https://github.com/python/cpython/commit/e7c93b7e5c2cb3ffe076845edd6b5c287af3bc4c
commit: e7c93b7e5c2cb3ffe076845edd6b5c287af3bc4c
branch: main
author: Serhiy Storchaka <[email protected]>
committer: serhiy-storchaka <[email protected]>
date: 2026-09-02T15:30:21+03:00
summary:
gh-156658: Only XML white space characters are treated as white space
(GH-156659)
XML defines white space as " \t\r\n" (see XML 1.0, 2.3), but str.strip()
also strips other characters, such as U+00A0. Such characters could be
lost in ElementTree.indent(), in canonicalize(strip_text=True), and when
parsing with the whitespace-in-element-content feature turned off.
files:
A Misc/NEWS.d/next/Library/2026-08-30-19-30-00.gh-issue-156658.Xq5Nt7.rst
M Lib/test/test_minidom.py
M Lib/test/test_xml_etree.py
M Lib/xml/dom/expatbuilder.py
M Lib/xml/dom/minidom.py
M Lib/xml/etree/ElementTree.py
diff --git a/Lib/test/test_minidom.py b/Lib/test/test_minidom.py
index 3735a6046891ea9..e204bdc7dc672db 100644
--- a/Lib/test/test_minidom.py
+++ b/Lib/test/test_minidom.py
@@ -642,6 +642,30 @@ def test_toprettyxml_preserves_content_of_text_node(self):
dom.getElementsByTagName('B')[0].childNodes[0].toxml(),
dom2.getElementsByTagName('B')[0].childNodes[0].toxml())
+ def test_isWhitespaceInElementContent(self):
+ # only " \t\r\n" are whitespace in XML (see XML 1.0, 2.3)
+ dom = parseString('<!DOCTYPE a [<!ELEMENT a (b)*><!ELEMENT b
(#PCDATA)>]>'
+ '<a> <b>x</b>\xa0</a>')
+ children = dom.documentElement.childNodes
+ self.assertTrue(children[0].isWhitespaceInElementContent)
+ self.assertFalse(children[2].isWhitespaceInElementContent)
+ dom.unlink()
+
+ def test_remove_whitespace_in_element_content(self):
+ from xml.dom.xmlbuilder import DOMBuilder, DOMInputSource
+ builder = DOMBuilder()
+ builder.setFeature("whitespace-in-element-content", False)
+ source = DOMInputSource()
+ source.byteStream = io.BytesIO(
+ b'<!DOCTYPE a [<!ELEMENT a (b)*><!ELEMENT b (#PCDATA)>]>'
+ b'<a> <b>x</b>\xc2\xa0</a>')
+ dom = builder.parse(source)
+ children = dom.documentElement.childNodes
+ # ignorable whitespace is removed, other characters are not
+ self.assertEqual([node.nodeName for node in children], ['b', '#text'])
+ self.assertEqual(children[1].data, '\xa0')
+ dom.unlink()
+
def testProcessingInstruction(self):
dom = parseString('<e><?mypi \t\n data \t\n ?></e>')
pi = dom.documentElement.firstChild
diff --git a/Lib/test/test_xml_etree.py b/Lib/test/test_xml_etree.py
index a0337de58b23ae4..f9ff8c4c3541eda 100644
--- a/Lib/test/test_xml_etree.py
+++ b/Lib/test/test_xml_etree.py
@@ -845,6 +845,15 @@ def test_indent_space_caching(self):
len({id(el.tail) for el in elem.iter()}),
)
+ def test_indent_non_xml_whitespace(self):
+ # only " \t\r\n" are whitespace in XML (see XML 1.0, 2.3)
+ elem = ET.XML('<html>\xa0<body><p>text</p>\xa0</body></html>')
+ ET.indent(elem)
+ self.assertEqual(
+ ET.tostring(elem),
+ b'<html> <body>\n <p>text</p> </body>\n</html>'
+ )
+
def test_indent_level(self):
elem =
ET.XML("<html><body><p>pre<br/>post</p><p>text</p></body></html>")
with self.assertRaises(ValueError):
@@ -4900,6 +4909,11 @@ def test_simple_roundtrip(self):
xml = '<X xmlns="http://nps/a";><Y xmlns:b="http://nsp/b";
b:targets="abc,xyz"></Y></X>'
self.assertEqual(c14n_roundtrip(xml), xml)
+ def test_c14n_strip_non_xml_whitespace(self):
+ # only " \t\r\n" are whitespace in XML (see XML 1.0, 2.3)
+ self.assertEqual(c14n_roundtrip("<a> \xa0x\xa0 </a>", strip_text=True),
+ "<a>\xa0x\xa0</a>")
+
def test_c14n_exclusion(self):
xml = textwrap.dedent("""\
<root xmlns:x="http://example.com/x";>
diff --git a/Lib/xml/dom/expatbuilder.py b/Lib/xml/dom/expatbuilder.py
index d56b2ddfdb25698..e3917c1cc682880 100644
--- a/Lib/xml/dom/expatbuilder.py
+++ b/Lib/xml/dom/expatbuilder.py
@@ -30,7 +30,8 @@
from xml.dom import xmlbuilder, minidom, Node
from xml.dom import EMPTY_NAMESPACE, EMPTY_PREFIX, XMLNS_NAMESPACE
from xml.parsers import expat
-from xml.dom.minidom import _append_child, _set_attribute_node
+from xml.dom.minidom import (_append_child, _set_attribute_node,
+ _XML_WHITESPACE)
from xml.dom.NodeFilter import NodeFilter
TEXT_NODE = Node.TEXT_NODE
@@ -413,7 +414,8 @@ def _handle_white_text_nodes(self, node, info):
