pywin32 capture thumbnail of first page of docx document

docx_thumbnail.py

import os.path
import time

from PIL import ImageGrab
from win32com.client import constants
import win32com.client as win32


def create_docx_thumbnail(docx_file, output_image):
    # Launch the Word application
    word = win32.Dispatch("Word.Application")

    # Hide the Word application interface (run in background)
    word.Visible = False

    # Open the specified Word document
    doc = word.Documents.Open(os.path.abspath(docx_file))

    selection = doc.ActiveWindow.Selection
    # Navigate to the first page of the document
    selection.GoTo(constants.wdGoToPage, constants.wdGoToAbsolute, 1)

    # Select the content of the first page
    selection.WholeStory()

    # Capture the content of the Word window and save it as an image
    selection.CopyAsPicture()

    # Or paste to a new document
    # new_doc = word.Documents.Add()
    # new_doc.Activate()
    # selection = word.Selection

    # Paste the first page content as an image
    selection.PasteSpecial(DataType=constants.wdPasteMetafilePicture)
    selection.Copy()
    # Add a short delay to ensure the copy operation is complete.
    time.sleep(1)

    # Close the document and the Word application
    doc.Close(SaveChanges=0)
    word.Quit()

    # Save image file from clipboard
    win32clipboard.OpenClipboard()
    try:
        if win32clipboard.IsClipboardFormatAvailable(win32clipboard.CF_DIB):
            data = win32clipboard.GetClipboardData(win32clipboard.CF_DIB)
            image = BmpImagePlugin.DibImageFile(BytesIO(data))
            image.save(output_image, format="PNG")
            return True
    finally:
        win32clipboard.CloseClipboard()