Fix false repo relocation warning on macOS due to NFC/NFD path differences, fixes #2913 (1.4-maint)

src/borg/cache.py

@@ -18,6 +18,7 @@
 from .helpers import Error
 from .helpers import Manifest
 from .helpers import get_cache_dir, get_security_dir
+from .helpers import normalize_local_path
 from .helpers import int_to_bigint, bigint_to_int, bin_to_hex, hex_to_bin, parse_stringified_list
 from .helpers import format_file_size
 from .helpers import safe_ns
@@ -119,7 +120,12 @@ def assert_location_matches(self, cache_config=None):
             logger.debug('security: using previous_location of cache: %r', previous_location)
 
         repository_location = self.repository._location.canonical_path()
-        if (previous_location and previous_location != repository_location and
+        # Normalize the unicode form of both paths before comparing them: on macOS the same path can be
+        # produced in different normalization forms (NFC when typed, NFD when derived from the filesystem),
+        # which would otherwise be mistaken for a repository relocation (see issue #2913). Normalizing is
+        # idempotent and applied symmetrically, so it never creates a spurious mismatch for remote repos.
+        if (previous_location and
+                normalize_local_path(previous_location) != normalize_local_path(repository_location) and
                 not self._is_same_local_directory(previous_location, repository_location)):
             msg = ("Warning: The repository at location {} was previously located at {}\n".format(
                 repository_location, previous_location) +

src/borg/helpers/parseformat.py

@@ -8,6 +8,7 @@
 import shlex
 import socket
 import stat
+import unicodedata
 import uuid
 from collections import Counter, OrderedDict
 from datetime import datetime, timezone
@@ -23,7 +24,7 @@
 from .. import __version__ as borg_version
 from .. import __version_tuple__ as borg_version_tuple
 from ..constants import *  # NOQA
-from ..platformflags import is_win32
+from ..platformflags import is_win32, is_darwin
 
 
 def bin_to_hex(binary):
@@ -322,6 +323,17 @@ def parse_stringified_list(s):
     return [item for item in l if item != '']
 
 
+def normalize_local_path(path):
+    """Normalize a local filesystem path's unicode form so paths compare equal regardless of how they were entered.
+
+    On macOS the filesystem returns paths in NFD (decomposed) form, while a path typed as a command line
+    argument is usually in NFC (composed) form. These look identical but are byte-different, which otherwise
+    makes borg believe a repository was relocated (see issue #2913). We normalize to NFD to match what the
+    macOS filesystem uses. On other platforms the path is returned unchanged.
+    """
+    return unicodedata.normalize('NFD', path) if is_darwin else path
+
+
 class Location:
     """Object representing a repository / archive location
     """
@@ -510,7 +522,7 @@ def host(self):
 
     def canonical_path(self):
         if self.proto == 'file':
-            return self.path
+            return normalize_local_path(self.path)
         else:
             if self.path and self.path.startswith('~'):
                 path = '/' + self.path  # /~/x = path x relative to home dir

src/borg/testsuite/helpers.py

@@ -3,6 +3,7 @@
 import shutil
 import signal
 import sys
+import unicodedata
 from argparse import ArgumentTypeError
 from datetime import datetime, timezone, timedelta
 from io import StringIO, BytesIO
@@ -12,7 +13,7 @@
 
 from .. import platform, __version__
 from ..constants import *  # NOQA
-from ..helpers import Location
+from ..helpers import Location, normalize_local_path
 from ..helpers import Buffer
 from ..helpers import partial_format, format_file_size, parse_file_size, format_timedelta, format_line, PlaceholderError, replace_placeholders
 from ..helpers import make_path_safe, clean_lines
@@ -34,7 +35,7 @@
 from ..helpers import iter_separated
 from ..helpers import eval_escapes
 from ..helpers import classify_ec, max_ec, add_warning, get_ec, get_reset_ec, init_ec_warnings
-from ..platform import is_win32, swidth, is_cygwin
+from ..platform import is_win32, swidth, is_cygwin, is_darwin
 
 from . import BaseTestCase, FakeInputs
 
@@ -255,6 +256,33 @@ def test_omit_archive(self):
         assert loc_without_archive.processed == "ssh://user@host:1234/repos/%s" % hostname
 
 
+def test_normalize_local_path():
+    nfc = unicodedata.normalize('NFC', 'tëst')  # composed: ë is a single codepoint
+    nfd = unicodedata.normalize('NFD', 'tëst')  # decomposed: e + combining diaeresis
+    assert nfc != nfd  # sanity: the two unicode forms really differ byte-wise
+    if is_darwin:
+        # on macOS both forms normalize to the same (NFD) form the filesystem uses
+        assert normalize_local_path(nfc) == normalize_local_path(nfd) == nfd
+    else:
+        # elsewhere it is a no-op
+        assert normalize_local_path(nfc) == nfc
+        assert normalize_local_path(nfd) == nfd
+
+
+def test_canonical_path_unicode_normalization(monkeypatch):
+    # regression test for #2913: a file repo path entered in NFC form (e.g. typed as an argument) and
+    # the same path in NFD form (e.g. derived from os.getcwd() on macOS) must yield the same canonical
+    # path, so borg does not mistake identical-looking paths for a repository relocation.
+    monkeypatch.delenv('BORG_REPO', raising=False)
+    nfc = unicodedata.normalize('NFC', '/absolute/tëst')
+    nfd = unicodedata.normalize('NFD', '/absolute/tëst')
+    if is_darwin:
+        assert Location(nfc).canonical_path() == Location(nfd).canonical_path()
+    else:
+        # on other platforms the filesystem does not normalize, so the forms stay distinct
+        assert Location(nfc).canonical_path() != Location(nfd).canonical_path()
+
+
 class TestLocationWithEnv:
     def test_ssh(self, monkeypatch):
         monkeypatch.setenv('BORG_REPO', 'ssh://user@host:1234/some/path')