map_chars: deal invalid chars in paths on windows

Author: ThomasWaldmann

src/borg/helpers/fs.py

@@ -252,7 +252,7 @@ def make_path_safe(path):
     if "\\.." in path or "..\\" in path:
         raise ValueError(f"unexpected '..' element in path {path!r}")
 
-    path = percentify(path)
+    path = map_chars(path)
 
     path = path.lstrip("/")
     if path.startswith("../") or "/../" in path or path.endswith("/..") or path == "..":
@@ -270,15 +270,32 @@ def slashify(path):
     return path.replace("\\", "/") if is_win32 else path
 
 
-def percentify(path):
+# Bijective mapping to Unicode Private Use Area (like cifs mapchars)
+WINDOWS_MAP_CHARS = str.maketrans(
+    {
+        "<": "\uF03C",
+        ">": "\uF03E",
+        ":": "\uF03A",
+        '"': "\uF022",
+        "\\": "\uF05C",
+        "|": "\uF07C",
+        "?": "\uF03F",
+        "*": "\uF02A",
+    }
+)
+
+
+def map_chars(path):
     """
-    Replace backslashes with percent signs if running on Windows.
+    Map reserved characters if running on Windows.
 
-    Use case: if an archived path contains backslashes (which is not a path separator on POSIX
-    and could appear as a normal character in POSIX paths), we need to replace them with percent
-    signs to make the path usable on Windows.
+    Use case: if an archived path contains reserved characters (that are not reserved on POSIX)
+    we need to replace them with replacements to make the path usable on Windows.
     """
-    return path.replace("\\", "%") if is_win32 else path
+    if not is_win32:
+        return path
+
+    return path.translate(WINDOWS_MAP_CHARS)
 
 
 def get_strip_prefix(path):

src/borg/item.pyx

@@ -7,7 +7,7 @@ from cpython.bytes cimport PyBytes_AsStringAndSize
 from .constants import ITEM_KEYS, ARCHIVE_KEYS
 from .helpers import StableDict
 from .helpers import format_file_size
-from .helpers.fs import assert_sanitized_path, to_sanitized_path, percentify, slashify
+from .helpers.fs import assert_sanitized_path, to_sanitized_path, map_chars, slashify
 from .helpers.msgpack import timestamp_to_int, int_to_timestamp, Timestamp
 from .helpers.time import OutputTimestamp, safe_timestamp
 
@@ -265,7 +265,7 @@ cdef class Item(PropDict):
 
     path = PropDictProperty(str, 'surrogate-escaped str', encode=assert_sanitized_path, decode=to_sanitized_path)
     source = PropDictProperty(str, 'surrogate-escaped str')  # legacy borg 1.x. borg 2: see .target
-    target = PropDictProperty(str, 'surrogate-escaped str', encode=slashify, decode=percentify)
+    target = PropDictProperty(str, 'surrogate-escaped str', encode=slashify, decode=map_chars)
     user = PropDictProperty(str, 'surrogate-escaped str')
     group = PropDictProperty(str, 'surrogate-escaped str')
 

src/borg/testsuite/helpers/fs_test.py

@@ -20,6 +20,7 @@
     safe_unlink,
     remove_dotdot_prefixes,
     make_path_safe,
+    map_chars,
 )
 from ...platform import is_win32, is_darwin, is_haiku
 from .. import are_hardlinks_supported
@@ -441,3 +442,24 @@ def open_dir(path):
             assert dir_is_tagged(dir_fd=fd, exclude_caches=True, exclude_if_present=[".NOBACKUP"]) == [".NOBACKUP"]
         with open_dir(str(normal_dir)) as fd:
             assert dir_is_tagged(dir_fd=fd, exclude_caches=True, exclude_if_present=[".NOBACKUP"]) == []
+
+
+def test_map_chars(monkeypatch):
+    # Test behavior on non-Windows (should return path unchanged)
+    monkeypatch.setattr("borg.helpers.fs.is_win32", False)
+    assert map_chars("foo/bar") == "foo/bar"
+    assert map_chars("foo\\bar") == "foo\\bar"
+    assert map_chars("foo:bar") == "foo:bar"
+
+    # Test behavior on Windows
+    monkeypatch.setattr("borg.helpers.fs.is_win32", True)
+
+    # Reserved characters replacement
+    assert map_chars("foo:bar") == "foo\uf03abar"
+    assert map_chars("foo<bar") == "foo\uf03cbar"
+    assert map_chars("foo>bar") == "foo\uf03ebar"
+    assert map_chars('foo"bar') == "foo\uf022bar"
+    assert map_chars("foo\\bar") == "foo\uf05cbar"
+    assert map_chars("foo|bar") == "foo\uf07cbar"
+    assert map_chars("foo?bar") == "foo\uf03fbar"
+    assert map_chars("foo*bar") == "foo\uf02abar"