cleanup and fix runtime bugs

This commit is contained in:
qananasikq
2026-04-09 20:34:48 +03:00
parent eb9fb48ea0
commit f6d7c8ea60
28 changed files with 693 additions and 628 deletions

View File

@@ -1,14 +1,54 @@
import os
from dataclasses import dataclass, field
from pathlib import Path
from dotenv import load_dotenv
load_dotenv()
TRUE_VALUES = {"1", "true", "yes", "on"}
# Хелперы для чтения env-переменных с приведением типов
def _env_str(name: str, default: str) -> str:
value = os.getenv(name)
return value if value is not None else default
def _env_optional_str(name: str) -> str | None:
value = os.getenv(name)
if value is None:
return None
value = value.strip()
return value or None
def _env_path_str(name: str) -> str | None:
value = _env_optional_str(name)
if value is None:
return None
return str(Path(value).expanduser())
def _env_bool(name: str, default: bool) -> bool:
fallback = "true" if default else "false"
return _env_str(name, fallback).strip().lower() in TRUE_VALUES
def _env_int(name: str, default: int) -> int:
return int(_env_str(name, str(default)).strip())
def _env_float(name: str, default: float) -> float:
return float(_env_str(name, str(default)).strip())
# Конфиг браузерного отпечатка (User-Agent, viewport, timezone)
@dataclass(slots=True)
class FingerprintConfig:
# Браузерный отпечаток.
user_agent: str = (
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
"AppleWebKit/537.36 (KHTML, like Gecko) "
@@ -30,83 +70,93 @@ class FingerprintConfig:
sec_ch_ua: str = '"Google Chrome";v="135", "Chromium";v="135", "Not.A/Brand";v="24"'
# Конфиг перехвата сетевых запросов (лимиты на кол-во)
@dataclass(slots=True)
class CaptureConfig:
# Параметры capture.
capture_same_origin_only: bool = os.getenv("IAAI_CAPTURE_SAME_ORIGIN_ONLY", "true").strip().lower() in {"1", "true", "yes", "on"}
max_requests: int = int(os.getenv("IAAI_MAX_CAPTURED_REQUESTS", "40"))
max_json_responses: int = int(os.getenv("IAAI_MAX_CAPTURED_JSON_RESPONSES", "30"))
capture_same_origin_only: bool = _env_bool("IAAI_CAPTURE_SAME_ORIGIN_ONLY", True)
max_requests: int = _env_int("IAAI_MAX_CAPTURED_REQUESTS", 40)
max_json_responses: int = _env_int("IAAI_MAX_CAPTURED_JSON_RESPONSES", 30)
# Конфиг пауз между действиями (имитация человека)
@dataclass(slots=True)
class HumanPaceConfig:
# Паузы между действиями.
enabled: bool = os.getenv("IAAI_HUMAN_PACE_ENABLED", "true").strip().lower() in {"1", "true", "yes", "on"}
after_listing_open_min_s: float = float(os.getenv("IAAI_AFTER_LISTING_OPEN_MIN_S", "2.5"))
after_listing_open_max_s: float = float(os.getenv("IAAI_AFTER_LISTING_OPEN_MAX_S", "4.5"))
after_filter_action_min_s: float = float(os.getenv("IAAI_AFTER_FILTER_ACTION_MIN_S", "2.0"))
after_filter_action_max_s: float = float(os.getenv("IAAI_AFTER_FILTER_ACTION_MAX_S", "4.0"))
before_vehicle_open_min_s: float = float(os.getenv("IAAI_BEFORE_VEHICLE_OPEN_MIN_S", "0.5"))
