feat: add support for DOCX file format in converter

mineru/backend/office/__init__.py

@@ -0,0 +1 @@
+# Copyright (c) Opendatalab. All rights reserved.

mineru/backend/office/docx.py

@@ -0,0 +1 @@
+# Copyright (c) Opendatalab. All rights reserved.

mineru/cli/common.py

@@ -28,6 +28,7 @@ if os.getenv("MINERU_LMDEPLOY_DEVICE", "") == "maca":
 
 pdf_suffixes = ["pdf"]
 image_suffixes = ["png", "jpeg", "jp2", "webp", "gif", "bmp", "jpg", "tiff"]
+docx_suffixes = ["docx"]
 
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
@@ -39,7 +40,7 @@ def read_fn(path):
         file_suffix = guess_suffix_by_bytes(file_bytes, path)
         if file_suffix in image_suffixes:
             return images_bytes_to_pdf_bytes(file_bytes)
-        elif file_suffix in pdf_suffixes:
+        elif file_suffix in pdf_suffixes + docx_suffixes:
             return file_bytes
         else:
             raise Exception(f"Unknown file suffix: {file_suffix}")

mineru/model/docx/docx_converter.py

@@ -5,7 +5,6 @@ from pathlib import Path
 from typing import BinaryIO, Optional, Union, Any, Final
 
 import logging
-import filetype
 from PIL import Image, UnidentifiedImageError
 from loguru import logger
 from docx import Document
@@ -14,8 +13,6 @@ from docx.text.paragraph import Paragraph
 from docx.text.hyperlink import Hyperlink
 from docx.text.run import Run
 from lxml import etree
-from markdown_it.common.html_re import attribute
-from markdown_it.rules_block import list_block
 from pydantic import AnyUrl
 
 from mineru.model.utils.docx.mammoth import body_xml
@@ -87,23 +84,6 @@ class DocxConverter:
         self.equation_bookends: str = "<eq>{EQ}</eq>"  # 公式标记格式
         Path.mkdir(self.output_path, parents=True, exist_ok=True)
 
-    def accepts(
-        self,
-        file_stream: BinaryIO,
-    ) -> bool:
-        mimetype = filetype.guess_mime(file_stream).lower()
-        if mimetype is None:
-            logger.error(f"Failed to detect mimetype for {self.file_path}")
-        extension = os.path.splitext(self.file_path)[1].lower()
-
-        if extension in ACCEPTED_FILE_EXTENSIONS:
-            return True
-
-        for prefix in ACCEPTED_MIME_TYPE_PREFIXES:
-            if mimetype.startswith(prefix):
-                return True
-
-        return False
 
     def convert(
         self,

mineru/model/docx/main.py

@@ -9,8 +9,6 @@ def convert_path(file_path: str):
 
 def convert_binary(file_binary: BinaryIO, file_path: str):
     converter = DocxConverter(file_path=file_path, output_path="./output")
-    if not converter.accepts(file_binary):
-        raise Exception("Not a docx file")
     return converter.convert(file_binary)
 
 

pyproject.toml

@@ -41,10 +41,8 @@ dependencies = [
     "magika>=0.6.2,<1.1.0",
     "mineru-vl-utils>=0.1.19.1,<1",
     "qwen-vl-utils>=0.0.14,<1",
-    "mammoth>=1.11.0,<2",
-    "defusedxml",
-    "filetype",
-    "markdownify",
+    "defusedxml>=0.7.1,<1",
+    "python-docx>=1.2.0,<2",
 ]
 
 [project.optional-dependencies]