config.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328
  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 = "近期热点"
  116. wxindex_score_threshold: float = 1_000_000.0
  117. postprocess_batch_size: int = 20
  118. contribution_match_llm_model: str = ""
  119. contribution_match_llm_max_attempts: int = 3
  120. contribution_match_llm_retry_sleep_seconds: float = 1.0
  121. contribution_match_llm_max_tokens: int = 4000
  122. wxindex_llm_model: str = ""
  123. wxindex_llm_max_attempts: int = 3
  124. wxindex_llm_max_tokens: int = 1200
  125. wxindex_api_url: str = "http://crawapi.piaoquantv.com/crawler/wei_xin/wxindex"
  126. wxindex_lookback_days: int = 7
  127. @classmethod
  128. def from_env(cls) -> "Settings":
  129. defaults = cls()
  130. return cls(
  131. crawapi_base_url=_env("CRAWAPI_BASE_URL", defaults.crawapi_base_url),
  132. crawapi_hot_content_rank_path=_env(
  133. "CRAWAPI_HOT_CONTENT_RANK_PATH",
  134. defaults.crawapi_hot_content_rank_path,
  135. ),
  136. crawapi_keyword_search_path=_env(
  137. "CRAWAPI_KEYWORD_SEARCH_PATH",
  138. defaults.crawapi_keyword_search_path,
  139. ),
  140. decode_api_url=_env("DECODE_API_URL", defaults.decode_api_url),
  141. decode_result_api_url=_env(
  142. "DECODE_RESULT_API_URL",
  143. defaults.decode_result_api_url,
  144. ),
  145. decode_config_id=_env_int("DECODE_CONFIG_ID", defaults.decode_config_id),
  146. request_timeout_seconds=_env_int(
  147. "REQUEST_TIMEOUT_SECONDS",
  148. defaults.request_timeout_seconds,
  149. ),
  150. https_verify_ssl=_env_bool("HTTPS_VERIFY_SSL", defaults.https_verify_ssl),
  151. hot_flow_cron_hours=_env(
  152. "HOT_FLOW_CRON_HOURS",
  153. defaults.hot_flow_cron_hours,
  154. ),
  155. hot_flow_cron_minute=_env_int(
  156. "HOT_FLOW_CRON_MINUTE",
  157. defaults.hot_flow_cron_minute,
  158. ),
  159. hot_flow_interval_seconds=_env_int(
  160. "HOT_FLOW_INTERVAL_SECONDS",
  161. defaults.hot_flow_interval_seconds,
  162. ),
  163. decode_result_flow_interval_seconds=_env_int(
  164. "DECODE_RESULT_FLOW_INTERVAL_SECONDS",
  165. defaults.decode_result_flow_interval_seconds,
  166. ),
  167. decode_result_batch_size=_env_int(
  168. "DECODE_RESULT_BATCH_SIZE",
  169. defaults.decode_result_batch_size,
  170. ),
  171. contribution_score_threshold=float(
  172. _env(
  173. "CONTRIBUTION_SCORE_THRESHOLD",
  174. str(defaults.contribution_score_threshold),
  175. )
  176. ),
  177. hot_flow_sources=_env_json(
  178. "HOT_FLOW_SOURCES_JSON",
  179. defaults.hot_flow_sources,
  180. "HOT_FLOW_SOURCES_FILE",
  181. ),
  182. mysql_host=_env("MYSQL_HOST", defaults.mysql_host),
  183. mysql_port=_env_int("MYSQL_PORT", defaults.mysql_port),
  184. mysql_user=_env("MYSQL_USER", defaults.mysql_user),
  185. mysql_password=_env("MYSQL_PASSWORD", defaults.mysql_password),
  186. mysql_database=_env("MYSQL_DATABASE", defaults.mysql_database),
  187. mysql_charset=_env("MYSQL_CHARSET", defaults.mysql_charset),
  188. open_router_api_key=_env_first(
  189. ("OPEN_ROUTER_API_KEY", "OPENROUTER_API_KEY"),
  190. defaults.open_router_api_key,
  191. ),
  192. open_router_default_model=_env(
  193. "OPEN_ROUTER_DEFAULT_MODEL",
  194. defaults.open_router_default_model,
  195. ),
  196. open_router_timeout_seconds=_env_int(
  197. "OPEN_ROUTER_TIMEOUT_SECONDS",
