create_decode_tasks.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359
  1. import json
  2. from collections import defaultdict
  3. from typing import Dict, List
  4. from app.core.database import DatabaseManager
  5. from app.core.observability import LogService
  6. from ._const import DecodeCardConst
  7. from ._mapper import CardDecodeTaskMapper
  8. from ._utils import CardDecodeUtils
  9. class CreateCardsDecodeTask(DecodeCardConst):
  10. def __init__(self, pool: DatabaseManager, log_service: LogService):
  11. self.pool = pool
  12. self.log_service = log_service
  13. self.mapper = CardDecodeTaskMapper(self.pool)
  14. self.tool = CardDecodeUtils()
  15. async def _acquire_cards(self) -> List[Dict]:
  16. """获取待解构卡片并加锁(status INIT → PROCESSING)"""
  17. cards = await self.mapper.fetch_cards()
  18. locked = []
  19. for card in cards:
  20. card_id = card["id"]
  21. acquired = await self.mapper.update_card_status(
  22. card_id, self.TaskStatus.INIT, self.TaskStatus.PROCESSING
  23. )
  24. if acquired:
  25. locked.append(card)
  26. else:
  27. await self.log_service.log(
  28. contents={
  29. "card_id": card_id,
  30. "task": "create_card_decode_task",
  31. "status": "skip",
  32. "message": "acquire lock failed",
  33. }
  34. )
  35. return locked
  36. async def _ensure_cover_ids(self, cards: List[Dict]):
  37. """为缺少 card_cover_id 的卡片下载封面并计算 MD5"""
  38. for card in cards:
  39. if card.get("card_cover_id"):
  40. continue
  41. if not card.get("share_cover"):
  42. continue # 空封面留给 _submit_and_record 标记 FAILED
  43. try:
  44. cover_id = self.tool.compute_cover_md5(card["share_cover"])
  45. await self.mapper.set_card_cover_id(card["id"], cover_id)
  46. card["card_cover_id"] = cover_id
  47. await self.log_service.log(
  48. contents={
  49. "card_id": card["id"],
  50. "task": "create_card_decode_task",
  51. "status": "info",
  52. "message": f"computed cover_id={cover_id}",
  53. }
  54. )
  55. except Exception as e:
  56. await self.mapper.update_card_status(
  57. card["id"],
  58. self.TaskStatus.PROCESSING,
  59. self.TaskStatus.FAILED,
  60. )
  61. await self.log_service.log(
  62. contents={
  63. "card_id": card["id"],
  64. "share_cover": card.get("share_cover"),
  65. "task": "create_card_decode_task",
  66. "status": "fail",
  67. "message": f"compute cover md5 failed, marked as FAILED: {e}",
  68. }
  69. )
  70. @staticmethod
  71. def _group_cards_by_channel(cards: List[Dict]) -> Dict[int, List[Dict]]:
  72. """按 auto_reply_top_cards_daily.channel 分组,映射到对应的 config_id"""
  73. grouped = defaultdict(list)
  74. for card in cards:
  75. source_channel = card.get("channel", "")
  76. config_id = DecodeCardConst.CHANNEL_CONFIG_MAP.get(source_channel)
  77. if config_id:
  78. grouped[config_id].append(card)
  79. return dict(grouped)
  80. async def _submit_and_record_for_config(
  81. self, cards: List[Dict], config_id: int, source: int
  82. ):
  83. """对同一 config_id 的卡片执行提交与落库"""
  84. if not cards:
  85. return
  86. # 过滤已有任务记录的卡片(按 card_cover_id 去重)
  87. cards_with_cid = [c for c in cards if c.get("card_cover_id")]
  88. if not cards_with_cid:
  89. return
  90. # 跨批次去重:查 DB 已有任务
  91. all_source_ids = [str(c["card_cover_id"]) for c in cards_with_cid]
  92. existing = await self.mapper.fetch_existing_source_ids(
  93. all_source_ids, config_id, source
  94. )
  95. # 同批次去重:相同 card_cover_id 只保留第一条
  96. seen = set()
  97. deduped_cards = []
  98. intra_dups = []
  99. for c in cards_with_cid:
  100. cid = str(c["card_cover_id"])
  101. if cid in existing or cid in seen:
  102. intra_dups.append(c)
  103. else:
