upload_workflow.py 29 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645
  1. """Logout and upload workflows without HTTP response objects (stage 5)."""
  2. from __future__ import annotations
  3. import hashlib
  4. import tempfile
  5. from dataclasses import dataclass
  6. from pathlib import Path
  7. from typing import Any
  8. from starlette.concurrency import run_in_threadpool
  9. from fithealth_agent.runtime import deps
  10. from fithealth_agent.runtime.deps import (
  11. _sign_analysis_confidence,
  12. logger,
  13. )
  14. from fithealth_agent.runtime.upload_io import (
  15. FIT_FILE_MAX_BYTES, HEALTH_CSV_MAX_BYTES, HEALTH_UPLOAD_MAX_FILES,
  16. HEALTH_ZIP_MAX_BYTES, PLAN_FILE_CONFIRM_BYTES, PLAN_FILE_MAX_BYTES,
  17. read_upload_with_limit,
  18. )
  19. from fithealth_agent.food_analysis import FoodAnalysisError, MAX_IMAGE_BYTES
  20. from fithealth_agent.health_importer import HealthImportError, extract_activity_fit
  21. from fithealth_agent.info_store import memory_retention_class, normalize_memory_facts
  22. from fithealth_agent.plan_classifier import validate_training_plan
  23. from fithealth_agent.domain.plan_validation import infer_training_subject
  24. from fithealth_agent import workout_store
  25. @dataclass(frozen=True)
  26. class WorkflowResult:
  27. body: dict[str, object]
  28. status_code: int = 200
  29. def result(*args: object, content: dict[str, object] | None = None, status_code: int = 200, **kwargs: object) -> WorkflowResult:
  30. body = content if content is not None else (args[0] if args else {})
  31. if not isinstance(body, dict):
  32. body = {"detail": body}
  33. return WorkflowResult(body=body, status_code=status_code)
  34. async def analyze_food(file: Any, context: str = "") -> WorkflowResult:
  35. if not deps.external_model_settings_store.get()["external_models_enabled"]:
  36. return result(
  37. {"error": "外部模型已关闭,餐盘照片不会被发送到视觉服务。"},
  38. status_code=503,
  39. )
  40. content = await read_upload_with_limit(file, MAX_IMAGE_BYTES)
  41. if content is None:
  42. return result({"error": "餐盘照片不能超过 10 MiB"}, status_code=413)
  43. try:
  44. analysis = await run_in_threadpool(deps.analyze_food_image, content, file.content_type or "", context)
  45. except FoodAnalysisError as exc:
  46. return result({"error": str(exc)}, status_code=400)
  47. except Exception as exc: # noqa: BLE001
  48. # 将视觉服务和响应解析异常转换成前端可展示的统一错误。
  49. logger.exception("餐盘照片分析失败")
  50. return result(
  51. {"error": f"餐盘照片分析失败:{exc}"}, status_code=502
  52. )
  53. confidence = str(analysis.get("confidence") or "low")
  54. return result({
  55. **analysis,
  56. "analysis_token": _sign_analysis_confidence(confidence),
  57. })
  58. def logout(payload: dict | None = None) -> WorkflowResult:
  59. """Handle logout: decide whether to save a partial summary of the conversation.
