENABLED: ${IAAI_STARTUP_SYNC_ENABLED:-true}
IAAI_INTER_BATCH_DELAY_SECONDS: ${IAAI_INTER_BATCH_DELAY_SECONDS:-0}
IAAI_FAIL_RATE_THRESHOLD: ${IAAI_FAIL_RATE_THRESHOLD:-0.9}
IAAI_PARALLEL_TABS: ${IAAI_PARALLEL_TABS:-8}
@@ -41,13 +62,13 @@ x-app-env: &app-env
IAAI_TOKENS_FILE: ${IAAI_TOKENS_FILE:-/home/app/tokens.json}
IAAI_RUNTIME_CONFIG_FILE: ${IAAI_RUNTIME_CONFIG_FILE:-/app/runtime_config.json}
IAAI_BROWSER_ENGINE: chromium
# Self-heal для worker.
# Self-heal для worker.
IAAI_SELF_HEAL_ENABLED: ${IAAI_SELF_HEAL_ENABLED:-true}
IAAI_SELF_HEAL_CHECK_INTERVAL_SECONDS: ${IAAI_SELF_HEAL_CHECK_INTERVAL_SECONDS:-30}
IAAI_SELF_HEAL_STALL_SECONDS: ${IAAI_SELF_HEAL_STALL_SECONDS:-900}
IAAI_SELF_HEAL_STARTUP_GRACE_SECONDS: ${IAAI_SELF_HEAL_STARTUP_GRACE_SECONDS:-300}
IAAI_SELF_HEAL_RESTART_COOLDOWN_SECONDS: ${IAAI_SELF_HEAL_RESTART_COOLDOWN_SECONDS:-300}
# Если в Postgres нет записей дольше 60 минут при активной работе — полный restart с segment 1.
# Если в Postgres нет записей дольше 60 минут при активной работе — полный restart с segment 1.
IAAI_DB_IDLE_RESTART_SECONDS: ${IAAI_DB_IDLE_RESTART_SECONDS:-3600}
TZ: ${TZ:-UTC}
@@ -72,18 +93,18 @@ services:
# PostgreSQL.
postgres:
image: postgres:16-alpine
container_name: iaai-postgres
container_name: mobilede-postgres
restart: unless-stopped
environment:
POSTGRES_USER: iaai
POSTGRES_PASSWORD: iaai
POSTGRES_DB: iaai_scraper
POSTGRES_USER: ${POSTGRES_USER:-mobilede}
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-mobilede}
POSTGRES_DB: ${POSTGRES_DB:-mobilede_scraper}
ports:
- "127.0.0.1:5432:5432"
volumes:
- pgdata:/var/lib/postgresql/data
healthcheck:
test: ["CMD-SHELL", "pg_isready -U iaai -d iaai_scraper"]
test: ["CMD-SHELL", "pg_isready -U ${POSTGRES_USER:-mobilede} -d ${POSTGRES_DB:-mobilede_scraper}"]
interval: 5s
timeout: 3s
retries: 5
@@ -96,7 +117,7 @@ services:
# Redis.
redis:
image: redis:7-alpine
container_name: iaai-redis
container_name: mobilede-redis
restart: unless-stopped
ports:
- "127.0.0.1:6379:6379"
@@ -117,10 +138,10 @@ services:
max-size: "10m"
max-file: "5"
# Миграции.
# Миграции.
migrate:
<<: *app-service
container_name: iaai-migrate
container_name: mobilede-migrate
restart: "no"
depends_on:
postgres:
@@ -130,7 +151,7 @@ services:
# API.
api:
<<: *app-service
container_name: iaai-api
container_name: mobilede-api
restart: unless-stopped
ports:
- "127.0.0.1:8000:8000"
@@ -158,6 +179,7 @@ services:
# Worker.
worker:
<<: *worker-service
container_name: mobilede-worker
restart: unless-stopped
init: true
depends_on:
@@ -170,9 +192,9 @@ services:
celery -A iaai_scraper.worker.celery_app worker
--loglevel=info --concurrency=${CELERY_WORKER_CONCURRENCY:-4} --pool=${CELERY_WORKER_POOL:-prefork}
--pidfile=/tmp/celery-worker.pid
-Q iaai_sync --max-tasks-per-child=${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
-Q mobilede_sync,iaai_sync --max-tasks-per-child=${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
healthcheck:
# Проверка worker.
# 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"]
interval: 60s
timeout: 10s
@@ -187,7 +209,7 @@ services:
# Beat.
beat:
<<: *worker-service
container_name: iaai-beat
container_name: mobilede-beat
restart: unless-stopped
init: true
depends_on:
@@ -212,6 +234,61 @@ services:
max-size: "10m"
max-file: "5"
# One-shot CLI: collect mobile.de search pages into artifacts/json.
mobilede-search:
<<: *app-service
container_name: mobilede-search
profiles: ["cli"]
restart: "no"
depends_on:
migrate:
condition: service_completed_successfully
volumes:
- ./runtime_config.json:/app/runtime_config.json:ro
- ./artifacts:/app/artifacts
command: >
mobilede search
--max-pages ${MOBILEDE_SEARCH_MAX_PAGES:-1}
--page ${MOBILEDE_SEARCH_START_PAGE:-1}
--delay ${MOBILEDE_REQUEST_DELAY_SECONDS:-0.7}
--output /app/artifacts/json/mobilede_search.json
# One-shot CLI: collect and upsert mobile.de search pages into Postgres.
mobilede-sync-search:
<<: *app-service
container_name: mobilede-sync-search
profiles: ["cli"]
restart: "no"
depends_on:
migrate:
condition: service_completed_successfully
volumes:
- ./runtime_config.json:/app/runtime_config.json:ro
- ./artifacts:/app/artifacts
command: >
mobilede sync-search
--max-pages ${MOBILEDE_SEARCH_MAX_PAGES:-1}
--page ${MOBILEDE_SEARCH_START_PAGE:-1}
--delay ${MOBILEDE_REQUEST_DELAY_SECONDS:-0.7}
--lane ${MOBILEDE_SYNC_LANE:-mobile_de_cars}
--output /app/artifacts/json/mobilede_sync_search.json
# One-shot CLI: collect one detail page by listing id.
mobilede-detail:
<<: *app-service
container_name: mobilede-detail
profiles: ["cli"]
restart: "no"
depends_on:
migrate:
condition: service_completed_successfully
volumes:
- ./runtime_config.json:/app/runtime_config.json:ro
- ./artifacts:/app/artifacts
command: >
mobilede detail ${MOBILEDE_LISTING_ID:-449929602}
--output /app/artifacts/json/mobilede_detail.json
volumes:
pgdata:
redisdata: