Subclassing cElementTree.XMLTreeBuilder

[mukappa]

I found an earlier post about subclassing cElementTree.Element which
can't
be done because it is a factory method. I am trying to subclass
XMLTreeBuilder
with success using the python implementation, but not with
cElementTree.

[1013]$ python
Python 2.3.4 (#1, Feb 22 2005, 04:09:37)
[GCC 3.4.3 20041212 (Red Hat 3.4.3-9.EL4)] on linux2
Type "help", "copyright", "credits" or "license" for more information..... pass
........ pass
....
Traceback (most recent call last):

Is it possible to subclass cElementTree.XMLTreeBuilder?

Thanks,
Mike

 

[Fredrik Lundh]

Is it possible to subclass cElementTree.XMLTreeBuilder?

no, it's a factory function. if you need to extend it, you'll have to wrap it. but
I'm not sure I see the use case; what is it you're trying to do here?

</F>

 

[mukappa]

I'm not sure I see the use case; what is it you're trying to do here?

</F>

I'm trying to parse xmpp stanzas coming in over a socket. Since
cElementTree is
more efficient, I would like to use it when available.

Here is what I have now:

import logging
from elementtree import ElementTree

class StanzaTreeBuilder(ElementTree.XMLTreeBuilder):
"""Capture stanza elements (<stream> children) from parser events.

Private methods are lifted from the old
elementtree.XMLTreeBuilder.FancyTreeBuilder
"""

def __init__(self, html=0):
"""Initialize state variables."""
ElementTree.XMLTreeBuilder.__init__(self, html)
self._parser.StartNamespaceDeclHandler = self._start_ns
self._parser.EndNamespaceDeclHandler = self._end_ns
self.namespaces = []
self.level = 0
self.stanza = None
self.streamelem = None

def _start(self, tag, attrib_in):
elem = ElementTree.XMLTreeBuilder._start(self, tag, attrib_in)
self.start(elem)

def _start_list(self, tag, attrib_in):
elem = ElementTree.XMLTreeBuilder._start_list(self, tag,
attrib_in)
self.start(elem)

def _end(self, tag):
elem = ElementTree.XMLTreeBuilder._end(self, tag)
self.end(elem)

def _start_ns(self, prefix, value):
self.namespaces.insert(0, (prefix, value))

def _end_ns(self, prefix):
assert self.namespaces.pop(0)[0] == prefix, "implementation
confused"

def start(self, element):
"""Track nesting level (capture open <stream> on first
call)."""
logging.debug("start(%s): %s" %
(self.level, ElementTree.tostring(element)))
if self.streamelem is None:
assert element.tag == "http://etherx.jabber.org/streams"
self.streamelem = element
self.stanza = element
self.level += 1

def end(self, element):
"""Capture stanza when nesting level is 1."""
self.level -= 1
if self.level == 1:
self.stanza = element
self.streamelem.clear()
logging.debug("end(%s): %s" %
(self.level, ElementTree.tostring(element)))

def stanzagen(socket):
"""Yield open <stream ...> on first call, then complete stanzas.

Caller needs 'id' from the opening <stream>, then he needs stanzas,
one per call. Works by parsing the socket one byte at a time, and
yielding complete stanza at earliest opportunity (or None if none
waiting), exits on EOF. Stanzas are returned as ElementTree
Elements
to give them a natural pythonic feel.

see
http://online.effbot.org/2004_12_01_archive.htm#element-generator"""
p = StanzaTreeBuilder()
data = None
while 1:
try:
data = socket.recv(1)
if not data:
logging.info("EOF from peer")
p.close()
break
p.feed(data)
if p.stanza is not None:
yield p.stanza
p.stanza = None
data = None
except:
if data is not None:
logging.error("error reading: %s" % (data))
raise
else:
yield None