gh-150821: Skip URL parsing in mimetypes.guess_type() for file paths

Lib/mimetypes.py

@@ -23,6 +23,10 @@
 read_mime_types(file) -- parse one file, return a dictionary or None
 """
 
+lazy import os
+lazy import posixpath
+lazy import urllib.parse
+
 try:
     from _winapi import _mimetypes_read_windows_registry
 except ImportError:
@@ -121,12 +125,13 @@ def guess_type(self, url, strict=True):
         Optional 'strict' argument when False adds a bunch of commonly found,
         but non-standard types.
         """
-        # Lazy import to improve module import time
-        import os
-        import urllib.parse
-
         # TODO: Deprecate accepting file paths (in particular path-like objects).
         url = os.fspath(url)
+        # Without a ':' the argument cannot carry a URL scheme, so it cannot
+        # be a URL; skip the relatively expensive urlparse() in that case.
+        if isinstance(url, str) and ':' not in url:
+            return self.guess_file_type(url, strict=strict)
+
         p = urllib.parse.urlparse(url)
         if p.scheme and len(p.scheme) > 1:
             scheme = p.scheme
@@ -153,19 +158,13 @@ def guess_type(self, url, strict=True):
                 type = 'text/plain'
             return type, None           # never compressed, so encoding is None
 
-        # Lazy import to improve module import time
-        import posixpath
-
         return self._guess_file_type(url, strict, posixpath.splitext)
 
     def guess_file_type(self, path, *, strict=True):
         """Guess the type of a file based on its path.
 
         Similar to guess_type(), but takes file path instead of URL.
         """
-        # Lazy import to improve module import time
-        import os
-
         path = os.fsdecode(path)
         path = os.path.splitdrive(path)[1]
         return self._guess_file_type(path, strict, os.path.splitext)
@@ -412,9 +411,6 @@ def init(files=None):
     else:
         db = _db
 
-    # Lazy import to improve module import time
-    import os
-
     for file in files:
         if os.path.isfile(file):
             db.read(file)

Lib/test/test_mimetypes.py

@@ -375,6 +375,15 @@ def test_url(self):
         result = self.db.guess_type('http://example.com/host.html?q=x.tar')
         self.assertSequenceEqual(result, ('text/html', None))
 
+    def test_path_with_colon_but_no_url_scheme(self):
+        # A ':' that does not introduce a real URL scheme -- a single-letter
+        # Windows drive, or a colon elsewhere in the name -- is treated as a
+        # file path rather than a URL.
+        eq = self.assertSequenceEqual
+        eq(self.db.guess_type("c:fake.html"), ("text/html", None))
+        eq(self.db.guess_type(r"c:\dir\fake.html"), ("text/html", None))
+        eq(self.db.guess_type("note 12:30.txt"), ("text/plain", None))
+
     def test_guess_all_types(self):
         # First try strict.  Use a set here for testing the results because if
         # test_urllib2 is run before test_mimetypes, global state is modified

Misc/NEWS.d/next/Library/2026-06-02-15-45-02.gh-issue-150821.yKimWm.rst

@@ -0,0 +1,2 @@
+Speed up :func:`mimetypes.guess_type` for plain file paths by skipping URL
+parsing when the argument has no scheme. Patch by Bernát Gábor.