merge: 合并 origin/xuhd，解决 message-list-item 和 WorkspaceRoutes 冲突

- message-list-item: 保留 EmbeddedPageContext + 消息内联编辑两套功能 - WorkspaceRoutes: 保留 IframeChatPage + NotebookDetailPageWithKey Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-03 11:41:49 +08:00
commit 712326387f
@@ -1,8 +1,9 @@
-"""素材收集专家节点：关键词提取 → 并发搜索 → 相关度排序。"""
+"""素材收集专家节点：关键词提取 → 并发搜索（含笔记） → 相关度排序。"""
 from __future__ import annotations

 import asyncio
 import logging
+import os
 import re
 import time
 import uuid
@@ -21,6 +22,10 @@ from ._utils import get_ai_writing_config, llm_json, safe_get_stream_writer

 logger = logging.getLogger(__name__)

+_NOTEBOOK_SEARCH_PATH = "/api/v1/source/mentioned-documents/search"
+_NOTEBOOK_MAX_KEYWORDS = 3   # 每次写作最多对笔记发起的检索轮数
+_NOTEBOOK_TOP_K = 5          # 每次检索取回的 chunk 数
+
 # 素材总量上限边界：config.yaml 的 ai_writing.max_materials，非法值回退默认 50
 _DEFAULT_MAX_MATERIALS = 50
 _MIN_MAX_MATERIALS = 1
@@ -95,6 +100,67 @@ def _notice(title: str, detail: str, raw: str = "") -> dict:
    return event


+async def _search_notebook(
+    keywords: list[str],
+    search_space_id: int,
+    source_ids: list[int],
+    token: str,
+) -> list[Material]:
+    """从笔记 API 按关键词检索内容，返回 source_type='notebook' 的 Material 列表。"""
+    import httpx  # 懒导入：仅在笔记模式下才需要
+
+    base = (os.getenv("DEER_FLOW_NOTEBOOK_API_BASE") or "").rstrip("/")
+    if not base:
+        logger.warning("DEER_FLOW_NOTEBOOK_API_BASE not set, skipping notebook search")
+        return []
+
+    results: list[Material] = []
+    mat_idx = 0
+    try:
+        async with httpx.AsyncClient(timeout=20) as client:
+            for kw in keywords[:_NOTEBOOK_MAX_KEYWORDS]:
+                try:
+                    resp = await client.post(
+                        f"{base}{_NOTEBOOK_SEARCH_PATH}",
+                        headers={"Authorization": f"Bearer {token}"},
+                        json={
+                            "query": kw,
+                            "source_ids": source_ids,
+                            "search_space_id": search_space_id,
+                            "top_k": _NOTEBOOK_TOP_K,
+                        },
+                    )
+                    resp.raise_for_status()
+                    data = resp.json()
+                    # 解析 documents 列表，每个 chunk 作为独立素材
+                    documents: list[dict] = data.get("documents") if isinstance(data, dict) else []
+                    for doc in documents or []:
+                        doc_title: str = doc.get("title") or "笔记文档"
+                        doc_url: str = doc.get("url") or ""
+                        for chunk in doc.get("chunks") or []:
+                            chunk_content: str = (chunk.get("content") or "").strip()
+                            if not chunk_content:
+                                continue
+                            mat_idx += 1
+                            raw_score = chunk.get("score")
+                            chunk_score = float(raw_score) if raw_score is not None else max(0.5, 0.95 - (mat_idx - 1) * 0.03)
+                            results.append(Material(
+                                id=f"nb{mat_idx:03d}",
+                                title=doc_title,
+                                content=chunk_content[:500],
+                                source=f"笔记·{doc_title}",
+                                source_type="notebook",
+                                relevance_score=chunk_score,
+                                url=doc_url,
+                            ))
+                except Exception as e:
+                    logger.warning("notebook search kw=%r failed: %s", kw, e)
+    except Exception as e:
+        logger.warning("notebook search client error: %s", e)
+
+    return results
+
+
 async def researcher_node(state: AIWritingState, config: RunnableConfig) -> dict:
    intent = state["user_intent"]
    article_type = state.get("article_type", "analysis")
@@ -181,6 +247,22 @@ async def researcher_node(state: AIWritingState, config: RunnableConfig) -> dict
        if extra_keywords:
            keywords = list(dict.fromkeys(keywords + extra_keywords))

+    # Step 1.5: 笔记检索（如果传入了 notebook_search_space_id）
+    notebook_search_space_id: int | None = state.get("notebook_search_space_id")
+    notebook_materials: list[Material] = []
+    if notebook_search_space_id:
+        notebook_source_ids: list[int] = state.get("notebook_source_ids") or []
+        notebook_token: str = cfg.get("notebook_token") or ""
+        _emit("正在检索笔记内容…")
+        notebook_materials = await _search_notebook(
+            keywords[:_NOTEBOOK_MAX_KEYWORDS],
+            notebook_search_space_id,
+            notebook_source_ids,
+            notebook_token,
+        )
+        if notebook_materials:
+            _emit(f"从笔记中获取到 {len(notebook_materials)} 条素材")
+
    # Step 2: 搜索
    # 技能检索：技能不是素材，是"检索策略 / 视角"。流程：
    #   a. SkillPicker 选出按相关度排序的 skills（推 list_skills + select_skills 事件）
@@ -330,10 +412,17 @@ async def researcher_node(state: AIWritingState, config: RunnableConfig) -> dict
    existing_pkg = state.get("material_package")
    if is_skill_search:
        # 技能编排循环已经完成素材累积，直接用 raw_materials_pre 当 raw_materials
+        # 笔记素材追加在技能素材之后
        raw_materials: list[Material] = raw_materials_pre  # type: ignore[name-defined]
+        raw_materials.extend(notebook_materials)
    else:
        seen_urls: set[str] = set()
-        raw_materials = []
+        # 笔记素材优先排在最前面，赋予高相关度
+        raw_materials = list(notebook_materials)
+        for m in notebook_materials:
+            if m.get("url"):
+                seen_urls.add(m["url"])
+
        preserve_existing = existing_pkg and (is_supplement or is_append_only)
        if preserve_existing:
            for m in existing_pkg.get("materials", []):
@@ -140,6 +140,9 @@ class AIWritingState(TypedDict):
    # 经 checkpoint 持久化 —— interrupt/resume 是新 invocation、其 config 不带原
    # model_name，存进 state 才能让 resume 后才运行的节点拿到用户模型。
    model_name: str
+    # 笔记来源（可选）：传入后 researcher 阶段会额外从笔记检索素材
+    notebook_search_space_id: Optional[int]
+    notebook_source_ids: Optional[List[int]]

    # Agent A 输出
    material_package: Optional[MaterialPackage]
@@ -56,7 +56,7 @@ def notebook_search_tool(query: str, top_k: int = 10) -> str:
            headers={"Authorization": f"Bearer {token}"},
            json={
                "query": query,
-                "source_ids": [source_ids],
+                "source_ids": source_ids,
                "search_space_id": search_space_id,
                "top_k": top_k,
            },