config.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394
  1. """应用配置。
  2. 配置优先级:环境变量 > 这里的默认值。
  3. """
  4. from __future__ import annotations
  5. import json
  6. import os
  7. from dataclasses import dataclass, field
  8. from pathlib import Path
  9. from typing import Any
  10. PROJECT_ROOT = Path(__file__).resolve().parents[2]
  11. def _env(name: str, default: str) -> str:
  12. value = os.getenv(name)
  13. if value is None or value == "":
  14. return default
  15. return value
  16. def _env_int(name: str, default: int) -> int:
  17. value = os.getenv(name)
  18. if value is None or value == "":
  19. return default
  20. return int(value)
  21. def _env_int_optional(name: str, default: int | None = None) -> int | None:
  22. value = os.getenv(name)
  23. if value is None or value == "":
  24. return default
  25. return int(value)
  26. def _env_float(name: str, default: float) -> float:
  27. value = os.getenv(name)
  28. if value is None or value == "":
  29. return default
  30. return float(value)
  31. def _env_float_optional(name: str, default: float | None = None) -> float | None:
  32. value = os.getenv(name)
  33. if value is None or value == "":
  34. return default
  35. return float(value)
  36. def _env_first(names: tuple[str, ...], default: str) -> str:
  37. for name in names:
  38. value = os.getenv(name)
  39. if value is not None and value != "":
  40. return value
  41. return default
  42. def _env_bool(name: str, default: bool) -> bool:
  43. value = os.getenv(name)
  44. if value is None or value == "":
  45. return default
  46. return value.strip().lower() in {"1", "true", "yes", "y", "on"}
  47. def _load_json_file(path_value: str) -> Any:
  48. path = Path(path_value).expanduser()
  49. if not path.is_absolute():
  50. path = PROJECT_ROOT / path
  51. return json.loads(path.read_text(encoding="utf-8"))
  52. def _env_json(name: str, default: Any, file_env_name: str | None = None) -> Any:
  53. if file_env_name:
  54. file_path = os.getenv(file_env_name)
  55. if file_path:
  56. return _load_json_file(file_path)
  57. value = os.getenv(name)
  58. if value is None or value == "":
  59. return default
  60. return json.loads(value)
  61. @dataclass(frozen=True)
  62. class Settings:
  63. crawapi_base_url: str = "http://crawapi.piaoquantv.com"
  64. crawapi_hot_content_rank_path: str = "/crawler/jin_ri_re_bang/content_rank"
  65. crawapi_keyword_search_path: str = "/crawler/bai_du/keyword"
  66. decode_api_url: str = "https://aigc-api.aiddit.com/aigc/api/task/decode"
  67. decode_result_api_url: str = "https://aigc-api.aiddit.com/aigc/api/task/decode/result"
  68. decode_config_id: int = 70
  69. request_timeout_seconds: int = 180
  70. https_verify_ssl: bool = False
  71. hot_flow_cron_hours: str = "6,12,18"
  72. hot_flow_cron_minute: int = 0
  73. hot_flow_interval_seconds: int = 1800
  74. decode_result_flow_interval_seconds: int = 1800
  75. decode_result_batch_size: int = 50
  76. contribution_score_threshold: float = 0.6
  77. hot_flow_sources: list[dict[str, Any]] = field(
  78. default_factory=lambda: [
  79. {
  80. "source": "百度",
  81. "count": 10,
  82. },
  83. {
  84. "source": "微博",
  85. "count": 10,
  86. },
  87. {
  88. "source": "微信",
  89. "count": 10,
  90. }
  91. ]
  92. )
  93. mysql_host: str = "rm-t4nh1xx6o2a6vj8qu3o.mysql.singapore.rds.aliyuncs.com"
