po/htm2po.py

   1 #!/usr/bin/python
   2 # -*- coding: utf-8 -*-
   3 import sys, re
   4 from BeautifulSoup import BeautifulSoup
   5
   6 help = """
   7 arg 1 : input file
   8 arg 2 : the locale field which you wanna extract
   9 arg 3 : output po file
  10 """
  11
  12 DESIGN_ID = "Design ID"
  13 dString = {}
  14
  15 def findindexof( word, tds ):
  16         for i, td in enumerate(tds):
  17                 try:
  18                         if td.contents[0]==word:
  19                                 return i
  20                 except IndexError:
  21                         pass
  22         return -1
  23
  24 if __name__=="__main__":
  25         if len(sys.argv)>3:
  26                 in_file = sys.argv[1]
  27                 locale = sys.argv[2]
  28                 po_file = sys.argv[3]
  29         else:
  30                 print help
  31                 sys.exit()
  32
  33         print 'input file :', in_file
  34         print 'locale :', locale
  35         print 'po file :', po_file
  36
  37         f = open( in_file, 'rt' )
  38         soup = BeautifulSoup( f.read() )
  39         f.close()
  40
  41         for i, tr in enumerate(soup('tr')):
  42                 tds = tr('td')
  43                 if i==0:
  44                         iID = findindexof( DESIGN_ID, tds )
  45                         iLocale = findindexof( locale, tds )
  46                         print 'index of design id :',iID, ", index of",locale,":", iLocale
  47                         if iID<0 or iLocale<0:
  48                                 print 'index failed'
  49                                 sys.exit()
  50                         continue
  51
  52                 bPass = True;
  53                 for j,td in enumerate(tds):
  54                         try:
  55                                 if j==iID: ID = td.contents[0]
  56                                 elif j==iLocale: szLocale = td.contents[0]
  57                         except IndexError:
  58                                 bPass = False;
  59                                 pass
  60
  61                 if bPass:
  62                         dString[ID] = szLocale
  63                         #print i, '\t', ID, '\t' ,szLocale
  64
  65         #modify po file
  66         f = open( po_file, 'rt' )
  67         po = f.read().decode('utf-8')
  68         f.close()
  69
  70         for k, v in dString.items():
  71                 #find key
  72                 msg = 'msgid "'+ k + '"'
  73                 sp = po.find( msg )
  74
  75                 #if yes, replace
  76                 if sp>-1:
  77                         lines = po.split('\n')
  78                         for i, line in enumerate(lines):
  79                                 if line.find( msg )>-1:
  80                                         #replace next line
  81                                         print 'replace :',k
  82                                         v = v.replace('\r', '')                 #remove linefeed
  83                                         v = v.replace('\n ', '\\n') #remove carriage return
  84                                         lines[i+1] = 'msgstr "' + v + '"'
  85                         po = "\n".join( lines )
  86
  87                 #if no, append
  88                 #if sp<0:
  89                 #       print 'append', k
  90                 #       po += 'msgid "' + k + '"\n'
  91                 #       po += 'msgstr "' + v + '"\n\n'
  92
  93         newfile = po_file+'.new'
  94         f = open( newfile , 'wt' )
  95         f.write( po.encode('utf-8') )
  96         f. close()
  97         print newfile, 'is created'