from win32com import client as wc
def doc2docx(file_path,file_name):
word = wc.Dispatch("Word.Application")
doc = word.Documents.Open(file_path)
doc.SaveAs(path_chen+file_name+'.docx', 12) #12为docx
doc.Close()
word.Quit()
all_file = os.listdir(path_chen)
for file_name in all_file:
if file_name.endswith(’.doc’) or file_name.endswith(‘DOC’):
file_name_only=file_name[:-4]
doc2docx(path_chen+file_name,file_name_only)
批量把doc转docx
def docx2excel(file,name):
doc=Document(file)
tables=doc.tables
comments=[]
zd=[]
for table in tqdm(tables):
for i in range(1,len(table.rows)):
per_comments=table.cell(i,2).text
per_zd=table.cell(i,1).text
comments.append(per_comments)
zd.append(per_zd)
jiegbc=pd.DataFrame({'注释':comments,'字段':zd},index=range(len(zd)))
jiegbc.to_excel(path_xian+name+'.xlsx',index=False)
all_file = os.listdir(path_xian)
for file_name in all_file:
if file_name.endswith(‘docx’):
file_name_only=file_name[:-5]
docx2excel(path_xian+file_name,file_name_only)
docx表格转excel