# whitespace.
L = []
for child in node.childNodes:
- if child.nodeType == TEXT_NODE and not child.data.strip():
+ if (child.nodeType == TEXT_NODE
+ and not child.data.strip(_XML_WHITESPACE)):
L.append(child)
# Remove ignorable whitespace from the tree.
diff --git a/Lib/xml/dom/minidom.py b/Lib/xml/dom/minidom.py
index 5fd3911bd3c9ebb..7639fa14c5050fb 100644
--- a/Lib/xml/dom/minidom.py
+++ b/Lib/xml/dom/minidom.py
@@ -31,6 +31,9 @@
_nodeTypes_with_children = (xml.dom.Node.ELEMENT_NODE,
xml.dom.Node.ENTITY_REFERENCE_NODE)
+# The white space characters of the XML specification (see XML 1.0, 2.3).
+_XML_WHITESPACE = " \t\r\n"
+
class Node(xml.dom.Node):
namespaceURI = None # this is non-null only for elements and attributes
@@ -1209,7 +1212,7 @@ def replaceWholeText(self, content):
return None
def _get_isWhitespaceInElementContent(self):
- if self.data.strip():
+ if self.data.strip(_XML_WHITESPACE):
return False
elem = _get_containing_element(self)
if elem is None:
diff --git a/Lib/xml/etree/ElementTree.py b/Lib/xml/etree/ElementTree.py
index 1cba3eae7ad629d..2fee57df59fdd83 100644
--- a/Lib/xml/etree/ElementTree.py
+++ b/Lib/xml/etree/ElementTree.py
@@ -101,6 +101,9 @@
from . import ElementPath
+# The white space characters of the XML specification (see XML 1.0, 2.3).
+_XML_WHITESPACE = " \t\r\n"
+
class ParseError(SyntaxError):
"""An error when parsing an XML document.
@@ -1209,17 +1212,17 @@ def _indent_children(elem, level):
child_indentation = indentations[level] + space
indentations.append(child_indentation)
- if not elem.text or not elem.text.strip():
+ if not elem.text or not elem.text.strip(_XML_WHITESPACE):
elem.text = child_indentation
for child in elem:
if len(child):
_indent_children(child, child_level)
- if not child.tail or not child.tail.strip():
+ if not child.tail or not child.tail.strip(_XML_WHITESPACE):
child.tail = child_indentation
# Dedent after the last child by overwriting the previous indentation.
- if not child.tail.strip():
+ if not child.tail.strip(_XML_WHITESPACE):
child.tail = indentations[level]
_indent_children(tree, 0)
@@ -1724,7 +1727,7 @@ def _default(self, text):
if prefix == ">":
self._doctype = None
return
- text = text.strip()
+ text = text.strip(_XML_WHITESPACE)
if not text:
return
self._doctype.append(text)
@@ -1940,7 +1943,7 @@ def _flush(self, _join_text=''.join):
data = _join_text(self._data)
del self._data[:]
if self._strip_text and not self._preserve_space[-1]:
- data = data.strip()
+ data = data.strip(_XML_WHITESPACE)
if self._pending_start is not None:
args, self._pending_start = self._pending_start, None
qname_text = data if data and _looks_like_prefix_name(data) else
None
diff --git
a/Misc/NEWS.d/next/Library/2026-08-30-19-30-00.gh-issue-156658.Xq5Nt7.rst
b/Misc/NEWS.d/next/Library/2026-08-30-19-30-00.gh-issue-156658.Xq5Nt7.rst
new file mode 100644
index 000000000000000..3f2f105116c1838
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-08-30-19-30-00.gh-issue-156658.Xq5Nt7.rst
@@ -0,0 +1,5 @@
+:mod:`xml.dom` and :mod:`xml.etree.ElementTree` no longer treat characters
+which are not white space in XML (such as U+00A0) as white space. Previously
+they could be lost in :func:`~xml.etree.ElementTree.indent`,
+:func:`~xml.etree.ElementTree.canonicalize` with ``strip_text=True``, and when
+parsing with the ``whitespace-in-element-content`` feature turned off.
_______________________________________________
Python-checkins mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3//lists/python-checkins.python.org
Member address: [email protected]
