import xml.sax

[0] Fix | Delete

import xml.sax.handler

[1] Fix | Delete

[2] Fix | Delete

START_ELEMENT = "START_ELEMENT"

[3] Fix | Delete

END_ELEMENT = "END_ELEMENT"

[4] Fix | Delete

COMMENT = "COMMENT"

[5] Fix | Delete

START_DOCUMENT = "START_DOCUMENT"

[6] Fix | Delete

END_DOCUMENT = "END_DOCUMENT"

[7] Fix | Delete

PROCESSING_INSTRUCTION = "PROCESSING_INSTRUCTION"

[8] Fix | Delete

IGNORABLE_WHITESPACE = "IGNORABLE_WHITESPACE"

[9] Fix | Delete

CHARACTERS = "CHARACTERS"

[10] Fix | Delete

[11] Fix | Delete

class PullDOM(xml.sax.ContentHandler):

[12] Fix | Delete

_locator = None

[13] Fix | Delete

document = None

[14] Fix | Delete

[15] Fix | Delete

def __init__(self, documentFactory=None):

[16] Fix | Delete

from xml.dom import XML_NAMESPACE

[17] Fix | Delete

self.documentFactory = documentFactory

[18] Fix | Delete

self.firstEvent = [None, None]

[19] Fix | Delete

self.lastEvent = self.firstEvent

[20] Fix | Delete

self.elementStack = []

[21] Fix | Delete

self.push = self.elementStack.append

[22] Fix | Delete

try:

[23] Fix | Delete

self.pop = self.elementStack.pop

[24] Fix | Delete

except AttributeError:

[25] Fix | Delete

# use class' pop instead

[26] Fix | Delete

pass

[27] Fix | Delete

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

[28] Fix | Delete

self._current_context = self._ns_contexts[-1]

[29] Fix | Delete

self.pending_events = []

[30] Fix | Delete

[31] Fix | Delete

def pop(self):

[32] Fix | Delete

result = self.elementStack[-1]

[33] Fix | Delete

del self.elementStack[-1]

[34] Fix | Delete

return result

[35] Fix | Delete

[36] Fix | Delete

def setDocumentLocator(self, locator):

[37] Fix | Delete

self._locator = locator

[38] Fix | Delete

[39] Fix | Delete

def startPrefixMapping(self, prefix, uri):

[40] Fix | Delete

if not hasattr(self, '_xmlns_attrs'):

[41] Fix | Delete

self._xmlns_attrs = []

[42] Fix | Delete

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

[43] Fix | Delete

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

[44] Fix | Delete

self._current_context[uri] = prefix or None

[45] Fix | Delete

[46] Fix | Delete

def endPrefixMapping(self, prefix):

[47] Fix | Delete

self._current_context = self._ns_contexts.pop()

[48] Fix | Delete

[49] Fix | Delete

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

[50] Fix | Delete

# Retrieve xml namespace declaration attributes.

[51] Fix | Delete

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

[52] Fix | Delete

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

[53] Fix | Delete

if xmlns_attrs is not None:

[54] Fix | Delete

for aname, value in xmlns_attrs:

[55] Fix | Delete

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

[56] Fix | Delete

self._xmlns_attrs = []

[57] Fix | Delete

uri, localname = name

[58] Fix | Delete

if uri:

[59] Fix | Delete

# When using namespaces, the reader may or may not

[60] Fix | Delete

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

[61] Fix | Delete

# *a* valid tagName from the current context.

[62] Fix | Delete

if tagName is None:

[63] Fix | Delete

prefix = self._current_context[uri]

[64] Fix | Delete

if prefix:

[65] Fix | Delete

tagName = prefix + ":" + localname

[66] Fix | Delete

else:

[67] Fix | Delete

tagName = localname

[68] Fix | Delete

if self.document:

[69] Fix | Delete

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

[70] Fix | Delete

else:

[71] Fix | Delete

node = self.buildDocument(uri, tagName)

[72] Fix | Delete

else:

[73] Fix | Delete

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

[74] Fix | Delete

# localname

[75] Fix | Delete

if self.document:

[76] Fix | Delete

node = self.document.createElement(localname)

