python里面实现doc转html貌似有点麻烦,这里先把doc转为docx,然后再转为html,代码如下
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()