Merge pull request #230 from Gallaecio/credential-case

canonicalize_url: do not apply lowercase to userinfo

Author: wRAR

tests/test_url.py

@@ -1384,6 +1384,12 @@ def test_domains_are_case_insensitive(self):
             canonicalize_url("http://www.EXAMPLE.com/"), "http://www.example.com/"
         )
 
+    def test_userinfo_is_case_sensitive(self):
+        self.assertEqual(
+            canonicalize_url("sftp://UsEr:PaSsWoRd@www.EXAMPLE.com/"),
+            "sftp://UsEr:PaSsWoRd@www.example.com/",
+        )
+
     def test_canonicalize_idns(self):
         self.assertEqual(
             canonicalize_url("http://www.bücher.de?q=bücher"),

w3lib/url.py

@@ -654,10 +654,13 @@ def canonicalize_url(
 
     fragment = "" if not keep_fragments else fragment
 
+    # Apply lowercase to the domain, but not to the userinfo.
+    netloc_parts = netloc.split("@")
+    netloc_parts[-1] = netloc_parts[-1].lower().rstrip(":")
+    netloc = "@".join(netloc_parts)
+
     # every part should be safe already
-    return urlunparse(
-        (scheme, netloc.lower().rstrip(":"), path, params, query, fragment)
-    )
+    return urlunparse((scheme, netloc, path, params, query, fragment))
 
 
 def _unquotepath(path: str) -> bytes: