Merge branch 'main' into pythongh-126211-fix-cases-gen

Lib/nturl2path.py

@@ -55,16 +55,11 @@ def pathname2url(p):
     if p[:4] == '\\\\?\\':
         p = p[4:]
         if p[:4].upper() == 'UNC\\':
-            p = '\\' + p[4:]
+            p = '\\\\' + p[4:]
         elif p[1:2] != ':':
             raise OSError('Bad path: ' + p)
     if not ':' in p:
         # No drive specifier, just convert slashes and quote the name
-        if p[:2] == '\\\\':
-        # path is something like \\host\path\on\remote\host
-        # convert this to ////host/path/on/remote/host
-        # (notice doubling of slashes at the start of the path)
-            p = '\\\\' + p
         components = p.split('\\')
         return urllib.parse.quote('/'.join(components))
     comp = p.split(':', maxsplit=2)

Lib/tarfile.py

@@ -1760,6 +1760,8 @@ def __init__(self, name=None, mode="r", fileobj=None, format=None,
                                 # current position in the archive file
         self.inodes = {}        # dictionary caching the inodes of
                                 # archive members already added
+        self._unames = {}       # Cached mappings of uid -> uname
+        self._gnames = {}       # Cached mappings of gid -> gname
 
         try:
             if self.mode == "r":
@@ -2138,16 +2140,23 @@ def gettarinfo(self, name=None, arcname=None, fileobj=None):
         tarinfo.mtime = statres.st_mtime
         tarinfo.type = type
         tarinfo.linkname = linkname
+
+        # Calls to pwd.getpwuid() and grp.getgrgid() tend to be expensive. To
+        # speed things up, cache the resolved usernames and group names.
         if pwd:
-            try:
-                tarinfo.uname = pwd.getpwuid(tarinfo.uid)[0]
-            except KeyError:
-                pass
+            if tarinfo.uid not in self._unames:
+                try:
+                    self._unames[tarinfo.uid] = pwd.getpwuid(tarinfo.uid)[0]
+                except KeyError:
+                    self._unames[tarinfo.uid] = ''
+            tarinfo.uname = self._unames[tarinfo.uid]
         if grp:
-            try:
-                tarinfo.gname = grp.getgrgid(tarinfo.gid)[0]
-            except KeyError:
-                pass
+            if tarinfo.gid not in self._gnames:
+                try:
+                    self._gnames[tarinfo.gid] = grp.getgrgid(tarinfo.gid)[0]
+                except KeyError:
+                    self._gnames[tarinfo.gid] = ''
+            tarinfo.gname = self._gnames[tarinfo.gid]
 
         if type in (CHRTYPE, BLKTYPE):
             if hasattr(os, "major") and hasattr(os, "minor"):

Lib/test/test_urllib.py

@@ -1524,7 +1524,7 @@ def test_pathname2url_win(self):
         # Test special prefixes are correctly handled in pathname2url()
         fn = urllib.request.pathname2url
         self.assertEqual(fn('\\\\?\\C:\\dir'), '///C:/dir')
-        self.assertEqual(fn('\\\\?\\unc\\server\\share\\dir'), '/server/share/dir')
+        self.assertEqual(fn('\\\\?\\unc\\server\\share\\dir'), '//server/share/dir')
         self.assertEqual(fn("C:"), '///C:')
         self.assertEqual(fn("C:\\"), '///C:')
         self.assertEqual(fn('C:\\a\\b.c'), '///C:/a/b.c')
@@ -1535,14 +1535,14 @@ def test_pathname2url_win(self):
         self.assertRaises(IOError, fn, "XX:\\")
         # No drive letter
         self.assertEqual(fn("\\folder\\test\\"), '/folder/test/')
-        self.assertEqual(fn("\\\\folder\\test\\"), '////folder/test/')
-        self.assertEqual(fn("\\\\\\folder\\test\\"), '/////folder/test/')
-        self.assertEqual(fn('\\\\some\\share\\'), '////some/share/')
-        self.assertEqual(fn('\\\\some\\share\\a\\b.c'), '////some/share/a/b.c')
-        self.assertEqual(fn('\\\\some\\share\\a\\b%#c\xe9'), '////some/share/a/b%25%23c%C3%A9')
+        self.assertEqual(fn("\\\\folder\\test\\"), '//folder/test/')
+        self.assertEqual(fn("\\\\\\folder\\test\\"), '///folder/test/')
+        self.assertEqual(fn('\\\\some\\share\\'), '//some/share/')
+        self.assertEqual(fn('\\\\some\\share\\a\\b.c'), '//some/share/a/b.c')
+        self.assertEqual(fn('\\\\some\\share\\a\\b%#c\xe9'), '//some/share/a/b%25%23c%C3%A9')
         # Round-tripping
         urls = ['///C:',
-                '/////folder/test/',
+                '///folder/test/',
                 '///C:/foo/bar/spam.foo']
         for url in urls:
             self.assertEqual(fn(urllib.request.url2pathname(url)), url)

Misc/NEWS.d/next/Library/2024-07-02-15-56-42.gh-issue-121267.yFBWkh.rst

@@ -0,0 +1,2 @@
+Improve the performance of :mod:`tarfile` when writing files, by caching user names
+and group names.

Misc/NEWS.d/next/Library/2024-10-30-20-45-17.gh-issue-126205.CHEmtx.rst

@@ -0,0 +1,2 @@
+Fix issue where :func:`urllib.request.pathname2url` generated URLs beginning
+with four slashes (rather than two) when given a Windows UNC path.