fix docker scraping

improve batch sync

add postgres upsert

fix sync locking

improve listing sync

speed up scraper

clean up project

prepare for github

update docker setup
This commit is contained in:
qananasikq
2026-04-13 16:35:08 +03:00
parent b9557922ed
commit a4c93a1df1
30 changed files with 2239 additions and 624 deletions

View File

@@ -10,25 +10,25 @@ IAAI_MAX_CAPTURED_JSON_RESPONSES=30
# Sequential listing-first mode.
IAAI_CARS_LISTING_URL=https://www.iaai.com/Vehiclelisting/Cars
IAAI_MAX_PAGES_PER_RUN=10
IAAI_MAX_VEHICLES_PER_RUN=30
IAAI_PAGE_LINK_LIMIT=100
IAAI_MAX_VEHICLES_PER_RUN=50000
IAAI_PAGE_LINK_LIMIT=500
IAAI_INCLUDE_PAGINATION=true
IAAI_COLLECT_CURRENT_PAGE_ONLY=false
# Human-like pacing.
IAAI_HUMAN_PACE_ENABLED=true
IAAI_AFTER_LISTING_OPEN_MIN_S=2.5
IAAI_AFTER_LISTING_OPEN_MAX_S=4.5
IAAI_AFTER_FILTER_ACTION_MIN_S=2.0
IAAI_AFTER_FILTER_ACTION_MAX_S=4.0
IAAI_BEFORE_VEHICLE_OPEN_MIN_S=1.5
IAAI_BEFORE_VEHICLE_OPEN_MAX_S=3.0
IAAI_AFTER_VEHICLE_OPEN_MIN_S=1.0
IAAI_AFTER_VEHICLE_OPEN_MAX_S=2.5
IAAI_BETWEEN_VEHICLES_MIN_S=3.0
IAAI_BETWEEN_VEHICLES_MAX_S=6.0
IAAI_AFTER_PAGE_CHANGE_MIN_S=3.0
IAAI_AFTER_PAGE_CHANGE_MAX_S=6.0
# Паузы (отключены для максимальной скорости; включи на VPS если попадаешь под блокировки).
IAAI_HUMAN_PACE_ENABLED=false
IAAI_AFTER_LISTING_OPEN_MIN_S=0.2
IAAI_AFTER_LISTING_OPEN_MAX_S=0.5
IAAI_AFTER_FILTER_ACTION_MIN_S=0.2
IAAI_AFTER_FILTER_ACTION_MAX_S=0.5
IAAI_BEFORE_VEHICLE_OPEN_MIN_S=0.02
IAAI_BEFORE_VEHICLE_OPEN_MAX_S=0.08
IAAI_AFTER_VEHICLE_OPEN_MIN_S=0.02
IAAI_AFTER_VEHICLE_OPEN_MAX_S=0.08
IAAI_BETWEEN_VEHICLES_MIN_S=0.02
IAAI_BETWEEN_VEHICLES_MAX_S=0.08
IAAI_AFTER_PAGE_CHANGE_MIN_S=0.2
IAAI_AFTER_PAGE_CHANGE_MAX_S=0.5
# Sync settings
IAAI_SYNC_ONLY_NEW=true
@@ -40,16 +40,6 @@ IAAI_RETRY_DELAY_SECONDS=2.5
IAAI_RETRY_BACKOFF_MULTIPLIER=2.0
IAAI_RETRY_JITTER_SECONDS=0.25
# Session persistence (cookies + localStorage)
# Keeps browser session alive between runs (up to 30 days).
IAAI_STORAGE_STATE_PATH=storage_state.json
IAAI_SESSION_MAX_AGE_DAYS=30
IAAI_SESSION_SAVE_ON_EXIT=true
# IAAI login credentials (for authenticated scraping)
# IAAI_LOGIN_EMAIL=your@email.com
# IAAI_LOGIN_PASSWORD=your_password
# Proxy (HTTP/HTTPS preferred for Playwright)
# Chromium does not support SOCKS5 proxy authentication directly.
# Use residential or mobile USA proxy.
@@ -60,8 +50,8 @@ IAAI_SESSION_SAVE_ON_EXIT=true
# Database (PostgreSQL).
IAAI_DATABASE_URL=postgresql+psycopg2://iaai:iaai@postgres:5432/iaai_scraper
IAAI_DATABASE_ECHO=false
IAAI_DATABASE_POOL_SIZE=5
IAAI_DATABASE_MAX_OVERFLOW=10
IAAI_DATABASE_POOL_SIZE=10
IAAI_DATABASE_MAX_OVERFLOW=20
# ─── Redis (Celery broker) ─────────────────────────────────
IAAI_REDIS_URL=redis://redis:6379/0