  94. mysql_port: int = 3306
  95. mysql_user: str = "content_rw"
  96. mysql_password: str = "bC1aH4bA1lB0"
  97. mysql_database: str = "external_demand"
  98. mysql_charset: str = "utf8mb4"
  99. open_router_api_key: str = "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"
  100. open_router_default_model: str = "anthropic/claude-haiku-4-5"
  101. open_router_timeout_seconds: int = 60
  102. open_router_http_referer: str = ""
  103. open_router_app_title: str = "external_demand"
  104. open_router_base_url: str = "https://openrouter.ai/api/v1"
  105. open_router_temperature: float | None = 0.7
  106. open_router_max_tokens: int | None = 20000
  107. odps_access_id: str = "LTAI9EBa0bd5PrDa"
  108. odps_access_key: str = "vAalxds7YxhfOA2yVv8GziCg3Y87v5"
  109. odps_project: str = "loghubods"
  110. odps_endpoint: str = "http://service.odps.aliyun.com/api"
  111. odps_tunnel_endpoint: str = ""
  112. demand_pool_source_table: str = "dwd_multi_demand_pool_di"
  113. demand_pool_excluded_strategy: str = "当下供需gap-分词"
  114. demand_pool_top_n: int = 200
  115. hot_demand_pool_strategy: str = "新热事件"
  116. wxindex_score_threshold: float = 100_000.0
  117. odps_daily_write_limit: int = 100
  118. postprocess_batch_size: int = 20
  119. contribution_match_llm_model: str = ""
  120. contribution_match_llm_max_attempts: int = 3
  121. contribution_match_llm_retry_sleep_seconds: float = 1.0
  122. contribution_match_llm_max_tokens: int = 4000
  123. wxindex_llm_model: str = "anthropic/claude-haiku-4-5"
  124. wxindex_llm_max_attempts: int = 3
  125. wxindex_llm_max_tokens: int = 4000
  126. wxindex_api_url: str = "http://crawapi.piaoquantv.com/crawler/wei_xin/wxindex"
  127. wxindex_lookback_days: int = 7
  128. demand_event_sense_threshold: float = 6.0
  129. demand_senior_fit_threshold: float = 6.0
  130. demand_quality_llm_model: str = "anthropic/claude-haiku-4-5"
  131. demand_quality_llm_max_attempts: int = 3
  132. demand_quality_llm_retry_sleep_seconds: float = 1.0
  133. demand_quality_llm_max_tokens: int = 4000
  134. category_filter_llm_model: str = "anthropic/claude-haiku-4-5"
  135. category_filter_llm_max_attempts: int = 3
  136. category_filter_llm_retry_sleep_seconds: float = 1.0
  137. category_filter_llm_max_tokens: int = 1024
  138. category_filter_body_max_chars: int = 2000
  139. category_filter_item_sleep_seconds: float = 0.0
  140. @classmethod
  141. def from_env(cls) -> "Settings":
  142. defaults = cls()
  143. return cls(
  144. crawapi_base_url=_env("CRAWAPI_BASE_URL", defaults.crawapi_base_url),
  145. crawapi_hot_content_rank_path=_env(
  146. "CRAWAPI_HOT_CONTENT_RANK_PATH",
  147. defaults.crawapi_hot_content_rank_path,
  148. ),
  149. crawapi_keyword_search_path=_env(
  150. "CRAWAPI_KEYWORD_SEARCH_PATH",
  151. defaults.crawapi_keyword_search_path,
  152. ),
  153. decode_api_url=_env("DECODE_API_URL", defaults.decode_api_url),
  154. decode_result_api_url=_env(
  155. "DECODE_RESULT_API_URL",
  156. defaults.decode_result_api_url,
  157. ),
  158. decode_config_id=_env_int("DECODE_CONFIG_ID", defaults.decode_config_id),
  159. request_timeout_seconds=_env_int(
  160. "REQUEST_TIMEOUT_SECONDS",
  161. defaults.request_timeout_seconds,
  162. ),
  163. https_verify_ssl=_env_bool("HTTPS_VERIFY_SSL", defaults.https_verify_ssl),
  164. hot_flow_cron_hours=_env(
  165. "HOT_FLOW_CRON_HOURS",
  166. defaults.hot_flow_cron_hours,
