word 转pdf 再转图片--用在轻社群发文章

#! /usr/bin/env python

# -*- coding: utf-8 -*-

import fitz

import glob

import os

from win32com.client import Dispatch, constants, gencache

import sys,shutil

import datetime

print sys.getdefaultencoding()

#################word

def doc2pdf(input, output):
  w = Dispatch("Word.Application")
    
  try:
     doc = w.Documents.Open(input, ReadOnly = 1)
     doc.ExportAsFixedFormat(output, constants.wdExportFormatPDF, 
      Item = constants.wdExportDocumentWithMarkup, CreateBookmarks = constants.wdExportCreateHeadingBookmarks)
     return 0
  except Exception ,e:
     print str(e)
     return 1
  finally:
     w.Quit(constants.wdDoNotSaveChanges)
    

# Generate all the support we can.

def GenerateSupport():
  # 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)
     

#####################

inputdoc = sys.argv[1]

store_tmp = unicode(os.path.splitext(inputdoc)[0],'GBK')

print inputdoc

outputpdf = os.path.splitext(inputdoc)[0]+'.pdf'

if (not os.path.isabs(inputdoc)):
     inputdoc = os.path.abspath(inputdoc)

if (not os.path.isabs(outputpdf)):
     outputpdf = os.path.abspath(outputpdf)

try:
     GenerateSupport()
     rc = doc2pdf(inputdoc, outputpdf)
    

except Exception,e:
     print str(e)

def rightinput(desc):
     flag=True
     while(flag):
         instr = raw_input(desc)
         try:
             intnum = eval(instr)
             if type(intnum)==int:
                 flag = False
         except:
             print('ÇëÊäÈëÕýÕûÊý£¡')
             pass
     return intnum

#pdffile = glob.glob("*.pdf")[0]

pdffile = outputpdf

pdffile =  unicode(pdffile , "GBK")

print pdffile

doc = fitz.open(pdffile)

#flag = rightinput("ÊäÈ룺1£ºÈ«²¿Ò³Ã棻2£ºÑ¡ÔñÒ³Ãæ\t")

flag = 1

if flag == 1:
     strat = 0
     totaling = doc.pageCount

# else:

#   strat = rightinput('ÊäÈëÆðʼҳÃ棺') - 1

#   totaling = rightinput('ÊäÈë½áÊøÒ³Ã棺')

pdf2png = os.path.join(os.path.dirname(__file__),'pdf2png')

pdf2png_old = os.path.join(pdf2png,'old')

if not os.path.isdir(pdf2png):
     os.mkdir(pdf2png)

if not os.path.isdir(pdf2png_old):
     os.mkdir(pdf2png_old)

n_t = datetime.datetime.now().strftime('%Y%m%d%H%M%S')

for  f in os.listdir(pdf2png):
         #print f.endswith('png')
         if f.endswith('png'):
             try:
                 src = os.path.join(pdf2png,f)
                 dst = os.path.join(pdf2png_old,f)
                 shutil.move(src,dst)
             except Exception,e:
                 pass

for pg in range(strat, totaling):
     page = doc[pg]
     zoom = int(100)
     rotate = int(0)
     trans = fitz.Matrix(zoom / 100.0, zoom / 100.0).preRotate(rotate)
     pm = page.getPixmap(matrix=trans, alpha=False)
     pm.writePNG(pdf2png+'/%s_%s_%s.png' % (store_tmp,n_t,str(pg+1)))