]> git.ipfire.org Git - thirdparty/Python/cpython.git/commitdiff
[3.9] gh-115398: Expose Expat >=2.6.0 reparse deferral API (CVE-2023-52425) (GH-11562...
authorSebastian Pipping <sebastian@pipping.org>
Wed, 6 Mar 2024 23:03:30 +0000 (00:03 +0100)
committerGitHub <noreply@github.com>
Wed, 6 Mar 2024 23:03:30 +0000 (00:03 +0100)
Allow controlling Expat >=2.6.0 reparse deferral (CVE-2023-52425) by adding five new methods:

- `xml.etree.ElementTree.XMLParser.flush`
- `xml.etree.ElementTree.XMLPullParser.flush`
- `xml.parsers.expat.xmlparser.GetReparseDeferralEnabled`
- `xml.parsers.expat.xmlparser.SetReparseDeferralEnabled`
- `xml.sax.expatreader.ExpatParser.flush`

Based on the "flush" idea from https://github.com/python/cpython/pull/115138#issuecomment-1932444270 .

Includes code suggested-by: Snild Dolkow <snild@sony.com>
and by core dev Serhiy Storchaka.

Co-authored-by: Gregory P. Smith <greg@krypto.org>
14 files changed:
Doc/library/pyexpat.rst
Doc/library/xml.etree.elementtree.rst
Include/pyexpat.h
Lib/test/test_pyexpat.py
Lib/test/test_sax.py
Lib/test/test_xml_etree.py
Lib/xml/etree/ElementTree.py
Lib/xml/sax/expatreader.py
Misc/NEWS.d/next/Security/2024-02-18-03-14-40.gh-issue-115398.tzvxH8.rst [new file with mode: 0644]
Modules/_elementtree.c
Modules/clinic/_elementtree.c.h
Modules/clinic/pyexpat.c.h
Modules/expat/pyexpatns.h
Modules/pyexpat.c

index 034e579315de00f9d6516534b46fb583f8ddb87e..fbf698456c2637b9892a85c18b7e3bd99c4309f2 100644 (file)
@@ -196,6 +196,42 @@ XMLParser Objects
    :exc:`ExpatError` to be raised with the :attr:`code` attribute set to
    ``errors.codes[errors.XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING]``.
 
+.. method:: xmlparser.SetReparseDeferralEnabled(enabled)
+
+   .. warning::
+
+      Calling ``SetReparseDeferralEnabled(False)`` has security implications,
+      as detailed below; please make sure to understand these consequences
+      prior to using the ``SetReparseDeferralEnabled`` method.
+
+   Expat 2.6.0 introduced a security mechanism called "reparse deferral"
+   where instead of causing denial of service through quadratic runtime
+   from reparsing large tokens, reparsing of unfinished tokens is now delayed
+   by default until a sufficient amount of input is reached.
+   Due to this delay, registered handlers may — depending of the sizing of
+   input chunks pushed to Expat — no longer be called right after pushing new
+   input to the parser.  Where immediate feedback and taking over responsiblity
+   of protecting against denial of service from large tokens are both wanted,
+   calling ``SetReparseDeferralEnabled(False)`` disables reparse deferral
+   for the current Expat parser instance, temporarily or altogether.
+   Calling ``SetReparseDeferralEnabled(True)`` allows re-enabling reparse
+   deferral.
+
+   Note that :meth:`SetReparseDeferralEnabled` has been backported to some
+   prior releases of CPython as a security fix.  Check for availability of
+   :meth:`SetReparseDeferralEnabled` using :func:`hasattr` if used in code
+   running across a variety of Python versions.
+
+   .. versionadded:: 3.9.19
+
+.. method:: xmlparser.GetReparseDeferralEnabled()
+
+   Returns whether reparse deferral is currently enabled for the given
+   Expat parser instance.
+
+   .. versionadded:: 3.9.19
+
+
 :class:`xmlparser` objects have the following attributes:
 
 
index 0d96a78cbbfd51ce484688229e6d92d10909d3d2..86ceb6ef36fe0805b6aa4127aba107a60ec4eac3 100644 (file)
@@ -165,6 +165,11 @@ data but would still like to have incremental parsing capabilities, take a look
 at :func:`iterparse`.  It can be useful when you're reading a large XML document
 and don't want to hold it wholly in memory.
 
