|
 |
10d7b0 |
diff --git a/src/lxml/apihelpers.pxi b/src/lxml/apihelpers.pxi
|
|
|
10d7b0 |
index 5eb3416..88a031d 100644
|
|
|
10d7b0 |
--- a/src/lxml/apihelpers.pxi
|
|
|
10d7b0 |
+++ b/src/lxml/apihelpers.pxi
|
|
|
10d7b0 |
@@ -246,9 +246,10 @@ cdef dict _build_nsmap(xmlNode* c_node):
|
|
|
10d7b0 |
while c_node is not NULL and c_node.type == tree.XML_ELEMENT_NODE:
|
|
|
10d7b0 |
c_ns = c_node.nsDef
|
|
|
10d7b0 |
while c_ns is not NULL:
|
|
|
10d7b0 |
- prefix = funicodeOrNone(c_ns.prefix)
|
|
|
10d7b0 |
- if prefix not in nsmap:
|
|
|
10d7b0 |
- nsmap[prefix] = funicodeOrNone(c_ns.href)
|
|
|
10d7b0 |
+ if c_ns.prefix or c_ns.href:
|
|
|
10d7b0 |
+ prefix = funicodeOrNone(c_ns.prefix)
|
|
|
10d7b0 |
+ if prefix not in nsmap:
|
|
|
10d7b0 |
+ nsmap[prefix] = funicodeOrNone(c_ns.href)
|
|
|
10d7b0 |
c_ns = c_ns.next
|
|
|
10d7b0 |
c_node = c_node.parent
|
|
|
10d7b0 |
return nsmap
|
|
|
10d7b0 |
diff --git a/src/lxml/includes/xmlparser.pxd b/src/lxml/includes/xmlparser.pxd
|
|
|
10d7b0 |
index a196e34..45acfc8 100644
|
|
|
10d7b0 |
--- a/src/lxml/includes/xmlparser.pxd
|
|
|
10d7b0 |
+++ b/src/lxml/includes/xmlparser.pxd
|
|
|
10d7b0 |
@@ -144,6 +144,7 @@ cdef extern from "libxml/parser.h":
|
|
|
10d7b0 |
void* userData
|
|
|
10d7b0 |
int* spaceTab
|
|
|
10d7b0 |
int spaceMax
|
|
|
10d7b0 |
+ int nsNr
|
|
|
10d7b0 |
bint html
|
|
|
10d7b0 |
bint progressive
|
|
|
10d7b0 |
int inSubset
|
|
|
10d7b0 |
diff --git a/src/lxml/iterparse.pxi b/src/lxml/iterparse.pxi
|
|
|
10d7b0 |
index 4c20506..3da7485 100644
|
|
|
10d7b0 |
--- a/src/lxml/iterparse.pxi
|
|
|
10d7b0 |
+++ b/src/lxml/iterparse.pxi
|
|
|
10d7b0 |
@@ -419,7 +419,7 @@ cdef int _countNsDefs(xmlNode* c_node):
|
|
|
10d7b0 |
count = 0
|
|
|
10d7b0 |
c_ns = c_node.nsDef
|
|
|
10d7b0 |
while c_ns is not NULL:
|
|
|
10d7b0 |
- count += 1
|
|
|
10d7b0 |
+ count += (c_ns.href is not NULL)
|
|
|
10d7b0 |
c_ns = c_ns.next
|
|
|
10d7b0 |
return count
|
|
|
10d7b0 |
|
|
|
10d7b0 |
@@ -430,9 +430,10 @@ cdef int _appendStartNsEvents(xmlNode* c_node, list event_list) except -1:
|
|
|
10d7b0 |
count = 0
|
|
|
10d7b0 |
c_ns = c_node.nsDef
|
|
|
10d7b0 |
while c_ns is not NULL:
|
|
|
10d7b0 |
- ns_tuple = (funicode(c_ns.prefix) if c_ns.prefix is not NULL else '',
|
|
|
10d7b0 |
- funicode(c_ns.href))
|
|
|
10d7b0 |
- event_list.append( (u"start-ns", ns_tuple) )
|
|
|
10d7b0 |
- count += 1
|
|
|
10d7b0 |
+ if c_ns.href:
|
|
|
10d7b0 |
