import xml.sax

[0] Fix | Delete

import xml.sax.handler

[1] Fix | Delete

import types

[2] Fix | Delete

[3] Fix | Delete

try:

[4] Fix | Delete

_StringTypes = [types.StringType, types.UnicodeType]

[5] Fix | Delete

except AttributeError:

[6] Fix | Delete

_StringTypes = [types.StringType]

[7] Fix | Delete

[8] Fix | Delete

START_ELEMENT = "START_ELEMENT"

[9] Fix | Delete

END_ELEMENT = "END_ELEMENT"

[10] Fix | Delete

COMMENT = "COMMENT"

[11] Fix | Delete

START_DOCUMENT = "START_DOCUMENT"

[12] Fix | Delete

END_DOCUMENT = "END_DOCUMENT"

[13] Fix | Delete

PROCESSING_INSTRUCTION = "PROCESSING_INSTRUCTION"

[14] Fix | Delete

IGNORABLE_WHITESPACE = "IGNORABLE_WHITESPACE"

[15] Fix | Delete

CHARACTERS = "CHARACTERS"

[16] Fix | Delete

[17] Fix | Delete

class PullDOM(xml.sax.ContentHandler):

[18] Fix | Delete

_locator = None

[19] Fix | Delete

document = None

[20] Fix | Delete

[21] Fix | Delete

def __init__(self, documentFactory=None):

[22] Fix | Delete

from xml.dom import XML_NAMESPACE

[23] Fix | Delete

self.documentFactory = documentFactory

[24] Fix | Delete

self.firstEvent = [None, None]

[25] Fix | Delete

self.lastEvent = self.firstEvent

[26] Fix | Delete

self.elementStack = []

[27] Fix | Delete

self.push = self.elementStack.append

[28] Fix | Delete

try:

[29] Fix | Delete

self.pop = self.elementStack.pop

[30] Fix | Delete

except AttributeError:

[31] Fix | Delete

# use class' pop instead

[32] Fix | Delete

pass

[33] Fix | Delete

self._ns_contexts = [{XML_NAMESPACE:'xml'}] # contains uri -> prefix dicts

[34] Fix | Delete

self._current_context = self._ns_contexts[-1]

[35] Fix | Delete

self.pending_events = []

[36] Fix | Delete

[37] Fix | Delete

def pop(self):

[38] Fix | Delete

result = self.elementStack[-1]

[39] Fix | Delete

del self.elementStack[-1]

[40] Fix | Delete

return result

[41] Fix | Delete

[42] Fix | Delete

def setDocumentLocator(self, locator):

[43] Fix | Delete

self._locator = locator

[44] Fix | Delete

[45] Fix | Delete

def startPrefixMapping(self, prefix, uri):

[46] Fix | Delete

if not hasattr(self, '_xmlns_attrs'):

[47] Fix | Delete

self._xmlns_attrs = []

[48] Fix | Delete

self._xmlns_attrs.append((prefix or 'xmlns', uri))

[49] Fix | Delete

self._ns_contexts.append(self._current_context.copy())

[50] Fix | Delete

self._current_context[uri] = prefix or None

[51] Fix | Delete

[52] Fix | Delete

def endPrefixMapping(self, prefix):

[53] Fix | Delete

self._current_context = self._ns_contexts.pop()

[54] Fix | Delete

[55] Fix | Delete

def startElementNS(self, name, tagName , attrs):

[56] Fix | Delete

# Retrieve xml namespace declaration attributes.

[57] Fix | Delete

xmlns_uri = 'http://www.w3.org/2000/xmlns/'

[58] Fix | Delete

xmlns_attrs = getattr(self, '_xmlns_attrs', None)

[59] Fix | Delete

if xmlns_attrs is not None:

[60] Fix | Delete

for aname, value in xmlns_attrs:

[61] Fix | Delete

attrs._attrs[(xmlns_uri, aname)] = value

[62] Fix | Delete

self._xmlns_attrs = []

[63] Fix | Delete

uri, localname = name

[64] Fix | Delete

if uri:

[65] Fix | Delete

# When using namespaces, the reader may or may not

[66] Fix | Delete

# provide us with the original name. If not, create

[67] Fix | Delete

# *a* valid tagName from the current context.

[68] Fix | Delete

if tagName is None:

[69] Fix | Delete

prefix = self._current_context[uri]

[70] Fix | Delete

if prefix:

[71] Fix | Delete

tagName = prefix + ":" + localname

[72] Fix | Delete

else:

[73] Fix | Delete

tagName = localname

[74] Fix | Delete

if self.document:

[75] Fix | Delete

node = self.document.createElementNS(uri, tagName)

[76] Fix | Delete

else:

[77] Fix | Delete

node = self.buildDocument(uri, tagName)

[78] Fix | Delete

else:

[79] Fix | Delete

# When the tagname is not prefixed, it just appears as

[80] Fix | Delete

# localname

[81] Fix | Delete

if self.document:

[82] Fix | Delete

node = self.document.createElement(localname)

[83] Fix | Delete

else:

[84] Fix | Delete

node = self.buildDocument(None, localname)

[85] Fix | Delete

[86] Fix | Delete

for aname,value in attrs.items():

[87] Fix | Delete

a_uri, a_localname = aname

[88] Fix | Delete

if a_uri == xmlns_uri:

[89] Fix | Delete

if a_localname == 'xmlns':

[90] Fix | Delete

qname = a_localname

[91] Fix | Delete

else:

[92] Fix | Delete

qname = 'xmlns:' + a_localname

[93] Fix | Delete

attr = self.document.createAttributeNS(a_uri, qname)

[94] Fix | Delete

node.setAttributeNodeNS(attr)

[95] Fix | Delete

elif a_uri:

[96] Fix | Delete

prefix = self._current_context[a_uri]

[97] Fix | Delete

if prefix:

[98] Fix | Delete

qname = prefix + ":" + a_localname

[99] Fix | Delete

else:

[100] Fix | Delete

qname = a_localname

[101] Fix | Delete

attr = self.document.createAttributeNS(a_uri, qname)

[102] Fix | Delete

node.setAttributeNodeNS(attr)

[103] Fix | Delete

else:

[104] Fix | Delete

attr = self.document.createAttribute(a_localname)

[105] Fix | Delete

node.setAttributeNode(attr)

[106] Fix | Delete

attr.value = value

[107] Fix | Delete

[108] Fix | Delete

self.lastEvent[1] = [(START_ELEMENT, node), None]

[109] Fix | Delete

self.lastEvent = self.lastEvent[1]

[110] Fix | Delete

self.push(node)

[111] Fix | Delete

[112] Fix | Delete

def endElementNS(self, name, tagName):

[113] Fix | Delete

self.lastEvent[1] = [(END_ELEMENT, self.pop()), None]

[114] Fix | Delete

self.lastEvent = self.lastEvent[1]

[115] Fix | Delete

[116] Fix | Delete

def startElement(self, name, attrs):

[117] Fix | Delete

if self.document:

[118] Fix | Delete

node = self.document.createElement(name)

[119] Fix | Delete

else:

[120] Fix | Delete

node = self.buildDocument(None, name)

[121] Fix | Delete

[122] Fix | Delete

for aname,value in attrs.items():

[123] Fix | Delete

attr = self.document.createAttribute(aname)

[124] Fix | Delete

attr.value = value

[125] Fix | Delete

node.setAttributeNode(attr)

[126] Fix | Delete

[127] Fix | Delete

self.lastEvent[1] = [(START_ELEMENT, node), None]

[128] Fix | Delete

self.lastEvent = self.lastEvent[1]

[129] Fix | Delete

self.push(node)

[130] Fix | Delete

[131] Fix | Delete

def endElement(self, name):

[132] Fix | Delete

self.lastEvent[1] = [(END_ELEMENT, self.pop()), None]

[133] Fix | Delete

self.lastEvent = self.lastEvent[1]

[134] Fix | Delete

[135] Fix | Delete

def comment(self, s):