  104. seen.add(cid)
  105. deduped_cards.append(c)
  106. if intra_dups:
  107. await self.log_service.log(
  108. contents={
  109. "task": "create_card_decode_task",
  110. "config_id": config_id,
  111. "message": f"Skipped {len(intra_dups)} duplicate cards (already submitted or same-batch)",
  112. }
  113. )
  114. for card in intra_dups:
  115. await self.mapper.update_card_status(
  116. card["id"],
  117. self.TaskStatus.PROCESSING,
  118. self.TaskStatus.SUCCESS,
  119. )
  120. new_cards = deduped_cards
  121. if not new_cards:
  122. return
  123. posts = self.tool.prepare_posts(new_cards)
  124. submit_results = await self.tool.submit_decode_batch(posts, config_id=config_id)
  125. posts_by_cid = {p["channelContentId"]: p for p in posts}
  126. for card in new_cards:
  127. source_id = str(card["card_cover_id"])
  128. card_id = card["id"]
  129. result = submit_results.get(source_id)
  130. if not result:
  131. await self.mapper.update_card_status(
  132. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.INIT
  133. )
  134. await self.log_service.log(
  135. contents={
  136. "card_id": card_id,
  137. "source_id": source_id,
  138. "config_id": config_id,
  139. "task": "create_card_decode_task",
  140. "status": "fail",
  141. "message": "no response for source_id, rolled back to INIT",
  142. }
  143. )
  144. continue
  145. status = result.get("status")
  146. if status == self.SubmitStatus.FAILED:
  147. await self.mapper.update_card_status(
  148. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.INIT
  149. )
  150. await self.log_service.log(
  151. contents={
  152. "card_id": card_id,
  153. "source_id": source_id,
  154. "config_id": config_id,
  155. "task": "create_card_decode_task",
  156. "status": "fail",
  157. "data": result,
  158. }
  159. )
  160. continue
  161. if status == self.SubmitStatus.SUCCESS:
  162. query_results = await self.tool.query_decode_results_batch(
  163. [source_id], config_id=config_id
  164. )
  165. result_data = query_results.get(source_id)
  166. if (
  167. result_data
  168. and result_data.get("status") == self.QueryStatus.SUCCESS
  169. ):
  170. data_content = result_data.get("dataContent") or "{}"
  171. html = result_data.get("html")
  172. await self.mapper.insert_decode_task(
  173. source_id=source_id,
  174. config_id=config_id,
  175. source=source,
  176. payload=json.dumps(
  177. posts_by_cid.get(source_id, {}), ensure_ascii=False
  178. ),
  179. remark="提交时已有解构结果,直接落库",
  180. )
  181. await self.mapper.set_decode_result(
  182. source_id=source_id,
  183. config_id=config_id,
  184. result=json.dumps(
  185. {"dataContent": data_content, "html": html},
  186. ensure_ascii=False,
  187. ),
  188. remark="提交时已返回 SUCCESS,结果已落库",
  189. )
  190. await self.mapper.update_card_status(
  191. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.SUCCESS
  192. )
  193. await self.log_service.log(
  194. contents={
  195. "card_id": card_id,
  196. "source_id": source_id,
  197. "config_id": config_id,
  198. "task": "create_card_decode_task",
  199. "status": "success",
  200. "message": "decode result already available on submit",
  201. }
  202. )
  203. else:
  204. await self.mapper.insert_decode_task(
  205. source_id=source_id,
  206. config_id=config_id,
  207. source=source,
  208. payload=json.dumps(
  209. posts_by_cid.get(source_id, {}), ensure_ascii=False
  210. ),
  211. remark="提交返回SUCCESS,查询未果,等待轮询",
  212. status=self.TaskStatus.PROCESSING,
  213. )
  214. await self.mapper.update_card_status(
  215. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.SUCCESS
  216. )