before_vehicle_open_max_s: float = float(os.getenv("IAAI_BEFORE_VEHICLE_OPEN_MAX_S", "1.5"))
after_vehicle_open_min_s: float = float(os.getenv("IAAI_AFTER_VEHICLE_OPEN_MIN_S", "0.3"))
after_vehicle_open_max_s: float = float(os.getenv("IAAI_AFTER_VEHICLE_OPEN_MAX_S", "1.0"))
between_vehicles_min_s: float = float(os.getenv("IAAI_BETWEEN_VEHICLES_MIN_S", "0.5"))
between_vehicles_max_s: float = float(os.getenv("IAAI_BETWEEN_VEHICLES_MAX_S", "1.5"))
after_page_change_min_s: float = float(os.getenv("IAAI_AFTER_PAGE_CHANGE_MIN_S", "3.0"))
after_page_change_max_s: float = float(os.getenv("IAAI_AFTER_PAGE_CHANGE_MAX_S", "6.0"))
enabled: bool = _env_bool("IAAI_HUMAN_PACE_ENABLED", True)
after_listing_open_min_s: float = _env_float("IAAI_AFTER_LISTING_OPEN_MIN_S", 2.5)
after_listing_open_max_s: float = _env_float("IAAI_AFTER_LISTING_OPEN_MAX_S", 4.5)
after_filter_action_min_s: float = _env_float("IAAI_AFTER_FILTER_ACTION_MIN_S", 2.0)
after_filter_action_max_s: float = _env_float("IAAI_AFTER_FILTER_ACTION_MAX_S", 4.0)
before_vehicle_open_min_s: float = _env_float("IAAI_BEFORE_VEHICLE_OPEN_MIN_S", 0.5)
before_vehicle_open_max_s: float = _env_float("IAAI_BEFORE_VEHICLE_OPEN_MAX_S", 1.5)
after_vehicle_open_min_s: float = _env_float("IAAI_AFTER_VEHICLE_OPEN_MIN_S", 0.3)
after_vehicle_open_max_s: float = _env_float("IAAI_AFTER_VEHICLE_OPEN_MAX_S", 1.0)
between_vehicles_min_s: float = _env_float("IAAI_BETWEEN_VEHICLES_MIN_S", 0.5)
between_vehicles_max_s: float = _env_float("IAAI_BETWEEN_VEHICLES_MAX_S", 1.5)
after_page_change_min_s: float = _env_float("IAAI_AFTER_PAGE_CHANGE_MIN_S", 3.0)
after_page_change_max_s: float = _env_float("IAAI_AFTER_PAGE_CHANGE_MAX_S", 6.0)
# Конфиг сбора листинга (URL, лимиты страниц и машин)
@dataclass(slots=True)
class ListingConfig:
# Лимиты листинга.
cars_url: str = os.getenv("IAAI_CARS_LISTING_URL", "https://www.iaai.com/Vehiclelisting/Cars")
max_pages_per_run: int = int(os.getenv("IAAI_MAX_PAGES_PER_RUN", "5"))
max_vehicles_per_run: int = int(os.getenv("IAAI_MAX_VEHICLES_PER_RUN", "100"))
page_link_limit: int = int(os.getenv("IAAI_PAGE_LINK_LIMIT", "200"))
include_pagination: bool = os.getenv("IAAI_INCLUDE_PAGINATION", "true").strip().lower() in {"1", "true", "yes", "on"}
collect_current_page_only: bool = os.getenv("IAAI_COLLECT_CURRENT_PAGE_ONLY", "false").strip().lower() in {"1", "true", "yes", "on"}
cars_url: str = _env_str("IAAI_CARS_LISTING_URL", "https://www.iaai.com/Vehiclelisting/Cars")
max_pages_per_run: int = _env_int("IAAI_MAX_PAGES_PER_RUN", 5)
max_vehicles_per_run: int = _env_int("IAAI_MAX_VEHICLES_PER_RUN", 100)
page_link_limit: int = _env_int("IAAI_PAGE_LINK_LIMIT", 200)
include_pagination: bool = _env_bool("IAAI_INCLUDE_PAGINATION", True)
collect_current_page_only: bool = _env_bool("IAAI_COLLECT_CURRENT_PAGE_ONLY", False)
# - Конфиг PostgreSQL (URL, пул соединений, pool_recycle)
@dataclass(slots=True)
class DatabaseConfig:
