refactor mobile.de parser, fix country mapping, update README

This commit is contained in:
qananasikq
2026-05-04 18:12:10 +03:00
parent edddfc884c
commit 800867b4e6
46 changed files with 2887 additions and 10900 deletions

View File

@@ -1,13 +1,11 @@
x-app-env: &app-env
# 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}
x-app-env: &app-env
MOBILEDE_DATABASE_URL: ${MOBILEDE_DATABASE_URL:-postgresql+psycopg2://mobilede:mobilede@postgres:5432/mobilede_scraper}
IAAI_REDIS_URL: ${IAAI_REDIS_URL:-redis://redis:6379/0}
MOBILEDE_REDIS_URL: ${MOBILEDE_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}
IAAI_DATABASE_POOL_RECYCLE_SECONDS: ${IAAI_DATABASE_POOL_RECYCLE_SECONDS:-1800}
IAAI_DATABASE_POOL_SIZE: ${IAAI_DATABASE_POOL_SIZE:-20}
IAAI_DATABASE_MAX_OVERFLOW: ${IAAI_DATABASE_MAX_OVERFLOW:-40}
MOBILEDE_DATABASE_POOL_RECYCLE_SECONDS: ${MOBILEDE_DATABASE_POOL_RECYCLE_SECONDS:-1800}
MOBILEDE_DATABASE_POOL_SIZE: ${MOBILEDE_DATABASE_POOL_SIZE:-20}
MOBILEDE_DATABASE_MAX_OVERFLOW: ${MOBILEDE_DATABASE_MAX_OVERFLOW:-40}
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}
@@ -21,7 +19,7 @@ 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}
@@ -31,6 +29,8 @@ x-app-env: &app-env
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_FILTERED_SEARCH_URL: ${MOBILEDE_FILTERED_SEARCH_URL:-}
MOBILEDE_FILTERED_SEARCH_URLS: ${MOBILEDE_FILTERED_SEARCH_URLS:-}
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:-3600}
@@ -43,12 +43,21 @@ x-app-env: &app-env
MOBILEDE_MAX_PAGE_NUMBER: ${MOBILEDE_MAX_PAGE_NUMBER:-50}
MOBILEDE_SEGMENT_TARGET_RESULTS: ${MOBILEDE_SEGMENT_TARGET_RESULTS:-1000}
MOBILEDE_COMPACT_SEGMENTS: ${MOBILEDE_COMPACT_SEGMENTS:-true}
MOBILEDE_PREPLAN_SEGMENT_PROBES: ${MOBILEDE_PREPLAN_SEGMENT_PROBES:-false}
MOBILEDE_PREPLAN_MAX_SEGMENTS: ${MOBILEDE_PREPLAN_MAX_SEGMENTS:-1000}
MOBILEDE_PREPLAN_MAX_PROBES: ${MOBILEDE_PREPLAN_MAX_PROBES:-40}
MOBILEDE_DYNAMIC_SEGMENT_PROBES: ${MOBILEDE_DYNAMIC_SEGMENT_PROBES:-false}
MOBILEDE_SPLIT_SEGMENTS_BY_MILEAGE: ${MOBILEDE_SPLIT_SEGMENTS_BY_MILEAGE:-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}
MOBILEDE_FLARESOLVERR_ENABLED: ${MOBILEDE_FLARESOLVERR_ENABLED:-false}
MOBILEDE_FLARESOLVERR_URL: ${MOBILEDE_FLARESOLVERR_URL:-http://flaresolverr:8191/v1}
MOBILEDE_FLARESOLVERR_TIMEOUT_SECONDS: ${MOBILEDE_FLARESOLVERR_TIMEOUT_SECONDS:-120}
MOBILEDE_FLARESOLVERR_MAX_TIMEOUT_MS: ${MOBILEDE_FLARESOLVERR_MAX_TIMEOUT_MS:-60000}
MOBILEDE_FLARESOLVERR_SESSION: ${MOBILEDE_FLARESOLVERR_SESSION:-}
MOBILEDE_FLARESOLVERR_STATUSES: ${MOBILEDE_FLARESOLVERR_STATUSES:-403,429,503}
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}
@@ -65,13 +74,13 @@ 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.
# Автовосстановление 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.
# При простое БД перезапускаем worker.
IAAI_DB_IDLE_RESTART_SECONDS: ${IAAI_DB_IDLE_RESTART_SECONDS:-3600}
TZ: ${TZ:-UTC}
@@ -93,7 +102,7 @@ x-worker-service: &worker-service
- tokens_data:/data
services:
# PostgreSQL.
# База данных.
postgres:
image: postgres:16-alpine
container_name: mobilede-postgres
@@ -117,7 +126,7 @@ services:
max-size: "10m"
max-file: "5"
# Redis.
# Очередь Redis.
redis:
image: redis:7-alpine
container_name: mobilede-redis
@@ -141,7 +150,24 @@ services:
max-size: "10m"
max-file: "5"
# Миграции.
flaresolverr:
image: ghcr.io/flaresolverr/flaresolverr:latest
container_name: mobilede-flaresolverr
restart: unless-stopped
environment:
LOG_LEVEL: ${FLARESOLVERR_LOG_LEVEL:-info}
LOG_HTML: ${FLARESOLVERR_LOG_HTML:-false}
CAPTCHA_SOLVER: ${FLARESOLVERR_CAPTCHA_SOLVER:-none}
TZ: ${TZ:-UTC}
ports:
- "127.0.0.1:${MOBILEDE_HOST_FLARESOLVERR_PORT:-8191}:8191"
logging:
driver: json-file
options:
max-size: "10m"
max-file: "5"
# Миграции.
migrate:
<<: *app-service
container_name: mobilede-migrate
@@ -151,7 +177,7 @@ services:
condition: service_healthy
command: alembic upgrade head
# API.
# API сервис.
api:
<<: *app-service
container_name: mobilede-api
@@ -179,7 +205,7 @@ services:
max-size: "10m"
max-file: "5"
# Worker.
# Worker сервис.
worker:
<<: *worker-service
container_name: mobilede-worker
@@ -197,7 +223,7 @@ services:
--pidfile=/tmp/celery-worker.pid
-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 'mobilede_scraper.worker.self_heal' >/dev/null"]
interval: 60s
timeout: 10s
@@ -209,7 +235,7 @@ services:
max-size: "10m"
max-file: "5"
# Beat.
# Beat сервис.
beat:
<<: *worker-service
container_name: mobilede-beat
@@ -237,7 +263,7 @@ services:
max-size: "10m"
max-file: "5"
# One-shot CLI: collect mobile.de search pages into artifacts/json.
# Одноразовый сбор страниц поиска.
mobilede-search:
<<: *app-service
container_name: mobilede-search
@@ -256,7 +282,7 @@ services:
--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.
# Одноразовый sync поиска в БД.
mobilede-sync-search:
<<: *app-service
container_name: mobilede-sync-search
@@ -276,7 +302,7 @@ services:
--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.
# Одноразовый сбор карточки по ID.
mobilede-detail:
<<: *app-service
container_name: mobilede-detail