#!/usr/bin/python # tz_word_doc.py # --copyright-- Copyright 2007 (C) Tranzoa, Co. All rights reserved. Warranty: You're free and on your own here. This code is not necessarily up-to-date or of public quality. # --url-- http://www.tranzoa.net/tzpython/ # --email-- pycode is the name to send to. tranzoa.com is the place to send to. # --bodstamps-- # January 18, 2008 bar # January 23, 2008 bar handle errors # March 19, 2008 bar save_as() # May 17, 2008 bar email adr # August 23, 2008 bar don't error out on linux systems # October 8, 2008 bar try both word app names # November 22, 2008 bar back-slash-ize the file name # November 29, 2011 bar pyflake cleanup # May 27, 2012 bar doxygen namespace # January 10, 2013 bar update_docx_image() # January 11, 2013 bar move that zip file update logic over to zipem # --eodstamps-- ## \file # \namespace tzpython.tz_word_doc # # # Operate on Word documents # # import os import sys import replace_file import zipem noway = True try : import win32com.client import pywintypes noway = False except : e = sys.exc_info() # print e[0], e[1], e[2] pass import tzlib class a_word_doc_exception(Exception) : pass class a_word_doc(object) : def __init__(me, file_name) : me.app = None me.doc = None me.file_name = os.path.normpath(os.path.abspath(file_name)) if noway : raise a_word_doc_exception("I can use Word only under Windows for now!") for wa in [ 'Word.Application', 'Word.Application.8' ] : try : me.app = win32com.client.Dispatch(wa) break except AttributeError : e = sys.exc_info() print e[0], e[1], e[2] es = "Cannot find " + wa except pywintypes.com_error : e = sys.exc_info() print e[0], e[1], e[2] es = "Cannot find '" + wa + "' name" pass if not me.app : raise a_word_doc_exception(es) try : me.doc = me.app.Documents.Open(FileName = me.file_name, ConfirmConversions = False, ReadOnly = True, AddToRecentFiles = False, PasswordDocument = "", PasswordTemplate = "", Revert = False, WritePasswordDocument = "", WritePasswordTemplate = "", # Format = win32com.client.constants.wdOpenFormatAuto ) except AttributeError : me.close() e = sys.exc_info() print e[0], e[1], e[2] raise a_word_doc_exception("Cannot open " + wa) except pywintypes.com_error : me.close() raise IOError, "Cannot use " + wa + " to open %s" % ( me.file_name ) pass def as_unicode(me) : if not me.doc : return(None) return(me.doc.Content.Text) def as_utf8(me) : if not me.doc : return(None) return(me.doc.Content.Text.encode('utf-8')) def as_ascii(me) : if not me.doc : return(None) return(tzlib.lf_only(tzlib.ascii(me.as_unicode()))) def as_quiet_ascii(me) : if not me.doc : return(None) return(me.as_ascii().replace('\007', "")) def __str__(me) : return(me.as_quiet_ascii()) fmts = {} fmts['wdFormatDocument'] = 0 fmts['wdFormatDocument97'] = 0 fmts['wdFormatTemplate'] = 1 fmts['wdFormatTemplate97'] = 1 fmts['wdFormatText'] = 2 fmts['wdFormatTextLineBreaks'] = 3 fmts['wdFormatDOSText'] = 4 fmts['wdFormatDOSTextLineBreaks'] = 5 fmts['wdFormatRTF'] = 6 fmts['wdFormatEncodedText'] = 7 fmts['wdFormatUnicodeText'] = 7 fmts['wdFormatHTML'] = 8 fmts['wdFormatWebArchive'] = 9 fmts['wdFormatFilteredHTML'] = 10 fmts['wdFormatXML'] = 11 fmts['wdFormatXMLDocument'] = 12 fmts['wdFormatXMLDocumentMacroEnabled'] = 13 fmts['wdFormatXMLTemplate'] = 14 fmts['wdFormatXMLTemplateMacroEnabled'] = 