sandbox/docbook/scripts/rstxml2xml.py

   1 #!/usr/bin/env 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 io import StringIO
   8
   9
  10 import asciitomathml.asciitomathml
  11 from xml.etree.ElementTree import Element, tostring
  12 import xml.etree.cElementTree as etree
  13 import tempfile, subprocess, os
  14 from xml.sax import InputSource
  15
  16 class InvaidXml(Exception):
  17     pass
  18
  19
  20 class FixTree(xml.sax.ContentHandler):
  21
  22   def __init__(self, mathml=True, raw_xml=True):
  23         self._characters = ''
  24         self._mathml = mathml
  25         self._raw_xml = raw_xml
  26         self._write_raw = False
  27         self._ns_dict = {'http://www.w3.org/XML/1998/namespace': "xml"}
  28
  29
  30   def startDocument(self):
  31       pass
  32
  33
  34   def characters (self, characters):
  35     self._characters += characters
  36
  37
  38   def startElementNS(self, name, qname, attrs):
  39         self._write_text()
  40         ns = name[0]
  41         el_name = name[1]
  42         self._write_string('<')
  43         if el_name == 'raw':
  44             if attrs.get((None, 'format')) == 'xml' and self._raw_xml:
  45                 self._write_raw = True
  46         if ns:
  47             self._write_string('ns1:%s' % el_name)
  48         else:
  49             self._write_string(el_name)
  50         if ns:
  51             self._write_string(' xmlns:ns1="%s"' % ns)
  52
  53         the_keys = list(attrs.keys())
  54         counter = 1
  55         for the_key in the_keys:
  56             counter +=1
  57             ns_att = the_key[0]
  58             att_name = the_key[1]
  59             value = attrs[the_key]
  60             ns_prefix = self._ns_dict.get(ns_att)
  61             if ns_att and not ns_prefix:
  62                 raise InvaidXml('No namespace for "%s"\n' % (ns_att))
  63             if ns_att and ns_prefix == 'xml':
  64                 self._write_string(' xml:%s="%s"' % (att_name, value))
  65             elif ns_att:
  66                 raise InvaidXml('Sorry, but don\'t know what to do with ns "%s"\n' % (ns_prefix))
  67             else:
  68                 self._write_string(' %s="%s"' % (att_name, value))
  69         self._write_string('>')
  70
  71
  72
  73   def _write_text(self, raw = False):
  74         if raw:
  75             text = self._characters
  76         else:
  77             text =  xml.sax.saxutils.escape(self._characters)
  78         self._write_string(text)
  79         self._characters = ''
  80
  81   def _write_string(self, the_string):
  82       sys.stdout.write(the_string)
  83
  84   def endElementNS(self, name, qname):
  85         ns = name[0]
  86         el_name = name[1]
  87         if (el_name == 'math_block' and  self._mathml) or (el_name == 'math' and self._mathml) :
  88             # math_obj =  asciitomathml.asciitomathml.AsciiMathML()
  89             if el_name == 'math_block':
  90                 math_obj =  asciitomathml.asciitomathml.AsciiMathML(mstyle={'displaystyle':'true'})
  91             else:
  92                 math_obj =  asciitomathml.asciitomathml.AsciiMathML()
  93             math_obj.parse_string(self._characters)
  94             math_tree = math_obj.get_tree()
  95             math_string = tostring(math_tree, encoding="us-ascii")
  96             self._write_string(math_string)
  97             self._characters = ''
  98         elif el_name == 'raw' and self._write_raw:
  99             self._write_text(raw = True)
 100             self._write_raw = False
 101         else:
 102             self._write_text()
 103         if ns:
 104             raise InvaidXml('Should not be namespace "%s" here\n' % (ns))
 105         else:
 106             self._write_string('</%s>' % el_name)
 107
 108
 109
 110 class ConverttoMathml:
 111
 112
 113     def _init_(self):
 114         pass
 115
 116     def _parse_args(self):
 117         desc = """Inserts Mathmx elements into an rst document.
 118 In order to use the script, first run rs2txml.py on the RST file.
 119 Then run this script on that resulting file
 120 Or, in one pass: rst2xml.py <infile> | python3 rstxml2mathml.py
 121         """
 122         parser = argparse.ArgumentParser(description=desc)
 123         parser.add_argument('--no-mathml', action="store_const", const=True, dest='no_mathml')
 124         parser.add_argument('--no-rawxml', action="store_const", const=True, dest='no_rawxml')
 125         parser.add_argument('--fix-sh', action="store_const", const=True, dest='fix_sh', default=False)
 126         parser.add_argument('in_file', default = sys.stdin, nargs='?',
 127                 help = 'the file to input; default is standard in')
 128         args =  parser.parse_args()
 129         return args
 130
 131
 132     def convert_to_mathml(self):
 133         args = self._parse_args()
 134         standard_in = False
 135         in_file = args.in_file
 136         no_mathml = args.no_mathml
 137         if no_mathml:
 138             mathml = False
 139         else:
 140             mathml = True
 141         no_rawxml = args.no_rawxml
 142         if no_rawxml:
 143             raw_xml = False
 144         else:
 145             raw_xml = True
 146         fix_sh = args.fix_sh
 147         if not isinstance(in_file, str):
 148             standard_in = True
 149             the_string = sys.stdin.read()
 150         if standard_in:
 151             read_obj = StringIO(the_string)
 152         else:
 153             read_obj = open(in_file, 'r')
 154         the_handle=FixTree(mathml= mathml, raw_xml = raw_xml)
 155         parser = xml.sax.make_parser()
 156         parser.setFeature(feature_namespaces, 1)
 157         parser.setContentHandler(the_handle)
 158         parser.setFeature("http://xml.org/sax/features/external-general-entities", True)
 159         try:
 160             parser.parse(read_obj)
 161         except xml.sax._exceptions.SAXParseException as error:
 162             msg = error.args[0]
 163             raise InvaidXml(msg)
 164         except InvaidXml as error:
 165             msg = error.args[0]
 166             raise InvaidXml(msg)
 167         read_obj.close()
 168
 169 if __name__ == '__main__':
 170     main_obj = ConverttoMathml()
 171     main_obj.convert_to_mathml()