compoundfiles

import os from compoundfiles import CompoundFile # IMPORTANT: Replace 'path/to/your/ole_document.doc' with the actual path # to an OLE Compound Document (e.g., old .doc, .xls, .ppt files). # This library is for reading existing OLE files. # A sample OLE file (e.g., an old .doc) is required for this example to run meaningfully. file_path = os.environ.get('OLE_FILE_PATH', './example.doc') # Placeholder for agent if not os.path.exists(file_path): print(f"Warning: OLE file not found at '{file_path}'. Please provide a valid path.") print("Example usage requires an actual OLE Compound Document file.") print("You can provide it via OLE_FILE_PATH environment variable or replace the hardcoded path.") else: try: with CompoundFile(file_path) as cf: print(f"Successfully opened OLE Compound Document: {file_path}") print("\nRoot entries:") for entry in cf.listdir('/'): print(f"- {entry}") # Example: Try to read a specific stream if it exists # 'WordDocument' is a common stream name in OLE Word documents. if 'WordDocument' in cf.listdir('/'): with cf.open('WordDocument') as stream: # Read a portion of the stream (actual content is often binary/complex) content = stream.read(50) print(f"\nContent of '/WordDocument' stream (first 50 bytes): {content!r}") else: print("\nNo 'WordDocument' stream found. Listing other common streams if available...") for common_stream in ['Workbook', 'PowerPoint Document', 'Contents']: # Add other common stream names if common_stream in cf.listdir('/'): print(f"Found stream: '{common_stream}'") except Exception as e: print(f"\nError processing OLE file '{file_path}': {e}") print("Ensure the file is a valid OLE Compound Document and not corrupted.")