add openlane scraper

This commit is contained in:
qananasikq
2026-04-21 23:49:56 +03:00
parent 7f72056289
commit 74f12da10a
32 changed files with 4713 additions and 0 deletions

View File

@@ -0,0 +1,194 @@
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:
user_agent: str = (
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
"AppleWebKit/537.36 (KHTML, like Gecko) "
"Chrome/135.0.0.0 Safari/537.36"
)
viewport_presets: tuple[dict[str, int], ...] = (
{"width": 1920, "height": 1080},
{"width": 1600, "height": 900},
{"width": 1536, "height": 864},
{"width": 1440, "height": 900},
{"width": 1366, "height": 768},
)
timezone_candidates: tuple[str, ...] = (
"America/New_York",
"America/Chicago",
"America/Los_Angeles",
)
locale: str = "en-US"
sec_ch_ua: str = '"Google Chrome";v="135", "Chromium";v="135", "Not.A/Brand";v="24"'
# --- Конфиг OpenLane (auth, paging, throttle, retry, checkpoint/output) ---
@dataclass(slots=True)
class OpenLaneConfig:
sign_in_url: str = _env_str("OPENLANE_SIGN_IN_URL", "https://app.openlane.com/sign_in")
search_url: str = _env_str("OPENLANE_SEARCH_URL", "https://app.openlane.com/search?tab=marketplace")
api_search_url: str = _env_str("OPENLANE_API_SEARCH_URL", "https://app.openlane.com/api/v4/search")
username: str = _env_str("OPENLANE_USERNAME", "")
password: str = _env_str("OPENLANE_PASSWORD", "")
source_tab: str = _env_str("OPENLANE_SOURCE_TAB", "marketplace")
sale_types: str = _env_str("OPENLANE_SALE_TYPES", "marketplace")
page_size: int = _env_int("OPENLANE_PAGE_SIZE", 30)
max_pages: int = _env_int("OPENLANE_MAX_PAGES", 512)
concurrency: int = _env_int("OPENLANE_CONCURRENCY", 4)
throttle_min_seconds: float = _env_float("OPENLANE_THROTTLE_MIN_SECONDS", 0.3)
throttle_max_seconds: float = _env_float("OPENLANE_THROTTLE_MAX_SECONDS", 0.8)
retry_schedule_seconds: tuple[float, ...] = tuple(
float(part.strip())
for part in _env_str("OPENLANE_RETRY_SCHEDULE_SECONDS", "2,5,10").split(",")
if part.strip()
) or (2.0, 5.0, 10.0)
checkpoint_every_pages: int = _env_int("OPENLANE_CHECKPOINT_EVERY_PAGES", 10)
output_dir: str = _env_str("OPENLANE_OUTPUT_DIR", "artifacts/openlane")
jsonl_output: str = _env_str("OPENLANE_JSONL_OUTPUT", "artifacts/openlane/openlane_search.jsonl")
aggregated_output: str = _env_str("OPENLANE_AGGREGATED_OUTPUT", "artifacts/openlane/openlane_search_aggregated.json")
checkpoint_file: str = _env_str("OPENLANE_CHECKPOINT_FILE", "artifacts/openlane/openlane_checkpoint.json")
storage_state_file: str = _env_str("OPENLANE_STORAGE_STATE_FILE", "artifacts/openlane/openlane_storage_state.json")
persist_storage_state: bool = _env_bool("OPENLANE_PERSIST_STORAGE_STATE", True)
progress_log_every_pages: int = _env_int("OPENLANE_PROGRESS_LOG_EVERY_PAGES", 10)
request_timeout_ms: int = _env_int("OPENLANE_REQUEST_TIMEOUT_MS", 45000)
max_cars_limit: int = _env_int("OPENLANE_MAX_CARS_LIMIT", 20)
refresh_token: str = _env_str("OPENLANE_REFRESH_TOKEN", "")
okta_client_id: str = _env_str("OPENLANE_OKTA_CLIENT_ID", "")
okta_token_endpoint: str = _env_str(
"OPENLANE_OKTA_TOKEN_ENDPOINT",
"https://okta.iam.karglobal.com/oauth2/default/v1/token",
)
okta_redirect_uri: str = _env_str("OPENLANE_OKTA_REDIRECT_URI", "https://app.openlane.com/sign_in")
