From 40fbc8ee7e9941c5bd8487a24370f6b734dcf451 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C5=A0ar=C5=ABnas=20Nejus?= <snejus@protonmail.com>
Date: Mon, 10 Mar 2025 03:16:37 +0000
Subject: [PATCH] Fix and simplify path truncation

---
 beets/util/__init__.py | 32 ++++++++++++++++++--------------
 test/test_util.py      | 27 ++++++++++++---------------
 2 files changed, 30 insertions(+), 29 deletions(-)

diff --git a/beets/util/__init__.py b/beets/util/__init__.py
index d8340a978..ff0a5d273 100644
--- a/beets/util/__init__.py
+++ b/beets/util/__init__.py
@@ -696,22 +696,26 @@ def sanitize_path(path: str, replacements: Replacements | None = None) -> str:
     return os.path.join(*comps)
 
 
-def truncate_path(path: AnyStr) -> AnyStr:
-    """Given a bytestring path or a Unicode path fragment, truncate the
-    components to a legal length. In the last component, the extension
-    is preserved.
+def truncate_str(s: str, length: int) -> str:
+    """Truncate the string to the given byte length.
+
+    If we end up truncating a unicode character in the middle (rendering it invalid),
+    it is removed:
+
+    >>> s = "🎹🎶"  # 8 bytes
+    >>> truncate_str(s, 6)
+    '🎹'
     """
+    return os.fsencode(s)[:length].decode(sys.getfilesystemencoding(), "ignore")
+
+
+def truncate_path(str_path: str) -> str:
+    """Truncate each path part to a legal length preserving the extension."""
     max_length = get_max_filename_length()
-    comps = components(path)
-
-    out = [c[:length] for c in comps]
-    base, ext = os.path.splitext(comps[-1])
-    if ext:
-        # Last component has an extension.
-        base = base[: max_length - len(ext)]
-        out[-1] = base + ext
-
-    return os.path.join(*out)
+    path = Path(str_path)
+    parent_parts = [truncate_str(p, max_length) for p in path.parts[:-1]]
+    stem = truncate_str(path.stem, max_length - len(path.suffix))
+    return str(Path(*parent_parts, stem).with_suffix(path.suffix))
 
 
 def _legalize_stage(
diff --git a/test/test_util.py b/test/test_util.py
index f5b4fd102..a4b224ee3 100644
--- a/test/test_util.py
+++ b/test/test_util.py
@@ -175,18 +175,15 @@ class PathConversionTest(BeetsTestCase):
         assert outpath == "C:\\caf\xe9".encode()
 
 
-class PathTruncationTest(BeetsTestCase):
-    def test_truncate_bytestring(self):
-        with _common.platform_posix():
-            p = util.truncate_path(b"abcde/fgh", 4)
-        assert p == b"abcd/fgh"
-
-    def test_truncate_unicode(self):
-        with _common.platform_posix():
-            p = util.truncate_path("abcde/fgh", 4)
-        assert p == "abcd/fgh"
-
-    def test_truncate_preserves_extension(self):
-        with _common.platform_posix():
-            p = util.truncate_path("abcde/fgh.ext", 5)
-        assert p == "abcde/f.ext"
+@patch("beets.util.get_max_filename_length", lambda: 5)
+@pytest.mark.parametrize(
+    "path, expected",
+    [
+        ("abcdeX/fgh", "abcde/fgh"),
+        ("abcde/fXX.ext", "abcde/f.ext"),
+        ("a🎹/a.ext", "a🎹/a.ext"),
+        ("ab🎹/a.ext", "ab/a.ext"),
+    ],
+)
+def test_truncate_path(path, expected):
+    assert util.truncate_path(path) == expected