[77] Fix | Delete

else:

[78] Fix | Delete

node = self.buildDocument(None, localname)

[79] Fix | Delete

[80] Fix | Delete

for aname,value in attrs.items():

[81] Fix | Delete

a_uri, a_localname = aname

[82] Fix | Delete

if a_uri == xmlns_uri:

[83] Fix | Delete

if a_localname == 'xmlns':

[84] Fix | Delete

qname = a_localname

[85] Fix | Delete

else:

[86] Fix | Delete

qname = 'xmlns:' + a_localname

[87] Fix | Delete

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

[88] Fix | Delete

node.setAttributeNodeNS(attr)

[89] Fix | Delete

elif a_uri:

[90] Fix | Delete

prefix = self._current_context[a_uri]

[91] Fix | Delete

if prefix:

[92] Fix | Delete

qname = prefix + ":" + a_localname

[93] Fix | Delete

else:

[94] Fix | Delete

qname = a_localname

[95] Fix | Delete

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

[96] Fix | Delete

node.setAttributeNodeNS(attr)

[97] Fix | Delete

else:

[98] Fix | Delete

attr = self.document.createAttribute(a_localname)

[99] Fix | Delete

node.setAttributeNode(attr)

[100] Fix | Delete

attr.value = value

[101] Fix | Delete

[102] Fix | Delete

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

[103] Fix | Delete

self.lastEvent = self.lastEvent[1]

[104] Fix | Delete

self.push(node)

[105] Fix | Delete

[106] Fix | Delete

def endElementNS(self, name, tagName):

[107] Fix | Delete

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

[108] Fix | Delete

self.lastEvent = self.lastEvent[1]

[109] Fix | Delete

[110] Fix | Delete

def startElement(self, name, attrs):

[111] Fix | Delete

if self.document:

[112] Fix | Delete

node = self.document.createElement(name)

[113] Fix | Delete

else:

[114] Fix | Delete

node = self.buildDocument(None, name)

[115] Fix | Delete

[116] Fix | Delete

for aname,value in attrs.items():

[117] Fix | Delete

attr = self.document.createAttribute(aname)

[118] Fix | Delete

attr.value = value

[119] Fix | Delete

node.setAttributeNode(attr)

[120] Fix | Delete

[121] Fix | Delete

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

[122] Fix | Delete

self.lastEvent = self.lastEvent[1]

[123] Fix | Delete

self.push(node)

[124] Fix | Delete

[125] Fix | Delete

def endElement(self, name):

[126] Fix | Delete

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

[127] Fix | Delete

self.lastEvent = self.lastEvent[1]

[128] Fix | Delete

[129] Fix | Delete

def comment(self, s):

[130] Fix | Delete

if self.document:

[131] Fix | Delete

node = self.document.createComment(s)

[132] Fix | Delete

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

[133] Fix | Delete

self.lastEvent = self.lastEvent[1]

[134] Fix | Delete

else:

[135] Fix | Delete

event = [(COMMENT, s), None]

[136] Fix | Delete

self.pending_events.append(event)

[137] Fix | Delete

[138] Fix | Delete

def processingInstruction(self, target, data):

[139] Fix | Delete

if self.document:

[140] Fix | Delete

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

[141] Fix | Delete

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

[142] Fix | Delete

self.lastEvent = self.lastEvent[1]

[143] Fix | Delete

else:

[144] Fix | Delete

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

[145] Fix | Delete

self.pending_events.append(event)

[146] Fix | Delete

[147] Fix | Delete

def ignorableWhitespace(self, chars):

[148] Fix | Delete

node = self.document.createTextNode(chars)

[149] Fix | Delete

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

[150] Fix | Delete

self.lastEvent = self.lastEvent[1]

[151] Fix | Delete

[152] Fix | Delete

def characters(self, chars):

[153] Fix | Delete

node = self.document.createTextNode(chars)

[154] Fix | Delete

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

[155] Fix | Delete

self.lastEvent = self.lastEvent[1]

[156] Fix | Delete

[157] Fix | Delete

def startDocument(self):

[158] Fix | Delete

if self.documentFactory is None:

[159] Fix | Delete

import xml.dom.minidom

[160] Fix | Delete

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

[161] Fix | Delete

[162] Fix | Delete

def buildDocument(self, uri, tagname):

[163] Fix | Delete

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

[164] Fix | Delete

# XXX: obtain DocumentType

[165] Fix | Delete

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

[166] Fix | Delete

self.document = node

[167] Fix | Delete

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

[168] Fix | Delete

self.lastEvent = self.lastEvent[1]

[169] Fix | Delete

self.push(node)

[170] Fix | Delete

# Put everything we have seen so far into the document

[171] Fix | Delete

for e in self.pending_events:

[172] Fix | Delete

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

[173] Fix | Delete

_,target,data = e[0]

[174] Fix | Delete

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

[175] Fix | Delete

e[0] = (PROCESSING_INSTRUCTION, n)

[176] Fix | Delete

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

[177] Fix | Delete

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

[178] Fix | Delete

e[0] = (COMMENT, n)

[179] Fix | Delete

else:

[180] Fix | Delete

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

[181] Fix | Delete

self.lastEvent[1] = e

[182] Fix | Delete

self.lastEvent = e

[183] Fix | Delete

self.pending_events = None

[184] Fix | Delete

return node.firstChild

[185] Fix | Delete

[186] Fix | Delete

def endDocument(self):

[187] Fix | Delete

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

[188] Fix | Delete

self.pop()

[189] Fix | Delete

[190] Fix | Delete

def clear(self):

[191] Fix | Delete

"clear(): Explicitly release parsing structures"

[192] Fix | Delete

self.document = None

[193] Fix | Delete

[194] Fix | Delete

class ErrorHandler:

[195] Fix | Delete

def warning(self, exception):

[196] Fix | Delete

print(exception)

[197] Fix | Delete

def error(self, exception):

[198] Fix | Delete

raise exception

[199] Fix | Delete

def fatalError(self, exception):

[200] Fix | Delete

raise exception

[201] Fix | Delete

[202] Fix | Delete

class DOMEventStream:

[203] Fix | Delete

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

[204] Fix | Delete

self.stream = stream

[205] Fix | Delete

self.parser = parser

[206] Fix | Delete

self.bufsize = bufsize

[207] Fix | Delete

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

[208] Fix | Delete

self.getEvent = self._slurp

[209] Fix | Delete

self.reset()

[210] Fix | Delete

[211] Fix | Delete

def reset(self):

[212] Fix | Delete

self.pulldom = PullDOM()

[213] Fix | Delete

# This content handler relies on namespace support

[214] Fix | Delete

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

[215] Fix | Delete

self.parser.setContentHandler(self.pulldom)

[216] Fix | Delete

[217] Fix | Delete

def __getitem__(self, pos):

[218] Fix | Delete

rc = self.getEvent()

[219] Fix | Delete

if rc:

[220] Fix | Delete

return rc

[221] Fix | Delete

raise IndexError

[222] Fix | Delete

[223] Fix | Delete

def __next__(self):

[224] Fix | Delete

rc = self.getEvent()

[225] Fix | Delete

if rc:

[226] Fix | Delete

return rc

[227] Fix | Delete

raise StopIteration

[228] Fix | Delete

[229] Fix | Delete

def __iter__(self):

[230] Fix | Delete

return self

[231] Fix | Delete

[232] Fix | Delete

def expandNode(self, node):

[233] Fix | Delete

event = self.getEvent()

[234] Fix | Delete

parents = [node]

[235] Fix | Delete

while event:

[236] Fix | Delete

token, cur_node = event

[237] Fix | Delete

if cur_node is node:

[238] Fix | Delete

return

[239] Fix | Delete

if token != END_ELEMENT:

[240] Fix | Delete

parents[-1].appendChild(cur_node)

[241] Fix | Delete

if token == START_ELEMENT:

[242] Fix | Delete

parents.append(cur_node)

[243] Fix | Delete

elif token == END_ELEMENT:

[244] Fix | Delete

del parents[-1]

[245] Fix | Delete

event = self.getEvent()

[246] Fix | Delete

[247] Fix | Delete

def getEvent(self):