okta_timeout_seconds: int = _env_int("OPENLANE_OKTA_TIMEOUT_SECONDS", 15)
# --- Конфиг PostgreSQL (URL, пул соединений, pool_recycle) ---
@dataclass(slots=True)
class DatabaseConfig:
url: str = _env_str("OPENLANE_DATABASE_URL", "postgresql+psycopg2://openlane:openlane@localhost:5432/openlane_scraper")
echo: bool = _env_bool("OPENLANE_DATABASE_ECHO", False)
pool_size: int = _env_int("OPENLANE_DATABASE_POOL_SIZE", 5)
max_overflow: int = _env_int("OPENLANE_DATABASE_MAX_OVERFLOW", 10)
pool_recycle_seconds: int = _env_int("OPENLANE_DATABASE_POOL_RECYCLE_SECONDS", 1800)
auto_create_tables: bool = _env_bool("OPENLANE_DATABASE_AUTO_CREATE_TABLES", False)
# --- Конфиг Redis (URL для Celery broker) ---
@dataclass(slots=True)
class RedisConfig:
url: str = _env_str("OPENLANE_REDIS_URL", "redis://localhost:6379/0")
socket_timeout_seconds: float = _env_float("OPENLANE_REDIS_SOCKET_TIMEOUT_SECONDS", 10.0)
socket_connect_timeout_seconds: float = _env_float("OPENLANE_REDIS_SOCKET_CONNECT_TIMEOUT_SECONDS", 5.0)
health_check_interval_seconds: int = _env_int("OPENLANE_REDIS_HEALTH_CHECK_INTERVAL_SECONDS", 30)
# --- Конфиг Celery (лимиты задач, concurrency, beat-расписание) ---
@dataclass(slots=True)
class CeleryConfig:
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", 3300)
task_time_limit: int = _env_int("CELERY_TASK_TIME_LIMIT", 3600)
worker_concurrency: int = _env_int("CELERY_WORKER_CONCURRENCY", 4)
worker_max_tasks_per_child: int = _env_int("CELERY_WORKER_MAX_TASKS_PER_CHILD", 5)
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 | None = _env_int("CELERY_BEAT_SYNC_LIMIT", 0) or None
# --- Конфиг прокси (server, username, password) ---
@dataclass(slots=True)
class ProxyConfig:
server: str | None = _env_optional_str("OPENLANE_PROXY_SERVER")
username: str | None = _env_optional_str("OPENLANE_PROXY_USERNAME")
password: str | None = _env_optional_str("OPENLANE_PROXY_PASSWORD")
@property
def enabled(self) -> bool:
return bool(self.server)
def to_playwright_dict(self) -> dict[str, str] | None:
if not self.server:
return None
result: dict[str, str] = {"server": self.server}
if self.username:
result["username"] = self.username
if self.password:
result["password"] = self.password
return result
# Главный объект настроек: собирает все блоки конфигурации
@dataclass(slots=True)
class Settings:
default_timeout_ms: int = _env_int("OPENLANE_TIMEOUT_MS", 45000)
headless: bool = _env_bool("OPENLANE_HEADLESS", True)
browser_engine: str = _env_str("OPENLANE_BROWSER_ENGINE", "auto")
log_level: str = _env_str("OPENLANE_LOG_LEVEL", "INFO")
log_file: str | None = _env_optional_str("OPENLANE_LOG_FILE")
enable_trace_id_logs: bool = _env_bool("OPENLANE_ENABLE_TRACE_ID_LOGS", True)
runtime_config_file: str | None = _env_path_str("OPENLANE_RUNTIME_CONFIG_FILE")
fingerprint: FingerprintConfig = field(default_factory=FingerprintConfig)
openlane: OpenLaneConfig = field(default_factory=OpenLaneConfig)
database: DatabaseConfig = field(default_factory=DatabaseConfig)
redis: RedisConfig = field(default_factory=RedisConfig)
celery: CeleryConfig = field(default_factory=CeleryConfig)
proxy: ProxyConfig = field(default_factory=ProxyConfig)
# Глобальный синглтон — используется по умолчанию во всех модулях.
settings = Settings()