config.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468
  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. {"source": "百度"},
  80. {"source": "微博"},
  81. {"source": "微信"},
  82. {
  83. "source": "快手",
  84. "hot_rank_base_url": "http://crawler.aiddit.com",
  85. "hot_rank_path": "/crawler/kuai_shou/hot_rank",
  86. "hot_rank_payload": {},
  87. },
  88. {
  89. "source": "抖音",
  90. "hot_rank_base_url": "http://8.217.190.241:8888",
  91. "hot_rank_path": "/crawler/dou_yin/hot_rank",
  92. "hot_rank_payload": {"tab_name": "热点榜"},
  93. },
  94. ]
  95. )
  96. mysql_host: str = "rm-t4nh1xx6o2a6vj8qu3o.mysql.singapore.rds.aliyuncs.com"
  97. mysql_port: int = 3306
  98. mysql_user: str = "content_rw"
  99. mysql_password: str = "bC1aH4bA1lB0"
  100. mysql_database: str = "external_demand"
  101. mysql_charset: str = "utf8mb4"
  102. open_router_api_key: str = "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"
  103. open_router_default_model: str = "anthropic/claude-haiku-4-5"
  104. open_router_timeout_seconds: int = 60
  105. open_router_http_referer: str = ""
  106. open_router_app_title: str = "external_demand"
  107. open_router_base_url: str = "https://openrouter.ai/api/v1"
  108. open_router_temperature: float | None = 0.7
  109. open_router_max_tokens: int | None = 20000
  110. odps_access_id: str = "LTAI9EBa0bd5PrDa"
  111. odps_access_key: str = "vAalxds7YxhfOA2yVv8GziCg3Y87v5"
  112. odps_project: str = "loghubods"
  113. odps_endpoint: str = "http://service.odps.aliyun.com/api"
  114. odps_tunnel_endpoint: str = ""
  115. demand_pool_source_table: str = "dwd_multi_demand_pool_di"
  116. demand_pool_excluded_strategy: str = "当下供需gap-分词"
  117. demand_pool_top_n: int = 200
  118. hot_demand_pool_strategy: str = "新热事件"
  119. wxindex_score_threshold: float = 100_000.0
  120. odps_daily_write_limit: int = 500
  121. postprocess_batch_size: int = 20
  122. contribution_match_llm_model: str = ""
  123. contribution_match_llm_max_attempts: int = 3
  124. contribution_match_llm_retry_sleep_seconds: float = 1.0
  125. contribution_match_llm_max_tokens: int = 4000
  126. wxindex_llm_model: str = "anthropic/claude-haiku-4-5"
  127. wxindex_llm_max_attempts: int = 3
  128. wxindex_llm_max_tokens: int = 4000
  129. wxindex_api_url: str = "http://crawapi.piaoquantv.com/crawler/wei_xin/wxindex"
  130. wxindex_lookback_days: int = 7
  131. wxindex_words_cron_hours: str = "10,14"
  132. wxindex_words_cron_minute: int = 0
  133. wxindex_heat_pattern_cron_hours: str = "11,15"
  134. wxindex_heat_pattern_cron_minute: int = 0
  135. demand_event_sense_threshold: float = 6.0
  136. demand_senior_fit_threshold: float = 6.0
  137. demand_quality_llm_model: str = "anthropic/claude-haiku-4-5"
  138. demand_quality_llm_max_attempts: int = 3
  139. demand_quality_llm_retry_sleep_seconds: float = 1.0
  140. demand_quality_llm_max_tokens: int = 4000
  141. category_filter_llm_model: str = "anthropic/claude-haiku-4-5"
  142. category_filter_llm_max_attempts: int = 3
  143. category_filter_llm_retry_sleep_seconds: float = 1.0
  144. category_filter_llm_max_tokens: int = 1024
  145. category_filter_body_max_chars: int = 2000
  146. category_filter_item_sleep_seconds: float = 0.0
  147. festival_demand_cron_hours: str = "7"
  148. festival_demand_cron_minute: int = 0
  149. festival_demand_llm_model: str = "anthropic/claude-haiku-4-5"