+Where *immediate* feedback through events is wanted, calling method
+:meth:`XMLPullParser.flush` can help reduce delay;
+please make sure to study the related security notes.
+
+
 Finding interesting elements
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
@@ -1352,6 +1357,24 @@ XMLParser Objects
 
       Feeds data to the parser.  *data* is encoded data.
 
+
+   .. method:: flush()
+
+      Triggers parsing of any previously fed unparsed data, which can be
+      used to ensure more immediate feedback, in particular with Expat >=2.6.0.
+      The implementation of :meth:`flush` temporarily disables reparse deferral
+      with Expat (if currently enabled) and triggers a reparse.
+      Disabling reparse deferral has security consequences; please see
+      :meth:`xml.parsers.expat.xmlparser.SetReparseDeferralEnabled` for details.
+
+      Note that :meth:`flush` has been backported to some prior releases of
+      CPython as a security fix.  Check for availability of :meth:`flush`
+      using :func:`hasattr` if used in code running across a variety of Python
+      versions.
+
+      .. versionadded:: 3.9.19
+
+
    :meth:`XMLParser.feed` calls *target*\'s ``start(tag, attrs_dict)`` method
    for each opening tag, its ``end(tag)`` method for each closing tag, and data
    is processed by method ``data(data)``.  For further supported callback
@@ -1413,6 +1436,22 @@ XMLPullParser Objects
 
       Feed the given bytes data to the parser.
 
+   .. method:: flush()
+
+      Triggers parsing of any previously fed unparsed data, which can be
+      used to ensure more immediate feedback, in particular with Expat >=2.6.0.
+      The implementation of :meth:`flush` temporarily disables reparse deferral
+      with Expat (if currently enabled) and triggers a reparse.
+      Disabling reparse deferral has security consequences; please see
+      :meth:`xml.parsers.expat.xmlparser.SetReparseDeferralEnabled` for details.
+
+      Note that :meth:`flush` has been backported to some prior releases of
+      CPython as a security fix.  Check for availability of :meth:`flush`
+      using :func:`hasattr` if used in code running across a variety of Python
+      versions.
+
+      .. versionadded:: 3.9.19
+
    .. method:: close()
 
       Signal the parser that the data stream is terminated. Unlike
index 07020b5dc964cb3abeeb7bd14131dce55acc737a..9824d099c3df7d08083ee6f9232dedcbe9d846a0 100644 (file)
@@ -48,8 +48,10 @@ struct PyExpat_CAPI
     enum XML_Status (*SetEncoding)(XML_Parser parser, const XML_Char *encoding);
     int (*DefaultUnknownEncodingHandler)(
         void *encodingHandlerData, const XML_Char *name, XML_Encoding *info);
-    /* might be none for expat < 2.1.0 */
+    /* might be NULL for expat < 2.1.0 */
     int (*SetHashSalt)(XML_Parser parser, unsigned long hash_salt);
+    /* might be NULL for expat < 2.6.0 */
+    XML_Bool (*SetReparseDeferralEnabled)(XML_Parser parser, XML_Bool enabled);
     /* always add new stuff to the end! */
 };
 
index b2b4dea060532d8cad2db4a0f29b154a8688c41c..5212c7a704725e8f3d1117896c7000123f53a6a0 100644 (file)
@@ -730,5 +730,59 @@ class ForeignDTDTests(unittest.TestCase):
         self.assertEqual(handler_call_args, [("bar", "baz")])
 