  60. Expected payload: { messages: [ { role: 'user'|'bot', text: '...'}, ... ] }
  61. Returns:
  62. {
  63. status: 'saved' | 'not_saved' | 'no_messages' | 'error',
  64. saved: bool,
  65. reason: str,
  66. summary: str, # present when saved=True
  67. pipeline_stage: str, # level1 / level2 / level3 / early
  68. expired_removed: int, # number of expired entries cleaned up
  69. }
  70. """
  71. try:
  72. messages = (payload or {}).get("messages") if payload else None
  73. if not messages or not isinstance(messages, list):
  74. return result(
  75. {"status": "no_messages", "saved": False, "reason": "未提供消息", "summary": "", "expired_removed": 0}
  76. )
  77. # 运行三级信息路由
  78. decision = deps.route_information(
  79. messages,
  80. allow_external_models=deps.external_model_settings_store.get()[
  81. "external_models_enabled"
  82. ],
  83. )
  84. stage = decision.get("pipeline_stage", "unknown")
  85. if decision.get("save"):
  86. from datetime import datetime, timedelta, timezone
  87. summary: str = decision.get("summary") or ""
  88. facts = normalize_memory_facts(decision.get("facts"))
  89. if not facts:
  90. removed = deps.info_store.cleanup_expired()
  91. return result({
  92. "status": "not_saved", "saved": False,
  93. "reason": "没有通过校验的结构化事实,未保存记忆",
  94. "summary": "", "pipeline_stage": stage,
  95. "rejected_facts": decision.get("rejected_facts", 0),
  96. "fact_validation": decision.get("fact_validation", ""),
  97. "expired_removed": removed,
  98. })
  99. retention = memory_retention_class(str(decision.get("type", "training_feedback")), facts)
  100. expires = None if retention == "long_term" else datetime.now(timezone.utc) + timedelta(days=(30 if retention == "medium" else 7 if retention == "short" else 3))
  101. entry = deps.info_store.add_entry(
  102. summary=summary,
  103. metadata={
  104. "reason": decision.get("reason", ""),
  105. "pipeline_stage": stage,
  106. },
  107. expires_at=expires,
  108. memory_type=str(decision.get("type", "training_feedback")),
  109. importance=int(decision.get("importance", 1)),
  110. user_confirmed=False,
  111. facts=facts,
  112. )
  113. removed = deps.info_store.cleanup_expired()
  114. return result(
  115. {
  116. "status": "saved",
  117. "saved": True,
  118. "reason": decision.get("reason", ""),
  119. "summary": summary,
  120. "pipeline_stage": stage,
  121. "entry_id": entry["id"],
  122. "memory_type": entry["type"],
  123. "importance": entry["importance"],
  124. "user_confirmed": entry["user_confirmed"],
  125. "rejected_facts": decision.get("rejected_facts", 0),
  126. "fact_validation": decision.get("fact_validation", ""),
  127. "expired_removed": removed,
  128. }
  129. )
  130. removed = deps.info_store.cleanup_expired()
  131. return result(
  132. {
  133. "status": "not_saved",
  134. "saved": False,
  135. "reason": decision.get("reason", ""),
  136. "summary": "",
  137. "pipeline_stage": stage,
  138. "rejected_facts": decision.get("rejected_facts", 0),
  139. "fact_validation": decision.get("fact_validation", ""),
  140. "expired_removed": removed,
  141. }
  142. )
  143. except Exception as e: # noqa: BLE001
  144. return result(
  145. {"status": "error", "saved": False, "reason": str(e), "summary": "", "expired_removed": 0},
  146. status_code=500,
  147. )
  148. async def upload_fit(file: Any, overwrite_pending: bool = False) -> WorkflowResult:
  149. """接收 .fit 二进制文件,解析后返回训练展示 JSON。
  150. 返回格式:
  151. {
  152. status: 'ok' | 'error',
  153. message: str, # 给对话的提示文本
