truncation incorporates extension length (GC-461)

beets/library.py

@@ -1152,7 +1152,6 @@ class Library(BaseLibrary):
         # Encode for the filesystem.
         if not fragment:
             subpath = bytestring_path(subpath)
-        subpath = util.truncate_path(subpath, pathmod)
 
         # Preserve extension.
         _, extension = pathmod.splitext(item.path)
@@ -1161,6 +1160,9 @@ class Library(BaseLibrary):
             extension = extension.decode('utf8', 'ignore')
         subpath += extension.lower()
 
+        # Truncate too-long components.
+        subpath = util.truncate_path(subpath, pathmod)
+
         if fragment:
             return subpath
         else:

beets/util/__init__.py

@@ -457,13 +457,22 @@ def sanitize_path(path, pathmod=None, replacements=None):
         comps[i] = comp
     return pathmod.join(*comps)
 
-def truncate_path(path, pathmod=None):
+def truncate_path(path, pathmod=None, length=MAX_FILENAME_LENGTH):
     """Given a bytestring path or a Unicode path fragment, truncate the
-    components to a legal length.
+    components to a legal length. In the last component, the extension
+    is preserved.
     """
     pathmod = pathmod or os.path
-    comps = [c[:MAX_FILENAME_LENGTH] for c in components(path, pathmod)]
-    return pathmod.join(*comps)
+    comps = components(path, pathmod)
+
+    out = [c[:length] for c in comps]
+    base, ext = pathmod.splitext(comps[-1])
+    if ext:
+        # Last component has an extension.
+        base = base[:length - len(ext)]
+        out[-1] = base + ext
+
+    return pathmod.join(*out)
 
 def sanitize_for_path(value, pathmod, key=None):
     """Sanitize the value for inclusion in a path: replace separators

docs/changelog.rst

@@ -60,6 +60,7 @@ Changelog
   question is now logged (thanks to Mike Kazantsev).
 * Truncate long filenames based on their *bytes* rather than their Unicode
   *characters*, fixing situations where encoded names could be too long.
+* Filename truncation now incorporates the length of the extension.
 * Fix an assertion failure when the MusicBrainz main database and search server
   disagree.
 * Fix a bug that caused the :doc:`/plugins/lastgenre` and other plugins not to

test/test_db.py

@@ -918,6 +918,19 @@ class PathStringTest(unittest.TestCase):
         alb = self.lib.get_album(alb.id)
         self.assert_(isinstance(alb.artpath, str))
 
+class PathTruncationTest(unittest.TestCase):
+    def test_truncate_bytestring(self):
+        p = util.truncate_path('abcde/fgh', posixpath, 4)
+        self.assertEqual(p, 'abcd/fgh')
+
+    def test_truncate_unicode(self):
+        p = util.truncate_path(u'abcde/fgh', posixpath, 4)
+        self.assertEqual(p, u'abcd/fgh')
+
+    def test_truncate_preserves_extension(self):
+        p = util.truncate_path(u'abcde/fgh.ext', posixpath, 5)
+        self.assertEqual(p, u'abcde/f.ext')
+
 class MtimeTest(unittest.TestCase):
     def setUp(self):
         self.ipath = os.path.join(_common.RSRC, 'testfile.mp3')