Merge branch 'gzy_iframedf' into 'master'

feat: iframe传 文件进入

See merge request knowledge-base-bj/deerflow-server!93

frontend-web/src/pages/iframe-chat-submit.test.ts

@@ -100,3 +100,37 @@ test("maps iframe model selection into stream context model_name", () => {
     model_name: "OpenAI3",
   });
 });
+
+test("maps externally uploaded files into stream additional_kwargs.files", () => {
+  const result = parseBrainyChatSubmitMessage({
+    source: "brainy-extension",
+    type: "brainy:chat-submit",
+    requestId: "iframe-789",
+    payload: {
+      text: "网摘",
+      prompt: "网摘",
+      uploadedFiles: [
+        {
+          filename: "LLM Wiki 接口说明.md",
+          size: "7002",
+          path: "/mnt/user-data/uploads/LLM Wiki 接口说明.md",
+          status: "uploaded",
+          artifact_url:
+            "/api/threads/445ca7a7-9737-44f0-8eea-6887776f4f9c/artifacts/mnt/user-data/uploads/LLM%20Wiki%20%E6%8E%A5%E5%8F%A3%E8%AF%B4%E6%98%8E.md",
+        },
+      ],
+    },
+  });
+
+  assert.equal(result.submitText, "网摘");
+  assert.deepEqual(result.additionalKwargs?.files, [
+    {
+      filename: "LLM Wiki 接口说明.md",
+      size: "7002",
+      path: "/mnt/user-data/uploads/LLM Wiki 接口说明.md",
+      status: "uploaded",
+      artifact_url:
+        "/api/threads/445ca7a7-9737-44f0-8eea-6887776f4f9c/artifacts/mnt/user-data/uploads/LLM%20Wiki%20%E6%8E%A5%E5%8F%A3%E8%AF%B4%E6%98%8E.md",
+    },
+  ]);
+});

frontend-web/src/pages/iframe-chat-submit.ts

@@ -26,6 +26,14 @@ type BrainyModelSelection = {
   value?: string;
 };
 
+type BrainyUploadedFile = {
+  filename?: string;
+  size?: string | number;
+  path?: string;
+  status?: string;
+  artifact_url?: string;
+};
+
 export type BrainyChatSubmitMessage = {
   source?: string;
   type?: string;
@@ -35,6 +43,7 @@ export type BrainyChatSubmitMessage = {
     pageContext?: BrainyPageContext;
     screenshot?: BrainyScreenshot;
     model?: BrainyModelSelection;
+    uploadedFiles?: BrainyUploadedFile[];
   };
   requestId?: string;
 };
@@ -253,6 +262,36 @@ function buildScreenshotFileParts(
   return [];
 }
 
+function normalizeUploadedFiles(
+  uploadedFiles: BrainyUploadedFile[] | undefined,
+): Record<string, unknown>[] {
+  if (!Array.isArray(uploadedFiles)) {
+    return [];
+  }
+
+  return uploadedFiles
+    .map((file) => {
+      const filename = trimToUndefined(file.filename);
+      if (!filename) {
+        return null;
+      }
+
+      const size = String(file.size ?? "0");
+      const path = trimToUndefined(file.path);
+      const status = trimToUndefined(file.status);
+      const artifactUrl = trimToUndefined(file.artifact_url);
+
+      return {
+        filename,
+        size,
+        ...(path ? { path } : {}),
+        ...(status ? { status } : {}),
+        ...(artifactUrl ? { artifact_url: artifactUrl } : {}),
+      };
+    })
+    .filter((file): file is Record<string, unknown> => file !== null);
+}
+
 export function parseBrainyChatSubmitMessage(
   data: BrainyChatSubmitMessage,
 ): ParsedBrainyChatSubmitMessage {
@@ -268,6 +307,7 @@ export function parseBrainyChatSubmitMessage(
   });
   const requestId = trimToUndefined(data.requestId);
   const modelName = trimToUndefined(data.payload?.model?.value);
+  const uploadedFiles = normalizeUploadedFiles(data.payload?.uploadedFiles);
 
   const iframeContext: Record<string, unknown> = {};
   if (requestId) {
@@ -281,6 +321,13 @@ export function parseBrainyChatSubmitMessage(
     submitText,
     files,
     ...(Object.keys(iframeContext).length > 0 ? { iframeContext } : {}),
-    ...(promptPrefix ? { additionalKwargs: { prompt_prefix: promptPrefix } } : {}),
+    ...(promptPrefix || uploadedFiles.length > 0
+      ? {
+          additionalKwargs: {
+            ...(promptPrefix ? { prompt_prefix: promptPrefix } : {}),
+            ...(uploadedFiles.length > 0 ? { files: uploadedFiles } : {}),
+          },
+        }
+      : {}),
   };
 }

offline-backend-20260512/backend/packages/harness/deerflow/agents/middlewares/uploads_middleware.py

@@ -1,8 +1,10 @@
 """Middleware to inject uploaded files information into agent context."""
 
