1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
1 ##############################################################################
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
2 #
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
3 # Copyright (c) 2001, 2002 Zope Corporation and Contributors.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
4 # All Rights Reserved.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
5 #
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
6 # This software is subject to the provisions of the Zope Public License,
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
7 # Version 2.0 (ZPL). A copy of the ZPL should accompany this distribution.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
8 # THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
9 # WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
10 # WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
11 # FOR A PARTICULAR PURPOSE
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
12 #
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
13 ##############################################################################
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
14 """
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
15 Generic expat-based XML parser base class.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
16 """
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
17
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
18 class XMLParser:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
19
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
20 ordered_attributes = 0
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
21
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
22 handler_names = [
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
23 "StartElementHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
24 "EndElementHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
25 "ProcessingInstructionHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
26 "CharacterDataHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
27 "UnparsedEntityDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
28 "NotationDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
29 "StartNamespaceDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
30 "EndNamespaceDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
31 "CommentHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
32 "StartCdataSectionHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
33 "EndCdataSectionHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
34 "DefaultHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
35 "DefaultHandlerExpand",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
36 "NotStandaloneHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
37 "ExternalEntityRefHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
38 "XmlDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
39 "StartDoctypeDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
40 "EndDoctypeDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
41 "ElementDeclHandler",
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
42 "AttlistDeclHandler"
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
43 ]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
44
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
45 def __init__(self, encoding=None):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
46 self.parser = p = self.createParser()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
47 if self.ordered_attributes:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
48 try:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
49 self.parser.ordered_attributes = self.ordered_attributes
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
50 except AttributeError:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
51 #zLOG.LOG("TAL.XMLParser", zLOG.INFO,
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
52 # "Can't set ordered_attributes")
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
53 self.ordered_attributes = 0
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
54 for name in self.handler_names:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
55 method = getattr(self, name, None)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
56 if method is not None:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
57 try:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
58 setattr(p, name, method)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
59 except AttributeError:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
60 #zLOG.LOG("TAL.XMLParser", zLOG.PROBLEM,
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
61 # "Can't set expat handler %s" % name)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
62 pass
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
63
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
64 def createParser(self, encoding=None):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
65 global XMLParseError
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
66 try:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
67 from Products.ParsedXML.Expat import pyexpat
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
68 XMLParseError = pyexpat.ExpatError
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
69 return pyexpat.ParserCreate(encoding, ' ')
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
70 except ImportError:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
71 from xml.parsers import expat
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
72 XMLParseError = expat.ExpatError
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
73 return expat.ParserCreate(encoding, ' ')
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
74
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
75 def parseFile(self, filename):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
76 self.parseStream(open(filename))
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
77
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
78 def parseString(self, s):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
79 self.parser.Parse(s, 1)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
80
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
81 def parseURL(self, url):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
82 import urllib
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
83 self.parseStream(urllib.urlopen(url))
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
84
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
85 def parseStream(self, stream):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
86 self.parser.ParseFile(stream)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
87
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
88 def parseFragment(self, s, end=0):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
89 self.parser.Parse(s, end)
|