  154. workout: {...} # to_public_dict() 结果
  155. }
  156. """
  157. if not file.filename or not file.filename.lower().endswith(".fit"):
  158. return result(
  159. {"status": "error", "message": "请上传 .fit 格式的文件"},
  160. status_code=400,
  161. )
  162. tmp_path: str | None = None
  163. try:
  164. content = await read_upload_with_limit(file, FIT_FILE_MAX_BYTES)
  165. if content is None:
  166. return result(
  167. {"status": "error", "message": "FIT 文件不能超过 50 MiB。"},
  168. status_code=413,
  169. )
  170. fit_source = await run_in_threadpool(deps.inspect_fit_source, file.filename, content)
  171. if fit_source.get("warnings") and fit_source.get("kind") == "fit_unknown":
  172. return result(
  173. {"status": "error", "message": fit_source["warnings"][0]}, status_code=400
  174. )
  175. if fit_source.get("kind") != "activity":
  176. imported = await run_in_threadpool(
  177. deps.health_import_service.import_file, file.filename, content
  178. )
  179. return result(
  180. {
  181. "status": "ok",
  182. "message": _health_import_message([imported], []),
  183. "health_import": imported,
  184. "workout": None,
  185. }
  186. )
  187. # 保存到临时文件
  188. suffix = Path(file.filename).suffix
  189. with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as tmp:
  190. tmp.write(content)
  191. tmp_path = tmp.name
  192. workout = deps.parse_fit_file(tmp_path)
  193. workout.source_file = Path(file.filename).name
  194. workout.source_sha256 = hashlib.sha256(content).hexdigest()
  195. if workout_store.get_current() is not None and not overwrite_pending:
  196. return result({"error": "已有待确认训练,请确认覆盖后再导入", "code": "PENDING_WORKOUT_EXISTS"}, status_code=409)
  197. # 将解析结果存入内存状态
  198. workout_store.set_current(workout)
  199. # 构建按运动类型自适应的 Agent 提示文本。
  200. sess = workout.session
  201. segs = workout.segments
  202. sport = sess.sport
  203. def _fmt_dur(s: float) -> str:
  204. s = int(s)
  205. if s >= 3600:
  206. return f"{s//3600}h{(s%3600)//60}m{s%60}s"
  207. if s >= 60:
  208. return f"{s//60}m{s%60}s"
  209. return f"{s}s"
  210. def _fmt_spd(mps: float) -> str:
  211. if mps <= 0:
  212. return "--"
  213. kmh = mps * 3.6
  214. return f"{kmh:.1f} km/h"
  215. def _session_summary_line(sess) -> str:
  216. parts = [f"运动类型:{sess.sport}"]
  217. if sess.total_elapsed_s:
  218. parts.append(f"总时长:{_fmt_dur(sess.total_elapsed_s)}")
  219. if sess.total_distance_m:
  220. parts.append(f"总距离:{sess.total_distance_m/1000:.2f} km")
  221. if sess.total_calories:
  222. parts.append(f"卡路里:{sess.total_calories} kcal")
  223. if sess.avg_hr:
  224. parts.append(f"平均心率:{sess.avg_hr} bpm")
  225. if sess.max_hr:
  226. parts.append(f"最大心率:{sess.max_hr} bpm")
  227. if sess.avg_speed_mps:
  228. parts.append(f"均速:{_fmt_spd(sess.avg_speed_mps)}")
  229. if sess.total_ascent_m:
  230. parts.append(f"总爬升:{sess.total_ascent_m:.0f} m")
  231. if sess.training_effect:
  232. parts.append(f"训练效果:{sess.training_effect}")
  233. return " ".join(parts)
  234. # 判断主要运动模式:以**实际解析出的分段类型**为准,而不是 sport_raw。
  235. # DATA-04:`sport=training` 是大类,有氧/HIIT/瑜伽都用它;这些活动现在
  236. # 会降级成 lap 分段,若仍按 sport_raw 走力量分支,就会把 lap 当成"动作组"
  237. # 渲染出一串重量 0、次数 0 的假数据。
  238. sport_raw_lower = (sess.sport_raw or "").lower()
  239. has_sets = any(s.segment_type in ("set_active", "set_rest") for s in segs)
  240. is_strength = has_sets or (
  241. not segs and sport_raw_lower in ("strength_training", "training")
  242. )
  243. if is_strength:
  244. # ── 力量训练:展示 set 表格 ──