-import logging
-from pathlib import Path
-from typing import NotRequired, override
+import logging
+import shutil
+from pathlib import Path
+from typing import NotRequired, override
+from urllib.parse import unquote
 
 from langchain.agents import AgentState
 from langchain.agents.middleware import AgentMiddleware
@@ -13,7 +15,7 @@ from deerflow.config.paths import Paths, get_paths
 from deerflow.runtime.user_context import get_effective_user_id
 from deerflow.utils.file_conversion import extract_outline
 
-logger = logging.getLogger(__name__)
+logger = logging.getLogger(__name__)
 
 
 _OUTLINE_PREVIEW_LINES = 5
@@ -22,7 +24,8 @@ _OUTLINE_PREVIEW_LINES = 5
 # configured vision model), never by reading their bytes. Listing them with the
 # generic read_file/grep guidance made the agent try to parse image binaries
 # with bash/python, which fails and wastes turns.
-_IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".webp", ".gif", ".bmp"}
+_IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".webp", ".gif", ".bmp"}
+_ARTIFACT_URL_MARKER = "/artifacts/"
 
 
 def _extract_outline_for_file(file_path: Path) -> tuple[list[dict], list[str]]:
@@ -69,7 +72,7 @@ class UploadsMiddlewareState(AgentState):
     uploaded_files: NotRequired[list[dict] | None]
 
 
-class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
+class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
     """Middleware to inject uploaded files information into the agent context.
 
     Reads file metadata from the current message's additional_kwargs.files
@@ -122,7 +125,7 @@ class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
             lines.append("  Use `grep` to search for keywords (e.g. `grep(pattern='keyword', path='/mnt/user-data/uploads/')`).")
         lines.append("")
 
-    def _create_files_message(self, new_files: list[dict], historical_files: list[dict]) -> str:
+    def _create_files_message(self, new_files: list[dict], historical_files: list[dict]) -> str:
         """Create a formatted message listing uploaded files.
 
         Args:
@@ -160,9 +163,61 @@ class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
         lines.append("- Only fall back to web search if the file content is clearly insufficient to answer the question.")
         lines.append("</uploaded_files>")
 
-        return "\n".join(lines)
-
-    def _files_from_kwargs(self, message: HumanMessage, uploads_dir: Path | None = None) -> list[dict] | None:
+        return "\n".join(lines)
+
+    def _import_file_from_artifact_url(
+        self,
+        file_info: dict,
+        uploads_dir: Path | None,
+    ) -> bool:
+        """Copy a previously uploaded file into the current thread when possible.
+
+        iframe-style callers may pass file metadata that points at an upload
+        from another DeerFlow thread. The current run thread still needs a real
+        file under its own ``/mnt/user-data/uploads`` directory, otherwise this
+        middleware will filter the entry out during existence checks.
+        """
+        filename = file_info.get("filename") or ""
+        artifact_url = str(file_info.get("artifact_url") or "").strip()
+        if not filename or uploads_dir is None or not artifact_url:
+            return False
+
+        if not artifact_url.startswith("/api/threads/") or _ARTIFACT_URL_MARKER not in artifact_url:
+            return False
+
+        prefix, encoded_virtual_path = artifact_url.split(_ARTIFACT_URL_MARKER, 1)
+        source_thread_id = prefix.removeprefix("/api/threads/").strip("/")
+        if not source_thread_id or not encoded_virtual_path:
+            return False
+
+        virtual_path = "/" + unquote(encoded_virtual_path.lstrip("/"))
+        if Path(virtual_path).name != filename:
+            return False
+
+        try:
+            source_path = self._paths.resolve_virtual_path(
+                source_thread_id,
+                virtual_path,
+                user_id=get_effective_user_id(),
+            )
+        except ValueError:
+            logger.debug("Failed to resolve uploaded artifact path: %s", artifact_url, exc_info=True)
+            return False
+
+        if not source_path.is_file():
+            return False
+
+        uploads_dir.mkdir(parents=True, exist_ok=True)
+        target_path = uploads_dir / filename
+        shutil.copy2(source_path, target_path)
+
+        source_markdown = source_path.with_suffix(".md")
+        if source_markdown.is_file() and source_markdown != source_path:
+            shutil.copy2(source_markdown, target_path.with_suffix(".md"))
+
+        return True
+
+    def _files_from_kwargs(self, message: HumanMessage, uploads_dir: Path | None = None) -> list[dict] | None:
         """Extract file info from message additional_kwargs.files.
 
         The frontend sends uploaded file metadata in additional_kwargs.files
@@ -183,16 +238,17 @@ class UploadsMiddleware(AgentMiddleware[UploadsMiddlewareState]):
 
         files = []
         for f in kwargs_files:
-            if not isinstance(f, dict):
-                continue
-            filename = f.get("filename") or ""
-            if not filename or Path(filename).name != filename:
-                continue
-            if uploads_dir is not None and not (uploads_dir / filename).is_file():
-                continue
-            files.append(
-                {
-                    "filename": filename,
+            if not isinstance(f, dict):
+                continue
+            filename = f.get("filename") or ""
+            if not filename or Path(filename).name != filename:
+                continue
+            if uploads_dir is not None and not (uploads_dir / filename).is_file():
+                if not self._import_file_from_artifact_url(f, uploads_dir):
+                    continue
+            files.append(
+                {
+                    "filename": filename,
                     "size": int(f.get("size") or 0),
                     "path": f"/mnt/user-data/uploads/{filename}",
                     "extension": Path(filename).suffix,