feat/ 导出开发容器内的数据用于评估

api/app/features/evaluation/session_catalog_service.py

@@ -0,0 +1,80 @@
+"""历史 session 目录：列表 + 导出为用户轮次列表（用于评测快照）。"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.features.evaluation.session_catalog_repo import SessionCatalogRepo
+
+
+@dataclass
+class SessionSummary:
+    id: str
+    user_id: str
+    started_at: object | None
+    conversation_stage: str | None
+    current_topic: str | None
+    status: str | None
+
+
+@dataclass
+class SessionTranscript:
+    conversation_id: str
+    user_id: str
+    user_utterances_from_segments: list[str]
+    user_utterances_from_messages: list[str]
+
+
+class SessionCatalogService:
+    def __init__(self, db: AsyncSession) -> None:
+        self._repo = SessionCatalogRepo(db)
+
+    async def list_sessions(
+        self,
+        *,
+        offset: int = 0,
+        limit: int = 50,
+        user_id: str | None = None,
+        q: str | None = None,
+    ) -> tuple[list[SessionSummary], int]:
+        total = await self._repo.count_conversations()
+        rows = await self._repo.list_conversations(
+            offset=offset, limit=limit, user_id=user_id, q_text=q
+        )
+        out = [
+            SessionSummary(
+                id=c.id,
+                user_id=c.user_id,
+                started_at=c.started_at,
+                conversation_stage=c.conversation_stage,
+                current_topic=c.current_topic,
+                status=c.status,
+            )
+            for c in rows
+        ]
+        return out, total
+
+    async def get_transcript(self, conversation_id: str) -> SessionTranscript | None:
+        c = await self._repo.get_conversation(conversation_id)
+        if not c or c.deleted_at:
+            return None
+        segs = await self._repo.list_segments_for_conversation(conversation_id)
+        msgs = await self._repo.list_messages_for_conversation(conversation_id)
+        from_segments = [
+            (s.user_input_text or "").strip()
+            for s in segs
+            if (s.user_input_text or "").strip()
+        ]
+        from_messages = [
+            m.content.strip()
+            for m in msgs
+            if m.role == "human" and (m.content or "").strip()
+        ]
+        return SessionTranscript(
+            conversation_id=c.id,
+            user_id=c.user_id,
+            user_utterances_from_segments=from_segments,
+            user_utterances_from_messages=from_messages,
+        )