 
+class ReparseDeferralTest(unittest.TestCase):
+    def test_getter_setter_round_trip(self):
+        parser = expat.ParserCreate()
+        enabled = (expat.version_info >= (2, 6, 0))
+
+        self.assertIs(parser.GetReparseDeferralEnabled(), enabled)
+        parser.SetReparseDeferralEnabled(False)
+        self.assertIs(parser.GetReparseDeferralEnabled(), False)
+        parser.SetReparseDeferralEnabled(True)
+        self.assertIs(parser.GetReparseDeferralEnabled(), enabled)
+
+    def test_reparse_deferral_enabled(self):
+        if expat.version_info < (2, 6, 0):
+            self.skipTest(f'Expat {expat.version_info} does not '
+                          'support reparse deferral')
+
+        started = []
+
+        def start_element(name, _):
+            started.append(name)
+
+        parser = expat.ParserCreate()
+        parser.StartElementHandler = start_element
+        self.assertTrue(parser.GetReparseDeferralEnabled())
+
+        for chunk in (b'<doc', b'/>'):
+            parser.Parse(chunk, False)
+
+        # The key test: Have handlers already fired?  Expecting: no.
+        self.assertEqual(started, [])
+
+        parser.Parse(b'', True)
+
+        self.assertEqual(started, ['doc'])
+
+    def test_reparse_deferral_disabled(self):
+        started = []
+
+        def start_element(name, _):
+            started.append(name)
+
+        parser = expat.ParserCreate()
+        parser.StartElementHandler = start_element
+        if expat.version_info >= (2, 6, 0):
+            parser.SetReparseDeferralEnabled(False)
+        self.assertFalse(parser.GetReparseDeferralEnabled())
+
+        for chunk in (b'<doc', b'/>'):
+            parser.Parse(chunk, False)
+
+        # The key test: Have handlers already fired?  Expecting: yes.
+        self.assertEqual(started, ['doc'])
+
+
 if __name__ == "__main__":
     unittest.main()
index 3e7fedaa6da702316d32233d9e41625246aee228..e6b9349fcddb6f2f7dd0996fdd7944586da52458 100644 (file)
@@ -18,6 +18,7 @@ from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl
 from io import BytesIO, StringIO
 import codecs
 import os.path
+import pyexpat
 import shutil
 from urllib.error import URLError
 import urllib.request
@@ -1210,6 +1211,56 @@ class ExpatReaderTest(XmlTestBase):
 
         self.assertEqual(result.getvalue(), start + b"<doc>text</doc>")
 
+    def test_flush_reparse_deferral_enabled(self):
+        if pyexpat.version_info < (2, 6, 0):
+            self.skipTest(f'Expat {pyexpat.version_info} does not support reparse deferral')
+
+        result = BytesIO()
+        xmlgen = XMLGenerator(result)
+        parser = create_parser()
+        parser.setContentHandler(xmlgen)
+
+        for chunk in ("<doc", ">"):
+            parser.feed(chunk)
+
+        self.assertEqual(result.getvalue(), start)  # i.e. no elements started
+        self.assertTrue(parser._parser.GetReparseDeferralEnabled())
+
+        parser.flush()
+
+        self.assertTrue(parser._parser.GetReparseDeferralEnabled())
+        self.assertEqual(result.getvalue(), start + b"<doc>")
+
+        parser.feed("</doc>")
+        parser.close()
+
+        self.assertEqual(result.getvalue(), start + b"<doc></doc>")
+
+    def test_flush_reparse_deferral_disabled(self):
+        result = BytesIO()
+        xmlgen = XMLGenerator(result)
+        parser = create_parser()
+        parser.setContentHandler(xmlgen)
+
+        for chunk in ("<doc", ">"):
+            parser.feed(chunk)
+
+        if pyexpat.version_info >= (2, 6, 0):
+            parser._parser.SetReparseDeferralEnabled(False)
+
+        self.assertEqual(result.getvalue(), start)  # i.e. no elements started
+        self.assertFalse(parser._parser.GetReparseDeferralEnabled())
+
+        parser.flush()
+
+        self.assertFalse(parser._parser.GetReparseDeferralEnabled())
+        self.assertEqual(result.getvalue(), start + b"<doc>")
+
+        parser.feed("</doc>")
+        parser.close()
+
+        self.assertEqual(result.getvalue(), start + b"<doc></doc>")
+
     # ===== Locator support
 
     def test_expat_locator_noinfo(self):
index 48e7c79932ad77649431166de33393a603c150bf..a811444848867599201426201d291ccbc3c129fb 100644 (file)
@@ -104,11 +104,6 @@ EXTERNAL_ENTITY_XML = """\
 """
 
