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