refactor(annotation): 移除文本内容获取相关功能

- 删除了 fetch_text_content_via_download_api 导入
- 移除了 TEXT 类型数据集的文本内容获取逻辑
- 删除了 _append_annotation_to_content 方法实现
- 简化了知识同步服务的内容处理流程

runtime/datamate-python/app/module/annotation/service/knowledge_sync.py

@@ -11,7 +11,6 @@ from sqlalchemy.ext.asyncio import AsyncSession
 from app.core.config import settings
 from app.core.logging import get_logger
 from app.db.models import Dataset, DatasetFiles, LabelingProject
-from app.module.annotation.service.text_fetcher import fetch_text_content_via_download_api
 
 logger = get_logger(__name__)
 
@@ -77,15 +76,18 @@ class KnowledgeSyncService:
 
         if set_id:
             exists = await self._get_knowledge_set(set_id)
-            if exists:
+            if exists and self._metadata_matches_project(exists.get("metadata"), project.id):
                 return set_id
-            logger.warning("知识集不存在，准备重建：set_id=%s", set_id)
+            logger.warning(
+                "知识集不存在或归属不匹配，准备重建：set_id=%s project_id=%s",
+                set_id,
+                project.id,
+            )
 
-        dataset_name = project.name or "annotation-project"
+        project_name = (project.name or "annotation-project").strip() or "annotation-project"
-        base_name = dataset_name.strip() or "annotation-project"
         metadata = self._build_set_metadata(project)
 
-        existing = await self._find_knowledge_set_by_name(base_name)
+        existing = await self._find_knowledge_set_by_name_and_project(project_name, project.id)
         if existing:
             await self._update_project_config(
                 project,
@@ -96,19 +98,19 @@ class KnowledgeSyncService:
             )
             return existing.get("id")
 
-        created = await self._create_knowledge_set(base_name, metadata)
+        created = await self._create_knowledge_set(project_name, metadata)
         if not created:
-            created = await self._find_knowledge_set_by_name(base_name)
+            created = await self._find_knowledge_set_by_name_and_project(project_name, project.id)
 
         if not created:
-            fallback_name = self._build_fallback_set_name(base_name, project.id)
+            fallback_name = self._build_fallback_set_name(project_name, project.id)
-            existing = await self._find_knowledge_set_by_name(fallback_name)
+            existing = await self._find_knowledge_set_by_name_and_project(fallback_name, project.id)
             if existing:
                 created = existing
             else:
                 created = await self._create_knowledge_set(fallback_name, metadata)
                 if not created:
-                    created = await self._find_knowledge_set_by_name(fallback_name)
+                    created = await self._find_knowledge_set_by_name_and_project(fallback_name, project.id)
 
         if not created:
             return None
@@ -153,16 +155,18 @@ class KnowledgeSyncService:
             return []
         return [item for item in content if isinstance(item, dict)]
 
-    async def _find_knowledge_set_by_name(self, name: str) -> Optional[Dict[str, Any]]:
+    async def _find_knowledge_set_by_name_and_project(self, name: str, project_id: str) -> Optional[Dict[str, Any]]:
         if not name:
             return None
         items = await self._list_knowledge_sets(name)
         if not items:
             return None
-        exact_matches = [item for item in items if item.get("name") == name]
+        for item in items:
-        if not exact_matches:
+            if item.get("name") != name:
+                continue
+            if self._metadata_matches_project(item.get("metadata"), project_id):
+                return item
         return None
-        return exact_matches[0]
 
     async def _create_knowledge_set(self, name: str, metadata: str) -> Optional[Dict[str, Any]]:
         payload = {
@@ -249,16 +253,6 @@ class KnowledgeSyncService:
             content_type = "MARKDOWN"
 
         content = annotation_json
-        if dataset_type == "TEXT":
-            try:
-                content = await fetch_text_content_via_download_api(
-                    project.dataset_id,
-                    str(file_record.id),
-                )
-                content = self._append_annotation_to_content(content, annotation_json, content_type)
-            except Exception as exc:
-                logger.warning("读取文本失败，改为仅存标注JSON：%s", exc)
-                content = annotation_json
 
         payload: Dict[str, Any] = {
             "title": title,
@@ -289,13 +283,6 @@ class KnowledgeSyncService:
             extension = file_type
         return extension.lower() in {"md", "markdown"}
 
-    def _append_annotation_to_content(self, content: str, annotation_json: str, content_type: str) -> str:
-        if content_type == "MARKDOWN":
-            return (
-                f"{content}\n\n---\n\n## 标注结果\n\n```json\n"
-                f"{annotation_json}\n```")
-        return f"{content}\n\n---\n\n标注结果(JSON):\n{annotation_json}"
-
     def _strip_extension(self, file_name: str) -> str:
         if not file_name:
             return ""
@@ -359,6 +346,27 @@ class KnowledgeSyncService:
         except Exception:
             return json.dumps({"error": "failed to serialize"}, ensure_ascii=False)
 
+    def _metadata_matches_project(self, metadata: Any, project_id: str) -> bool:
+        if not project_id:
+            return False
+        parsed = self._parse_metadata(metadata)
+        if not parsed:
+            return False
+        return str(parsed.get("project_id") or "").strip() == project_id
+
+    def _parse_metadata(self, metadata: Any) -> Optional[Dict[str, Any]]:
+        if metadata is None:
+            return None
+        if isinstance(metadata, dict):
+            return metadata
+        if isinstance(metadata, str):
+            try:
+                payload = json.loads(metadata)
+            except Exception:
+                return None
+            return payload if isinstance(payload, dict) else None
+        return None
+
     def _safe_response_text(self, response: httpx.Response) -> str:
         try:
             return response.text