GH-123599: url2pathname(): don't call gethostbyname() by default

Doc/library/urllib.request.rst

@@ -175,7 +175,7 @@ The :mod:`urllib.request` module defines the following functions:
       The *add_scheme* argument was added.
 
 
-.. function:: url2pathname(url, *, require_scheme=False, resolve_netloc=False)
+.. function:: url2pathname(url, *, require_scheme=False, resolve_host=False)
 
    Convert the given ``file:`` URL to a local path. This function uses
    :func:`~urllib.parse.unquote` to decode the URL.
@@ -185,12 +185,12 @@ The :mod:`urllib.request` module defines the following functions:
    value should include the prefix; a :exc:`~urllib.error.URLError` is raised
    if it doesn't.
 
-   The URL authority is discarded if it empty, ``localhost``, or the local
-   hostname. Otherwise, if *resolve_netloc* is set to true, the authority is
+   The URL authority is discarded if it is empty, ``localhost``, or the local
+   hostname. Otherwise, if *resolve_host* is set to true, the authority is
    resolved using :func:`socket.gethostbyname` and discarded if it matches a
-   local IP address. If the authority is still unhandled, then on Windows a
-   UNC path is returned, and on other platforms a
-   :exc:`~urllib.error.URLError` is raised.
+   local IP address (as per :rfc:`RFC 8089 §3 <8089#section-3>`). If the
+   authority is still unhandled, then on Windows a UNC path is returned, and
+   on other platforms a :exc:`~urllib.error.URLError` is raised.
 
    This example shows the function being used on Windows::
 
@@ -211,10 +211,7 @@ The :mod:`urllib.request` module defines the following functions:
       :exc:`~urllib.error.URLError` is raised.
 
    .. versionchanged:: next
-      The *require_scheme* argument was added.
-
-   .. versionchanged:: next
-      The *resolve_netloc* argument was added.
+      The *require_scheme* and *resolve_host* arguments were added.
 
 
 .. function:: getproxies()

Doc/whatsnew/3.14.rst

@@ -1232,7 +1232,7 @@ urllib
     true.
   - Discard URL authority if it matches the local hostname.
   - Discard URL authority if it resolves to a local IP address when the new
-    *resolve_netloc* argument is set to true.
+    *resolve_host* argument is set to true.
   - Raise :exc:`~urllib.error.URLError` if a URL authority isn't local,
     except on Windows where we return a UNC path as before.
 

Lib/test/test_urllib.py

@@ -1569,12 +1569,12 @@ def test_url2pathname_require_scheme_errors(self):
                     urllib.request.url2pathname,
                     url, require_scheme=True)
 
-    def test_url2pathname_resolve_netloc(self):
+    def test_url2pathname_resolve_host(self):
         fn = urllib.request.url2pathname
         sep = os.path.sep
-        self.assertEqual(fn('//127.0.0.1/foo/bar', resolve_netloc=True), f'{sep}foo{sep}bar')
+        self.assertEqual(fn('//127.0.0.1/foo/bar', resolve_host=True), f'{sep}foo{sep}bar')
         self.assertEqual(fn(f'//{socket.gethostname()}/foo/bar'), f'{sep}foo{sep}bar')
-        self.assertEqual(fn(f'//{socket.gethostname()}/foo/bar', resolve_netloc=True), f'{sep}foo{sep}bar')
+        self.assertEqual(fn(f'//{socket.gethostname()}/foo/bar', resolve_host=True), f'{sep}foo{sep}bar')
 
     @unittest.skipUnless(sys.platform == 'win32',
                          'test specific to Windows pathnames.')

Lib/urllib/request.py

@@ -1466,7 +1466,7 @@ def get_names(self):
     def open_local_file(self, req):
         import email.utils
         import mimetypes
-        localfile = url2pathname(req.full_url, require_scheme=True, resolve_netloc=True)
+        localfile = url2pathname(req.full_url, require_scheme=True, resolve_host=True)
         try:
             stats = os.stat(localfile)
             size = stats.st_size
@@ -1643,24 +1643,24 @@ def data_open(self, req):
         return addinfourl(io.BytesIO(data), headers, url)
 
 
-# Code move from the old urllib module
+# Code moved from the old urllib module
 
-def url2pathname(url, *, require_scheme=False, resolve_netloc=False):
+def url2pathname(url, *, require_scheme=False, resolve_host=False):
     """Convert the given file URL to a local file system path.
 
     The 'file:' scheme prefix must be omitted unless *require_scheme*
     is set to true.
 
     The URL authority may be resolved with gethostbyname() if
-    *resolve_netloc* is set to true.
+    *resolve_host* is set to true.
     """
     if require_scheme:
         scheme, url = _splittype(url)
         if scheme != 'file':
             raise URLError("URL is missing a 'file:' scheme")
     authority, url = _splithost(url)
     if os.name == 'nt':
-        if not _is_local_authority(authority, resolve_netloc):
+        if not _is_local_authority(authority, resolve_host):
             # e.g. file://server/share/file.txt
             url = '//' + authority + url
         elif url[:3] == '///':
@@ -1674,7 +1674,7 @@ def url2pathname(url, *, require_scheme=False, resolve_netloc=False):
                 # Older URLs use a pipe after a drive letter
                 url = url[:1] + ':' + url[2:]
         url = url.replace('/', '\\')
-    elif not _is_local_authority(authority, resolve_netloc):
+    elif not _is_local_authority(authority, resolve_host):
         raise URLError("file:// scheme is supported only on localhost")
     encoding = sys.getfilesystemencoding()
     errors = sys.getfilesystemencodeerrors()

Misc/NEWS.d/next/Library/2024-11-14-21-17-48.gh-issue-126838.Yr5vKF.rst

@@ -1,9 +1,3 @@
-Fix issue where :func:`urllib.request.url2pathname` mishandled file URLs with
-authorities. The process now works as follows:
-
-1. Discard authority if it is empty or ``localhost``; otherwise
-2. (New) Discard authority if it matches the local hostname; otherwise
-3. (New) If the new *resolve_netloc* keyword-only argument is set to true,
-   discard authority if it resolves to a local IP address; otherwise
-4. On Windows, return a UNC path; otherwise
-5. (New) Raise :exc:`urllib.error.URLError`.
+Add *resolve_host* keyword-only argument to
+:func:`urllib.request.url2pathname`, and fix handling of file URLs with
+authorities.