  150. festival_demand_llm_max_attempts: int = 3
  151. festival_demand_llm_retry_sleep_seconds: float = 1.0
  152. festival_demand_llm_max_tokens: int = 4000
  153. festival_demand_llm_temperature: float = 0.0
  154. festival_demand_pool_strategy: str = "去年同期阳历"
  155. festival_demand_output_strategy: str = "去年同期阳历-节点事件"
  156. festival_demand_match_batch_size: int = 200
  157. @classmethod
  158. def from_env(cls) -> "Settings":
  159. defaults = cls()
  160. return cls(
  161. crawapi_base_url=_env("CRAWAPI_BASE_URL", defaults.crawapi_base_url),
  162. crawapi_hot_content_rank_path=_env(
  163. "CRAWAPI_HOT_CONTENT_RANK_PATH",
  164. defaults.crawapi_hot_content_rank_path,
  165. ),
  166. crawapi_keyword_search_path=_env(
  167. "CRAWAPI_KEYWORD_SEARCH_PATH",
  168. defaults.crawapi_keyword_search_path,
  169. ),
  170. decode_api_url=_env("DECODE_API_URL", defaults.decode_api_url),
  171. decode_result_api_url=_env(
  172. "DECODE_RESULT_API_URL",
  173. defaults.decode_result_api_url,
  174. ),
  175. decode_config_id=_env_int("DECODE_CONFIG_ID", defaults.decode_config_id),
  176. request_timeout_seconds=_env_int(
  177. "REQUEST_TIMEOUT_SECONDS",
  178. defaults.request_timeout_seconds,
  179. ),
  180. https_verify_ssl=_env_bool("HTTPS_VERIFY_SSL", defaults.https_verify_ssl),
  181. hot_flow_cron_hours=_env(
  182. "HOT_FLOW_CRON_HOURS",
  183. defaults.hot_flow_cron_hours,
  184. ),
  185. hot_flow_cron_minute=_env_int(
  186. "HOT_FLOW_CRON_MINUTE",
  187. defaults.hot_flow_cron_minute,
  188. ),
  189. hot_flow_interval_seconds=_env_int(
  190. "HOT_FLOW_INTERVAL_SECONDS",
  191. defaults.hot_flow_interval_seconds,
  192. ),
  193. decode_result_flow_interval_seconds=_env_int(
  194. "DECODE_RESULT_FLOW_INTERVAL_SECONDS",
  195. defaults.decode_result_flow_interval_seconds,
  196. ),
  197. decode_result_batch_size=_env_int(
  198. "DECODE_RESULT_BATCH_SIZE",
  199. defaults.decode_result_batch_size,
  200. ),
  201. contribution_score_threshold=float(
  202. _env(
  203. "CONTRIBUTION_SCORE_THRESHOLD",
  204. str(defaults.contribution_score_threshold),
  205. )
  206. ),
  207. hot_flow_sources=_env_json(
  208. "HOT_FLOW_SOURCES_JSON",
  209. defaults.hot_flow_sources,
  210. "HOT_FLOW_SOURCES_FILE",
  211. ),
  212. mysql_host=_env("MYSQL_HOST", defaults.mysql_host),
  213. mysql_port=_env_int("MYSQL_PORT", defaults.mysql_port),
  214. mysql_user=_env("MYSQL_USER", defaults.mysql_user),
  215. mysql_password=_env("MYSQL_PASSWORD", defaults.mysql_password),
  216. mysql_database=_env("MYSQL_DATABASE", defaults.mysql_database),
  217. mysql_charset=_env("MYSQL_CHARSET", defaults.mysql_charset),
  218. open_router_api_key=_env_first(
  219. ("OPEN_ROUTER_API_KEY", "OPENROUTER_API_KEY"),
  220. defaults.open_router_api_key,
  221. ),
  222. open_router_default_model=_env(
  223. "OPEN_ROUTER_DEFAULT_MODEL",
  224. defaults.open_router_default_model,
  225. ),
  226. open_router_timeout_seconds=_env_int(
  227. "OPEN_ROUTER_TIMEOUT_SECONDS",
  228. defaults.open_router_timeout_seconds,
  229. ),
  230. open_router_http_referer=_env_first(
  231. ("OPEN_ROUTER_HTTP_REFERER", "OPENROUTER_HTTP_REFERER"),
  232. defaults.open_router_http_referer,
  233. ),
  234. open_router_app_title=_env_first(
  235. ("OPEN_ROUTER_APP_TITLE", "OPENROUTER_X_OPEN_ROUTER_TITLE"),
  236. defaults.open_router_app_title,
  237. ),
  238. open_router_base_url=_env(
