Normalize regex query patterns/values (fix beetbox#1482)

beets/dbcore/query.py

@@ -21,6 +21,7 @@
 from operator import mul
 from beets import util
 from datetime import datetime, timedelta
+import unicodedata
 
 
 class ParsingError(ValueError):
@@ -201,6 +202,7 @@ class RegexpQuery(StringFieldQuery):
     """
     def __init__(self, field, pattern, fast=True):
         super(RegexpQuery, self).__init__(field, pattern, fast)
+        pattern = self._normalize(pattern)
         try:
             self.pattern = re.compile(self.pattern)
         except re.error as exc:
@@ -209,9 +211,16 @@ def __init__(self, field, pattern, fast=True):
                                                 "a regular expression",
                                                 format(exc))
 
+    @staticmethod
+    def _normalize(s):
+        """Normalize a Unicode string's representation (used on both
+        patterns and matched values).
+        """
+        return unicodedata.normalize('NFC', s)
+
     @classmethod
     def string_match(cls, pattern, value):
-        return pattern.search(value) is not None
+        return pattern.search(cls._normalize(value)) is not None
 
 
 class BooleanQuery(MatchQuery):

docs/changelog.rst

@@ -44,6 +44,8 @@ Fixes:
   :bug:`825`
 * Fix some logging errors when filenames and other user-provided strings
   contain curly braces. :bug:`1481`
+* Regular expression queries over paths now work more reliably with non-ASCII
+  characters in filenames. :bug:`1482`
 
 
 1.3.13 (April 24, 2015)