2000-06-29 16:33:43 -03:00
|
|
|
"""Simple API for XML (SAX) implementation for Python.
|
|
|
|
|
|
|
|
This module provides an implementation of the SAX 2 interface;
|
|
|
|
information about the Java version of the interface can be found at
|
|
|
|
http://www.megginson.com/SAX/. The Python version of the interface is
|
|
|
|
documented at <...>.
|
|
|
|
|
|
|
|
This package contains the following modules:
|
|
|
|
|
2000-09-18 13:31:08 -03:00
|
|
|
handler -- Base classes and constants which define the SAX 2 API for
|
|
|
|
the 'client-side' of SAX for Python.
|
2000-06-29 16:33:43 -03:00
|
|
|
|
2000-09-18 13:31:08 -03:00
|
|
|
saxutils -- Implementation of the convenience classes commonly used to
|
|
|
|
work with SAX.
|
2000-06-29 16:33:43 -03:00
|
|
|
|
2000-09-18 13:31:08 -03:00
|
|
|
xmlreader -- Base classes and constants which define the SAX 2 API for
|
|
|
|
the parsers used with SAX for Python.
|
|
|
|
|
2000-09-24 17:38:18 -03:00
|
|
|
expatreader -- Driver that allows use of the Expat parser with SAX.
|
2000-06-29 16:33:43 -03:00
|
|
|
"""
|
|
|
|
|
2000-09-24 15:54:49 -03:00
|
|
|
from xmlreader import InputSource
|
2000-07-01 10:44:04 -03:00
|
|
|
from handler import ContentHandler, ErrorHandler
|
|
|
|
from _exceptions import SAXException, SAXNotRecognizedException, \
|
2000-09-18 13:31:08 -03:00
|
|
|
SAXParseException, SAXNotSupportedException
|
|
|
|
|
2000-07-01 01:58:47 -03:00
|
|
|
|
2000-09-24 15:54:49 -03:00
|
|
|
def parse(source, handler, errorHandler=ErrorHandler()):
|
2000-09-18 13:31:08 -03:00
|
|
|
parser = ExpatParser()
|
|
|
|
parser.setContentHandler(handler)
|
|
|
|
parser.setErrorHandler(errorHandler)
|
2000-09-24 15:54:49 -03:00
|
|
|
parser.parse(source)
|
2000-07-01 01:58:47 -03:00
|
|
|
|
2000-09-18 13:31:08 -03:00
|
|
|
def parseString(string, handler, errorHandler=ErrorHandler()):
|
2000-07-01 01:58:47 -03:00
|
|
|
try:
|
2000-09-18 13:31:08 -03:00
|
|
|
from cStringIO import StringIO
|
2000-07-01 01:58:47 -03:00
|
|
|
except ImportError:
|
2000-09-18 13:31:08 -03:00
|
|
|
from StringIO import StringIO
|
2000-07-01 01:58:47 -03:00
|
|
|
|
2000-09-18 13:31:08 -03:00
|
|
|
if errorHandler is None:
|
|
|
|
errorHandler = ErrorHandler()
|
|
|
|
parser = ExpatParser()
|
|
|
|
parser.setContentHandler(handler)
|
|
|
|
parser.setErrorHandler(errorHandler)
|
2000-09-24 15:54:49 -03:00
|
|
|
|
|
|
|
inpsrc = InputSource()
|
|
|
|
inpsrc.setByteStream(StringIO(string))
|
|
|
|
parser.parse(inpsrc)
|
2000-09-21 05:34:04 -03:00
|
|
|
|
|
|
|
# this is the parser list used by the make_parser function if no
|
|
|
|
# alternatives are given as parameters to the function
|
|
|
|
|
|
|
|
default_parser_list = ["xml.sax.expatreader"]
|
|
|
|
|
|
|
|
import os, string, sys
|
|
|
|
if os.environ.has_key("PY_SAX_PARSER"):
|
|
|
|
default_parser_list = string.split(os.environ["PY_SAX_PARSER"], ",")
|
|
|
|
del os
|
|
|
|
|
|
|
|
_key = "python.xml.sax.parser"
|
|
|
|
if sys.platform[:4] == "java" and sys.registry.containsKey(_key):
|
|
|
|
default_parser_list = string.split(sys.registry.getProperty(_key), ",")
|
|
|
|
|
|
|
|
|
|
|
|
def make_parser(parser_list = []):
|
|
|
|
"""Creates and returns a SAX parser.
|
|
|
|
|
|
|
|
Creates the first parser it is able to instantiate of the ones
|
|
|
|
given in the list created by doing parser_list +
|
|
|
|
default_parser_list. The lists must contain the names of Python
|
|
|
|
modules containing both a SAX parser and a create_parser function."""
|
|
|
|
|
|
|
|
for parser_name in parser_list + default_parser_list:
|
|
|
|
try:
|
2000-09-22 02:07:56 -03:00
|
|
|
return _create_parser(parser_name)
|
2000-09-21 05:34:04 -03:00
|
|
|
except ImportError,e:
|
|
|
|
pass
|
|
|
|
|
|
|
|
raise SAXException("No parsers found", None)
|
|
|
|
|
|
|
|
# --- Internal utility methods used by make_parser
|
|
|
|
|
|
|
|
if sys.platform[ : 4] == "java":
|
|
|
|
def _create_parser(parser_name):
|
2000-09-22 06:30:29 -03:00
|
|
|
from org.python.core import imp
|
|
|
|
drv_module = imp.importName(parser_name, 0, globals())
|
|
|
|
return drv_module.create_parser()
|
2000-09-21 05:34:04 -03:00
|
|
|
|
|
|
|
else:
|
|
|
|
import imp as _imp
|
2000-09-22 02:07:56 -03:00
|
|
|
|
2000-09-21 05:34:04 -03:00
|
|
|
def _rec_find_module(module):
|
2000-09-22 06:30:29 -03:00
|
|
|
"Improvement over imp.find_module which finds submodules."
|
|
|
|
path=""
|
|
|
|
for mod in string.split(module,"."):
|
2000-09-21 05:34:04 -03:00
|
|
|
if path == "":
|
2000-09-22 06:30:29 -03:00
|
|
|
info = (mod,) + _imp.find_module(mod)
|
2000-09-21 05:34:04 -03:00
|
|
|
else:
|
2000-09-22 06:30:29 -03:00
|
|
|
info = (mod,) + _imp.find_module(mod, [path])
|
|
|
|
|
2000-09-21 11:46:56 -03:00
|
|
|
lastmod = _imp.load_module(*info)
|
2000-09-22 02:07:56 -03:00
|
|
|
|
2000-09-21 05:34:04 -03:00
|
|
|
try:
|
2000-09-22 06:30:29 -03:00
|
|
|
path = lastmod.__path__[0]
|
2000-09-21 05:34:04 -03:00
|
|
|
except AttributeError, e:
|
2000-09-22 06:30:29 -03:00
|
|
|
pass
|
2000-09-22 02:07:56 -03:00
|
|
|
|
2000-09-22 06:30:29 -03:00
|
|
|
return info
|
2000-09-22 02:07:56 -03:00
|
|
|
|
2000-09-21 05:34:04 -03:00
|
|
|
def _create_parser(parser_name):
|
2000-09-22 06:30:29 -03:00
|
|
|
info = _rec_find_module(parser_name)
|
|
|
|
drv_module = _imp.load_module(*info)
|
|
|
|
return drv_module.create_parser()
|
2000-09-21 05:34:04 -03:00
|
|
|
|
|
|
|
del sys
|