bpo-36742: Fixes handling of pre-normalization characters in urlsplit… · python/cpython@d537ab0 (original) (raw)

File tree

3 files changed

lines changed

3 files changed

lines changed

Original file line number Diff line number Diff line change
@@ -1011,6 +1011,12 @@ def test_urlsplit_normalization(self):
1011 1011 self.assertIn('\u2100', denorm_chars)
1012 1012 self.assertIn('\uFF03', denorm_chars)
1013 1013
1014 +# bpo-36742: Verify port separators are ignored when they
1015 +# existed prior to decomposition
1016 +urllib.parse.urlsplit('http://\u30d5\u309a:80')
1017 +with self.assertRaises(ValueError):
1018 +urllib.parse.urlsplit('http://\u30d5\u309a\ufe1380')
1019 +
1014 1020 for scheme in ["http", "https", "ftp"]:
1015 1021 for c in denorm_chars:
1016 1022 url = "{}://netloc{}false.netloc/path".format(scheme, c)
Original file line number Diff line number Diff line change
@@ -402,13 +402,16 @@ def _checknetloc(netloc):
402 402 # looking for characters like \u2100 that expand to 'a/c'
403 403 # IDNA uses NFKC equivalence, so normalize for this check
404 404 import unicodedata
405 -netloc2 = unicodedata.normalize('NFKC', netloc)
406 -if netloc == netloc2:
405 +n = netloc.rpartition('@')[2] # ignore anything to the left of '@'
406 +n = n.replace(':', '') # ignore characters already included
407 +n = n.replace('#', '') # but not the surrounding text
408 +n = n.replace('?', '')
409 +netloc2 = unicodedata.normalize('NFKC', n)
410 +if n == netloc2:
407 411 return
408 -_, _, netloc = netloc.rpartition('@') # anything to the left of '@' is okay
409 412 for c in '/?#@:':
410 413 if c in netloc2:
411 -raise ValueError("netloc '" + netloc2 + "' contains invalid " +
414 +raise ValueError("netloc '" + netloc + "' contains invalid " +
412 415 "characters under NFKC normalization")
413 416
414 417 def urlsplit(url, scheme='', allow_fragments=True):
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
1 +Fixes mishandling of pre-normalization characters in urlsplit().