sandbox/docbook/scripts/rstxml2xml.py

   1 #!/usr/bin/env python
   2 import sys
   3 import argparse
   4 if sys.version_info < (3,0):
   5     sys.stderr.write('Need python 3.0 or greater\n')
   6     sys.stderr.wrtie('Script quiting\n')
   7     sys.exit(1)
   8 import xml.etree.cElementTree as etree
   9 import asciitomathml.asciitomathml
  10 from xml.etree.ElementTree import Element, tostring
  11
  12 class InvaidXml(Exception):
  13     # not used at this point
  14     pass
  15
  16 class ConverttoMathml():
  17
  18     def __init__(self):
  19         pass
  20
  21     def _parse_args(self):
  22         desc = """Inserts Mathml elements into an rst document.
  23 In order to use the script, first run rs2txml.py on the RST file.
  24 Then run this script on that resulting file
  25 Or, in one pass: rst2xml.py <infile> | python3 rstxml2mathml.py
  26         """
  27         parser = argparse.ArgumentParser(description=desc)
  28         parser.add_argument('in_file', default = sys.stdin, nargs='?',
  29                 help = 'the file to input; default is standard in')
  30         parser.add_argument('-o,--out', nargs=1, dest="out_file" )
  31         parser.add_argument('--out-encoding', nargs=1, dest="out_encoding" )
  32         parser.add_argument('--in-encoding', nargs=1, dest="in_encoding" )
  33         args =  parser.parse_args()
  34         return args
  35
  36     def convert_to_mathml(self):
  37         args = self._parse_args()
  38         out_encoding = args.out_encoding
  39         if not out_encoding:
  40             out_encoding = 'utf8'
  41         else:
  42             out_encoding = out_encoding[0]
  43         in_encoding = args.in_encoding
  44         if not in_encoding:
  45             in_encoding = 'utf8'
  46         else:
  47             in_encoding = in_encoding[0]
  48         standard_in = False
  49         in_file = args.in_file
  50         out_file = args.out_file
  51         if out_file:
  52             out_file = out_file[0]
  53         if not isinstance(in_file, str):
  54             orig_string = sys.stdin.read()
  55         else:
  56             read_obj = open(in_file, mode='r', encoding=in_encoding)
  57             lines = read_obj.readlines()
  58             orig_string = ''.join(lines)
  59         old_tree = etree.XML(orig_string)
  60         for ma in ['math_block', 'math']:
  61             for e in old_tree.iter(ma):
  62                 math_text = e.text
  63                 e.text = ''
  64                 if ma == 'math_block':
  65                     math_obj =  asciitomathml.asciitomathml.AsciiMathML(mstyle={'displaystyle':'true'})
  66                 else:
  67                     math_obj =  asciitomathml.asciitomathml.AsciiMathML()
  68                 math_obj.parse_string(math_text)
  69                 math_tree = math_obj.get_tree()
  70                 math_string = tostring(math_tree, encoding='ascii') # <encoding=utf8> cannot appear here
  71                 math_tree = etree.XML(math_string)
  72                 e.insert(0, math_tree)
  73         xml_string = tostring(old_tree, encoding=out_encoding)
  74         if out_file:
  75             write_obj = open(out_file, mode='w', encoding=out_encoding)
  76             write_obj.write(xml_string.decode(out_encoding))
  77             write_obj.close()
  78         else:
  79             sys.stdout.write(xml_string.decode(out_encoding))
  80
  81
  82
  83 if __name__ == '__main__':
  84     main_obj = ConverttoMathml()
  85     main_obj.convert_to_mathml()