 
-fails_with_expat_2_6_0 = (unittest.expectedFailure
-                        if pyexpat.version_info >= (2, 6, 0) else
-                        lambda test: test)
-
-
 def checkwarnings(*filters, quiet=False):
     def decorator(test):
         def newtest(*args, **kwargs):
@@ -1375,12 +1370,14 @@ class ElementTreeTest(unittest.TestCase):
 
 class XMLPullParserTest(unittest.TestCase):
 
-    def _feed(self, parser, data, chunk_size=None):
+    def _feed(self, parser, data, chunk_size=None, flush=False):
         if chunk_size is None:
             parser.feed(data)
         else:
             for i in range(0, len(data), chunk_size):
                 parser.feed(data[i:i+chunk_size])
+        if flush:
+            parser.flush()
 
     def assert_events(self, parser, expected, max_events=None):
         self.assertEqual(
@@ -1398,34 +1395,32 @@ class XMLPullParserTest(unittest.TestCase):
         self.assertEqual([(action, elem.tag) for action, elem in events],
                          expected)
 
-    def test_simple_xml(self, chunk_size=None):
+    def test_simple_xml(self, chunk_size=None, flush=False):
         parser = ET.XMLPullParser()
         self.assert_event_tags(parser, [])
-        self._feed(parser, "<!-- comment -->\n", chunk_size)
+        self._feed(parser, "<!-- comment -->\n", chunk_size, flush)
         self.assert_event_tags(parser, [])
         self._feed(parser,
                    "<root>\n  <element key='value'>text</element",
-                   chunk_size)
+                   chunk_size, flush)
         self.assert_event_tags(parser, [])
-        self._feed(parser, ">\n", chunk_size)
+        self._feed(parser, ">\n", chunk_size, flush)
         self.assert_event_tags(parser, [('end', 'element')])
-        self._feed(parser, "<element>text</element>tail\n", chunk_size)
-        self._feed(parser, "<empty-element/>\n", chunk_size)
+        self._feed(parser, "<element>text</element>tail\n", chunk_size, flush)
+        self._feed(parser, "<empty-element/>\n", chunk_size, flush)
         self.assert_event_tags(parser, [
             ('end', 'element'),
             ('end', 'empty-element'),
             ])
-        self._feed(parser, "</root>\n", chunk_size)
+        self._feed(parser, "</root>\n", chunk_size, flush)
         self.assert_event_tags(parser, [('end', 'root')])
         self.assertIsNone(parser.close())
 
-    @fails_with_expat_2_6_0
     def test_simple_xml_chunk_1(self):
-        self.test_simple_xml(chunk_size=1)
+        self.test_simple_xml(chunk_size=1, flush=True)
 
-    @fails_with_expat_2_6_0
     def test_simple_xml_chunk_5(self):
-        self.test_simple_xml(chunk_size=5)
+        self.test_simple_xml(chunk_size=5, flush=True)
 
     def test_simple_xml_chunk_22(self):
         self.test_simple_xml(chunk_size=22)
@@ -1624,6 +1619,57 @@ class XMLPullParserTest(unittest.TestCase):
         with self.assertRaises(ValueError):
             ET.XMLPullParser(events=('start', 'end', 'bogus'))
 
