sandbox/paultremblay/scripts/rstxml2mathml_sax.py

   1 #!/usr/bin/python
   2 #   $Id: sax_complete_copy.py 54 2011-04-17 15:44:41Z cynthia $
   3
   4 import os, sys, argparse, io
   5 import xml.sax.handler
   6 from xml.sax.handler import feature_namespaces
   7 from StringIO import StringIO
   8
   9 """
  10 if sys.version_info < (3,):
  11 else:
  12     from io import StringIO
  13 """
  14
  15 import asciimathml
  16 from xml.etree.ElementTree import Element, tostring
  17 import xml.etree.cElementTree as etree
  18 import tempfile, subprocess, os
  19 # import docutils.math.latex2mathml
  20 from xml.sax import InputSource
  21
  22 """
  23 if sys.version_info < (3,):
  24     sys.stderr.write('Only run with pyton 3\n')
  25     sys.stderr.write('Script now quiting\n')
  26     sys.exit(1)
  27 """
  28
  29
  30 class CopyTree(xml.sax.ContentHandler):
  31
  32
  33
  34   def __init__(self, mathml):
  35         self.__characters = ''
  36         self.__mathml = mathml
  37         self.__ns_dict = {'http://www.w3.org/XML/1998/namespace': "xml"}
  38         self.__raw = False
  39         self.__fix_soft_hyphens = False
  40
  41
  42   def startDocument(self):
  43       pass
  44
  45   def startElement(self, name, qname):
  46       print('okay')
  47
  48   def characters (self, characters):
  49     self.__characters += characters
  50
  51
  52   def startElementNS(self, name, qname, attrs):
  53         self.__write_text()
  54         ns = name[0]
  55         el_name = name[1]
  56         sys.stdout.write('<')
  57         if el_name == 'raw':
  58             if attrs.get((None, 'format')) == 'xml':
  59                 self.__raw = True
  60         if ns:
  61             sys.stdout.write('ns1:%s' % el_name)
  62         else:
  63             sys.stdout.write(el_name)
  64         if ns:
  65             sys.stdout.write(' xmlns:ns1="%s"' % ns)
  66
  67         the_keys = list(attrs.keys())
  68         counter = 1
  69         for the_key in the_keys:
  70             counter +=1
  71             ns_att = the_key[0]
  72             att_name = the_key[1]
  73             value = attrs[the_key]
  74             ns_prefix = self.__ns_dict.get(ns_att)
  75             if ns_att and not ns_prefix:
  76                 sys.stderr.write('No name space for "%s"\n' % (ns_att))
  77                 sys.exit(1)
  78             if ns_att and ns_prefix == 'xml':
  79                 sys.stdout.write(' xml:%s="%s"' % (att_name, value))
  80             elif ns_att:
  81                 sys.stderr.write('Sorry, but don\'t know what to do with ns "%s"\n' % (ns_prefix))
  82                 sys.exit(1)
  83 #            if ns_att and ns_att != ns:
  84 #                sys.stdout.write(' xmlns:ns%s="%s"' % (counter,ns_att))
  85 #            if ns_att and ns_att == ns:
  86 #                sys.stdout.write(' ns1:%s="%s"' % (att_name, value))
  87             else:
  88                 sys.stdout.write(' %s="%s"' % (att_name, value))
  89         sys.stdout.write('>')
  90
  91
  92
  93   def __write_text(self, raw = False):
  94         soft_hyphen = chr(173)
  95         if raw:
  96             text = self.__characters
  97         else:
  98             text =  xml.sax.saxutils.escape(self.__characters)
  99             if self.__fix_soft_hyphens:
 100                 text = text.replace(soft_hyphen, '-')
 101         sys.stdout.write(text)
 102         self.__characters = ''
 103
 104   def endElementNS(self, name, qname):
 105         ns = name[0]
 106         el_name = name[1]
 107         if (el_name == 'math_block' and  self.__mathml == 'ascii') or (el_name == 'math' and self.__mathml == 'ascii'):
 108             raw_tree  = asciimathml.parse(self.__characters)[0]
 109             math_tree = Element('math', title="%s" % self.__characters, xmlns="http://www.w3.org/1998/Math/MathML")
 110             math_tree.append(raw_tree)
 111             string_tree = tostring(math_tree, encoding="utf-8")
 112             sys.stdout.write(string_tree.decode('utf8'))
 113             """
 114             if sys.version_info < (3,):
 115                 print(type(string_tree))
 116                 print()
 117                 sys.stdout.write(string_tree.decode('utf8'))
 118                 # sys.stdout.write(line.encode('utf8'))
 119             else:
 120                 sys.stdout.write(string_tree.decode())
 121             """
 122             self.__characters = ''
 123         elif (el_name == 'math_block' and  self.__mathml == 'latex') or (el_name == 'math' and self.__mathml == 'latex'):
 124             raw_tree = self.__tralics()
 125             if raw_tree == None:
 126                 self.__write_text()
 127             else:
 128                 raw_tree = raw_tree[0]
