tools/dev/create_unimap.py

   1 #!/usr/bin/env python
   2
   3 # $Id$
   4 # Author: Lea Wiemann <LeWiemann@gmail.com>
   5 # Copyright: This file has been placed in the public domain.
   6
   7 # Call: create_unimap.py < unicode.xml > unicode_latex.py
   8 #
   9 # Get unicode.xml from
  10 # <http://www.w3.org/2003/entities/xml/unicode.xml>.
  11
  12 from xml.dom import minidom
  13 import sys
  14 import pprint
  15
  16 def w(s):
  17     if isinstance(s, unicode):
  18         s = s.encode('utf8')
  19     sys.stdout.write(s)
  20
  21 text_map = {}
  22 math_map = {}
  23
  24 class Visitor:
  25
  26     """Node visitor for contents of unicode.xml."""
  27
  28     def visit_character(self, node):
  29         for n in node.childNodes:
  30             if n.nodeName == 'latex':
  31                 code = node.attributes['dec'].value
  32                 if '-' in code:
  33                     # I don't know what this means, but we probably
  34                     # don't need it....
  35                     continue
  36                 if int(code) < 128:
  37                     # Wrong (maps "-" to "$-$", which is too wide) and
  38                     # unnecessary (maps "a" to "{a}").
  39                     continue
  40                 latex_code = n.childNodes[0].nodeValue.encode('ascii').strip()
  41                 if node.attributes['mode'].value == 'math':
  42                     math_map[unichr(int(code))] = '$%s$' % latex_code
  43                 else:
  44                     text_map[unichr(int(code))] = '{%s}' % latex_code
  45
  46 def call_visitor(node, visitor=Visitor()):
  47     if isinstance(node, minidom.Text):
  48         name = 'Text'
  49     else:
  50         name = node.nodeName.replace('#', '_')
  51     if hasattr(visitor, 'visit_' + name):
  52         getattr(visitor, 'visit_' + name)(node)
  53     for child in node.childNodes:
  54         call_visitor(child)
  55     if hasattr(visitor, 'depart_' + name):
  56         getattr(visitor, 'depart_' + name)(node)
  57
  58 document = minidom.parse(sys.stdin)
  59 call_visitor(document)
  60
  61 unicode_map = math_map
  62 unicode_map.update(text_map)
  63 # Now unicode_map contains the text entries plus dollar-enclosed math
  64 # entries for those chars for which no text entry exists.
  65
  66 print '# $%s$' % 'Id'
  67 print '# Author: Lea Wiemann <LeWiemann@gmail.com>'
  68 print '# Copyright: This file has been placed in the public domain.'
  69 print
  70 print '# This is a mapping of Unicode characters to LaTeX equivalents.'
  71 print '# The information has been extracted from'
  72 print '# <http://www.w3.org/2003/entities/xml/unicode.xml>, written by'
  73 print '# David Carlisle and Sebastian Rahtz.'
  74 print '#'
  75 print '# The extraction has been done by the "create_unimap.py" script'
  76 print '# located at <http://docutils.sf.net/tools/dev/create_unimap.py>.'
  77 print
  78 print 'unicode_map = %s' % pprint.pformat(unicode_map, indent=0)