  167. ),
  168. hot_flow_cron_minute=_env_int(
  169. "HOT_FLOW_CRON_MINUTE",
  170. defaults.hot_flow_cron_minute,
  171. ),
  172. hot_flow_interval_seconds=_env_int(
  173. "HOT_FLOW_INTERVAL_SECONDS",
  174. defaults.hot_flow_interval_seconds,
  175. ),
  176. decode_result_flow_interval_seconds=_env_int(
  177. "DECODE_RESULT_FLOW_INTERVAL_SECONDS",
  178. defaults.decode_result_flow_interval_seconds,
  179. ),
  180. decode_result_batch_size=_env_int(
  181. "DECODE_RESULT_BATCH_SIZE",
  182. defaults.decode_result_batch_size,
  183. ),
  184. contribution_score_threshold=float(
  185. _env(
  186. "CONTRIBUTION_SCORE_THRESHOLD",
  187. str(defaults.contribution_score_threshold),
  188. )
  189. ),
  190. hot_flow_sources=_env_json(
  191. "HOT_FLOW_SOURCES_JSON",
  192. defaults.hot_flow_sources,
  193. "HOT_FLOW_SOURCES_FILE",
  194. ),
  195. mysql_host=_env("MYSQL_HOST", defaults.mysql_host),
  196. mysql_port=_env_int("MYSQL_PORT", defaults.mysql_port),
  197. mysql_user=_env("MYSQL_USER", defaults.mysql_user),
  198. mysql_password=_env("MYSQL_PASSWORD", defaults.mysql_password),
  199. mysql_database=_env("MYSQL_DATABASE", defaults.mysql_database),
  200. mysql_charset=_env("MYSQL_CHARSET", defaults.mysql_charset),
  201. open_router_api_key=_env_first(
  202. ("OPEN_ROUTER_API_KEY", "OPENROUTER_API_KEY"),
  203. defaults.open_router_api_key,
  204. ),
  205. open_router_default_model=_env(
  206. "OPEN_ROUTER_DEFAULT_MODEL",
  207. defaults.open_router_default_model,
  208. ),
  209. open_router_timeout_seconds=_env_int(
  210. "OPEN_ROUTER_TIMEOUT_SECONDS",
  211. defaults.open_router_timeout_seconds,
  212. ),
  213. open_router_http_referer=_env_first(
  214. ("OPEN_ROUTER_HTTP_REFERER", "OPENROUTER_HTTP_REFERER"),
  215. defaults.open_router_http_referer,
  216. ),
  217. open_router_app_title=_env_first(
  218. ("OPEN_ROUTER_APP_TITLE", "OPENROUTER_X_OPEN_ROUTER_TITLE"),
  219. defaults.open_router_app_title,
  220. ),
  221. open_router_base_url=_env(
  222. "OPEN_ROUTER_BASE_URL",
  223. defaults.open_router_base_url,
  224. ),
  225. open_router_temperature=_env_float_optional(
  226. "OPEN_ROUTER_TEMPERATURE",
  227. defaults.open_router_temperature,
  228. ),
  229. open_router_max_tokens=_env_int_optional(
  230. "OPEN_ROUTER_MAX_TOKENS",
  231. defaults.open_router_max_tokens,
  232. ),
  233. odps_access_id=_env("ODPS_ACCESS_ID", defaults.odps_access_id),
  234. odps_access_key=_env("ODPS_ACCESS_KEY", defaults.odps_access_key),
  235. odps_project=_env("ODPS_PROJECT", defaults.odps_project),
  236. odps_endpoint=_env("ODPS_ENDPOINT", defaults.odps_endpoint),
  237. odps_tunnel_endpoint=_env(
  238. "ODPS_TUNNEL_ENDPOINT",
  239. defaults.odps_tunnel_endpoint,
  240. ),
  241. demand_pool_source_table=_env(
  242. "DEMAND_POOL_SOURCE_TABLE",
  243. defaults.demand_pool_source_table,
  244. ),
  245. demand_pool_excluded_strategy=_env(
  246. "DEMAND_POOL_EXCLUDED_STRATEGY",
  247. defaults.demand_pool_excluded_strategy,
  248. ),
  249. demand_pool_top_n=_env_int(
  250. "DEMAND_POOL_TOP_N",
  251. defaults.demand_pool_top_n,
  252. ),
  253. hot_demand_pool_strategy=_env(
  254. "HOT_DEMAND_POOL_STRATEGY",
  255. defaults.hot_demand_pool_strategy,
  256. ),
  257. wxindex_score_threshold=_env_float(
  258. "WXINDEX_SCORE_THRESHOLD",
  259. _env_float(
  260. "HOT_DEMAND_POOL_WXINDEX_THRESHOLD",