  245. active_sets = [s for s in segs if not s.is_rest]
  246. if not segs:
  247. msg = f"[已解析 FIT 文件 {file.filename},未找到任何训练组数据]"
  248. else:
  249. header = "序号 | 类型 | 动作名 | 重量(kg) | 次数 | 时长 | 均心率 | 峰心率"
  250. sep = "-" * 72
  251. rows = []
  252. for s in segs:
  253. if s.is_rest:
  254. dur = _fmt_dur(s.duration_s)
  255. rows.append(f"{s.index:<4} | 休息 | 组间休息 | -- | -- | {dur:<7} | -- | --")
  256. else:
  257. avg = f"{s.avg_hr}" if s.avg_hr else "--"
  258. mx = f"{s.max_hr}" if s.max_hr else "--"
  259. rows.append(
  260. f"{s.index:<4} | 动作 | {s.category:<14} | {s.weight_kg:<8} | "
  261. f"{s.repetitions:<4} | {_fmt_dur(s.duration_s):<7} | {avg:<6} | {mx}"
  262. )
  263. table = "\n".join(rows)
  264. msg = (
  265. f"[FIT 解析完成] {file.filename} | {sport}\n"
  266. f"概况:{_session_summary_line(sess)}\n\n"
  267. f"{header}\n{sep}\n{table}\n\n"
  268. f"共 {len(active_sets)} 组动作、{len(segs)-len(active_sets)} 段休息。"
  269. f"以上识别结果是否准确?如有动作识别错误或需要合并(手表自动暂停导致拆分),请告知我。"
  270. )
  271. else:
  272. # ── 有氧 / 间歇 / 自定义运动:展示 session 摘要 + lap 表格 ──
  273. summary_line = _session_summary_line(sess)
  274. if not segs:
  275. msg = f"[FIT 解析完成] {file.filename} | {sport}\n概况:{summary_line}"
  276. else:
  277. # 根据有无距离决定显示列
  278. has_dist = any(s.distance_m > 0 for s in segs)
  279. if has_dist:
  280. header = "段 | 时长 | 距离(km) | 均速 | 均心率 | 峰心率"
  281. rows = []
  282. for s in segs:
  283. avg = f"{s.avg_hr}" if s.avg_hr else "--"
  284. mx = f"{s.max_hr}" if s.max_hr else "--"
  285. rows.append(
  286. f"{s.index:<3} | {_fmt_dur(s.duration_s):<7} | {s.distance_m/1000:<8.2f} | "
  287. f"{_fmt_spd(s.avg_speed_mps):<9} | {avg:<6} | {mx}"
  288. )
  289. else:
  290. header = "段 | 时长 | 节拍(rpm) | 均心率 | 峰心率 | 卡路里"
  291. rows = []
  292. for s in segs:
  293. avg = f"{s.avg_hr}" if s.avg_hr else "--"
  294. mx = f"{s.max_hr}" if s.max_hr else "--"
  295. cad = f"{s.avg_cadence}" if s.avg_cadence else "--"
  296. cal = f"{s.calories}" if s.calories else "--"
  297. rows.append(
  298. f"{s.index:<3} | {_fmt_dur(s.duration_s):<7} | {cad:<9} | "
  299. f"{avg:<6} | {mx:<6} | {cal}"
  300. )
  301. sep = "-" * 60
  302. table = "\n".join(rows)
  303. msg = (
  304. f"[FIT 解析完成] {file.filename} | {sport}\n"
  305. f"概况:{summary_line}\n\n"
  306. f"{header}\n{sep}\n{table}\n\n"
  307. f"共 {len(segs)} 段记录。数据是否符合你的预期?"
  308. )
  309. # 文件可读取但没有可用分段时,返回明确的空结果而不是成功状态。
  310. if not segs:
  311. logger.warning(
  312. "FIT 解析未产出任何分段: %s | sport_raw=%s sub_sport=%s hr=%d",
  313. file.filename, sess.sport_raw, sess.sub_sport, len(workout.hr_records),
  314. )
  315. workout_store.clear_current()
  316. return result(
  317. {
  318. "status": "empty",
  319. "message": (
  320. f"[FIT 已读取,但没有解析出任何训练分段] {file.filename} | {sport}\n"
  321. f"概况:{_session_summary_line(sess)}\n\n"
  322. "文件本身可以读取,但里面既没有动作组也没有分段记录"
  323. "(常见于中断或未正常结束的活动)。**请保留这个 FIT 文件**,"
  324. "它没有被保存,重新导入同一份文件不会有不同结果。"
  325. ),
  326. "workout": None,
  327. }
  328. )
  329. return result(
  330. {
  331. "status": "ok",
  332. "message": msg,
  333. "workout": workout.to_public_dict(),
  334. }
  335. )
  336. except HealthImportError as exc:
  337. # 保留健康数据导入错误携带的具体原因和状态码。
  338. logger.info("FIT 文件按健康数据导入失败: %s | %s", file.filename, exc)
  339. return result(
  340. {"status": "error", "message": str(exc)}, status_code=exc.status_code
  341. )
  342. except Exception: # noqa: BLE001
  343. logger.exception("FIT 文件解析失败: %s", file.filename)
  344. return result(
  345. {
  346. "status": "error",
