CVE-2025-68463

@@ -354,7 +354,7 @@ def __init__(self, validate, escape, ignore_errors):
         self.dtd_urls = []
         self.element = None
         self.level = 0
-        self.secure = False
+        self.bypass_url_security = False
         self.data = []
         self.attributes = None
         self.allowed_tags = None
@@ -1081,21 +1081,20 @@ def save_xsd_file(self, filename, text):
             handle.write(text)
             handle.close()
 
-    def verify_security(self, url):
-        """Check if the url is from a trustable sournce."""
-        if not self.secure:
+    def verify_security(self, url, verify_hostname=True):
+        """Check if the given URL is from a trustable source.
+
+        When ``self.bypass_url_security`` evaluates to ``True``,
+        all URL security checks will be skipped.
+        """
+        if not self.bypass_url_security:
             parts = urlparse(url)
             scheme = parts.scheme
             hostname = parts.hostname
-            hostnames = (
-                "www.ncbi.nlm.nih.gov",
-                "dtd.nlm.nih.gov",
-                "eutils.ncbi.nlm.nih.gov",
-            )
-            if scheme != "https" or hostname not in hostnames:
-                raise ValueError(f"expected secure URL to NCBI, found {url}")
-            # Trust URLs linked from NCBI
-            self.secure = True
+            if scheme != "https" or (
+                verify_hostname and not hostname.endswith(".nlm.nih.gov")
+            ):
+                raise ValueError(f"Expected secure URL to NCBI, found {url!r}")
 
     def externalEntityRefHandler(self, context, base, systemId, publicId):
         """Handle external entity reference in order to cache DTD locally.
@@ -1125,27 +1124,42 @@ def externalEntityRefHandler(self, context, base, systemId, publicId):
             url = source.rstrip("/") + "/" + systemId
         else:
             raise ValueError("Unexpected URL scheme %r" % urlinfo.scheme)
+
+        # NOTE: This trusts any external references from a trusted parent,
+        #       even if these external references go to unknown hosts,
+        #       e.g. when NCBI starts referencing things on a new host
+        #       from existing DTD files.
+        #       Needs to be checked *prior* to appending to ``self.dtd_urls``.
+        self.verify_security(url, verify_hostname=not self.dtd_urls)
+
+        # NOTE: Since ``self.dtd_urls`` being non-empty has security
+        #       consequences with the check above, we use a ``finally`` wrap
+        #       here, in order to guarantee that push and pop are matched.
         self.dtd_urls.append(url)
-        self.verify_security(url)
-        # First, try to load the local version of the DTD file
-        location, filename = os.path.split(systemId)
-        handle = self.open_dtd_file(filename)
-        if not handle:
-            # DTD is not available as a local file. Try accessing it through
-            # the internet instead.
-            try:
-                handle = urlopen(url)
-            except OSError:
-                raise RuntimeError(f"Failed to access {filename} at {url}") from None
-            text = handle.read()
+        try:
+            # First, try to load the local version of the DTD file
+            location, filename = os.path.split(systemId)
+            handle = self.open_dtd_file(filename)
+            if not handle:
+                # DTD is not available as a local file. Try accessing it through
+                # the internet instead.
+                try:
+                    handle = urlopen(url)
+                except OSError:
+                    raise RuntimeError(
+                        f"Failed to access {filename} at {url}"
+                    ) from None
+                text = handle.read()
+                handle.close()
+                self.save_dtd_file(filename, text)
+                handle = BytesIO(text)
+
+            parser = self.parser.ExternalEntityParserCreate(context)
+            parser.ElementDeclHandler = self.elementDecl
+            parser.ParseFile(handle)
             handle.close()
-            self.save_dtd_file(filename, text)
-            handle = BytesIO(text)
-
-        parser = self.parser.ExternalEntityParserCreate(context)
-        parser.ElementDeclHandler = self.elementDecl
-        parser.ParseFile(handle)
-        handle.close()
-        self.dtd_urls.pop()
+        finally:
+            self.dtd_urls.pop()
+
         self.parser.StartElementHandler = self.startElementHandler
         return 1

@@ -314,6 +314,7 @@ please open an issue on GitHub or mention it on the mailing list.
 - Sean Johnson <https://github.com/seanrjohnson>
 - Sean Workman <https://github.com/sean-workman>
 - Sebastian Bassi <https://about.me/bassi>
+- Sebastian Pipping <https://blog.hartwork.org/>
 - Sergei Lebedev <https://github.com/superbobry>
 - Sergio Valqui <https://github.com/svalqui>
 - Seth Sims <https://github.com/xzy3>

@@ -23,6 +23,7 @@ possible, especially the following contributors:
 - Timothy Dennis (first contribution)
 - Ziyan Rao (first contribution)
 - Manuel Lera-Ramirez
+- Sebastian Pipping
 
 28 October 2025: Biopython 1.86
 ===============================

@@ -9,6 +9,8 @@
 import pickle
 import unittest
 from io import BytesIO
+from textwrap import dedent
+from unittest.mock import call, Mock
 
 from Bio import Entrez
 from Bio import StreamModeError
@@ -8867,6 +8869,48 @@ def test_truncated_xml(self):
         self.assertRaises(CorruptedXMLError, next, records)
 
 
+class UrlSecurityCheckTest(unittest.TestCase):
+    """Test for DTD and XSL URL validation."""
+
+    def test_continued_url_security_checking(self):
+        content = dedent(
+            """\
+                <?xml version="1.0" encoding="UTF-8"?>
+                <!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "https://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">
+                <IPGReportSet xmlns:xsi="https://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="https://host.invalid/404.dtd" />
+                <!--                                                                                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ -->
+            """
+        ).encode("utf-8")
+
+        handler = Entrez.Parser.DataHandler(
+            validate=True, escape=False, ignore_errors=False
+        )
+        handler.verify_security = Mock(side_effect=handler.verify_security)
+
+        with self.assertRaises(ValueError) as caught:
+            handler.read(BytesIO(content))
+
+        self.assertIn("Expected secure URL to NCBI", caught.exception.args[0])
+        self.assertEqual(
+            handler.verify_security.call_args_list,
+            [
+                call(
+                    "https://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd",
+                    verify_hostname=True,
+                ),
+                call(
+                    "https://www.ncbi.nlm.nih.gov/dtd/NCBI_Entity.mod.dtd",
+                    verify_hostname=False,
+                ),
+                call(
+                    "https://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.mod.dtd",
+                    verify_hostname=False,
+                ),
+                call("https://host.invalid/404.dtd"),
+            ],
+        )
+
+
 if __name__ == "__main__":
     runner = unittest.TextTestRunner(verbosity=2)
     unittest.main(testRunner=runner)