parseTxt.py

   1 # -*- coding: utf-8 -*-
   2 import os, sys
   3
   4 print 'AO7L TEXT PARSER'
   5 print sys.argv[1]+" -> "+sys.argv[2]
   6
   7 def fixBBCode(line,code,tagname):
   8         line=line.replace(u'['+code.lower()+']',u'<'+tagname+u'>')
   9         line=line.replace(u'[/'+code.lower()+u']',u'</'+tagname+u'>')
  10         line=line.replace(u'['+code.upper()+']',u'<'+tagname+u'>')
  11         line=line.replace(u'[/'+code.upper()+u']',u'</'+tagname+u'>')
  12         return line
  13
  14 with open(sys.argv[2],'w') as out:
  15         out.write(u'text = ""'.encode('utf8'))
  16         with open(sys.argv[1],'r') as inf:
  17                 bracketOpen=False
  18                 for ln, _line in enumerate(inf):
  19                         line = _line.decode('utf8').replace(u'\n',u'')
  20                         hasText=False
  21                         commentlevel=0
  22                         comment=''
  23                         buffer=''
  24                         line=line.replace(u'. . .',u'...') # Fix ellipses
  25
  26                         #BBCode
  27                         line=fixBBCode(line,'b','b')
  28                         line=fixBBCode(line,'i','i')
  29
  30                         for i, c in enumerate(line):
  31                                 #print u'[%d:%d] = %x' % (ln,i,c.encode('utf8'))
  32                                 if c == u']':
  33                                         commentlevel-=1
  34                                         if commentlevel < 0:
  35                                                 commentlevel=0
  36                                         print 'Line %d, char %d: CLEVEL %d' % (ln,i,commentlevel)
  37                                         if commentlevel==0:
  38                                                 if comment is not '':
  39                                                         out.write((u"\n/* "+comment.strip()+u" */").encode('utf8'))
  40                                                         comment=''
  41                                                 continue
  42                                 if c == u'[':
  43                                         commentlevel+=1
  44                                         print 'Line %d, char %d: CLEVEL %d' % (ln,i,commentlevel)
  45                                         if buffer is not '':
  46                                                 out.write((u'\ntext += "'+buffer.replace(u'"',u'&quot;').replace(u'{BR}',u'";\ntext += "').strip()+u'";').encode('utf8'))
  47                                                 buffer=''
  48                                         if commentlevel == 1:
  49                                                 continue
  50                                 if c in (u'“',u'”'):
  51                                         c = u'"'
  52                                 if commentlevel > 0:
  53                                         comment += c
  54                                 else:
  55                                         buffer += c
  56                                         if (len(buffer)%90) == 0:
  57                                                 buffer += u'{BR}'
  58                         if buffer is not '':
  59                                 out.write((u"\ntext += \"<p>"+buffer.replace(u'"',u"&quot;").replace(u'{BR}',u'";\ntext += "').strip()+u"</p>\";").encode('utf8'))
  60                         if comment is not '':
  61                                 out.write((u"\n/* "+comment.strip()+u" */").encode('utf8'))