x1.125 speed up

Author: Gallaecio

tests/test_url.py

@@ -639,7 +639,7 @@ def test_safe_url_performance(url):
     # presummably due to caching by urllib.
     number = 1  # TODO: Increase? How much?
     # Make sure the new implementation is at most this number of times as slow.
-    multiplier = 45  # TODO: Lower as close to 1 as possible.
+    multiplier = 40  # TODO: Lower as close to 1 as possible.
 
     time1 = timeit(
         f"safe_url({url!r})", "from w3lib.url import safe_url", number=number

w3lib/_rfc5892.py

@@ -10,6 +10,8 @@
 
 
 def _check_contextj_rules(label: str) -> None:
+    if label.isascii():
+        return
     for i, code_point in enumerate(label):
         value = ord(code_point)
         if not intranges_contain(value, codepoint_classes["CONTEXTJ"]):

w3lib/_url.py

@@ -432,25 +432,6 @@ def _starts_with_windows_drive_letter(input: str) -> bool:
     )
 
 
-_ASCII_URL_CODE_POINTS = _ASCII_ALPHANUMERIC + "!$&'()*+,-./:;=?@_~"
-
-
-# https://url.spec.whatwg.org/commit-snapshots/a46cb9188a48c2c9d80ba32a9b1891652d6b4900/#url-code-points
-def _is_url_code_point(code_point: str) -> bool:
-    if code_point in _ASCII_URL_CODE_POINTS:
-        return True
-    code_point_id = ord(code_point)
-    if code_point_id < 0xA0:
-        return False
-    if code_point_id > 0x10FFFD:
-        return False
-    if _is_surrogate_code_point_id(code_point_id):
-        return False
-    if _is_noncharacter_code_point_id(code_point_id):
-        return False
-    return True
-
-
 # https://url.spec.whatwg.org/commit-snapshots/a46cb9188a48c2c9d80ba32a9b1891652d6b4900/#double-dot-path-segment
 def _is_double_dot_path_segment(input: str) -> bool:
     return input in (

w3lib/_utr46.py

@@ -256,7 +256,7 @@ def _process(
 
 
 def _convert_label(label: str) -> str:
-    if not any(ord(code_point) >= 0x80 for code_point in label):
+    if label.isascii():
         return label
     return f"xn--{label.encode('punycode').decode()}"