sandbox/paultremblay/scripts/rsttoxml.py

   1 import docutils.core
   2 from docutils import writers
   3
   4 import docutils
   5 from docutils import frontend, writers
   6 from docutils import languages, Component
   7 from docutils.transforms import universal
   8 from docutils.transforms import math
   9
  10 __docformat__ = 'reStructuredText'
  11
  12 class MWriter(writers.Writer, Component):
  13
  14     supported = ('xml',)
  15     """Formats this writer supports."""
  16
  17     settings_spec = (
  18         '"Docutils XML" Writer Options',
  19         'Warning: the --newlines and --indents options may adversely affect '
  20         'whitespace; use them only for reading convenience.',
  21         (('Generate XML with newlines before and after tags.',
  22           ['--newlines'],
  23           {'action': 'store_true', 'validator': frontend.validate_boolean}),
  24          ('Generate XML with indents and newlines.',
  25           ['--indents'],
  26           {'action': 'store_true', 'validator': frontend.validate_boolean}),
  27          ('Omit the XML declaration.  Use with caution.',
  28           ['--no-xml-declaration'],
  29           {'dest': 'xml_declaration', 'default': 1, 'action': 'store_false',
  30            'validator': frontend.validate_boolean}),
  31          ('Omit the DOCTYPE declaration.',
  32           ['--no-doctype'],
  33           {'dest': 'doctype_declaration', 'default': 1,
  34            'action': 'store_false', 'validator': frontend.validate_boolean}),
  35           ('Convert LaTeX math in math_block and math to MathML',
  36               ['--latex-mathml'],
  37               {'dest': 'latex_mathml', 'default':False,
  38                   'action': 'store_true', 'validator': frontend.validate_boolean}),
  39             ('Convert ASCII math in math_block and math to MathML',
  40               ['--ascii-mathml'],
  41               {'dest': 'ascii_mathml', 'default':False,
  42                   'action': 'store_true', 'validator': frontend.validate_boolean}),
  43
  44           ))
  45
  46     settings_defaults = {'output_encoding_error_handler': 'xmlcharrefreplace'}
  47
  48     config_section = 'docutils_xml writer'
  49     config_section_dependencies = ('writers',)
  50
  51     output = None
  52     """Final translated form of `document`."""
  53
  54     xml_declaration = '<?xml version="1.0" encoding="%s"?>\n'
  55     #xml_stylesheet = '<?xml-stylesheet type="text/xsl" href="%s"?>\n'
  56     doctype = (
  57         '<!DOCTYPE document PUBLIC'
  58         ' "+//IDN docutils.sourceforge.net//DTD Docutils Generic//EN//XML"'
  59         ' "http://docutils.sourceforge.net/docs/ref/docutils.dtd">\n')
  60     generator = '<!-- Generated by Docutils %s -->\n'
  61
  62     def get_transforms(self):
  63         return Component.get_transforms(self) + [
  64             universal.Messages,
  65             universal.FilterMessages,
  66             universal.StripClassesAndElements,
  67             math.LaTeXmath2MathML,
  68             math.Asciimath2MathML,
  69             ]
  70
  71     def translate(self):
  72         settings = self.document.settings
  73         indent = newline = ''
  74         if settings.newlines:
  75             newline = '\n'
  76         if settings.indents:
  77             newline = '\n'
  78             indent = '    '
  79         output_prefix = []
  80         if settings.xml_declaration:
  81             output_prefix.append(
  82                 self.xml_declaration % settings.output_encoding)
  83         if settings.doctype_declaration:
  84             output_prefix.append(self.doctype)
  85         output_prefix.append(self.generator % docutils.__version__)
  86         docnode = self.document.asdom().childNodes[0]
  87         self.output = (''.join(output_prefix)
  88                        + docnode.toprettyxml(indent, newline))
  89
  90
  91 class MPublisher(docutils.core.Publisher):
  92     pass
  93
  94
  95
  96     def set_writer(self, writer_name):
  97         """Set `self.writer` by name."""
  98         writer_class = writers.get_writer_class(writer_name)
  99         # self.writer = writer_class()
 100         self.writer = MWriter()
 101
 102 """
 103 A minimal front end to the Docutils Publisher, producing Docutils XML.
 104 """
 105
 106 try:
 107     import locale
 108     locale.setlocale(locale.LC_ALL, '')
 109 except:
 110     pass
 111
 112 # from docutils.core import publish_cmdline, default_description
 113 from docutils.core import  default_description
 114
 115
 116 description = ('Generates Docutils-native XML from standalone '
 117                'reStructuredText sources.  ' + default_description)
 118 default_usage = '%prog [options] [<source> [<destination>]]'
 119 default_description = ('Reads from <source> (default is stdin) and writes to '
 120                        '<destination> (default is stdout).  See '
 121                        '<http://docutils.sf.net/docs/user/config.html> for '
 122                        'the full reference.')
 123
 124
 125 def publish_cmdline(reader=None, reader_name='standalone',
 126                     parser=None, parser_name='restructuredtext',
 127                     writer=None, writer_name='pseudoxml',
 128                     settings=None, settings_spec=None,
 129                     settings_overrides=None, config_section=None,
 130                     enable_exit_status=1, argv=None,
 131                     usage=default_usage, description=default_description):
 132     """
 133     Set up & run a `Publisher` for command-line-based file I/O (input and
 134     output file paths taken automatically from the command line).  Return the
 135     encoded string output also.
 136
 137     Parameters: see `publish_programmatically` for the remainder.
 138
 139     - `argv`: Command-line argument list to use instead of ``sys.argv[1:]``.
 140     - `usage`: Usage string, output if there's a problem parsing the command
 141       line.
 142     - `description`: Program description, output for the "--help" option
 143       (along with command-line option descriptions).
 144     """
 145     pub = MPublisher(reader, parser, writer, settings=settings)
 146     pub.set_components(reader_name, parser_name, writer_name)
 147     output = pub.publish(
 148         argv, usage, description, settings_spec, settings_overrides,
 149         config_section=config_section, enable_exit_status=enable_exit_status)
 150     return output
 151
 152 publish_cmdline(writer_name='xml', description=description)