zeroinstall/injector/qdom.py

   1 """A quick DOM implementation.
   2
   3 Python's xml.dom is very slow. The xml.sax module is also slow (as it imports urllib2).
   4 This is our light-weight version.
   5 """
   6
   7 # Copyright (C) 2009, Thomas Leonard
   8 # See the README file for details, or visit http://0install.net.
   9
  10 from zeroinstall import _
  11 from xml.parsers import expat
  12
  13 class Element(object):
  14         """An XML element.
  15         @ivar uri: the element's namespace
  16         @type uri: str
  17         @ivar name: the element's localName
  18         @type name: str
  19         @ivar attrs: the element's attributes (key is in the form [namespace " "] localName
  20         @type attrs: {str: str}
  21         @ivar childNodes: children
  22         @type childNodes: [L{Element}]
  23         @ivar content: the text content
  24         @type content: str"""
  25         __slots__ = ['uri', 'name', 'attrs', 'childNodes', 'content']
  26         def __init__(self, uri, name, attrs):
  27                 self.uri = uri
  28                 self.name = name
  29                 self.attrs = attrs.copy()
  30                 self.childNodes = []
  31
  32         def __str__(self):
  33                 attrs = [n + '=' + self.attrs[n] for n in self.attrs]
  34                 start = '<{%s}%s %s' % (self.uri, self.name, ' '.join(attrs))
  35                 if self.childNodes:
  36                         return start + '>' + '\n'.join(map(str, self.childNodes)) + ('</%s>' % (self.name))
  37                 elif self.content:
  38                         return start + '>' + self.content + ('</%s>' % (self.name))
  39                 else:
  40                         return start + '/>'
  41
  42         def getAttribute(self, name):
  43                 return self.attrs.get(name, None)
  44
  45 class QSAXhandler:
  46         """SAXHandler that builds a tree of L{Element}s"""
  47         def __init__(self):
  48                 self.stack = []
  49
  50         def startElementNS(self, fullname, attrs):
  51                 split = fullname.split(' ', 1)
  52                 if len(split) == 2:
  53                         self.stack.append(Element(split[0], split[1], attrs))
  54                 else:
  55                         self.stack.append(Element(None, fullname, attrs))
  56                 self.contents = ''
  57
  58         def characters(self, data):
  59                 self.contents += data
  60
  61         def endElementNS(self, name):
  62                 contents = self.contents.strip()
  63                 self.stack[-1].content = contents
  64                 self.contents = ''
  65                 new = self.stack.pop()
  66                 if self.stack:
  67                         self.stack[-1].childNodes.append(new)
  68                 else:
  69                         self.doc = new
  70
  71 def parse(source):
  72         """Parse an XML stream into a tree of L{Element}s.
  73         @param source: data to parse
  74         @type source: file
  75         @return: the root
  76         @rtype: L{Element}"""
  77         handler = QSAXhandler()
  78         parser = expat.ParserCreate(namespace_separator = ' ')
  79
  80         parser.StartElementHandler = handler.startElementNS
  81         parser.EndElementHandler = handler.endElementNS
  82         parser.CharacterDataHandler = handler.characters
  83
  84         parser.ParseFile(source)
  85         return handler.doc