Initial MVP for QA asset backend

2026-07-05 17:44:15 +08:00
commit 95b5e09fd4
71 changed files with 6546 additions and 0 deletions
--- a/hy_qa_asset_backend/backend/app/services/task_runner.py
+++ b/hy_qa_asset_backend/backend/app/services/task_runner.py
@@ -0,0 +1,124 @@
+from sqlalchemy.orm import Session
+
+from app import models
+from app.services.ai_cleaner import AICleaner
+from app.services.feishu_client import FeishuClient
+from app.services.mock_data_service import create_mock_session
+from app.services.standard_qa_service import refresh_session_counts
+from app.utils.time_utils import utc_now
+
+
+class TaskRunner:
+    def __init__(self) -> None:
+        self.feishu = FeishuClient()
+        self.ai = AICleaner()
+
+    async def run(
+        self,
+        db: Session,
+        *,
+        task_type: models.TaskType = models.TaskType.manual_scan,
+        session_id: int | None = None,
+        reprocess: bool = False,
+    ) -> models.TaskRun:
+        task = models.TaskRun(task_type=task_type, task_status=models.TaskStatus.running, started_at=utc_now())
+        db.add(task)
+        db.commit()
+        db.refresh(task)
+        try:
+            sessions = self._resolve_sessions(db, session_id=session_id)
+            if not sessions and self.feishu.is_mock_mode and session_id is None:
+                sessions = [create_mock_session(db)]
+            task.sessions_scanned = len(sessions)
+
+            for session in sessions:
+                try:
+                    created_count = await self._process_session(db, session, reprocess=reprocess)
+                    task.sessions_processed += 1
+                    task.qa_created += created_count
+                except Exception as exc:  # noqa: BLE001
+                    task.qa_failed += 1
+                    session.process_status = models.ProcessStatus.failed
+                    session.failed_reason = str(exc)
+                    task.error_message = str(exc)
+                    db.add(session)
+                    db.commit()
+                    self.feishu.update_session_status(session, session.process_status.value)
+
+            task.task_status = (
+                models.TaskStatus.success
+                if task.qa_failed == 0
+                else models.TaskStatus.partial_success
+                if task.sessions_processed > 0
+                else models.TaskStatus.failed
+            )
+        except Exception as exc:  # noqa: BLE001
+            task.task_status = models.TaskStatus.failed
+            task.error_message = str(exc)
+        finally:
+            task.ended_at = utc_now()
+            db.add(task)
+            db.commit()
+            db.refresh(task)
+        return task
+
+    def _resolve_sessions(self, db: Session, session_id: int | None) -> list[models.FeishuSession]:
+        if session_id is not None:
+            session = db.get(models.FeishuSession, session_id)
+            return [session] if session else []
+        return self.feishu.scan_unprocessed_sessions(db)
+
+    async def _process_session(self, db: Session, session: models.FeishuSession, *, reprocess: bool = False) -> int:
+        if reprocess:
+            db.query(models.StandardQAItem).filter(models.StandardQAItem.session_id == session.id).delete()
+            db.query(models.RawQAItem).filter(models.RawQAItem.session_id == session.id).delete()
+            db.commit()
+
+        session.process_status = models.ProcessStatus.processing
+        session.failed_reason = None
+        db.add(session)
+        db.commit()
+        self.feishu.update_session_status(session, session.process_status.value)
+
+        transcript = self.feishu.fetch_transcript(session)
+        session.source_text = transcript
+        ai_items = await self.ai.clean_transcript(session.id, transcript)
+
+        created: list[models.RawQAItem] = []
+        for index, item in enumerate(ai_items, start=1):
+            raw = models.RawQAItem(
+                session_id=session.id,
+                qa_code=f"{session.session_code}-QA-{index:03d}",
+                raw_question=item["raw_question"],
+                normalized_question=item.get("normalized_question"),
+                raw_answer=item["raw_answer"],
+                normalized_answer=item.get("normalized_answer"),
+                suggested_standard_question=item.get("suggested_standard_question"),
+                suggested_standard_answer=item.get("suggested_standard_answer"),
+                answer_person=item.get("answer_person"),
+                primary_topic=item.get("primary_topic", "其他"),
+                problem_tags=item.get("problem_tags", []),
+                course_stage=item.get("course_stage", "不确定"),
+                audience_tags=item.get("audience_tags", []),
+                emotion_intensity=item.get("emotion_intensity"),
+                risk_level=item.get("risk_level", "medium"),
+                risk_types=item.get("risk_types", []),
+                risk_notes=item.get("risk_notes"),
+                need_desensitization=item.get("need_desensitization", False),
+                desensitization_status=item.get("desensitization_status", "unknown"),
+                review_status=models.ReviewStatus.pending,
+                review_notes=item.get("review_notes"),
+                suggested_to_standard_qa=item.get("suggested_to_standard_qa", True),
+                source_timestamp=item.get("source_timestamp"),
+            )
+            db.add(raw)
+            created.append(raw)
+
+        session.process_status = models.ProcessStatus.pending_review
+        db.add(session)
+        db.commit()
+        refresh_session_counts(db, session.id)
+        db.commit()
+        self.feishu.write_raw_qa_items(session, created)
+        self.feishu.update_session_status(session, session.process_status.value)
+        return len(created)