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) 2006, Thomas Leonard
   8 # See the README file for details, or visit http://0install.net.
   9
  10 from xml.parsers import expat
  11 import sys
  12
  13 class Element(object):
  14         __slots__ = ['uri', 'name', 'attrs', 'childNodes', 'content']
  15         def __init__(self, uri, name, attrs):
  16                 self.uri = uri
  17                 self.name = name
  18                 self.attrs = attrs.copy()
  19                 self.childNodes = []
  20
  21         def __str__(self):
  22                 attrs = [n + '=' + self.attrs[n] for n in self.attrs]
  23                 start = '<{%s}%s %s' % (self.uri, self.name, ' '.join(attrs))
  24                 if self.childNodes:
  25                         return start + '>' + '\n'.join(map(str, self.childNodes)) + ('</%s>' % (self.name))
  26                 elif self.content:
  27                         return start + '>' + self.content + ('</%s>' % (self.name))
  28                 else:
  29                         return start + '/>'
  30
  31         def getAttribute(self, name):
  32                 return self.attrs.get(name, None)
  33
  34 class QSAXhandler:
  35         def __init__(self):
  36                 self.stack = []
  37
  38         def startElementNS(self, fullname, attrs):
  39                 split = fullname.split(' ', 1)
  40                 if len(split) == 2:
  41                         self.stack.append(Element(split[0], split[1], attrs))
  42                 else:
  43                         self.stack.append(Element(None, fullname, attrs))
  44                 self.contents = ''
  45
  46         def characters(self, data):
  47                 self.contents += data
  48
  49         def endElementNS(self, name):
  50                 contents = self.contents.strip()
  51                 self.stack[-1].content = contents
  52                 self.contents = ''
  53                 new = self.stack.pop()
  54                 if self.stack:
  55                         self.stack[-1].childNodes.append(new)
  56                 else:
  57                         self.doc = new
  58
  59 def parse(source):
  60         handler = QSAXhandler()
  61         parser = expat.ParserCreate(namespace_separator = ' ')
  62
  63         parser.StartElementHandler = handler.startElementNS
  64         parser.EndElementHandler = handler.endElementNS
  65         parser.CharacterDataHandler = handler.characters
  66
  67         parser.ParseFile(source)
  68         return handler.doc