[248] Fix | Delete

# use IncrementalParser interface, so we get the desired

[249] Fix | Delete

# pull effect

[250] Fix | Delete

if not self.pulldom.firstEvent[1]:

[251] Fix | Delete

self.pulldom.lastEvent = self.pulldom.firstEvent

[252] Fix | Delete

while not self.pulldom.firstEvent[1]:

[253] Fix | Delete

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

[254] Fix | Delete

if not buf:

[255] Fix | Delete

self.parser.close()

[256] Fix | Delete

return None

[257] Fix | Delete

self.parser.feed(buf)

[258] Fix | Delete

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

[259] Fix | Delete

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

[260] Fix | Delete

return rc

[261] Fix | Delete

[262] Fix | Delete

def _slurp(self):

[263] Fix | Delete

""" Fallback replacement for getEvent() using the

[264] Fix | Delete

standard SAX2 interface, which means we slurp the

[265] Fix | Delete

SAX events into memory (no performance gain, but

[266] Fix | Delete

we are compatible to all SAX parsers).

[267] Fix | Delete

"""

[268] Fix | Delete

self.parser.parse(self.stream)

[269] Fix | Delete

self.getEvent = self._emit

[270] Fix | Delete

return self._emit()

[271] Fix | Delete

[272] Fix | Delete

def _emit(self):

[273] Fix | Delete

""" Fallback replacement for getEvent() that emits

[274] Fix | Delete

the events that _slurp() read previously.

[275] Fix | Delete

"""

[276] Fix | Delete

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

[277] Fix | Delete

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

[278] Fix | Delete

return rc

[279] Fix | Delete

[280] Fix | Delete

def clear(self):

[281] Fix | Delete

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

[282] Fix | Delete

self.pulldom.clear()

[283] Fix | Delete

del self.pulldom

[284] Fix | Delete

self.parser = None

[285] Fix | Delete

self.stream = None

[286] Fix | Delete

[287] Fix | Delete

class SAX2DOM(PullDOM):

[288] Fix | Delete

[289] Fix | Delete

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

[290] Fix | Delete

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

[291] Fix | Delete

curNode = self.elementStack[-1]

[292] Fix | Delete

parentNode = self.elementStack[-2]

[293] Fix | Delete

parentNode.appendChild(curNode)

[294] Fix | Delete

[295] Fix | Delete

def startElement(self, name, attrs):

[296] Fix | Delete

PullDOM.startElement(self, name, 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 processingInstruction(self, target, data):

[302] Fix | Delete

PullDOM.processingInstruction(self, target, data)

[303] Fix | Delete

node = self.lastEvent[0][1]

[304] Fix | Delete

parentNode = self.elementStack[-1]

[305] Fix | Delete

parentNode.appendChild(node)

[306] Fix | Delete

[307] Fix | Delete

def ignorableWhitespace(self, chars):

[308] Fix | Delete

PullDOM.ignorableWhitespace(self, chars)

[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 characters(self, chars):

[314] Fix | Delete

PullDOM.characters(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

[320] Fix | Delete

default_bufsize = (2 ** 14) - 20

[321] Fix | Delete

[322] Fix | Delete

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

[323] Fix | Delete

if bufsize is None:

[324] Fix | Delete

bufsize = default_bufsize

[325] Fix | Delete

if isinstance(stream_or_string, str):

[326] Fix | Delete

stream = open(stream_or_string, 'rb')

[327] Fix | Delete

else:

[328] Fix | Delete

stream = stream_or_string

[329] Fix | Delete

if not parser:

[330] Fix | Delete

parser = xml.sax.make_parser()

[331] Fix | Delete

return DOMEventStream(stream, parser, bufsize)

[332] Fix | Delete

[333] Fix | Delete

def parseString(string, parser=None):

[334] Fix | Delete

from io import StringIO

[335] Fix | Delete

[336] Fix | Delete

bufsize = len(string)

[337] Fix | Delete

buf = StringIO(string)

[338] Fix | Delete

if not parser:

[339] Fix | Delete

parser = xml.sax.make_parser()

[340] Fix | Delete

return DOMEventStream(buf, parser, bufsize)

[341] Fix | Delete

[342] Fix | Delete