  261. _env_float(
  262. "WXINDEX_LATEST_SCORE_THRESHOLD",
  263. defaults.wxindex_score_threshold,
  264. ),
  265. ),
  266. ),
  267. odps_daily_write_limit=_env_int(
  268. "ODPS_DAILY_WRITE_LIMIT",
  269. defaults.odps_daily_write_limit,
  270. ),
  271. postprocess_batch_size=_env_int(
  272. "POSTPROCESS_BATCH_SIZE",
  273. defaults.postprocess_batch_size,
  274. ),
  275. contribution_match_llm_model=_env(
  276. "CONTRIBUTION_MATCH_LLM_MODEL",
  277. defaults.contribution_match_llm_model,
  278. ),
  279. contribution_match_llm_max_attempts=_env_int(
  280. "CONTRIBUTION_MATCH_LLM_MAX_ATTEMPTS",
  281. defaults.contribution_match_llm_max_attempts,
  282. ),
  283. contribution_match_llm_retry_sleep_seconds=_env_float(
  284. "CONTRIBUTION_MATCH_LLM_RETRY_SLEEP_SECONDS",
  285. defaults.contribution_match_llm_retry_sleep_seconds,
  286. ),
  287. contribution_match_llm_max_tokens=_env_int(
  288. "CONTRIBUTION_MATCH_LLM_MAX_TOKENS",
  289. defaults.contribution_match_llm_max_tokens,
  290. ),
  291. wxindex_llm_model=_env(
  292. "WXINDEX_LLM_MODEL",
  293. defaults.wxindex_llm_model,
  294. ),
  295. wxindex_llm_max_attempts=_env_int(
  296. "WXINDEX_LLM_MAX_ATTEMPTS",
  297. defaults.wxindex_llm_max_attempts,
  298. ),
  299. wxindex_llm_max_tokens=_env_int(
  300. "WXINDEX_LLM_MAX_TOKENS",
  301. defaults.wxindex_llm_max_tokens,
  302. ),
  303. wxindex_api_url=_env("WXINDEX_API_URL", defaults.wxindex_api_url),
  304. wxindex_lookback_days=_env_int(
  305. "WXINDEX_LOOKBACK_DAYS",
  306. defaults.wxindex_lookback_days,
  307. ),
  308. demand_event_sense_threshold=_env_float(
  309. "DEMAND_EVENT_SENSE_THRESHOLD",
  310. defaults.demand_event_sense_threshold,
  311. ),
  312. demand_senior_fit_threshold=_env_float(
  313. "DEMAND_SENIOR_FIT_THRESHOLD",
  314. defaults.demand_senior_fit_threshold,
  315. ),
  316. demand_quality_llm_model=_env(
  317. "DEMAND_QUALITY_LLM_MODEL",
  318. defaults.demand_quality_llm_model,
  319. ),
  320. demand_quality_llm_max_attempts=_env_int(
  321. "DEMAND_QUALITY_LLM_MAX_ATTEMPTS",
  322. defaults.demand_quality_llm_max_attempts,
  323. ),
  324. demand_quality_llm_retry_sleep_seconds=_env_float(
  325. "DEMAND_QUALITY_LLM_RETRY_SLEEP_SECONDS",
  326. defaults.demand_quality_llm_retry_sleep_seconds,
  327. ),
  328. demand_quality_llm_max_tokens=_env_int(
  329. "DEMAND_QUALITY_LLM_MAX_TOKENS",
  330. defaults.demand_quality_llm_max_tokens,
  331. ),
  332. category_filter_llm_model=_env(
  333. "CATEGORY_FILTER_LLM_MODEL",
  334. defaults.category_filter_llm_model,
  335. ),
  336. category_filter_llm_max_attempts=_env_int(
  337. "CATEGORY_FILTER_LLM_MAX_ATTEMPTS",
  338. defaults.category_filter_llm_max_attempts,
  339. ),
  340. category_filter_llm_retry_sleep_seconds=_env_float(
  341. "CATEGORY_FILTER_LLM_RETRY_SLEEP_SECONDS",
  342. defaults.category_filter_llm_retry_sleep_seconds,
  343. ),
  344. category_filter_llm_max_tokens=_env_int(
  345. "CATEGORY_FILTER_LLM_MAX_TOKENS",
  346. defaults.category_filter_llm_max_tokens,
  347. ),
  348. category_filter_body_max_chars=_env_int(
  349. "CATEGORY_FILTER_BODY_MAX_CHARS",
  350. defaults.category_filter_body_max_chars,
  351. ),
  352. category_filter_item_sleep_seconds=_env_float(
  353. "CATEGORY_FILTER_ITEM_SLEEP_SECONDS",
  354. defaults.category_filter_item_sleep_seconds,
  355. ),
  356. )
  357. settings = Settings.from_env()