  239. "OPEN_ROUTER_BASE_URL",
  240. defaults.open_router_base_url,
  241. ),
  242. open_router_temperature=_env_float_optional(
  243. "OPEN_ROUTER_TEMPERATURE",
  244. defaults.open_router_temperature,
  245. ),
  246. open_router_max_tokens=_env_int_optional(
  247. "OPEN_ROUTER_MAX_TOKENS",
  248. defaults.open_router_max_tokens,
  249. ),
  250. odps_access_id=_env("ODPS_ACCESS_ID", defaults.odps_access_id),
  251. odps_access_key=_env("ODPS_ACCESS_KEY", defaults.odps_access_key),
  252. odps_project=_env("ODPS_PROJECT", defaults.odps_project),
  253. odps_endpoint=_env("ODPS_ENDPOINT", defaults.odps_endpoint),
  254. odps_tunnel_endpoint=_env(
  255. "ODPS_TUNNEL_ENDPOINT",
  256. defaults.odps_tunnel_endpoint,
  257. ),
  258. demand_pool_source_table=_env(
  259. "DEMAND_POOL_SOURCE_TABLE",
  260. defaults.demand_pool_source_table,
  261. ),
  262. demand_pool_excluded_strategy=_env(
  263. "DEMAND_POOL_EXCLUDED_STRATEGY",
  264. defaults.demand_pool_excluded_strategy,
  265. ),
  266. demand_pool_top_n=_env_int(
  267. "DEMAND_POOL_TOP_N",
  268. defaults.demand_pool_top_n,
  269. ),
  270. hot_demand_pool_strategy=_env(
  271. "HOT_DEMAND_POOL_STRATEGY",
  272. defaults.hot_demand_pool_strategy,
  273. ),
  274. wxindex_score_threshold=_env_float(
  275. "WXINDEX_SCORE_THRESHOLD",
  276. _env_float(
  277. "HOT_DEMAND_POOL_WXINDEX_THRESHOLD",
  278. _env_float(
  279. "WXINDEX_LATEST_SCORE_THRESHOLD",
  280. defaults.wxindex_score_threshold,
  281. ),
  282. ),
  283. ),
  284. odps_daily_write_limit=_env_int(
  285. "ODPS_DAILY_WRITE_LIMIT",
  286. defaults.odps_daily_write_limit,
  287. ),
  288. postprocess_batch_size=_env_int(
  289. "POSTPROCESS_BATCH_SIZE",
  290. defaults.postprocess_batch_size,
  291. ),
  292. contribution_match_llm_model=_env(
  293. "CONTRIBUTION_MATCH_LLM_MODEL",
  294. defaults.contribution_match_llm_model,
  295. ),
  296. contribution_match_llm_max_attempts=_env_int(
  297. "CONTRIBUTION_MATCH_LLM_MAX_ATTEMPTS",
  298. defaults.contribution_match_llm_max_attempts,
  299. ),
  300. contribution_match_llm_retry_sleep_seconds=_env_float(
  301. "CONTRIBUTION_MATCH_LLM_RETRY_SLEEP_SECONDS",
  302. defaults.contribution_match_llm_retry_sleep_seconds,
  303. ),
  304. contribution_match_llm_max_tokens=_env_int(
  305. "CONTRIBUTION_MATCH_LLM_MAX_TOKENS",
  306. defaults.contribution_match_llm_max_tokens,
  307. ),
  308. wxindex_llm_model=_env(
  309. "WXINDEX_LLM_MODEL",
  310. defaults.wxindex_llm_model,
  311. ),
  312. wxindex_llm_max_attempts=_env_int(
  313. "WXINDEX_LLM_MAX_ATTEMPTS",
  314. defaults.wxindex_llm_max_attempts,
  315. ),
  316. wxindex_llm_max_tokens=_env_int(
  317. "WXINDEX_LLM_MAX_TOKENS",
  318. defaults.wxindex_llm_max_tokens,
  319. ),
  320. wxindex_api_url=_env("WXINDEX_API_URL", defaults.wxindex_api_url),
  321. wxindex_lookback_days=_env_int(
  322. "WXINDEX_LOOKBACK_DAYS",
  323. defaults.wxindex_lookback_days,
  324. ),
  325. wxindex_words_cron_hours=_env(
  326. "WXINDEX_WORDS_CRON_HOURS",
  327. defaults.wxindex_words_cron_hours,
  328. ),
  329. wxindex_words_cron_minute=_env_int(
  330. "WXINDEX_WORDS_CRON_MINUTE",
  331. defaults.wxindex_words_cron_minute,
  332. ),
  333. wxindex_heat_pattern_cron_hours=_env(
  334. "WXINDEX_HEAT_PATTERN_CRON_HOURS",
  335. defaults.wxindex_heat_pattern_cron_hours,
  336. ),
  337. wxindex_heat_pattern_cron_minute=_env_int(