  347. "message": "FIT 文件解析失败。若文件能在 Garmin Connect 正常打开,请把它保留下来并反馈,这可能是解析器的问题。",
  348. },
  349. status_code=500,
  350. )
  351. finally:
  352. if tmp_path is not None:
  353. Path(tmp_path).unlink(missing_ok=True)
  354. async def upload_plan(
  355. file: Any,
  356. confirm_large: bool = False,
  357. ) -> WorkflowResult:
  358. """
  359. 处理上传的 .md / .txt 训练计划文件,并进行初步鉴定。
  360. 如果是训练计划,则返回其内容;否则返回拦截原因。
  361. """
  362. if not file.filename or Path(file.filename).suffix.lower() not in {".md", ".txt"}:
  363. return result(
  364. {"status": "error", "message": "请上传 .md 或 .txt 格式的训练计划。"},
  365. status_code=400,
  366. )
  367. try:
  368. content_bytes = await read_upload_with_limit(file, PLAN_FILE_MAX_BYTES)
  369. if content_bytes is None:
  370. return result(
  371. {"status": "error", "message": "训练计划文件不能超过 1 MiB。"},
  372. status_code=413,
  373. )
  374. if len(content_bytes) > PLAN_FILE_CONFIRM_BYTES and not confirm_large:
  375. return result(
  376. {
  377. "status": "confirmation_required",
  378. "message": "该训练计划超过 256 KiB,需要确认后才能继续解析。",
  379. "size_bytes": len(content_bytes),
  380. },
  381. status_code=409,
  382. )
  383. try:
  384. content = content_bytes.decode('utf-8')
  385. except UnicodeDecodeError:
  386. content = content_bytes.decode('gbk', errors='ignore')
  387. validation_result = validate_training_plan(
  388. content,
  389. allow_external_models=deps.external_model_settings_store.get()[
  390. "external_models_enabled"
  391. ],
  392. )
  393. if validation_result["is_plan"]:
  394. subject = infer_training_subject(content)
  395. return result(
  396. content={
  397. "status": "ok",
  398. "valid": True,
  399. "content": content,
  400. "reason": validation_result["reason"],
  401. "subject": subject,
  402. }
  403. )
  404. else:
  405. return result(
  406. content={
  407. "status": "ok",
  408. "valid": False,
  409. "reason": validation_result["reason"]
  410. }
  411. )
  412. except Exception: # noqa: BLE001
  413. logger.exception("训练计划解析失败: %s", file.filename)
  414. return result(
  415. content={"status": "error", "message": "训练计划解析失败,请检查文件编码和内容。"},
  416. status_code=500,
  417. )
  418. def _health_import_message(results: list[dict], errors: list[dict]) -> str:
  419. lines = ["[健康数据导入完成]"]
  420. for item in results:
  421. state = "已存在,未重复导入" if item.get("duplicate") else "已保存"
  422. details: list[str] = []
  423. if item.get("heart_rate_count"):
  424. details.append(f"有效心率 {item['heart_rate_count']} 条")
  425. if item.get("metric_count"):
  426. details.append(f"健康指标 {item['metric_count']} 条")
  427. if item.get("sleep_stage_count"):
  428. details.append(f"睡眠阶段 {item['sleep_stage_count']} 段")
  429. if item.get("data_types"):
  430. details.append("数据类型 " + "/".join(item["data_types"]))
  431. if item.get("sleep_date"):
  432. details.append(f"睡眠日期 {item['sleep_date']}")
  433. if item.get("source_count"):
  434. details.append(f"FIT 文件 {item['source_count']} 个")
  435. suffix = f"({','.join(details)})" if details else ""
  436. lines.append(f"- {item['filename']}:{state}{suffix}")
  437. for warning in item.get("warnings", [])[:3]:
  438. lines.append(f" - 提示:{warning}")
  439. for skipped in item.get("skipped_files", []):
  440. if not isinstance(skipped, dict):
  441. continue
  442. skipped_name = str(skipped.get("filename") or "未命名文件")
  443. reason = str(skipped.get("reason") or "格式不受支持")
  444. lines.append(f" - 未处理:{skipped_name}({reason})")
  445. for error in errors:
  446. lines.append(f"- {error['filename']}:导入失败,{error['message']}")
