Issue #10131: Fixed deep copying of minidom documents. Based on patch

by Marian Ganisin.
This commit is contained in:
Serhiy Storchaka 2015-11-26 23:49:42 +02:00
parent 747d48cf27
commit c472246d81
4 changed files with 94 additions and 45 deletions

View File

@ -1,5 +1,6 @@
# test for xml.dom.minidom
import copy
import pickle
from test.support import run_unittest, findfile
import unittest
@ -11,6 +12,13 @@ from xml.dom.minidom import getDOMImplementation
tstfile = findfile("test.xml", subdir="xmltestdata")
sample = ("<?xml version='1.0' encoding='us-ascii'?>\n"
"<!DOCTYPE doc PUBLIC 'http://xml.python.org/public'"
" 'http://xml.python.org/system' [\n"
" <!ELEMENT e EMPTY>\n"
" <!ENTITY ent SYSTEM 'http://xml.python.org/entity'>\n"
"]><doc attr='value'> text\n"
"<?pi sample?> <!-- comment --> <e/> </doc>")
# The tests of DocumentType importing use these helpers to construct
# the documents to work with, since not all DOM builders actually
@ -1466,52 +1474,54 @@ class MinidomTest(unittest.TestCase):
self.confirm(e.isSameNode(doc.getElementById("w"))
and a2.isId)
def assert_recursive_equal(self, doc, doc2):
stack = [(doc, doc2)]
while stack:
n1, n2 = stack.pop()
self.assertEqual(n1.nodeType, n2.nodeType)
self.assertEqual(len(n1.childNodes), len(n2.childNodes))
self.assertEqual(n1.nodeName, n2.nodeName)
self.assertFalse(n1.isSameNode(n2))
self.assertFalse(n2.isSameNode(n1))
if n1.nodeType == Node.DOCUMENT_TYPE_NODE:
len(n1.entities)
len(n2.entities)
len(n1.notations)
len(n2.notations)
self.assertEqual(len(n1.entities), len(n2.entities))
self.assertEqual(len(n1.notations), len(n2.notations))
for i in range(len(n1.notations)):
# XXX this loop body doesn't seem to be executed?
no1 = n1.notations.item(i)
no2 = n1.notations.item(i)
self.assertEqual(no1.name, no2.name)
self.assertEqual(no1.publicId, no2.publicId)
self.assertEqual(no1.systemId, no2.systemId)
stack.append((no1, no2))
for i in range(len(n1.entities)):
e1 = n1.entities.item(i)
e2 = n2.entities.item(i)
self.assertEqual(e1.notationName, e2.notationName)
self.assertEqual(e1.publicId, e2.publicId)
self.assertEqual(e1.systemId, e2.systemId)
stack.append((e1, e2))
if n1.nodeType != Node.DOCUMENT_NODE:
self.assertTrue(n1.ownerDocument.isSameNode(doc))
self.assertTrue(n2.ownerDocument.isSameNode(doc2))
for i in range(len(n1.childNodes)):
stack.append((n1.childNodes[i], n2.childNodes[i]))
def testPickledDocument(self):
doc = parseString("<?xml version='1.0' encoding='us-ascii'?>\n"
"<!DOCTYPE doc PUBLIC 'http://xml.python.org/public'"
" 'http://xml.python.org/system' [\n"
" <!ELEMENT e EMPTY>\n"
" <!ENTITY ent SYSTEM 'http://xml.python.org/entity'>\n"
"]><doc attr='value'> text\n"
"<?pi sample?> <!-- comment --> <e/> </doc>")
doc = parseString(sample)
for proto in range(2, pickle.HIGHEST_PROTOCOL + 1):
s = pickle.dumps(doc, proto)
doc2 = pickle.loads(s)
stack = [(doc, doc2)]
while stack:
n1, n2 = stack.pop()
self.confirm(n1.nodeType == n2.nodeType
and len(n1.childNodes) == len(n2.childNodes)
and n1.nodeName == n2.nodeName
and not n1.isSameNode(n2)
and not n2.isSameNode(n1))
if n1.nodeType == Node.DOCUMENT_TYPE_NODE:
len(n1.entities)
len(n2.entities)
len(n1.notations)
len(n2.notations)
self.confirm(len(n1.entities) == len(n2.entities)
and len(n1.notations) == len(n2.notations))
for i in range(len(n1.notations)):
# XXX this loop body doesn't seem to be executed?
no1 = n1.notations.item(i)
no2 = n1.notations.item(i)
self.confirm(no1.name == no2.name
and no1.publicId == no2.publicId
and no1.systemId == no2.systemId)
stack.append((no1, no2))
for i in range(len(n1.entities)):
e1 = n1.entities.item(i)
e2 = n2.entities.item(i)
self.confirm(e1.notationName == e2.notationName
and e1.publicId == e2.publicId
and e1.systemId == e2.systemId)
stack.append((e1, e2))
if n1.nodeType != Node.DOCUMENT_NODE:
self.confirm(n1.ownerDocument.isSameNode(doc)
and n2.ownerDocument.isSameNode(doc2))
for i in range(len(n1.childNodes)):
stack.append((n1.childNodes[i], n2.childNodes[i]))
self.assert_recursive_equal(doc, doc2)
def testDeepcopiedDocument(self):
doc = parseString(sample)
doc2 = copy.deepcopy(doc)
self.assert_recursive_equal(doc, doc2)
def testSerializeCommentNodeWithDoubleHyphen(self):
doc = create_doc_without_doctype()