+ ns_tuple = (funicodeOrEmpty(c_ns.prefix),
|
|
|
10d7b0 |
+ funicode(c_ns.href))
|
|
|
10d7b0 |
+ event_list.append( (u"start-ns", ns_tuple) )
|
|
|
10d7b0 |
+ count += 1
|
|
|
10d7b0 |
c_ns = c_ns.next
|
|
|
10d7b0 |
return count
|
|
|
10d7b0 |
diff --git a/src/lxml/parser.pxi b/src/lxml/parser.pxi
|
|
|
10d7b0 |
index 3ed223b..f5ff6b2 100644
|
|
|
10d7b0 |
--- a/src/lxml/parser.pxi
|
|
|
10d7b0 |
+++ b/src/lxml/parser.pxi
|
|
|
10d7b0 |
@@ -569,6 +569,9 @@ cdef class _ParserContext(_ResolverContext):
|
|
|
10d7b0 |
self._c_ctxt.disableSAX = 0 # work around bug in libxml2
|
|
|
10d7b0 |
else:
|
|
|
10d7b0 |
xmlparser.xmlClearParserCtxt(self._c_ctxt)
|
|
|
10d7b0 |
+ # work around bug in libxml2 [2.9.10 .. 2.9.14]:
|
|
|
10d7b0 |
+ # https://gitlab.gnome.org/GNOME/libxml2/-/issues/378
|
|
|
10d7b0 |
+ self._c_ctxt.nsNr = 0
|
|
|
10d7b0 |
|
|
|
10d7b0 |
cdef int prepare(self, bint set_document_loader=True) except -1:
|
|
|
10d7b0 |
cdef int result
|
|
|
10d7b0 |
diff --git a/src/lxml/tests/test_etree.py b/src/lxml/tests/test_etree.py
|
|
|
10d7b0 |
index 42613dc..db1f560 100644
|
|
|
10d7b0 |
--- a/src/lxml/tests/test_etree.py
|
|
|
10d7b0 |
+++ b/src/lxml/tests/test_etree.py
|
|
|
10d7b0 |
@@ -1459,6 +1459,27 @@ class ETreeOnlyTestCase(HelperTestCase):
|
|
|
10d7b0 |
[1,2,1,4],
|
|
|
10d7b0 |
counts)
|
|
|
10d7b0 |
|
|
|
10d7b0 |
+ def test_walk_after_parse_failure(self):
|
|
|
10d7b0 |
+ # This used to be an issue because libxml2 can leak empty namespaces
|
|
|
10d7b0 |
+ # between failed parser runs. iterwalk() failed to handle such a tree.
|
|
|
10d7b0 |
+ parser = etree.XMLParser()
|
|
|
10d7b0 |
+
|
|
|
10d7b0 |
+ try:
|
|
|
10d7b0 |
+ etree.XML('''<anot xmlns="1">''', parser=parser)
|
|
|
10d7b0 |
+ except etree.XMLSyntaxError:
|
|
|
10d7b0 |
+ pass
|
|
|
10d7b0 |
+ else:
|
|
|
10d7b0 |
+ assert False, "invalid input did not fail to parse"
|
|
|
10d7b0 |
+
|
|
|
10d7b0 |
+ et = etree.XML('''<root> </root>''', parser=parser)
|
|
|
10d7b0 |
+ try:
|
|
|
10d7b0 |
+ ns = next(etree.iterwalk(et, events=('start-ns',)))
|
|
|
10d7b0 |
+ except StopIteration:
|
|
|
10d7b0 |
+ # This would be the expected result, because there was no namespace
|
|
|
10d7b0 |
+ pass
|
|
|
10d7b0 |
+ else:
|
|
|
10d7b0 |
+ assert False, "Found unexpected namespace '%s'" % ns
|
|
|
10d7b0 |
+
|
|
|
10d7b0 |
def test_itertext_comment_pi(self):
|
|
|
10d7b0 |
# https://bugs.launchpad.net/lxml/+bug/1844674
|
|
|
10d7b0 |
XML = self.etree.XML
|