xml_resolve_entities.py

# -*- coding: utf-8 -*-

__all__ = ('resolve_entities', )


try:
    from cStringIO import StringIO
except ImportError:
    from StringIO import StringIO
    
from xml.sax import handler
from xml.sax import make_parser
from xml.sax import saxutils


def xml_resolve_entities(xmlstr, resolve_dtd=None, output_encoding='utf-8'):
    """Convert all XML entities to their applicable values.

    :param xmlstr: The input file-like object or string.
    :param resolve_dtd: The function to resolve the system identifier of an
	entity and return either the system identifier to read from as a
	string, or an file-like object to read from. The default implementation
	returns systemId.
    :rtype string: XML result

    For example::

	>>> xml_source = '''<?xml version="1.0"?>
	... <!DOCTYPE times SYSTEM "sample.dtd">
	... <math>&times;</math>'''

	>>> dtd_catalogue = {'sample.dtd': '<!ENTITY times "&#215;">', }

	>>> print xml_resolve_entities(xml_source, dtd_catalogue.get)
	<?xml version="1.0" encoding="utf-8"?>
	<math>×</math>
    """

    flxml = enshure_is_filelike(xmlstr)
    out = StringIO()
    try:
        parser = make_parser()
        if resolve_dtd:
            parser.setEntityResolver(EntityResolver(resolve_dtd))
        parser.setContentHandler(saxutils.XMLGenerator(out, output_encoding))
        parser.parse(flxml)
        output_xmlstr = out.getvalue()
    finally:
        out.close()
    return output_xmlstr

def enshure_is_filelike(filelike):
    if isinstance(filelike, basestring):
        return StringIO(filelike)
    else:
        # FIXME: как удостовериться, что filelike действительно похож на файл?
	# EAFP: верьте на слово. если это не так, то парсер крякнет.
        return filelike


class EntityResolver(handler.EntityResolver):
    def __init__(self, resolve):
        self._resolve = resolve

    def resolveEntity(self, unused_public_id, system_id):
        data = self._resolve(system_id)
        return enshure_is_filelike(data)