---

 Lib/test/test_urlparse.py                                             |   21 ++++++++++

 Lib/urlparse.py                                                       |    9 +++-

 Misc/NEWS.d/next/Library/2022-11-12-15-45-51.gh-issue-99418.FxfAXS.rs |    2 

 3 files changed, 30 insertions(+), 2 deletions(-)

​

Index: Python-2.7.18/Lib/test/test_urlparse.py

===================================================================

--- Python-2.7.18.orig/Lib/test/test_urlparse.py

+++ Python-2.7.18/Lib/test/test_urlparse.py

@@ -1,4 +1,5 @@

 from test import test_support

+from urlparse import isascii

 import sys

 import unicodedata

 import unittest

@@ -592,6 +593,26 @@ class UrlParseTestCase(unittest.TestCase

         self.assertEqual(p.netloc, "www.example.net:foo")

         self.assertRaises(ValueError, lambda: p.port)

+    def do_attributes_bad_scheme(self, bytes, parse, scheme):

+        url = scheme + "://www.example.net"

+        if bytes:

+            if isascii(url):

+                url = url.encode("ascii")

+            else:

+                return

+        p = parse(url)

+        if bytes:

+            self.assertEqual(p.scheme, b"")

+        else:

+            self.assertEqual(p.scheme, "")

+

+    def test_attributes_bad_scheme(self):

+        """Check handling of invalid schemes."""

+        for bytes in (False, True):

+            for parse in (urlparse.urlsplit, urlparse.urlparse):

+                for scheme in (".", "+", "-", "0", "http&"):

+                    self.do_attributes_bad_scheme(bytes, parse, scheme)

+

     def test_attributes_without_netloc(self):

         # This example is straight from RFC 3261.  It looks like it

         # should allow the username, hostname, and port to be filled

Index: Python-2.7.18/Lib/urlparse.py

===================================================================

--- Python-2.7.18.orig/Lib/urlparse.py

+++ Python-2.7.18/Lib/urlparse.py

@@ -31,7 +31,8 @@ test_urlparse.py provides a good indicat

 import re

 __all__ = ["urlparse", "urlunparse", "urljoin", "urldefrag",

-           "urlsplit", "urlunsplit", "parse_qs", "parse_qsl"]

+           "urlsplit", "urlunsplit", "parse_qs", "parse_qsl",

+           "isascii"]

 # A classification of schemes ('' means apply by default)

 uses_relative = ['ftp', 'http', 'gopher', 'nntp', 'imap',

@@ -68,6 +69,10 @@ _UNSAFE_URL_BYTES_TO_REMOVE = ['\t', '\r

 MAX_CACHE_SIZE = 20

 _parse_cache = {}

+# Py3k shim

+def isascii(word):

+    return all([ord(c) < 128 for c in word])

+

 def clear_cache():

     """Clear the parse cache."""

     _parse_cache.clear()

@@ -211,7 +216,7 @@ def urlsplit(url, scheme='', allow_fragm

         clear_cache()

     netloc = query = fragment = ''

     i = url.find(':')

-    if i > 0:

+    if i > 0 and isascii(url[0]) and url[0].isalpha():

         if url[:i] == 'http': # optimize the common case

             scheme = url[:i].lower()

             url = url[i+1:]

Index: Python-2.7.18/Misc/NEWS.d/next/Library/2022-11-12-15-45-51.gh-issue-99418.FxfAXS.rs

===================================================================

--- /dev/null

+++ Python-2.7.18/Misc/NEWS.d/next/Library/2022-11-12-15-45-51.gh-issue-99418.FxfAXS.rs

@@ -0,0 +1,2 @@

+Fix bug in :func:`urllib.parse.urlparse` that causes URL schemes that begin

+with a digit, a plus sign, or a minus sign to be parsed incorrectly.

​