  447. if any(item.get("data_types") or item.get("source_count") for item in results):
  448. lines.append("后续可以直接询问某天的心率、睡眠情况或一段日期内的趋势。")
  449. elif results:
  450. lines.append("该压缩包中没有识别出可导入的健康或训练数据。")
  451. return "\n".join(lines)
  452. def select_autoload_activity(
  453. activities: list[tuple[str, str, bytes]]
  454. ) -> tuple[str, str, bytes] | None:
  455. """只有**全局恰好一个**活动时才自动载入训练编辑区(BUG-08)。
  456. 原实现的判断是 `len(activities) == 1 and workout is None`,其中 activities
  457. 是**单个 ZIP 内**的列表、workout 是整个循环共用的单变量。于是一次上传两个
  458. 各含一个活动的 ZIP 时,第 1 个被载入,第 2 个只把名字塞进 activity_files
  459. 就没了;而前端在 `data.workout` 非空时直接走第一分支,第 2 个既不进选择器
  460. 也没有任何提示,用户以为两个都导入了。
  461. 改成全局计数:多于一个就一律交给前端选择器逐个处理,不再自动挑一个。
  462. """
  463. return activities[0] if len(activities) == 1 else None
  464. async def upload_health(files: list[Any]) -> WorkflowResult:
  465. if not files or len(files) > HEALTH_UPLOAD_MAX_FILES:
  466. return result(
  467. {"status": "error", "message": f"每次请上传 1-{HEALTH_UPLOAD_MAX_FILES} 个健康数据文件。"},
  468. status_code=400,
  469. )
  470. results: list[dict] = []
  471. errors: list[dict] = []
  472. # 收集全部 ZIP 中的活动后,再统一决定是否自动载入。
  473. activity_files: list[str] = []
  474. activities_index: list[dict[str, str]] = []
  475. pending_activities: list[tuple[str, str, bytes]] = []
  476. for file in files:
  477. filename = Path(file.filename or "").name
  478. suffix = Path(filename).suffix.lower()
  479. if suffix not in {".zip", ".csv"}:
  480. errors.append({"filename": filename or "未命名文件", "message": "只支持 .zip 或 .csv"})
  481. continue
  482. limit = HEALTH_ZIP_MAX_BYTES if suffix == ".zip" else HEALTH_CSV_MAX_BYTES
  483. content = await read_upload_with_limit(file, limit)
  484. if content is None:
  485. size_text = "50 MiB" if suffix == ".zip" else "2 MiB"
  486. errors.append({"filename": filename, "message": f"文件不能超过 {size_text}"})
  487. continue
  488. try:
  489. imported = await run_in_threadpool(deps.health_import_service.import_file, filename, content)
  490. results.append(imported)
  491. if suffix == ".zip":
  492. activities = await run_in_threadpool(deps.extract_activity_fits, content)
  493. for activity_name, activity_content in activities:
  494. activity_files.append(activity_name)
  495. # 带上来源 ZIP:选择器要重新上传对应的那个 ZIP 才能取到 FIT
  496. activities_index.append({"zip": filename, "name": activity_name})
  497. pending_activities.append((filename, activity_name, activity_content))
  498. except HealthImportError as exc:
  499. errors.append({"filename": filename, "message": str(exc), "status_code": exc.status_code})
  500. except Exception: # noqa: BLE001
  501. logger.exception("健康数据导入失败: %s", filename)
  502. errors.append({"filename": filename, "message": "解析或保存失败"})
  503. # 只有全局恰好一个活动时才自动载入编辑区;多于一个一律交给前端选择器,
  504. # 免得静默丢掉除第一个以外的训练。
  505. workout = None
  506. autoload = select_autoload_activity(pending_activities)
  507. if autoload is not None:
  508. _, activity_name, activity_content = autoload
  509. with tempfile.NamedTemporaryFile(delete=False, suffix=".fit") as temporary:
  510. temporary.write(activity_content)
  511. temporary_path = temporary.name
  512. try:
  513. workout = await run_in_threadpool(deps.parse_fit_file, temporary_path)
  514. workout.source_file = activity_name
  515. workout.source_sha256 = hashlib.sha256(activity_content).hexdigest()
  516. if workout_store.get_current() is not None:
  517. # Health data itself is still imported, but its optional FIT
  518. # must not silently replace the workout being edited.