 129                 math_tree = Element('math', title="%s" % self.__characters, xmlns="http://www.w3.org/1998/Math/MathML")
 130                 math_tree.append(raw_tree)
 131                 string_tree = tostring(math_tree, encoding="utf-8").decode()
 132                 sys.stdout.write(string_tree)
 133                 self.__characters = ''
 134         elif el_name == 'raw' and self.__raw:
 135             self.__write_text(raw = True)
 136         else:
 137             self.__write_text()
 138         if ns:
 139             sys.stderr.write('Should not be namespace "%s" here\n' % (ns))
 140             sys.exit(1)
 141             sys.stdout.write('</ns1:%s>' % el_name)
 142         else:
 143             sys.stdout.write('</%s>' % el_name)
 144
 145   def __python_latex_math(self):
 146         """
 147         Python code seriously broken
 148
 149         """
 150         try:
 151             mathml_tree = docutils.math.latex2mathml.parse_latex_math(self.__characters)
 152         except SyntaxError:
 153             return self.__characters
 154         math_code = ''.join(mathml_tree.xml())
 155         return math_code
 156
 157   def __tralics(self):
 158         num, tex_file = tempfile.mkstemp(suffix='.tex')
 159         write_obj = open(tex_file, 'w')
 160         write_obj.write('$')
 161         write_obj.write(self.__characters)
 162         write_obj.write('$')
 163         write_obj.close()
 164         num, bogus_out = tempfile.mkstemp()
 165         bogus_out = open(bogus_out, 'w')
 166         p = subprocess.call(['tralics', '-silent', '-utf8output', '-noentnames', tex_file], stdout=bogus_out)
 167         bogus_out.close()
 168         dir_name = os.path.dirname(tex_file)
 169         filename, ext = os.path.splitext(tex_file)
 170         xml_file = filename + '.xml'
 171         log_file = filename + '.log'
 172         xml_file = os.path.join(dir_name, xml_file)
 173         if not os.path.isfile(xml_file):
 174             sys.stderr.write('Cannot find file %s\n"' % xml_file)
 175             sys.stderr.write('Bug, program now quiting\n')
 176             sys.exit(1)
 177         tree = etree.ElementTree()
 178         read_obj = open(xml_file, 'r')
 179         xml_tree = tree.parse(xml_file)
 180         found = None
 181         while not found:
 182             for child in xml_tree:
 183                 if child.tag == '{http://www.w3.org/1998/Math/MathML}math':
 184                     found = 1
 185                     break
 186             try:
 187                 xml_tree = xml_tree[0]
 188             except IndexError:
 189                 sys.stderr.write('Could not find any latex math\n')
 190                 break
 191         if not found:
 192             return None
 193         return xml_tree
 194
 195 class ConverttoMathml:
 196
 197
 198     def __init__(self):
 199         pass
 200
 201     def __parse_args(self):
 202         desc = """Inserts Mathmx elements into an rst document.
 203 In order to use the script, first run rs2txml.py on the RST file.
 204 Then run this script on that resulting file
 205 Or, in one pass: rst2xml.py <infile> | python3 rstxml2mathml.py
 206         """
 207         parser = argparse.ArgumentParser(description=desc)
 208         parser.add_argument('--mathml', choices = ['latex', 'ascii'], nargs=1 ) # much better--demand an arg; the option is still optional
 209         parser.add_argument('in_file', default = sys.stdin, nargs='?',
 210                 help = 'the file to input; default is standard in')
 211         args =  parser.parse_args()
 212         return args
 213
 214
 215     def convert_to_mathml(self):
 216         args = self.__parse_args()
 217         standard_in = False
 218         in_file = args.in_file
 219         mathml = args.mathml
 220         if mathml:
 221             mathml = mathml[0]
 222         if not isinstance(in_file, str):
 223             standard_in = True
 224             the_string = sys.stdin.read()
 225         if standard_in:
 226             read_obj = StringIO(the_string)
 227         else:
 228             read_obj = open(in_file, 'r')
 229         the_handle=CopyTree(mathml)
 230         parser = xml.sax.make_parser()
 231         parser.setFeature(feature_namespaces, 1)
 232         parser.setContentHandler(the_handle)
 233         parser.setFeature("http://xml.org/sax/features/external-general-entities", True)
 234         try:
 235             parser.parse(read_obj)
 236         except xml.sax._exceptions.SAXParseException as msg:
 237             print(str(msg))
 238             sys.exit(1)
 239         read_obj.close()
 240
 241 if __name__ == '__main__':
 242     main_obj = ConverttoMathml()
 243     main_obj.convert_to_mathml()