office-to-md

Office 转 Markdown 技能

概述

此技能使用 markitdown 将各种 Office 格式转换为 Markdown。markitdown 是 Microsoft 开源的文档转 Markdown 工具。它非常适合将 Office 内容转换为可搜索、可进行版本控制且适合 AI 处理的内容。

使用方法

1. 提供 Office 文件（Word、Excel、PowerPoint、PDF 等）
2. 可选：指定转换选项
3. 我会将其转换为简洁的 Markdown

提示词示例：

• “将此 Word 文档转换为 Markdown”
• “将此 PowerPoint 转换为 Markdown 笔记”
• “将此 PDF 的内容提取为 Markdown”
• “将此 Excel 文件转换为 Markdown 表格”

领域知识

markitdown 基础知识

from markitdown import MarkItDown

# Initialize converter
md = MarkItDown()

# Convert file
result = md.convert("document.docx")
print(result.text_content)

# Save to file
with open("output.md", "w") as f:
    f.write(result.text_content)

支持的格式

格式	扩展名	说明
Word	.docx	全文、表格、基本格式
Excel	.xlsx	转换为 Markdown 表格
PowerPoint	.pptx	将幻灯片转换为各个章节
PDF	.pdf	文本提取
HTML	.html	简洁的 Markdown
图像	.jpg, .png	使用视觉模型进行 OCR
音频	.mp3, .wav	转录
ZIP	.zip	处理其中包含的文件

基本用法

Python API

from markitdown import MarkItDown

# Simple conversion
md = MarkItDown()
result = md.convert("document.docx")

# Access content
markdown_text = result.text_content

# With options
md = MarkItDown(
    llm_client=None,      # Optional LLM for enhanced processing
    llm_model=None        # Model name if using LLM
)

命令行

# Install
pip install markitdown

# Convert file
markitdown document.docx > output.md

# Or with output file
markitdown document.docx -o output.md

Word 文档转换

from markitdown import MarkItDown

md = MarkItDown()

# Convert Word document
result = md.convert("report.docx")

# Output preserves:
# - Headings (as # headers)
# - Bold/italic formatting
# - Lists (bulleted and numbered)
# - Tables (as markdown tables)
# - Hyperlinks

print(result.text_content)

输出示例：

# Annual Report 2024

## Executive Summary

This report summarizes the key achievements and challenges...

### Key Metrics

| Metric | 2023 | 2024 | Change |
|--------|------|------|--------|
| Revenue | $10M | $12M | +20% |
| Users | 50K | 75K | +50% |

## Detailed Analysis

The following sections provide...

Excel 转换

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("data.xlsx")

# Each sheet becomes a section
# Data becomes markdown tables
print(result.text_content)

输出示例：

## Sheet1

| Name | Department | Salary |
|------|------------|--------|
| John | Engineering | $80,000 |
| Jane | Marketing | $75,000 |

## Sheet2

| Product | Q1 | Q2 | Q3 | Q4 |
|---------|----|----|----|----|
| Widget A | 100 | 120 | 150 | 180 |

PowerPoint 转换

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("presentation.pptx")

# Each slide becomes a section
# Speaker notes included if present
print(result.text_content)

输出示例：

# Slide 1: Company Overview

Our mission is to...

## Key Points
- Innovation first
- Customer focused
- Global reach

---

# Slide 2: Market Analysis

The market opportunity is significant...

**Notes:** Mention the competitor analysis here

PDF 转换

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("document.pdf")

# Extracts text content
# Tables converted where detected
print(result.text_content)

图像转换（使用视觉模型）

from markitdown import MarkItDown
import anthropic

# Use Claude for image description
client = anthropic.Anthropic()

md = MarkItDown(
    llm_client=client,
    llm_model="claude-sonnet-4-20250514"
)

result = md.convert("diagram.png")
print(result.text_content)

# Output: Description of the image content

批量转换

from markitdown import MarkItDown
from pathlib import Path

def batch_convert(input_dir, output_dir):
    """Convert all Office files to Markdown."""
    md = MarkItDown()
    input_path = Path(input_dir)
    output_path = Path(output_dir)
    output_path.mkdir(exist_ok=True)
    
    extensions = ['.docx', '.xlsx', '.pptx', '.pdf']
    
    for ext in extensions:
        for file in input_path.glob(f'*{ext}'):
            try:
                result = md.convert(str(file))
                output_file = output_path / f"{file.stem}.md"
                
                with open(output_file, 'w') as f:
                    f.write(result.text_content)
                
                print(f"Converted: {file.name}")
            except Exception as e:
                print(f"Error converting {file.name}: {e}")

batch_convert('./documents', './markdown')

最佳实践

1. 检查输出质量：检查转换后的 Markdown 是否准确
2. 处理表格：复杂表格可能需要手动调整
3. 保留结构：在源文档中使用一致的标题级别
4. 处理图像：重要图像可考虑使用视觉模型
5. 版本控制：将转换后的 Markdown 存储在 Git 中进行跟踪

常见模式

文档归档

import os
from datetime import datetime
from markitdown import MarkItDown

def archive_document(doc_path, archive_dir):
    """Convert and archive Office document to Markdown."""
    md = MarkItDown()
    result = md.convert(doc_path)
    
    # Create archive structure
    date_str = datetime.now().strftime('%Y-%m-%d')
    filename = os.path.basename(doc_path)
    base_name = os.path.splitext(filename)[0]
    
