Feat: extract message output to file (#11251)

### What problem does this PR solve?

Feat: extract message output to file

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

Dockerfile

@@ -51,7 +51,9 @@ RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
     apt install -y libpython3-dev libgtk-4-1 libnss3 xdg-utils libgbm-dev && \
     apt install -y libjemalloc-dev && \
     apt install -y python3-pip pipx nginx unzip curl wget git vim less && \
-    apt install -y ghostscript
+    apt install -y ghostscript && \
+    apt install -y pandoc && \
+    apt install -y texlive
 
 RUN if [ "$NEED_MIRROR" == "1" ]; then \
         pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && \

agent/canvas.py

@@ -408,6 +408,10 @@ class Canvas(Graph):
                     else:
                         yield decorate("message", {"content": cpn_obj.output("content")})
                         cite = re.search(r"\[ID:[ 0-9]+\]",  cpn_obj.output("content"))
+
+                    if isinstance(cpn_obj.output("attachment"), tuple):
+                        yield decorate("message", {"attachment": cpn_obj.output("attachment")})
+                        
                     yield decorate("message_end", {"reference": self.get_reference() if cite else None})
 
                     while partials:

agent/component/message.py

@@ -17,6 +17,9 @@ import json
 import os
 import random
 import re
+import pypandoc
+import logging
+import tempfile
 from functools import partial
 from typing import Any
 
@@ -24,7 +27,8 @@ from agent.component.base import ComponentBase, ComponentParamBase
 from jinja2 import Template as Jinja2Template
 
 from common.connection_utils import timeout
-
+from common.misc_utils import get_uuid
+from common import settings
 
 class MessageParam(ComponentParamBase):
     """
@@ -34,6 +38,7 @@ class MessageParam(ComponentParamBase):
         super().__init__()
         self.content = []
         self.stream = True
+        self.output_format = None  # default output format
         self.outputs = {
             "content": {
                 "type": "str"
@@ -133,6 +138,7 @@ class Message(ComponentBase):
             yield rand_cnt[s: ]
 
         self.set_output("content", all_content)
+        self._convert_content(all_content)
 
     def _is_jinjia2(self, content:str) -> bool:
         patt = [
@@ -164,6 +170,68 @@ class Message(ComponentBase):
             content = re.sub(n, v, content)
 
         self.set_output("content", content)
+        self._convert_content(content)
 
     def thoughts(self) -> str:
         return ""
+
+    def _convert_content(self, content):
+        doc_id = get_uuid()
+        
+        if self._param.output_format.lower() not in {"markdown", "html", "pdf", "docx"}:
+            self._param.output_format = "markdown"
+
+        try:
+            if self._param.output_format in {"markdown", "html"}:
+                if isinstance(content, str):
+                    converted = pypandoc.convert_text(
+                        content,
+                        to=self._param.output_format,
+                        format="markdown",
+                    )
+                else:
+                    converted = pypandoc.convert_file(
+                        content,
+                        to=self._param.output_format,
+                        format="markdown",
+                    )
+
+                binary_content = converted.encode("utf-8")
+
+            else:  # pdf, docx
+                with tempfile.NamedTemporaryFile(suffix=f".{self._param.output_format}", delete=False) as tmp:
+                    tmp_name = tmp.name
+
+                try:
+                    if isinstance(content, str):
+                        pypandoc.convert_text(
+                            content,
+                            to=self._param.output_format,
+                            format="markdown",
+                            outputfile=tmp_name,
+                        )
+                    else:
+                        pypandoc.convert_file(
+                            content,
+                            to=self._param.output_format,
+                            format="markdown",
+                            outputfile=tmp_name,
+                        )
+
+                    with open(tmp_name, "rb") as f:
+                        binary_content = f.read()
+
+                finally:
+                    if os.path.exists(tmp_name):
+                        os.remove(tmp_name)
+
+            settings.STORAGE_IMPL.put(self._canvas._tenant_id, doc_id, binary_content)
+            self.set_output("attachment", {
+                "doc_id":doc_id, 
+                "format":self._param.output_format, 
+                "file_name":f"{doc_id[:8]}.{self._param.output_format}"})
+
+            logging.info(f"Converted content uploaded as {doc_id} (format={self._param.output_format})")
+
+        except Exception as e:
+            logging.error(f"Error converting content to {self._param.output_format}: {e}")

api/apps/document_app.py

@@ -508,6 +508,7 @@ def get(doc_id):
         ext = ext.group(1) if ext else None
         if ext:
             if doc.type == FileType.VISUAL.value:
+
                 content_type = CONTENT_TYPE_MAP.get(ext, f"image/{ext}")
             else:
                 content_type = CONTENT_TYPE_MAP.get(ext, f"application/{ext}")
@@ -517,6 +518,22 @@ def get(doc_id):
         return server_error_response(e)
 
 
+@manager.route("/download/<attachment_id>", methods=["GET"])  # noqa: F821
+@login_required
+def download_attachment(attachment_id):
+    try:
+        ext = request.args.get("ext", "markdown")
+        data = settings.STORAGE_IMPL.get(current_user.id, attachment_id)
+        # data = settings.STORAGE_IMPL.get("eb500d50bb0411f0907561d2782adda5", attachment_id)
+        response = flask.make_response(data)
+        response.headers.set("Content-Type", CONTENT_TYPE_MAP.get(ext, f"application/{ext}"))
+
+        return response
+
+    except Exception as e:
+        return server_error_response(e)
+
+
 @manager.route("/change_parser", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("doc_id")

pyproject.toml

@@ -145,6 +145,7 @@ dependencies = [
     "markdownify>=1.2.0",
     "captcha>=0.7.1",
     "pip>=25.2",
+    "pypandoc>=1.16",
 ]
 
 [dependency-groups]

uv.lock

@@ -4892,6 +4892,14 @@ wheels = [
     { url = "https://pypi.tuna.tsinghua.edu.cn/packages/80/28/2659c02301b9500751f8d42f9a6632e1508aa5120de5e43042b8b30f8d5d/pyopenssl-25.1.0-py3-none-any.whl", hash = "sha256:2b11f239acc47ac2e5aca04fd7fa829800aeee22a2eb30d744572a157bd8a1ab", size = 56771, upload-time = "2025-05-17T16:28:29.197Z" },
 ]
 
+[[package]]
+name = "pypandoc"
+version = "1.16"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/24/77/af1fc54740a0712988f9518e629d38edc7b8ffccd7549203f19c3d8a2db6/pypandoc-1.16-py3-none-any.whl", hash = "sha256:868f390d48388743e7a5885915cbbaa005dea36a825ecdfd571f8c523416c822", size = 19425, upload-time = "2025-11-08T15:44:38.429Z" },
+]
+
 [[package]]
 name = "pyparsing"
 version = "3.2.3"
@@ -5292,6 +5300,7 @@ dependencies = [
     { name = "pyicu" },
     { name = "pymysql" },
     { name = "pyodbc" },
+    { name = "pypandoc" },
     { name = "pypdf" },
     { name = "pypdf2" },
     { name = "python-calamine" },
@@ -5447,6 +5456,7 @@ requires-dist = [
     { name = "pyicu", specifier = ">=2.15.3,<3.0.0" },
     { name = "pymysql", specifier = ">=1.1.1,<2.0.0" },
     { name = "pyodbc", specifier = ">=5.2.0,<6.0.0" },
+    { name = "pypandoc", specifier = ">=1.16" },
     { name = "pypdf", specifier = "==6.0.0" },
     { name = "pypdf2", specifier = ">=3.0.1,<4.0.0" },
     { name = "python-calamine", specifier = ">=0.4.0" },