  519. errors.append({
  520. "filename": activity_name,
  521. "message": "已有待确认训练,活动未自动载入;请在活动选择器中确认覆盖后载入",
  522. "code": "PENDING_WORKOUT_EXISTS",
  523. })
  524. workout = None
  525. else:
  526. workout_store.set_current(workout)
  527. except Exception: # noqa: BLE001
  528. # 单个活动解析失败不该让整次健康数据导入失败——导入本身已经成功了
  529. logger.exception("压缩包内活动 FIT 解析失败: %s", activity_name)
  530. workout = None
  531. errors.append({"filename": activity_name, "message": "活动 FIT 解析失败,其余健康数据已导入"})
  532. finally:
  533. Path(temporary_path).unlink(missing_ok=True)
  534. if not results:
  535. status_code = max((int(item.get("status_code", 400)) for item in errors), default=400)
  536. return result(
  537. {
  538. "status": "error",
  539. "message": _health_import_message([], errors),
  540. "results": [],
  541. "errors": errors,
  542. "unprocessed_files": [
  543. {
  544. "zip": "",
  545. "filename": str(item.get("filename") or "未命名文件"),
  546. "reason": str(item.get("message") or "导入失败"),
  547. }
  548. for item in errors
  549. ],
  550. },
  551. status_code=status_code,
  552. )
  553. status = "partial" if errors or any(item.get("status") == "partial" for item in results) else "ok"
  554. unprocessed_files = [
  555. {
  556. "zip": str(result.get("filename") or ""),
  557. "filename": str(skipped.get("filename") or "未命名文件"),
  558. "reason": str(skipped.get("reason") or "格式不受支持"),
  559. }
  560. for result in results
  561. for skipped in result.get("skipped_files", [])
  562. if isinstance(skipped, dict)
  563. ]
  564. unprocessed_files.extend({
  565. "zip": "",
  566. "filename": str(error.get("filename") or "未命名文件"),
  567. "reason": str(error.get("message") or "导入失败"),
  568. } for error in errors)
  569. return result(
  570. {
  571. "status": status,
  572. "message": _health_import_message(results, errors),
  573. "results": results,
  574. "errors": errors,
  575. "unprocessed_files": unprocessed_files,
  576. "activity_files": activity_files,
  577. # 富结构:每个活动带上来源 ZIP,让前端跨多个 ZIP 也能逐个处理
  578. "activities": activities_index,
  579. "workout": workout.to_public_dict() if workout else None,
  580. }
  581. )
  582. async def upload_activity_from_health_zip(
  583. file: Any, activity_name: str, overwrite_pending: bool = False
  584. ) -> WorkflowResult:
  585. if not (file.filename or "").lower().endswith(".zip"):
  586. return result({"error": "请选择包含活动 FIT 的 ZIP 文件"}, status_code=400)
  587. content = await read_upload_with_limit(file, HEALTH_ZIP_MAX_BYTES)
  588. if content is None:
  589. return result({"error": "ZIP 文件不能超过 50 MiB"}, status_code=413)
  590. try:
  591. activity_content = await run_in_threadpool(extract_activity_fit, content, activity_name)
  592. with tempfile.NamedTemporaryFile(delete=False, suffix=".fit") as temporary:
  593. temporary.write(activity_content)
  594. temporary_path = temporary.name
  595. try:
  596. workout = await run_in_threadpool(deps.parse_fit_file, temporary_path)
  597. workout.source_file = activity_name
  598. workout.source_sha256 = hashlib.sha256(activity_content).hexdigest()
  599. if workout_store.get_current() is not None and not overwrite_pending:
  600. return result({"error": "已有待确认训练,请确认覆盖后再载入", "code": "PENDING_WORKOUT_EXISTS"}, status_code=409)
  601. workout_store.set_current(workout)
  602. finally:
  603. Path(temporary_path).unlink(missing_ok=True)
  604. return result({"status": "ok", "workout": workout.to_public_dict(), "filename": activity_name})
  605. except HealthImportError as exc:
  606. return result({"error": str(exc)}, status_code=400)
  607. except Exception: # noqa: BLE001
  608. logger.exception("压缩包活动 FIT 解析失败: %s", activity_name)
  609. return result({"error": "活动 FIT 解析失败,请确认文件完整"}, status_code=500)