+    def test_flush_reparse_deferral_enabled(self):
+        if pyexpat.version_info < (2, 6, 0):
+            self.skipTest(f'Expat {pyexpat.version_info} does not '
+                          'support reparse deferral')
+
+        parser = ET.XMLPullParser(events=('start', 'end'))
+
+        for chunk in ("<doc", ">"):
+            parser.feed(chunk)
+
+        self.assert_event_tags(parser, [])  # i.e. no elements started
+        if ET is pyET:
+            self.assertTrue(parser._parser._parser.GetReparseDeferralEnabled())
+
+        parser.flush()
+
+        self.assert_event_tags(parser, [('start', 'doc')])
+        if ET is pyET:
+            self.assertTrue(parser._parser._parser.GetReparseDeferralEnabled())
+
+        parser.feed("</doc>")
+        parser.close()
+
+        self.assert_event_tags(parser, [('end', 'doc')])
+
+    def test_flush_reparse_deferral_disabled(self):
+        parser = ET.XMLPullParser(events=('start', 'end'))
+
+        for chunk in ("<doc", ">"):
+            parser.feed(chunk)
+
+        if pyexpat.version_info >= (2, 6, 0):
+            if not ET is pyET:
+                self.skipTest(f'XMLParser.(Get|Set)ReparseDeferralEnabled '
+                              'methods not available in C')
+            parser._parser._parser.SetReparseDeferralEnabled(False)
+
+        self.assert_event_tags(parser, [])  # i.e. no elements started
+        if ET is pyET:
+            self.assertFalse(parser._parser._parser.GetReparseDeferralEnabled())
+
+        parser.flush()
+
+        self.assert_event_tags(parser, [('start', 'doc')])
+        if ET is pyET:
+            self.assertFalse(parser._parser._parser.GetReparseDeferralEnabled())
+
+        parser.feed("</doc>")
+        parser.close()
+
+        self.assert_event_tags(parser, [('end', 'doc')])
 
 #
 # xinclude tests (samples from appendix C of the xinclude specification)
index 46368bb2f9ebb1fbb3d12938441787d6996a050b..b695768d11bb4ac1349626c25d68f3a7df1395ef 100644 (file)
@@ -1325,6 +1325,11 @@ class XMLPullParser:
             else:
                 yield event
 
+    def flush(self):
+        if self._parser is None:
+            raise ValueError("flush() called after end of stream")
+        self._parser.flush()
+
 
 def XML(text, parser=None):
     """Parse XML document from string constant.
@@ -1733,6 +1738,15 @@ class XMLParser:
             del self.parser, self._parser
             del self.target, self._target
 
+    def flush(self):
+        was_enabled = self.parser.GetReparseDeferralEnabled()
+        try:
+            self.parser.SetReparseDeferralEnabled(False)
+            self.parser.Parse(b"", False)
+        except self._error as v:
+            self._raiseerror(v)
+        finally:
+            self.parser.SetReparseDeferralEnabled(was_enabled)
 
 # --------------------------------------------------------------------
 # C14N 2.0
index e334ac9fea0d362ce2dbc494eecd2187d1dec88f..2f7c87b126c9933a4071dc19ec797e2472f28e42 100644 (file)
@@ -220,6 +220,20 @@ class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator):
             # FIXME: when to invoke error()?
             self._err_handler.fatalError(exc)
 
+    def flush(self):
+        if self._parser is None:
+            return
+
+        was_enabled = self._parser.GetReparseDeferralEnabled()
+        try:
+            self._parser.SetReparseDeferralEnabled(False)
+            self._parser.Parse(b"", False)
+        except expat.error as e:
+            exc = SAXParseException(expat.ErrorString(e.code), e, self)
+            self._err_handler.fatalError(exc)
+        finally:
+            self._parser.SetReparseDeferralEnabled(was_enabled)
+
     def _close_source(self):
         source = self._source
         try:
diff --git a/Misc/NEWS.d/next/Security/2024-02-18-03-14-40.gh-issue-115398.tzvxH8.rst b/Misc/NEWS.d/next/Security/2024-02-18-03-14-40.gh-issue-115398.tzvxH8.rst
new file mode 100644 (file)
index 0000000..a40fcd3
--- /dev/null
@@ -0,0 +1,8 @@
+Allow controlling Expat >=2.6.0 reparse deferral (CVE-2023-52425) by adding
+five new methods:
+
+* :meth:`xml.etree.ElementTree.XMLParser.flush`
+* :meth:`xml.etree.ElementTree.XMLPullParser.flush`
+* :meth:`xml.parsers.expat.xmlparser.GetReparseDeferralEnabled`
+* :meth:`xml.parsers.expat.xmlparser.SetReparseDeferralEnabled`
+* :meth:`xml.sax.expatreader.ExpatParser.flush`
index 8d1b56b35d204f3db71b06f5824bbf39183f959c..63b483188a5c3d386df3b38d0b0f56e8256f01da 100644 (file)
@@ -3883,6 +3883,38 @@ _elementtree_XMLParser_close_impl(XMLParserObject *self)
     }
 }
 