[136] Fix | Delete

if self.document:

[137] Fix | Delete

node = self.document.createComment(s)

[138] Fix | Delete

self.lastEvent[1] = [(COMMENT, node), None]

[139] Fix | Delete

self.lastEvent = self.lastEvent[1]

[140] Fix | Delete

else:

[141] Fix | Delete

event = [(COMMENT, s), None]

[142] Fix | Delete

self.pending_events.append(event)

[143] Fix | Delete

[144] Fix | Delete

def processingInstruction(self, target, data):

[145] Fix | Delete

if self.document:

[146] Fix | Delete

node = self.document.createProcessingInstruction(target, data)

[147] Fix | Delete

self.lastEvent[1] = [(PROCESSING_INSTRUCTION, node), None]

[148] Fix | Delete

self.lastEvent = self.lastEvent[1]

[149] Fix | Delete

else:

[150] Fix | Delete

event = [(PROCESSING_INSTRUCTION, target, data), None]

[151] Fix | Delete

self.pending_events.append(event)

[152] Fix | Delete

[153] Fix | Delete

def ignorableWhitespace(self, chars):

[154] Fix | Delete

node = self.document.createTextNode(chars)

[155] Fix | Delete

self.lastEvent[1] = [(IGNORABLE_WHITESPACE, node), None]

[156] Fix | Delete

self.lastEvent = self.lastEvent[1]

[157] Fix | Delete

[158] Fix | Delete

def characters(self, chars):

[159] Fix | Delete

node = self.document.createTextNode(chars)

[160] Fix | Delete

self.lastEvent[1] = [(CHARACTERS, node), None]

[161] Fix | Delete

self.lastEvent = self.lastEvent[1]

[162] Fix | Delete

[163] Fix | Delete

def startDocument(self):

[164] Fix | Delete

if self.documentFactory is None:

[165] Fix | Delete

import xml.dom.minidom

[166] Fix | Delete

self.documentFactory = xml.dom.minidom.Document.implementation

[167] Fix | Delete

[168] Fix | Delete

def buildDocument(self, uri, tagname):

[169] Fix | Delete

# Can't do that in startDocument, since we need the tagname

[170] Fix | Delete

# XXX: obtain DocumentType

[171] Fix | Delete

node = self.documentFactory.createDocument(uri, tagname, None)

[172] Fix | Delete

self.document = node

[173] Fix | Delete

self.lastEvent[1] = [(START_DOCUMENT, node), None]

[174] Fix | Delete

self.lastEvent = self.lastEvent[1]

[175] Fix | Delete

self.push(node)

[176] Fix | Delete

# Put everything we have seen so far into the document

[177] Fix | Delete

for e in self.pending_events:

[178] Fix | Delete

if e[0][0] == PROCESSING_INSTRUCTION:

[179] Fix | Delete

_,target,data = e[0]

[180] Fix | Delete

n = self.document.createProcessingInstruction(target, data)

[181] Fix | Delete

e[0] = (PROCESSING_INSTRUCTION, n)

[182] Fix | Delete

elif e[0][0] == COMMENT:

[183] Fix | Delete

n = self.document.createComment(e[0][1])

[184] Fix | Delete

e[0] = (COMMENT, n)

[185] Fix | Delete

else:

[186] Fix | Delete

raise AssertionError("Unknown pending event ",e[0][0])

[187] Fix | Delete

self.lastEvent[1] = e

[188] Fix | Delete

self.lastEvent = e

[189] Fix | Delete

self.pending_events = None

[190] Fix | Delete

return node.firstChild

[191] Fix | Delete

[192] Fix | Delete

def endDocument(self):

[193] Fix | Delete

self.lastEvent[1] = [(END_DOCUMENT, self.document), None]

[194] Fix | Delete

self.pop()

[195] Fix | Delete

[196] Fix | Delete

def clear(self):

[197] Fix | Delete

"clear(): Explicitly release parsing structures"

[198] Fix | Delete

self.document = None

[199] Fix | Delete

[200] Fix | Delete

class ErrorHandler:

[201] Fix | Delete

def warning(self, exception):

[202] Fix | Delete

print exception

[203] Fix | Delete

def error(self, exception):

[204] Fix | Delete

raise exception

[205] Fix | Delete

def fatalError(self, exception):

[206] Fix | Delete

raise exception

[207] Fix | Delete

[208] Fix | Delete

class DOMEventStream:

[209] Fix | Delete

def __init__(self, stream, parser, bufsize):

[210] Fix | Delete

self.stream = stream

[211] Fix | Delete

self.parser = parser

[212] Fix | Delete

self.bufsize = bufsize

[213] Fix | Delete

if not hasattr(self.parser, 'feed'):

[214] Fix | Delete

self.getEvent = self._slurp

[215] Fix | Delete

self.reset()

[216] Fix | Delete

[217] Fix | Delete

def reset(self):

[218] Fix | Delete

self.pulldom = PullDOM()

[219] Fix | Delete

# This content handler relies on namespace support

[220] Fix | Delete

self.parser.setFeature(xml.sax.handler.feature_namespaces, 1)

[221] Fix | Delete

self.parser.setContentHandler(self.pulldom)

[222] Fix | Delete

[223] Fix | Delete

def __getitem__(self, pos):

[224] Fix | Delete

rc = self.getEvent()

[225] Fix | Delete

if rc:

[226] Fix | Delete

return rc

[227] Fix | Delete

raise IndexError

[228] Fix | Delete

[229] Fix | Delete

def next(self):

[230] Fix | Delete

rc = self.getEvent()

[231] Fix | Delete

if rc:

[232] Fix | Delete

return rc

[233] Fix | Delete

raise StopIteration

[234] Fix | Delete

[235] Fix | Delete

def __iter__(self):

[236] Fix | Delete

return self

[237] Fix | Delete

[238] Fix | Delete

def expandNode(self, node):

[239] Fix | Delete

event = self.getEvent()

[240] Fix | Delete

parents = [node]

[241] Fix | Delete

while event:

[242] Fix | Delete

token, cur_node = event

[243] Fix | Delete

if cur_node is node:

[244] Fix | Delete

return

[245] Fix | Delete

if token != END_ELEMENT:

[246] Fix | Delete

parents[-1].appendChild(cur_node)

[247] Fix | Delete

if token == START_ELEMENT:

[248] Fix | Delete

parents.append(cur_node)

[249] Fix | Delete

elif token == END_ELEMENT:

[250] Fix | Delete

del parents[-1]

[251] Fix | Delete

event = self.getEvent()

[252] Fix | Delete

[253] Fix | Delete

def getEvent(self):

[254] Fix | Delete

# use IncrementalParser interface, so we get the desired

[255] Fix | Delete

# pull effect

[256] Fix | Delete

if not self.pulldom.firstEvent[1]:

[257] Fix | Delete

self.pulldom.lastEvent = self.pulldom.firstEvent

[258] Fix | Delete

while not self.pulldom.firstEvent[1]:

[259] Fix | Delete

buf = self.stream.read(self.bufsize)

[260] Fix | Delete

if not buf:

[261] Fix | Delete

self.parser.close()

[262] Fix | Delete

return None

[263] Fix | Delete

self.parser.feed(buf)

[264] Fix | Delete

rc = self.pulldom.firstEvent[1][0]

[265] Fix | Delete

self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1]

[266] Fix | Delete

return rc

[267] Fix | Delete

[268] Fix | Delete

def _slurp(self):

[269] Fix | Delete

""" Fallback replacement for getEvent() using the

[270] Fix | Delete

standard SAX2 interface, which means we slurp the

[271] Fix | Delete

SAX events into memory (no performance gain, but

[272] Fix | Delete

we are compatible to all SAX parsers).

[273] Fix | Delete

"""

[274] Fix | Delete

self.parser.parse(self.stream)

[275] Fix | Delete

self.getEvent = self._emit

[276] Fix | Delete

return self._emit()

[277] Fix | Delete

[278] Fix | Delete

def _emit(self):

[279] Fix | Delete

""" Fallback replacement for getEvent() that emits

[280] Fix | Delete

the events that _slurp() read previously.