    # Save with metadata
    output_content = f"""---
source: {filename}
converted: {date_str}
---

{result.text_content}
"""
    
    output_path = os.path.join(archive_dir, f"{base_name}.md")
    with open(output_path, 'w') as f:
        f.write(output_content)
    
    return output_path

AI 就绪语料库

from markitdown import MarkItDown
from pathlib import Path
import json

def create_ai_corpus(doc_folder, output_file):
    """Convert documents to JSON corpus for AI training/RAG."""
    md = MarkItDown()
    corpus = []
    
    for doc in Path(doc_folder).glob('**/*'):
        if doc.suffix in ['.docx', '.pdf', '.pptx', '.xlsx']:
            try:
                result = md.convert(str(doc))
                corpus.append({
                    'source': str(doc),
                    'filename': doc.name,
                    'content': result.text_content,
                    'type': doc.suffix[1:]
                })
            except Exception as e:
                print(f"Skipped {doc.name}: {e}")
    
    with open(output_file, 'w') as f:
        json.dump(corpus, f, indent=2)
    
    print(f"Created corpus with {len(corpus)} documents")
    return corpus

示例

示例 1：转换文档套件

from markitdown import MarkItDown
from pathlib import Path

def convert_docs_to_wiki(docs_folder, wiki_folder):
    """Convert all Office docs to markdown wiki structure."""
    md = MarkItDown()
    docs_path = Path(docs_folder)
    wiki_path = Path(wiki_folder)
    
    # Create wiki structure
    wiki_path.mkdir(exist_ok=True)
    
    # Create index
    index_content = "# Documentation Index\n\n"
    
    for doc in sorted(docs_path.glob('**/*.docx')):
        try:
            result = md.convert(str(doc))
            
            # Create relative path in wiki
            rel_path = doc.relative_to(docs_path)
            output_file = wiki_path / rel_path.with_suffix('.md')
            output_file.parent.mkdir(parents=True, exist_ok=True)
            
            # Write markdown
            with open(output_file, 'w') as f:
                f.write(result.text_content)
            
            # Add to index
            link = str(rel_path.with_suffix('.md')).replace('\\', '/')
            index_content += f"- [{doc.stem}]({link})\n"
            
            print(f"Converted: {doc.name}")
            
        except Exception as e:
            print(f"Error: {doc.name} - {e}")
    
    # Write index
    with open(wiki_path / 'index.md', 'w') as f:
        f.write(index_content)

convert_docs_to_wiki('./company_docs', './wiki')

示例 2：会议笔记处理器

from markitdown import MarkItDown
import re
from datetime import datetime

def process_meeting_notes(pptx_path):
    """Extract and structure meeting notes from PowerPoint."""
    md = MarkItDown()
    result = md.convert(pptx_path)
    
    # Parse the markdown
    content = result.text_content
    
    # Extract sections
    sections = {
        'attendees': [],
        'agenda': [],
        'decisions': [],
        'action_items': []
    }
    
    current_section = None
    
    for line in content.split('\n'):
        line_lower = line.lower()
        
        if 'attendee' in line_lower or 'participant' in line_lower:
            current_section = 'attendees'
        elif 'agenda' in line_lower:
            current_section = 'agenda'
        elif 'decision' in line_lower:
            current_section = 'decisions'
        elif 'action' in line_lower:
            current_section = 'action_items'
        elif line.strip().startswith(('-', '*', '•')) and current_section:
            sections[current_section].append(line.strip()[1:].strip())
    
    # Generate structured output
    output = f"""# Meeting Notes

**Date:** {datetime.now().strftime('%Y-%m-%d')}
**Source:** {pptx_path}

## Attendees
{chr(10).join('- ' + a for a in sections['attendees'])}

## Agenda
{chr(10).join('- ' + a for a in sections['agenda'])}

## Decisions Made
{chr(10).join('- ' + d for d in sections['decisions'])}

## Action Items
{chr(10).join('- [ ] ' + a for a in sections['action_items'])}
"""
    
    return output

notes = process_meeting_notes('team_meeting.pptx')
print(notes)

示例 3：将 Excel 转换为文档

from markitdown import MarkItDown

def excel_to_data_dictionary(xlsx_path):
    """Convert Excel data model to data dictionary documentation."""
    md = MarkItDown()
    result = md.convert(xlsx_path)
    
    # Add documentation structure
    doc = f"""# Data Dictionary

Generated from: `{xlsx_path}`

{result.text_content}

## Usage Notes

- All tables are derived from the source Excel file
- Review data types and constraints before use
- Contact data team for clarifications

## Change Log

| Date | Change | Author |
|------|--------|--------|
| {datetime.now().strftime('%Y-%m-%d')} | Initial generation | Auto |
"""
    
    return doc

documentation = excel_to_data_dictionary('data_model.xlsx')
with open('data_dictionary.md', 'w') as f:
    f.write(documentation)

限制

• 复杂格式可能会被简化
• 不会嵌入图像（可使用视觉模型生成描述）
• 部分表格结构可能无法完美转换
• 不会保留 Word 中的修订
• 可能无法提取批注

安装

pip install markitdown

# For image/audio processing
pip install markitdown[all]

# For specific features
pip install markitdown[images]  # Image OCR
pip install markitdown[audio]   # Audio transcription

资源

• GitHub 仓库
• PyPI 软件包
• 支持的格式