+/*[clinic input]
+_elementtree.XMLParser.flush
+
+[clinic start generated code]*/
+
+static PyObject *
+_elementtree_XMLParser_flush_impl(XMLParserObject *self)
+/*[clinic end generated code: output=42fdb8795ca24509 input=effbecdb28715949]*/
+{
+    if (!_check_xmlparser(self)) {
+        return NULL;
+    }
+
+    if (EXPAT(SetReparseDeferralEnabled) == NULL) {
+        Py_RETURN_NONE;
+    }
+
+    // NOTE: The Expat parser in the C implementation of ElementTree is not
+    //       exposed to the outside; as a result we known that reparse deferral
+    //       is currently enabled, or we would not even have access to function
+    //       XML_SetReparseDeferralEnabled in the first place (which we checked
+    //       for, a few lines up).
+
+    EXPAT(SetReparseDeferralEnabled)(self->parser, XML_FALSE);
+
+    PyObject *res = expat_parse(self, "", 0, XML_FALSE);
+
+    EXPAT(SetReparseDeferralEnabled)(self->parser, XML_TRUE);
+
+    return res;
+}
+
 /*[clinic input]
 _elementtree.XMLParser.feed
 
@@ -4310,6 +4342,7 @@ static PyTypeObject TreeBuilder_Type = {
 static PyMethodDef xmlparser_methods[] = {
     _ELEMENTTREE_XMLPARSER_FEED_METHODDEF
     _ELEMENTTREE_XMLPARSER_CLOSE_METHODDEF
+    _ELEMENTTREE_XMLPARSER_FLUSH_METHODDEF
     _ELEMENTTREE_XMLPARSER__PARSE_WHOLE_METHODDEF
     _ELEMENTTREE_XMLPARSER__SETEVENTS_METHODDEF
     {NULL, NULL}
index 034c1081dccdf15e89a11f8f10892e03d8834959..f9210beffa8970a0e1060641f40b8ff7bbaf9c59 100644 (file)
@@ -870,6 +870,23 @@ _elementtree_XMLParser_close(XMLParserObject *self, PyObject *Py_UNUSED(ignored)
     return _elementtree_XMLParser_close_impl(self);
 }
 
+PyDoc_STRVAR(_elementtree_XMLParser_flush__doc__,
+"flush($self, /)\n"
+"--\n"
+"\n");
+
+#define _ELEMENTTREE_XMLPARSER_FLUSH_METHODDEF    \
+    {"flush", (PyCFunction)_elementtree_XMLParser_flush, METH_NOARGS, _elementtree_XMLParser_flush__doc__},
+
+static PyObject *
+_elementtree_XMLParser_flush_impl(XMLParserObject *self);
+
+static PyObject *
+_elementtree_XMLParser_flush(XMLParserObject *self, PyObject *Py_UNUSED(ignored))
+{
+    return _elementtree_XMLParser_flush_impl(self);
+}
+
 PyDoc_STRVAR(_elementtree_XMLParser_feed__doc__,
 "feed($self, data, /)\n"
 "--\n"
@@ -920,4 +937,4 @@ skip_optional:
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=18ddc3c84139443d input=a9049054013a1b77]*/
+/*[clinic end generated code: output=f3bf5972287d37e4 input=a9049054013a1b77]*/
index ee5907ca7e447bcd7261ef5d84175765059b87f8..f3bc36cf08bc4e5f71a3e73eaec810148834c277 100644 (file)
@@ -2,6 +2,53 @@
 preserve
 [clinic start generated code]*/
 