  198. defaults.open_router_timeout_seconds,
  199. ),
  200. open_router_http_referer=_env_first(
  201. ("OPEN_ROUTER_HTTP_REFERER", "OPENROUTER_HTTP_REFERER"),
  202. defaults.open_router_http_referer,
  203. ),
  204. open_router_app_title=_env_first(
  205. ("OPEN_ROUTER_APP_TITLE", "OPENROUTER_X_OPEN_ROUTER_TITLE"),
  206. defaults.open_router_app_title,
  207. ),
  208. open_router_base_url=_env(
  209. "OPEN_ROUTER_BASE_URL",
  210. defaults.open_router_base_url,
  211. ),
  212. open_router_temperature=_env_float_optional(
  213. "OPEN_ROUTER_TEMPERATURE",
  214. defaults.open_router_temperature,
  215. ),
  216. open_router_max_tokens=_env_int_optional(
  217. "OPEN_ROUTER_MAX_TOKENS",
  218. defaults.open_router_max_tokens,
  219. ),
  220. odps_access_id=_env("ODPS_ACCESS_ID", defaults.odps_access_id),
  221. odps_access_key=_env("ODPS_ACCESS_KEY", defaults.odps_access_key),
  222. odps_project=_env("ODPS_PROJECT", defaults.odps_project),
  223. odps_endpoint=_env("ODPS_ENDPOINT", defaults.odps_endpoint),
  224. odps_tunnel_endpoint=_env(
  225. "ODPS_TUNNEL_ENDPOINT",
  226. defaults.odps_tunnel_endpoint,
  227. ),
  228. demand_pool_source_table=_env(
  229. "DEMAND_POOL_SOURCE_TABLE",
  230. defaults.demand_pool_source_table,
  231. ),
  232. demand_pool_excluded_strategy=_env(
  233. "DEMAND_POOL_EXCLUDED_STRATEGY",
  234. defaults.demand_pool_excluded_strategy,
  235. ),
  236. demand_pool_top_n=_env_int(
  237. "DEMAND_POOL_TOP_N",
  238. defaults.demand_pool_top_n,
  239. ),
  240. hot_demand_pool_strategy=_env(
  241. "HOT_DEMAND_POOL_STRATEGY",
  242. defaults.hot_demand_pool_strategy,
  243. ),
  244. wxindex_score_threshold=_env_float(
  245. "WXINDEX_SCORE_THRESHOLD",
  246. _env_float(
  247. "HOT_DEMAND_POOL_WXINDEX_THRESHOLD",
  248. _env_float(
  249. "WXINDEX_LATEST_SCORE_THRESHOLD",
  250. defaults.wxindex_score_threshold,
  251. ),
  252. ),
  253. ),
  254. postprocess_batch_size=_env_int(
  255. "POSTPROCESS_BATCH_SIZE",
  256. defaults.postprocess_batch_size,
  257. ),
  258. contribution_match_llm_model=_env(
  259. "CONTRIBUTION_MATCH_LLM_MODEL",
  260. defaults.contribution_match_llm_model,
  261. ),
  262. contribution_match_llm_max_attempts=_env_int(
  263. "CONTRIBUTION_MATCH_LLM_MAX_ATTEMPTS",
  264. defaults.contribution_match_llm_max_attempts,
  265. ),
  266. contribution_match_llm_retry_sleep_seconds=_env_float(
  267. "CONTRIBUTION_MATCH_LLM_RETRY_SLEEP_SECONDS",
  268. defaults.contribution_match_llm_retry_sleep_seconds,
  269. ),
  270. contribution_match_llm_max_tokens=_env_int(
  271. "CONTRIBUTION_MATCH_LLM_MAX_TOKENS",
  272. defaults.contribution_match_llm_max_tokens,
  273. ),
  274. wxindex_llm_model=_env(
  275. "WXINDEX_LLM_MODEL",
  276. defaults.wxindex_llm_model,
  277. ),
  278. wxindex_llm_max_attempts=_env_int(
  279. "WXINDEX_LLM_MAX_ATTEMPTS",
  280. defaults.wxindex_llm_max_attempts,
  281. ),
  282. wxindex_llm_max_tokens=_env_int(
  283. "WXINDEX_LLM_MAX_TOKENS",
  284. defaults.wxindex_llm_max_tokens,
  285. ),
  286. wxindex_api_url=_env("WXINDEX_API_URL", defaults.wxindex_api_url),
  287. wxindex_lookback_days=_env_int(
  288. "WXINDEX_LOOKBACK_DAYS",
  289. defaults.wxindex_lookback_days,
  290. ),
  291. )
  292. settings = Settings.from_env()