Python读取Word文档中的Excel嵌入文件的方法详解

from win32com import client as win32
import os

word = win32.Dispatch("Word.Application")
word.Visible = True
wdDoc = word.Documents.Open(os.path.abspath("test.docx"))
try:
    for shape in wdDoc.InlineShapes:
        if shape.Type != 1 or not shape.OLEFormat.ProgID.startswith("Excel.Sheet"):
            # 要求形状类型为wdInlineShapeEmbeddedOLEObject，是Excel类型的OLE对象
            continue
        shape.OLEFormat.Open()
        xlApp = win32.GetActiveObject('Excel.Application')
        book = xlApp.Workbooks(1)
        print([sht.Name for sht in book.Sheets])
        print(book.Sheets("python").Range("A1").Value)
        book.Close()
finally:
    wdDoc.Close()
    xlApp.Quit()
    word.Quit()

import olefile
from zipfile import ZipFile
from openpyxl import load_workbook

filename = "test.docx"
with ZipFile(filename, "r") as zip_file:
    for name in zip_file.namelist():
        if not name.startswith("word/embeddings/"):
            continue
        with zip_file.open(name) as f:
            if not olefile.isOleFile(f):
                continue
            ole = olefile.OleFileIO(f)
            try:
                book = load_workbook(ole.openstream("package"))
                print(book.sheetnames)
                print(book["python"]["A1"].value)
            except Exception as e:
                print(name, "当前ole对象不是Excel文件:", e)