Optimize netloc extraction (#284)

Regex can be replaced with an if-else equivalent for tangible speed improvement.

---------

Co-authored-by: John Kurkowski <john.kurkowski@gmail.com>

tests/main_test.py

@@ -214,6 +214,13 @@ def test_empty():
 
 
 def test_scheme():
+ assert_extract("//", ("", "", "", ""))
+ assert_extract("://", ("", "", "", ""))
+ assert_extract("://example.com", ("", "", "", ""))
+ assert_extract("a+-.://example.com", ("example.com", "", "example", "com"))
+ assert_extract("a#//example.com", ("", "", "a", ""))
+ assert_extract("a@://example.com", ("", "", "", ""))
+ assert_extract("#//example.com", ("", "", "", ""))
  assert_extract(
  "https://mail.google.com/mail", ("mail.google.com", "mail", "google", "com")
  )

tldextract/remote.py

@@ -9,16 +9,15 @@
  r"{3}([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])$"
 )
 
-SCHEME_RE = re.compile(r"^([" + scheme_chars + "]+:)?//")
+scheme_chars_set = set(scheme_chars)
 
 
 def lenient_netloc(url: str) -> str:
  """Extract the netloc of a URL-like string, similar to the netloc attribute
  returned by urllib.parse.{urlparse,urlsplit}, but extract more leniently,
  without raising errors."""
-
  return (
- SCHEME_RE.sub("", url, 1)
+ _schemeless_url(url)
  .partition("/")[0]
  .partition("?")[0]
  .partition("#")[0]
@@ -29,6 +28,19 @@ def lenient_netloc(url: str) -> str:
  )
 
 
+def _schemeless_url(url: str) -> str:
+ double_slashes_start = url.find("//")
+ if double_slashes_start == 0:
+ return url[2:]
+ if (
+ double_slashes_start < 2
+ or not url[double_slashes_start - 1] == ":"
+ or set(url[: double_slashes_start - 1]) - scheme_chars_set
+ ):
+ return url
+ return url[double_slashes_start + 2 :]
+
+
 def looks_like_ip(maybe_ip: str) -> bool:
  """Does the given str look like an IP address?"""
  if not maybe_ip[0].isdigit():