View File

@ -1,5 +1,6 @@
# Tests for xml.dom.minicompat
import copy
import pickle
import unittest
@ -89,6 +90,7 @@ class NodeListTestCase(unittest.TestCase):
node_list = NodeList()
pickled = pickle.dumps(node_list, proto)
unpickled = pickle.loads(pickled)
self.assertIsNot(unpickled, node_list)
self.assertEqual(unpickled, node_list)
# Non-empty NodeList.
@ -96,7 +98,41 @@ class NodeListTestCase(unittest.TestCase):
node_list.append(2)
pickled = pickle.dumps(node_list, proto)
unpickled = pickle.loads(pickled)
self.assertIsNot(unpickled, node_list)
self.assertEqual(unpickled, node_list)
def test_nodelist_copy(self):
# Empty NodeList.
node_list = NodeList()
copied = copy.copy(node_list)
self.assertIsNot(copied, node_list)
self.assertEqual(copied, node_list)
# Non-empty NodeList.
node_list.append([1])
node_list.append([2])
copied = copy.copy(node_list)
self.assertIsNot(copied, node_list)
self.assertEqual(copied, node_list)
for x, y in zip(copied, node_list):
self.assertIs(x, y)
def test_nodelist_deepcopy(self):
# Empty NodeList.
node_list = NodeList()
copied = copy.deepcopy(node_list)
self.assertIsNot(copied, node_list)
self.assertEqual(copied, node_list)
# Non-empty NodeList.
node_list.append([1])
node_list.append([2])
copied = copy.deepcopy(node_list)
self.assertIsNot(copied, node_list)
self.assertEqual(copied, node_list)
for x, y in zip(copied, node_list):
self.assertIsNot(x, y)
self.assertEqual(x, y)
if __name__ == '__main__':
unittest.main()

View File

@ -64,10 +64,10 @@ class NodeList(list):
length = property(_get_length, _set_length,
doc="The number of nodes in the NodeList.")
def __getstate__(self):
return list(self)
# For backward compatibility
def __setstate__(self, state):
if state is None:
state = []
self[:] = state

View File

@ -113,6 +113,9 @@ Core and Builtins
Library
-------
- Issue #10131: Fixed deep copying of minidom documents. Based on patch
by Marian Ganisin.
- Issue #25725: Fixed a reference leak in pickle.loads() when unpickling
invalid data including tuple instructions.