15 fmts['wdFormatDocumentDefault'] = 16 fmts['wdFormatPDF'] = 17 fmts['wdFormatXPS'] = 18 fmts['wdFormatFlatXML'] = 19 fmts['wdFormatFlatXMLMacroEnabled'] = 20 fmts['wdFormatFlatXMLTemplate'] = 21 fmts['wdFormatFlatXMLTemplateMacroEnabled'] = 22 try : fmts['wdFormatDocument'] = win32com.client.constants.wdFormatDocument except AttributeError : pass except NameError : pass try : fmts['wdFormatText'] = win32com.client.constants.wdFormatText except AttributeError : pass except NameError : pass try : fmts['wdFormatRTF'] = win32com.client.constants.wdFormatRTF except AttributeError : pass except NameError : pass try : fmts['wdFormatHTML'] = win32com.client.constants.wdFormatHTML except AttributeError : pass except NameError : pass try : fmts['wdFormatHTML'] = win32com.client.constants.wdFormatHTML except AttributeError : pass except NameError : pass try : fmts['wdFormatPDF'] = win32com.client.constants.wdFormatPDF except AttributeError : pass except NameError : pass try : fmts['wdFormatXML'] = win32com.client.constants.wdFormatXML except AttributeError : pass except NameError : pass try : fmts['wdFormatXPS'] = win32com.client.constants.wdFormatXPS except AttributeError : pass except NameError : pass ext_fmts = { '.doc' : fmts['wdFormatDocument'], '.txt' : fmts['wdFormatText'], '.rtf' : fmts['wdFormatRTF'], '.htm' : fmts['wdFormatHTML'], '.html' : fmts['wdFormatHTML'], '.pdf' : fmts['wdFormatPDF'], '.xml' : fmts['wdFormatXML'], '.xps' : fmts['wdFormatXPS'], } def save_as(me, file_name, fmt = None) : file_name = os.path.abspath(file_name) ( fn, ext) = os.path.splitext(file_name) if fmt is None : fmt = me.ext_fmts.get(ext.lower(), me.fmts['wdFormatDocument']) elif isinstance(fmt, basestring) : fmt = me.fmts[fmt] tfn = fn + "_tmp" + ext me.doc.SaveAs(tfn, fmt) me.doc.Close() # unfortunately, replace file fails if we don't do this (timing?) when we write a non-.pdf file replace_file.replace_file(file_name, tfn, file_name + ".bak") def close(me) : me.doc = None a = me.app me.app = None if a : # a.Quit() # unfortunately, this closes word 2007 completely, including docs you have open. so we'll leave it running pass pass def __del__(me) : me.close() pass # a_word_doc def update_docx_image(dfn, img_name, ifn, min_size_factor = None, max_size_factor = None) : """ Update the given image (by base name - e.g. "image1.png") in a .docx file with the given image file. Do not update if the new image is smaller than the current image times the 'min_size_factor'. Do not update if the new image is lareger than the current image times the 'max_size_factor'. Return True if the image is updated OK, False otherwise. """ img_name = os.path.join('word', 'media', img_name) return(zipem.update_zip_file(dfn, img_name, ifn, min_size_factor = min_size_factor, max_size_factor = max_size_factor)) # # # Test # # if __name__ == '__main__' : import sys import TZCommandLineAtFile del(sys.argv[0]) TZCommandLineAtFile.expand_at_sign_command_line_files(sys.argv) while len(sys.argv) : ifn = sys.argv.pop(0) ofn = sys.argv.pop(0) doc = a_word_doc(ifn) if os.path.splitext(ofn)[1].lower() == '.txt' : txt = doc.as_quiet_ascii() doc.close() tfn = ofn + ".tmp" tzlib.write_whole_text_file(tfn, txt) replace_file.replace_file(ofn, tfn, ofn + ".bak") else : doc.save_as(ofn) pass pass # # # # eof