工作平台:window10
工作环境:python2.7.13
python 开发包:pywin32 /pypdf2
开发包下载命令:pip install pywin32/pip install pypdf2
若出现pywin32无法安装的情况,请下载whl包:https://www.lfd.uci.edu/~gohlke/pythonlibs/#pywin32,使用管理员权限进行安装
源代码:
# -*- coding:utf-8 -*-
# doc2pdf.py: python script to convert doc to pdf with bookmarks!
# Requires Office
# Requires python for win32 extension
# requires python for pypdf2
import os
from win32com.client import Dispatch, constants, gencache
from PyPDF2 import PdfFileWriter,PdfFileReader,PdfFileMerger
#word批量转pdf
def word2pdf_patch(fileIn_path,fileOut_path):
# from config import REPORT_DOC_PATH,REPORT_PDF_PATH
fileIn_path = fileIn_path.decode('utf-8')
path_list = os.listdir(fileIn_path)
for item in path_list:
print(item)
rc = word2pdf(os.path.join(fileIn_path,item))
#word转pdf
def word2pdf(input):
output = input
try:
print input, output
# enable python COM support for Word 2007
# this is generated by: makepy.py -i "Microsoft Word 12.0 Object Library"
gencache.EnsureModule('{00020905-0000-0000-C000-000000000046}', 0, 8, 4)
# 开始转换
w = Dispatch("Word.Application")
try:
doc = w.Documents.Open(input, ReadOnly=1)
doc.ExportAsFixedFormat(output, constants.wdExportFormatPDF, \
Item=constants.wdExportDocumentWithMarkup,
CreateBookmarks=constants.wdExportCreateHeadingBookmarks)
except:
print ' exception'
finally:
w.Quit(constants.wdDoNotSaveChanges)
if os.path.isfile(output):
print 'translate success'
return True
else:
print 'translate fail'
return False
except:
print ' exception'
return -1
#合并pdf为一个文件
def merge_pdf(infnList, outfn):
pdf_output = PdfFileWriter()
for infn in infnList:
pdf_input = PdfFileReader(open(infn, 'rb'))
# 获取 pdf 共用多少页
page_count = pdf_input.getNumPages()
print(page_count)
for i in range(page_count):
pdf_output.addPage(pdf_input.getPage(i))
pdf_output.write(open(outfn, 'wb'))
if __name__ == '__main__':
print 'starting dealing'