  338. "WXINDEX_HEAT_PATTERN_CRON_MINUTE",
  339. defaults.wxindex_heat_pattern_cron_minute,
  340. ),
  341. demand_event_sense_threshold=_env_float(
  342. "DEMAND_EVENT_SENSE_THRESHOLD",
  343. defaults.demand_event_sense_threshold,
  344. ),
  345. demand_senior_fit_threshold=_env_float(
  346. "DEMAND_SENIOR_FIT_THRESHOLD",
  347. defaults.demand_senior_fit_threshold,
  348. ),
  349. demand_quality_llm_model=_env(
  350. "DEMAND_QUALITY_LLM_MODEL",
  351. defaults.demand_quality_llm_model,
  352. ),
  353. demand_quality_llm_max_attempts=_env_int(
  354. "DEMAND_QUALITY_LLM_MAX_ATTEMPTS",
  355. defaults.demand_quality_llm_max_attempts,
  356. ),
  357. demand_quality_llm_retry_sleep_seconds=_env_float(
  358. "DEMAND_QUALITY_LLM_RETRY_SLEEP_SECONDS",
  359. defaults.demand_quality_llm_retry_sleep_seconds,
  360. ),
  361. demand_quality_llm_max_tokens=_env_int(
  362. "DEMAND_QUALITY_LLM_MAX_TOKENS",
  363. defaults.demand_quality_llm_max_tokens,
  364. ),
  365. category_filter_llm_model=_env(
  366. "CATEGORY_FILTER_LLM_MODEL",
  367. defaults.category_filter_llm_model,
  368. ),
  369. category_filter_llm_max_attempts=_env_int(
  370. "CATEGORY_FILTER_LLM_MAX_ATTEMPTS",
  371. defaults.category_filter_llm_max_attempts,
  372. ),
  373. category_filter_llm_retry_sleep_seconds=_env_float(
  374. "CATEGORY_FILTER_LLM_RETRY_SLEEP_SECONDS",
  375. defaults.category_filter_llm_retry_sleep_seconds,
  376. ),
  377. category_filter_llm_max_tokens=_env_int(
  378. "CATEGORY_FILTER_LLM_MAX_TOKENS",
  379. defaults.category_filter_llm_max_tokens,
  380. ),
  381. category_filter_body_max_chars=_env_int(
  382. "CATEGORY_FILTER_BODY_MAX_CHARS",
  383. defaults.category_filter_body_max_chars,
  384. ),
  385. category_filter_item_sleep_seconds=_env_float(
  386. "CATEGORY_FILTER_ITEM_SLEEP_SECONDS",
  387. defaults.category_filter_item_sleep_seconds,
  388. ),
  389. festival_demand_cron_hours=_env(
  390. "FESTIVAL_DEMAND_CRON_HOURS",
  391. defaults.festival_demand_cron_hours,
  392. ),
  393. festival_demand_cron_minute=_env_int(
  394. "FESTIVAL_DEMAND_CRON_MINUTE",
  395. defaults.festival_demand_cron_minute,
  396. ),
  397. festival_demand_llm_model=_env(
  398. "FESTIVAL_DEMAND_LLM_MODEL",
  399. defaults.festival_demand_llm_model,
  400. ),
  401. festival_demand_llm_max_attempts=_env_int(
  402. "FESTIVAL_DEMAND_LLM_MAX_ATTEMPTS",
  403. defaults.festival_demand_llm_max_attempts,
  404. ),
  405. festival_demand_llm_retry_sleep_seconds=_env_float(
  406. "FESTIVAL_DEMAND_LLM_RETRY_SLEEP_SECONDS",
  407. defaults.festival_demand_llm_retry_sleep_seconds,
  408. ),
  409. festival_demand_llm_max_tokens=_env_int(
  410. "FESTIVAL_DEMAND_LLM_MAX_TOKENS",
  411. defaults.festival_demand_llm_max_tokens,
  412. ),
  413. festival_demand_llm_temperature=_env_float(
  414. "FESTIVAL_DEMAND_LLM_TEMPERATURE",
  415. defaults.festival_demand_llm_temperature,
  416. ),
  417. festival_demand_pool_strategy=_env(
  418. "FESTIVAL_DEMAND_POOL_STRATEGY",
  419. defaults.festival_demand_pool_strategy,
  420. ),
  421. festival_demand_output_strategy=_env(
  422. "FESTIVAL_DEMAND_OUTPUT_STRATEGY",
  423. defaults.festival_demand_output_strategy,
  424. ),
  425. festival_demand_match_batch_size=_env_int(
  426. "FESTIVAL_DEMAND_MATCH_BATCH_SIZE",
  427. defaults.festival_demand_match_batch_size,
  428. ),
  429. )
  430. settings = Settings.from_env()