feat(api): 收敛对话与记忆流程边界，引入 LLM 网关与专用服务

- MemoryService 异步路径委托 MemoryIngestService / MemoryRetrievalService；富化派发经 MemoryEnrichmentScheduler
- WebSocket pipeline 经 ChatTurnService 与显式 DTO 编排单轮对话；回忆录片段入队由 MemoirIngestScheduler 封装
- 新增 LlmGateway（LlmUseCase），各 agent、任务与适配器对齐 ports
- 补充 memory 提示适配、runtime 类型、memory-retrieval 文档、ai-touchpoints 说明与扫描脚本及配套测试

Made-with: Cursor

api/app/features/memory/service.py

@@ -15,18 +15,14 @@ from app.core.logging import get_logger
 from app.features.conversation.lineage_schemas import (
     primary_user_message_id_from_lineage,
 )
-from app.features.memory.chunker import chunk_transcript
+from app.features.memory.enrichment_scheduler import MemoryEnrichmentScheduler
+from app.features.memory.ingest_service import MemoryIngestService
 from app.features.memory.repo import (
-    create_chunk,
     create_curation_action,
-    create_source,
     set_chunk_excluded,
     set_memory_fact_status,
-    update_chunk_embedding,
-)
-from app.features.conversation.lineage_schemas import (
-    primary_user_message_id_from_lineage,
 )
+from app.features.memory.retrieval_service import MemoryRetrievalService
 from app.features.memory.schemas import EvidenceBundle
 from app.ports.embedding import EmbeddingProvider
 
@@ -56,101 +52,20 @@ class MemoryService:
         Creates MemorySource, chunks, populates embedding.
         Returns source_id.
         """
-        if not transcript or not transcript.strip():
-            raise ValueError("transcript cannot be empty")
-
-        primary_mid = (
-            primary_user_message_id_from_lineage(lineage_json) if lineage_json else None
-        )
-        source = await create_source(
-            self._db,
-            user_id=user_id,
-            source_type="transcript",
-            raw_text=transcript.strip(),
-            conversation_id=conversation_id,
-            lineage_json=lineage_json,
-            primary_user_message_id=primary_mid,
-        )
-
-        chunks_text = chunk_transcript(transcript.strip())
-        chunk_records = []
-        for i, content in enumerate(chunks_text):
-            chunk = await create_chunk(
-                self._db,
-                source_id=source.id,
-                user_id=user_id,
-                content=content,
-                chunk_index=i,
-            )
-            chunk_records.append((chunk.id, content))
-
-        await self._db.flush()
-
-        from app.core.config import settings
-
-        vectors_written = 0
-        # Embedding: 若有 provider 则写入
-        if self._embedding and chunk_records:
-            texts = [c for _, c in chunk_records]
-            embeddings = await self._embedding.embed_texts(texts)
-            for (chunk_id, _), emb in zip(chunk_records, embeddings):
-                if emb:
-                    vectors_written += 1
-                    await update_chunk_embedding(self._db, chunk_id, emb)
-
-        await self._db.commit()
-        emb_ok = self._embedding.is_available() if self._embedding else False
-        enrichment_task_id: str | None = None
-        try:
-            from app.tasks.memory_enrichment_tasks import schedule_memory_enrichment
-
-            enrichment_task_id = schedule_memory_enrichment(
-                user_id, source.id, memoir_correlation_id=None
-            )
-        except Exception as e:
-            logger.warning(
-                "memory enrichment 派发跳过: {} exc_type={}", e, type(e).__name__
-            )
-
-        logger.info(
-            "event=memory_ingest_done user_id={} conversation_id={} source_id={} "
-            "chunks={} vectors_written={} embedding_available={} enrichment_enabled={} enrichment_task_id={}",
+        service = MemoryIngestService(self._db, embedding_provider=self._embedding)
+        return await service.ingest_transcript(
             user_id,
             conversation_id,
-            source.id,
-            len(chunk_records),
-            vectors_written,
-            emb_ok,
-            settings.memory_enrichment_enabled,
-            enrichment_task_id,
+            transcript,
+            lineage_json=lineage_json,
         )
-        return source.id
 
     async def retrieve(
         self, user_id: str, query: str, *, top_k: int = 10
     ) -> EvidenceBundle:
         """Retrieve relevant evidence. 委托 HybridRetriever。"""
-        from app.features.memory.retriever import HybridRetriever
-
-        retriever = HybridRetriever(self._db, embedding_provider=self._embedding)
-        raw = await retriever.retrieve(user_id=user_id, query=query, top_k=top_k)
-        bundle = EvidenceBundle.model_validate(raw)
-        bd = bundle.model_dump()
-        vec_ok = self._embedding.is_available() if self._embedding else False
-        logger.info(
-            "event=memory_retrieve_done user_id={} query_len={} top_k={} "
-            "chunks={} facts={} summaries={} timeline={} stories={} vector_ok={}",
-            user_id,
-            len((query or "").strip()),
-            top_k,
-            len(bd.get("relevant_chunks") or []),
-            len(bd.get("relevant_facts") or []),
-            len(bd.get("relevant_summaries") or []),
-            len(bd.get("timeline_hints") or []),
-            len(bd.get("relevant_stories") or []),
-            vec_ok,
-        )
-        return bundle
+        service = MemoryRetrievalService(self._db, embedding_provider=self._embedding)
+        return await service.retrieve(user_id, query, top_k=top_k)
 
     async def exclude_chunk(
         self, user_id: str, chunk_id: str, *, reason: str = ""
@@ -292,7 +207,9 @@ def ingest_transcript_sync(
             if chunk_records and embedding_provider is not None:
                 texts = [content for _, content in chunk_records]
                 embeddings = embedding_provider.embed_texts_sync(texts)
-                for (chunk_id, _), emb in zip(chunk_records, embeddings):
+                for (chunk_id, _), emb in zip(
+                    chunk_records, embeddings, strict=False
+                ):
                     if emb:
                         vectors_written += 1
                         update_chunk_embedding_sync(session, chunk_id, emb)
@@ -405,7 +322,9 @@ def ingest_transcripts_batch_sync(
             if all_chunk_records and embedding_provider is not None:
                 texts = [content for _, content in all_chunk_records]
                 embeddings = embedding_provider.embed_texts_sync(texts)
-                for (chunk_id, _), emb in zip(all_chunk_records, embeddings):
+                for (chunk_id, _), emb in zip(
+                    all_chunk_records, embeddings, strict=False
+                ):
                     if emb:
                         vectors_written += 1
                         update_chunk_embedding_sync(session, chunk_id, emb)
@@ -438,10 +357,8 @@ def schedule_enrichment_for_sources(
     memoir_correlation_id: str | None = None,
 ) -> None:
     """After successful ingest commit, enqueue LLM enrichment for each source (memory_idle queue)."""
-    from app.tasks.memory_enrichment_tasks import schedule_memory_enrichment
-
-    for sid in source_ids:
-        if sid:
-            schedule_memory_enrichment(
-                user_id, sid, memoir_correlation_id=memoir_correlation_id
-            )
+    MemoryEnrichmentScheduler().schedule_many(
+        user_id,
+        source_ids,
+        memoir_correlation_id=memoir_correlation_id,
+    )