convert_docx_to_txt.py

"""
Convert all .docx files in a directory and its subdirectories to .md (Markdown) files.
Preserves the original directory structure and creates .md files alongside the .docx files.
"""

import os
from pathlib import Path
from docx2md.convert import do_convert


def convert_docx_to_md(docx_path, md_path):
    """
    Convert a single .docx file to .md (Markdown) format.
    
    Args:
        docx_path: Path to the source .docx file
        md_path: Path where the .md file will be saved
    """
    try:
        # Convert .docx to markdown
        markdown_content = do_convert(str(docx_path))
        
        # Write to .md file
        with open(md_path, 'w', encoding='utf-8') as md_file:
            md_file.write(markdown_content)
        
        print(f"✓ Converted: {docx_path.name}")
        return True
        
    except Exception as e:
        print(f"✗ Error converting {docx_path.name}: {str(e)}")
        return False


def convert_all_docx_in_directory(root_dir):
    """
    Recursively find and convert all .docx files in a directory.
    
    Args:
        root_dir: Root directory to start searching from
    """
    root_path = Path(root_dir)
    
    if not root_path.exists():
        print(f"Error: Directory '{root_dir}' does not exist")
        return
    
    # Find all .docx files recursively
    docx_files = list(root_path.rglob("*.docx"))
    
    if not docx_files:
        print(f"No .docx files found in '{root_dir}'")
        return
    
    print(f"Found {len(docx_files)} .docx files to convert...\n")
    
    successful = 0
    failed = 0
    
    for docx_file in docx_files:
        # Skip temporary Word files (starting with ~$)
        if docx_file.name.startswith('~$'):
            continue
        
        # Create .md filename in the same directory
        md_file = docx_file.with_suffix('.md')
        
        # Convert the file
        if convert_docx_to_md(docx_file, md_file):
            successful += 1
        else:
            failed += 1
    
    print(f"\n{'='*50}")
    print(f"Conversion complete!")
    print(f"Successfully converted: {successful}")
    print(f"Failed: {failed}")
    print(f"{'='*50}")


if __name__ == "__main__":
    # Set the directory to convert
    # Change this to the path of your documentation folder
    directory_to_convert = "cisco-academy-documentation"
    
    print(f"Starting conversion of .docx files to Markdown in '{directory_to_convert}'...")
    print(f"{'='*50}\n")
    
    convert_all_docx_in_directory(directory_to_convert)