python word,ppt,excel转pdf(word转html)

装依赖

pip install comtypes==1.1.10 

转换

import comtypes.client
import os


def ppt_pdf(path):
	# PPT 转 PDF
    pdf_path = path.replace('ppt', 'pdf') # pdf保存路径 推荐使用绝对路径
    try:
        p = client.CreateObject("PowerPoint.Application")
        ppt = p.Presentations.Open(path)
        ppt.ExportAsFixedFormat(pdf_path, 2, PrintRange=None)
        ppt.Close()
        p.Quit()
    except Exception as e:
        pass
 

def word_pdf(path):
	# Word转pdf
    pdf_path = path.replace('doc', 'pdf')
    w = client.CreateObject("Word.Application")
    doc = w.documents.Open(path)
    doc.ExportAsFixedFormat(pdf_path, 17)
    doc.Close()
    w.Quit()


def excel_pdf(self, path):
	# Excel转pdf
    pdf_path = path.replace('xls', 'pdf')
    xlApp = client.CreateObject("Excel.Application")
    books = xlApp.Workbooks.Open(path)
    books.ExportAsFixedFormat(0, pdf_path)
    xlApp.Quit()

import win32com.client as wc   # doc转docx用
from pydocx import PyDocX      # docx转html用
 
'''
doc文件转docx文件
fullpath:路径+文件名(不带后缀)
如：D:\test\文件1
'''
def doc2docx(fullpath):
    word = wc.Dispatch("WORD.Application")   # 启动word进程
    word.displayalerts=0  # 不警告
    word.visible=0        # 不显示
    #print(fullpath)
    doc = word.documents.Open(fullpath + '.doc')  # D:\test\文件名1.doc
    doc.SaveAs(fullpath,12, False, "", True, "", False, False, False, False)  #转为docx
    doc.Close()
    word.Quit()
 
'''
docx转html
fullpath:路径+文件名(不带后缀)
如：D:\test\文件2
'''
def docx2html(fullpath):
    html = PyDocX.to_html(fullpath + ".docx")  # 转为html，如：D:\test\文件2.docx
    f = open(fullpath + ".html", 'w', encoding="utf-8")  # 变为如：D:\test\文件2.html
    f.write(html)
    f.close()