  217. await self.log_service.log(
  218. contents={
  219. "card_id": card_id,
  220. "source_id": source_id,
  221. "config_id": config_id,
  222. "task": "create_card_decode_task",
  223. "status": "pending",
  224. "message": "submit SUCCESS but query not ready, inserted for polling",
  225. }
  226. )
  227. elif status == self.SubmitStatus.PENDING:
  228. await self.mapper.insert_decode_task(
  229. source_id=source_id,
  230. config_id=config_id,
  231. source=source,
  232. payload=json.dumps(
  233. posts_by_cid.get(source_id, {}), ensure_ascii=False
  234. ),
  235. remark="卡片解构任务已提交,等待轮询",
  236. status=self.TaskStatus.PROCESSING,
  237. )
  238. await self.mapper.update_card_status(
  239. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.SUCCESS
  240. )
  241. await self.log_service.log(
  242. contents={
  243. "card_id": card_id,
  244. "source_id": source_id,
  245. "config_id": config_id,
  246. "task": "create_card_decode_task",
  247. "status": "pending",
  248. "message": "task submitted, waiting for polling",
  249. }
  250. )
  251. else:
  252. await self.mapper.update_card_status(
  253. card_id, self.TaskStatus.PROCESSING, self.TaskStatus.INIT
  254. )
  255. await self.log_service.log(
  256. contents={
  257. "card_id": card_id,
  258. "source_id": source_id,
  259. "config_id": config_id,
  260. "task": "create_card_decode_task",
  261. "status": "fail",
  262. "message": f"unexpected submit status: {status}, rolled back to INIT",
  263. "data": result,
  264. }
  265. )
  266. async def _submit_and_record(self, cards: List[Dict]):
  267. if not cards:
  268. return
  269. # 无 share_cover 的卡片直接标记失败
  270. valid_cards = []
  271. for card in cards:
  272. if not card.get("share_cover"):
  273. await self.mapper.update_card_status(
  274. card["id"],
  275. self.TaskStatus.PROCESSING,
  276. self.TaskStatus.FAILED,
  277. )
  278. await self.log_service.log(
  279. contents={
  280. "card_id": card["id"],
  281. "task": "create_card_decode_task",
  282. "status": "fail",
  283. "message": "share_cover is empty, marked as FAILED",
  284. }
  285. )
  286. else:
  287. valid_cards.append(card)
  288. if not valid_cards:
  289. return
  290. grouped = self._group_cards_by_channel(valid_cards)
  291. for config_id, group_cards in grouped.items():
  292. source_channel = group_cards[0]["channel"]
  293. source = self.CHANNEL_SOURCE_MAP[source_channel]
  294. await self._submit_and_record_for_config(group_cards, config_id, source)
  295. # 处理不在映射表中的卡片(回滚状态)
  296. mapped_ids = {id(c) for g in grouped.values() for c in g}
  297. for card in valid_cards:
  298. if id(card) not in mapped_ids:
  299. await self.mapper.update_card_status(
  300. card["id"],
  301. self.TaskStatus.PROCESSING,
  302. self.TaskStatus.INIT,
  303. )
  304. await self.log_service.log(
  305. contents={
  306. "card_id": card["id"],
  307. "channel": card.get("channel"),
  308. "task": "create_card_decode_task",
  309. "status": "fail",
  310. "message": "unknown channel, rolled back to INIT",
  311. }
  312. )
  313. async def deal(self):
  314. cards = await self._acquire_cards()
  315. if not cards:
  316. await self.log_service.log(
  317. contents={
  318. "task": "create_card_decode_task",
  319. "message": "No more cards to decode",
  320. }
  321. )
  322. return
  323. await self._ensure_cover_ids(cards)
  324. await self._submit_and_record(cards)
  325. await self.log_service.log(
  326. contents={
  327. "task": "create_card_decode_task",
  328. "message": f"Processed {len(cards)} cards",
  329. }
  330. )
  331. __all__ = ["CreateCardsDecodeTask"]