# Подключение к БД.
url: str = os.getenv("IAAI_DATABASE_URL", "postgresql+psycopg2://iaai:iaai@localhost:5432/iaai_scraper")
echo: bool = os.getenv("IAAI_DATABASE_ECHO", "false").strip().lower() in {"1", "true", "yes", "on"}
pool_size: int = int(os.getenv("IAAI_DATABASE_POOL_SIZE", "5"))
max_overflow: int = int(os.getenv("IAAI_DATABASE_MAX_OVERFLOW", "10"))
url: str = _env_str("IAAI_DATABASE_URL", "postgresql+psycopg2://iaai:iaai@localhost:5432/iaai_scraper")
echo: bool = _env_bool("IAAI_DATABASE_ECHO", False)
pool_size: int = _env_int("IAAI_DATABASE_POOL_SIZE", 5)
max_overflow: int = _env_int("IAAI_DATABASE_MAX_OVERFLOW", 10)
pool_recycle_seconds: int = _env_int("IAAI_DATABASE_POOL_RECYCLE_SECONDS", 1800)
auto_create_tables: bool = _env_bool("IAAI_DATABASE_AUTO_CREATE_TABLES", False)
# --- Конфиг Redis (URL для Celery broker) ---
@dataclass(slots=True)
class RedisConfig:
# Подключение к Redis.
url: str = os.getenv("IAAI_REDIS_URL", "redis://localhost:6379/0")
url: str = _env_str("IAAI_REDIS_URL", "redis://localhost:6379/0")
# --- Конфиг Celery (лимиты задач, concurrency, beat-расписание) ---
@dataclass(slots=True)
class CeleryConfig:
# Настройки Celery.
broker_url: str = os.getenv("CELERY_BROKER_URL", "") or ""
result_backend: str = os.getenv("CELERY_RESULT_BACKEND", "") or ""
task_soft_time_limit: int = int(os.getenv("CELERY_TASK_SOFT_TIME_LIMIT", "600"))
task_time_limit: int = int(os.getenv("CELERY_TASK_TIME_LIMIT", "900"))
worker_concurrency: int = int(os.getenv("CELERY_WORKER_CONCURRENCY", "1"))
beat_sync_interval_minutes: int = int(os.getenv("CELERY_BEAT_SYNC_INTERVAL_MINUTES", "60"))
beat_sync_limit: int = int(os.getenv("CELERY_BEAT_SYNC_LIMIT", "26"))
broker_url: str = _env_str("CELERY_BROKER_URL", "")
result_backend: str = _env_str("CELERY_RESULT_BACKEND", "")
task_soft_time_limit: int = _env_int("CELERY_TASK_SOFT_TIME_LIMIT", 600)
task_time_limit: int = _env_int("CELERY_TASK_TIME_LIMIT", 900)
worker_concurrency: int = _env_int("CELERY_WORKER_CONCURRENCY", 1)
worker_max_tasks_per_child: int = _env_int("CELERY_WORKER_MAX_TASKS_PER_CHILD", 20)
broker_visibility_timeout: int = _env_int("CELERY_BROKER_VISIBILITY_TIMEOUT", 7200)
beat_sync_interval_minutes: int = _env_int("CELERY_BEAT_SYNC_INTERVAL_MINUTES", 60)
beat_sync_limit: int = _env_int("CELERY_BEAT_SYNC_LIMIT", 26)
# --- Конфиг прокси (server, username, password) ---
@dataclass(slots=True)
class ProxyConfig:
# Настройки прокси.
server: str | None = (os.getenv("IAAI_PROXY_SERVER") or "").strip() or None
username: str | None = (os.getenv("IAAI_PROXY_USERNAME") or "").strip() or None
password: str | None = (os.getenv("IAAI_PROXY_PASSWORD") or "").strip() or None
server: str | None = _env_optional_str("IAAI_PROXY_SERVER")
username: str | None = _env_optional_str("IAAI_PROXY_USERNAME")
password: str | None = _env_optional_str("IAAI_PROXY_PASSWORD")
@property
def enabled(self) -> bool:
return bool(self.server)
def to_playwright_dict(self) -> dict[str, str] | None:
# Формат для Playwright.
if not self.server:
return None
result: dict[str, str] = {"server": self.server}
@@ -117,23 +167,26 @@ class ProxyConfig:
return result
# --- Главный объект настроек: собирает все блоки конфигурации ---
@dataclass(slots=True)
class Settings:
# Общие настройки.
home_url: str = "https://www.iaai.com/"
default_timeout_ms: int = int(os.getenv("IAAI_TIMEOUT_MS", "45000"))
network_settle_ms: int = int(os.getenv("IAAI_NETWORK_SETTLE_MS", "800"))
max_retries: int = int(os.getenv("IAAI_MAX_RETRIES", "3"))
retry_delay_seconds: float = float(os.getenv("IAAI_RETRY_DELAY_SECONDS", "2.5"))
retry_backoff_multiplier: float = float(os.getenv("IAAI_RETRY_BACKOFF_MULTIPLIER", "2.0"))
retry_jitter_seconds: float = float(os.getenv("IAAI_RETRY_JITTER_SECONDS", "0.25"))
headless: bool = os.getenv("IAAI_HEADLESS", "true").strip().lower() in {"1", "true", "yes", "on"}
log_level: str = os.getenv("IAAI_LOG_LEVEL", "INFO")
log_file: str | None = os.getenv("IAAI_LOG_FILE") or None
enable_trace_id_logs: bool = os.getenv("IAAI_ENABLE_TRACE_ID_LOGS", "true").strip().lower() in {"1", "true", "yes", "on"}
sync_only_new: bool = os.getenv("IAAI_SYNC_ONLY_NEW", "true").strip().lower() in {"1", "true", "yes", "on"}
raw_output_json: str | None = os.getenv("IAAI_RAW_OUTPUT_JSON") or None
scheduler_interval_minutes: int = int(os.getenv("IAAI_SCHEDULER_INTERVAL_MINUTES", "60"))
default_timeout_ms: int = _env_int("IAAI_TIMEOUT_MS", 45000)
network_settle_ms: int = _env_int("IAAI_NETWORK_SETTLE_MS", 800)
max_retries: int = _env_int("IAAI_MAX_RETRIES", 3)
retry_delay_seconds: float = _env_float("IAAI_RETRY_DELAY_SECONDS", 2.5)
retry_backoff_multiplier: float = _env_float("IAAI_RETRY_BACKOFF_MULTIPLIER", 2.0)
retry_jitter_seconds: float = _env_float("IAAI_RETRY_JITTER_SECONDS", 0.25)
headless: bool = _env_bool("IAAI_HEADLESS", True)
log_level: str = _env_str("IAAI_LOG_LEVEL", "INFO")
log_file: str | None = _env_optional_str("IAAI_LOG_FILE")
enable_trace_id_logs: bool = _env_bool("IAAI_ENABLE_TRACE_ID_LOGS", True)
sync_only_new: bool = _env_bool("IAAI_SYNC_ONLY_NEW", True)
raw_output_json: str | None = _env_optional_str("IAAI_RAW_OUTPUT_JSON")
tokens_file: str | None = _env_path_str("IAAI_TOKENS_FILE")
runtime_config_file: str | None = _env_path_str("IAAI_RUNTIME_CONFIG_FILE")
scheduler_interval_minutes: int = _env_int("IAAI_SCHEDULER_INTERVAL_MINUTES", 60)
fingerprint: FingerprintConfig = field(default_factory=FingerprintConfig)
capture: CaptureConfig = field(default_factory=CaptureConfig)
pace: HumanPaceConfig = field(default_factory=HumanPaceConfig)
@@ -143,6 +196,5 @@ class Settings:
celery: CeleryConfig = field(default_factory=CeleryConfig)
proxy: ProxyConfig = field(default_factory=ProxyConfig)
# Настройки по умолчанию.
# Глобальный синглтон — используется по умолчанию во всех модулях.
settings = Settings()