+PyDoc_STRVAR(pyexpat_xmlparser_SetReparseDeferralEnabled__doc__,
+"SetReparseDeferralEnabled($self, enabled, /)\n"
+"--\n"
+"\n"
+"Enable/Disable reparse deferral; enabled by default with Expat >=2.6.0.");
+
+#define PYEXPAT_XMLPARSER_SETREPARSEDEFERRALENABLED_METHODDEF    \
+    {"SetReparseDeferralEnabled", (PyCFunction)pyexpat_xmlparser_SetReparseDeferralEnabled, METH_O, pyexpat_xmlparser_SetReparseDeferralEnabled__doc__},
+
+static PyObject *
+pyexpat_xmlparser_SetReparseDeferralEnabled_impl(xmlparseobject *self,
+                                                 int enabled);
+
+static PyObject *
+pyexpat_xmlparser_SetReparseDeferralEnabled(xmlparseobject *self, PyObject *arg)
+{
+    PyObject *return_value = NULL;
+    int enabled;
+
+    enabled = PyObject_IsTrue(arg);
+    if (enabled < 0) {
+        goto exit;
+    }
+    return_value = pyexpat_xmlparser_SetReparseDeferralEnabled_impl(self, enabled);
+
+exit:
+    return return_value;
+}
+
+PyDoc_STRVAR(pyexpat_xmlparser_GetReparseDeferralEnabled__doc__,
+"GetReparseDeferralEnabled($self, /)\n"
+"--\n"
+"\n"
+"Retrieve reparse deferral enabled status; always returns false with Expat <2.6.0.");
+
+#define PYEXPAT_XMLPARSER_GETREPARSEDEFERRALENABLED_METHODDEF    \
+    {"GetReparseDeferralEnabled", (PyCFunction)pyexpat_xmlparser_GetReparseDeferralEnabled, METH_NOARGS, pyexpat_xmlparser_GetReparseDeferralEnabled__doc__},
+
+static PyObject *
+pyexpat_xmlparser_GetReparseDeferralEnabled_impl(xmlparseobject *self);
+
+static PyObject *
+pyexpat_xmlparser_GetReparseDeferralEnabled(xmlparseobject *self, PyObject *Py_UNUSED(ignored))
+{
+    return pyexpat_xmlparser_GetReparseDeferralEnabled_impl(self);
+}
+
 PyDoc_STRVAR(pyexpat_xmlparser_Parse__doc__,
 "Parse($self, data, isfinal=False, /)\n"
 "--\n"
@@ -402,4 +449,4 @@ exit:
 #ifndef PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF
     #define PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF
 #endif /* !defined(PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF) */
-/*[clinic end generated code: output=68ce25024280af41 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=f28e6c19662ac7be input=a9049054013a1b77]*/
index d45d9b6c4571595f4e9d676833fa24724647d6b7..8ee03ef079281576658ef714533dcdc400d41e79 100644 (file)
 #define XML_SetNotStandaloneHandler     PyExpat_XML_SetNotStandaloneHandler
 #define XML_SetParamEntityParsing       PyExpat_XML_SetParamEntityParsing
 #define XML_SetProcessingInstructionHandler PyExpat_XML_SetProcessingInstructionHandler
+#define XML_SetReparseDeferralEnabled   PyExpat_XML_SetReparseDeferralEnabled
 #define XML_SetReturnNSTriplet          PyExpat_XML_SetReturnNSTriplet
 #define XML_SetSkippedEntityHandler     PyExpat_XML_SetSkippedEntityHandler
 #define XML_SetStartCdataSectionHandler PyExpat_XML_SetStartCdataSectionHandler
index 12ae66d945bda8cb6b85944fd3de9e02f35245ac..e0c3e100d51d5702f2912cc01ad09467da124682 100644 (file)
@@ -1,6 +1,7 @@
 #include "Python.h"
 #include <ctype.h>
 