[281] Fix | Delete

"""

[282] Fix | Delete

rc = self.pulldom.firstEvent[1][0]

[283] Fix | Delete

self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1]

[284] Fix | Delete

return rc

[285] Fix | Delete

[286] Fix | Delete

def clear(self):

[287] Fix | Delete

"""clear(): Explicitly release parsing objects"""

[288] Fix | Delete

self.pulldom.clear()

[289] Fix | Delete

del self.pulldom

[290] Fix | Delete

self.parser = None

[291] Fix | Delete

self.stream = None

[292] Fix | Delete

[293] Fix | Delete

class SAX2DOM(PullDOM):

[294] Fix | Delete

[295] Fix | Delete

def startElementNS(self, name, tagName , attrs):

[296] Fix | Delete

PullDOM.startElementNS(self, name, tagName, attrs)

[297] Fix | Delete

curNode = self.elementStack[-1]

[298] Fix | Delete

parentNode = self.elementStack[-2]

[299] Fix | Delete

parentNode.appendChild(curNode)

[300] Fix | Delete

[301] Fix | Delete

def startElement(self, name, attrs):

[302] Fix | Delete

PullDOM.startElement(self, name, attrs)

[303] Fix | Delete

curNode = self.elementStack[-1]

[304] Fix | Delete

parentNode = self.elementStack[-2]

[305] Fix | Delete

parentNode.appendChild(curNode)

[306] Fix | Delete

[307] Fix | Delete

def processingInstruction(self, target, data):

[308] Fix | Delete

PullDOM.processingInstruction(self, target, data)

[309] Fix | Delete

node = self.lastEvent[0][1]

[310] Fix | Delete

parentNode = self.elementStack[-1]

[311] Fix | Delete

parentNode.appendChild(node)

[312] Fix | Delete

[313] Fix | Delete

def ignorableWhitespace(self, chars):

[314] Fix | Delete

PullDOM.ignorableWhitespace(self, chars)

[315] Fix | Delete

node = self.lastEvent[0][1]

[316] Fix | Delete

parentNode = self.elementStack[-1]

[317] Fix | Delete

parentNode.appendChild(node)

[318] Fix | Delete

[319] Fix | Delete

def characters(self, chars):

[320] Fix | Delete

PullDOM.characters(self, chars)

[321] Fix | Delete

node = self.lastEvent[0][1]

[322] Fix | Delete

parentNode = self.elementStack[-1]

[323] Fix | Delete

parentNode.appendChild(node)

[324] Fix | Delete

[325] Fix | Delete

[326] Fix | Delete

default_bufsize = (2 ** 14) - 20

[327] Fix | Delete

[328] Fix | Delete

def parse(stream_or_string, parser=None, bufsize=None):

[329] Fix | Delete

if bufsize is None:

[330] Fix | Delete

bufsize = default_bufsize

[331] Fix | Delete

if type(stream_or_string) in _StringTypes:

[332] Fix | Delete

stream = open(stream_or_string)

[333] Fix | Delete

else:

[334] Fix | Delete

stream = stream_or_string

[335] Fix | Delete

if not parser:

[336] Fix | Delete

parser = xml.sax.make_parser()

[337] Fix | Delete

return DOMEventStream(stream, parser, bufsize)

[338] Fix | Delete

[339] Fix | Delete

def parseString(string, parser=None):

[340] Fix | Delete

try:

[341] Fix | Delete

from cStringIO import StringIO

[342] Fix | Delete

except ImportError:

[343] Fix | Delete

from StringIO import StringIO

[344] Fix | Delete

[345] Fix | Delete

bufsize = len(string)

[346] Fix | Delete

buf = StringIO(string)

[347] Fix | Delete

if not parser:

[348] Fix | Delete

parser = xml.sax.make_parser()

[349] Fix | Delete

return DOMEventStream(buf, parser, bufsize)

[350] Fix | Delete

[351] Fix | Delete