support custom docx template file for styling

Author: bowenliang123

tools/md_to_docx/md_to_docx.py

@@ -1,8 +1,12 @@
 import logging
-from typing import Generator
+import os
+from pathlib import Path
+from tempfile import NamedTemporaryFile
+from typing import Generator, Optional
 
 from dify_plugin import Tool
 from dify_plugin.entities.tool import ToolInvokeMessage
+from dify_plugin.file.file import File
 
 from tools.utils.file_utils import get_meta_data
 from tools.utils.mimetype_utils import MimeType
@@ -19,8 +23,30 @@ def _invoke(self, tool_parameters: dict) -> Generator[ToolInvokeMessage, None, N
         """
         # get parameters
         md_text = get_md_text(tool_parameters, is_strip_wrapper=True)
+        docx_template_file: Optional[File] = tool_parameters.get("docx_template_file")
+        temp_pptx_template_file_path: Optional[str] = None
+        if docx_template_file and not isinstance(docx_template_file, File):
+            raise ValueError("Not a valid file for pptx template file")
+
         try:
-            result_file_bytes = pandoc_convert_file(md_text, "docx")
+            if docx_template_file:
+                temp_pptx_template_file = NamedTemporaryFile(delete=False)
+                temp_pptx_template_file.write(docx_template_file.blob)
+                temp_pptx_template_file.close()
+                temp_pptx_template_file_path = temp_pptx_template_file.name
+
+            current_script_folder = os.path.split(os.path.realpath(__file__))[0]
+            if temp_pptx_template_file_path:
+                template_docx_file = temp_pptx_template_file_path
+            else:
+                template_docx_file = f"{current_script_folder}/template/docx_template.docx"
+
+            # Options for pandoc
+            # https://pandoc.org/MANUAL.html#options
+            extra_args = [
+                "--reference-doc", template_docx_file,
+            ]
+            result_file_bytes = pandoc_convert_file(md_text, dst_format="docx", extra_args=extra_args)
             yield self.create_blob_message(
                 blob=result_file_bytes,
                 meta=get_meta_data(
@@ -32,3 +58,6 @@ def _invoke(self, tool_parameters: dict) -> Generator[ToolInvokeMessage, None, N
             self.logger.exception("Failed to convert markdown text to DOCX file")
             yield self.create_text_message(f"Failed to convert markdown text to DOCX file, error: {str(e)}")
             raise e
+        finally:
+            if temp_pptx_template_file_path:
+                Path(temp_pptx_template_file_path).unlink()

tools/md_to_docx/md_to_docx.yaml

@@ -20,6 +20,17 @@ parameters:
       en_US: Markdown text
       zh_Hans: Markdown格式文本
     form: llm
+  - name: docx_template_file
+    type: file
+    required: false
+    label:
+      en_US: DOCX Template File
+      zh_Hans: DOCX 模板文件
+    human_description:
+      en_US: Optional docx template file for styling。Use "Home"-"Styles Pane" in Microsoft Word to edit styles of the docx file.
+      zh_Hans: 可选的docx模板文件，用于样式控制。使用Microsoft Word中“开始”-“样式窗格”对docx文件的样式进行编辑。
+    llm_description: An optional .pptx file that serves as a template for the generated presentation
+    form: llm
   - name: output_filename
     type: string
     required: false

tools/md_to_docx/template/docx_template.docx

@@ -1,16 +1,23 @@
 from pathlib import Path
 from tempfile import NamedTemporaryFile
+from typing import Optional
 
 from pypandoc import convert_file, convert_text
 
 
-def pandoc_convert_file(md_text: str, dst_format: str) -> bytes:
+def pandoc_convert_file(md_text: str, dst_format: str, extra_args: Optional[list[str]]) -> bytes:
     with NamedTemporaryFile(suffix=".md", delete=True) as md_file:
         md_file.write(md_text.encode("utf-8"))
         md_file.flush()
 
         with NamedTemporaryFile(suffix=f".{dst_format}", delete=True) as target_file:
-            convert_file(source_file=md_file.name, format="markdown", to=dst_format, outputfile=target_file.name)
+            convert_file(
+                source_file=md_file.name,
+                format="markdown",
+                to=dst_format,
+                outputfile=target_file.name,
+                extra_args=extra_args,
+            )
             target_file.flush()
             return Path(target_file.name).read_bytes()