+#include <stdbool.h>
 #include "structmember.h"         // PyMemberDef
 #include "frameobject.h"
 #include "expat.h"
@@ -65,6 +66,12 @@ typedef struct {
                                 /* NULL if not enabled */
     int buffer_size;            /* Size of buffer, in XML_Char units */
     int buffer_used;            /* Buffer units in use */
+    bool reparse_deferral_enabled; /* Whether to defer reparsing of
+                                   unfinished XML tokens; a de-facto cache of
+                                   what Expat has the authority on, for lack
+                                   of a getter API function
+                                   "XML_GetReparseDeferralEnabled" in Expat
+                                   2.6.0 */
     PyObject *intern;           /* Dictionary to intern strings */
     PyObject **handlers;
 } xmlparseobject;
@@ -696,6 +703,40 @@ get_parse_result(xmlparseobject *self, int rv)
 
 #define MAX_CHUNK_SIZE (1 << 20)
 
+/*[clinic input]
+pyexpat.xmlparser.SetReparseDeferralEnabled
+
+    enabled: bool
+    /
+
+Enable/Disable reparse deferral; enabled by default with Expat >=2.6.0.
+[clinic start generated code]*/
+
+static PyObject *
+pyexpat_xmlparser_SetReparseDeferralEnabled_impl(xmlparseobject *self,
+                                                 int enabled)
+/*[clinic end generated code: output=5ec539e3b63c8c49 input=021eb9e0bafc32c5]*/
+{
+#if XML_COMBINED_VERSION >= 20600
+    XML_SetReparseDeferralEnabled(self->itself, enabled ? XML_TRUE : XML_FALSE);
+    self->reparse_deferral_enabled = (bool)enabled;
+#endif
+    Py_RETURN_NONE;
+}
+
+/*[clinic input]
+pyexpat.xmlparser.GetReparseDeferralEnabled
+
+Retrieve reparse deferral enabled status; always returns false with Expat <2.6.0.
+[clinic start generated code]*/
+
+static PyObject *
+pyexpat_xmlparser_GetReparseDeferralEnabled_impl(xmlparseobject *self)
+/*[clinic end generated code: output=4e91312e88a595a8 input=54b5f11d32b20f3e]*/
+{
+    return PyBool_FromLong(self->reparse_deferral_enabled);
+}
+
 /*[clinic input]
 pyexpat.xmlparser.Parse
 
@@ -1041,6 +1082,8 @@ static struct PyMethodDef xmlparse_methods[] = {
 #if XML_COMBINED_VERSION >= 19505
     PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF
 #endif
+    PYEXPAT_XMLPARSER_SETREPARSEDEFERRALENABLED_METHODDEF
+    PYEXPAT_XMLPARSER_GETREPARSEDEFERRALENABLED_METHODDEF
     {NULL, NULL}  /* sentinel */
 };
 
@@ -1123,6 +1166,11 @@ newxmlparseobject(const char *encoding, const char *namespace_separator, PyObjec
     self->handlers = NULL;
     self->intern = intern;
     Py_XINCREF(self->intern);
+#if XML_COMBINED_VERSION >= 20600
+    self->reparse_deferral_enabled = true;
+#else
+    self->reparse_deferral_enabled = false;
+#endif
 
     /* namespace_separator is either NULL or contains one char + \0 */
     self->itself = XML_ParserCreate_MM(encoding, &ExpatMemoryHandler,
@@ -1878,6 +1926,11 @@ MODULE_INITFUNC(void)
 #else
     capi.SetHashSalt = NULL;
 #endif
+#if XML_COMBINED_VERSION >= 20600
+    capi.SetReparseDeferralEnabled = XML_SetReparseDeferralEnabled;
+#else
+    capi.SetReparseDeferralEnabled = NULL;
+#endif
 
     /* export using capsule */
     capi_object = PyCapsule_New(&capi, PyExpat_CAPSULE_NAME, NULL);