* Replace custom Levenshtein distance implementation with difflib.SequenceMatcher

* Better files->album matching.

picard/album.py

@@ -159,7 +159,7 @@ class Album(DataObject):
             sim = file.orig_metadata.compare(track.metadata)
             if sim > bestMatch[0]:
                 bestMatch = sim, track
-                
+
         if bestMatch[1]:
             file.move_to_track(bestMatch[1])
 

picard/metadata.py

@@ -41,14 +41,11 @@ class Metadata(QtCore.QObject):
         parts = []
         
         tags = {
-            "musicbrainz_trackid": 10,
-            "musicbrainz_artistid": 10,
-            "musicbrainz_albumid": 10,
             "~#length": 16,
-            "title": 14,
-            "artist": 8,
-            "album": 10,
-            "tracknumber": 12,
+            "title": 20,
+            "artist": 6,
+            "album": 12,
+            "tracknumber": 5,
         }
 
         identical = [
@@ -60,25 +57,25 @@ class Metadata(QtCore.QObject):
             "discnumber",
             "totaldiscs",
         ]
-        
-        for tag in self.keys():
-            if tag not in tags and not tag.startswith("~"):
-                tags[tag] = 1
-        
-        for tag in other.keys():
-            if tag not in tags and not tag.startswith("~"):
-                tags[tag] = 1
-                
+
+        #for tag in self.keys():
+        #    if tag not in tags and not tag.startswith("~"):
+        #        tags[tag] = 1
+
+        #for tag in other.keys():
+        #    if tag not in tags and not tag.startswith("~"):
+        #        tags[tag] = 1
+
         for tag, weight in tags.items():
             if self[tag] and other[tag]:
                 if tag in identical:
                     sim = 1.0 - abs(cmp(self[tag], other[tag]))
-                elif tag in ["~#length"]:
+                elif tag == "~#length":
                     sim = 1.0 - min(abs(self[tag] - other[tag]), 30000) / 30000.0
                 else:
                     sim = similarity(self[tag], other[tag])
                 parts.append((sim, weight))
-            
+
         total = reduce(lambda x, y: x + y[1], parts, 0.0)
         return reduce(lambda x, y: x + y[0] * y[1] / total, parts, 0.0)
 

picard/similarity.py

@@ -19,45 +19,14 @@
 
 import math
 import re
-
-
-def distance(a,b):
-    """Calculates the Levenshtein distance between a and b."""
-    
-    n, m = len(a), len(b)
-    if n > m:
-        # Make sure n <= m, to use O(min(n,m)) space
-        a,b = b,a
-        n,m = m,n
-        
-    current = range(n+1)
-    for i in range(1,m+1):
-        previous, current = current, [i]+[0]*n
-        for j in range(1,n+1):
-            add, delete = previous[j]+1, current[j-1]+1
-            change = previous[j-1]
-            if a[j-1] != b[i-1]:
-                change = change + 1
-            current[j] = min(add, delete, change)
-            
-    return current[n]
-
-
-def boost(sim):
-    sim2 = sim
-    sim = min(1, (math.exp(sim) - 1) / (math.e - 1.2))
-    sim = math.pow(sim, 0.8)
-    sim = max(sim2, sim)
-    return sim
+from difflib import SequenceMatcher
+from picard.util import unaccent
 
 
 def raw_similarity(a, b):
     """Calculates raw similarity of strings ``a`` and ``b``."""
-    if not a or not b:
-        return 0.0
-    sim = 1 - distance(a, b) * 1.0 / max(len(a), len(b))
-    return boost(sim)
-
+    d = SequenceMatcher(None, a, b).ratio()
+    return d
 
 _split_re = re.compile("\W", re.UNICODE)
 _stop_words = ["the", "--", "in", "of", "a", "feat"]
@@ -73,19 +42,23 @@ _replace_words = {
     "disc 8": "CD8",
 }
 
-def similarity(a1, b1):
-    """Calculates "smart" similarity of strings ``a`` and ``b``."""
-    a2 = a1
-    b2 = b1
+def normalize(string):
     for w, r in _replace_words.items():
-        a2 = a2.replace(w, r)
-        b2 = b2.replace(w, r)
-    def flt(a):
-        def flt(a):
-            return a not in _stop_words and len(a) > 1
-        return u" ".join(filter(flt, _split_re.split(a.lower())))
-    a2 = flt(a2)
-    b2 = flt(b2)
+        string = string.replace(w, r)
+    string = string.lower()
+    string = " ".join(filter(lambda a: a not in _stop_words and len(a) > 1,
+                             _split_re.split(string)))
+    string = unaccent(string)
+    return string
+
+def similarity(a1, b1):
+    return raw_similarity(a1, b1)
+    """Calculates "smart" similarity of strings ``a`` and ``b``."""
+    a2 = normalize(a1)
+    if a2:
+        b2 = normalize(b1)
+    else:
+        b2 = ""
     sim1 = raw_similarity(a1, b1)
     if a2 or b2:
         sim2 = raw_similarity(a2, b2)

picard/tagger.py

@@ -96,20 +96,38 @@ class Tagger(QtGui.QApplication, ComponentManager, Component):
         self.connect(self.window, QtCore.SIGNAL("addDirectory"), self.onAddDirectory)
         self.connect(self.worker, QtCore.SIGNAL("statusBarMessage(const QString &)"), self.window.setStatusBarMessage)
         self.connect(self.window, QtCore.SIGNAL("file_updated(int)"), QtCore.SIGNAL("file_updated(int)"))
-        
+
         self.worker.start()
         self.browserIntegration.start()
-        
+
+    def match_files_to_album(self, files, album):
+        matches = []
+        for file in files:
+            for track in album.tracks:
+                sim = track.metadata.compare(file.orig_metadata)
+                matches.append((sim, file, track))
+        matches.sort(reverse=True)
+        matched = []
+        for sim, file, track in matches:
+            if sim <= 0.5:
+                continue
+            if file in matched:
+                continue
+            if track.linked_file and track.linked_file.similarity > sim:
+                continue
+            file.move_to_track(track)
+            matched.append(file)
+
     def exit(self):
         self.browserIntegration.stop()
         self.worker.stop()
-        
+
     def run(self):
         self.window.show()
         res = self.exec_()
         self.exit()
         return res
-        
+
     def setup_gettext(self, localeDir):
         """Setup locales, load translations, install gettext functions."""
         if sys.platform == "win32":

picard/ui/itemviews.py

@@ -147,7 +147,7 @@ class BaseTreeView(QtGui.QTreeWidget):
         mimeData.setData("application/picard.file-list", "\n".join(file_ids))
         print "\n".join(file_ids)
         return mimeData
-        
+
     def dropFiles(self, files, target):
         # File -> Track
         if isinstance(target, Track):
@@ -164,8 +164,7 @@ class BaseTreeView(QtGui.QTreeWidget):
                     file.move_to_cluster(target.cluster)
         # File -> Album
         elif isinstance(target, Album):
-            for file in files:
-                target.matchFile(file)
+            self.tagger.match_files_to_album(files, target)
 
     def dropAlbums(self, albums, target):
         # Album -> Cluster
@@ -289,7 +288,7 @@ class FileTreeView(BaseTreeView):
 
         file.lock_for_read()
         try:
-            metadata = file.metadata
+            metadata = file.orig_metadata
             item.setText(0, metadata["title"])
             item.setText(1, format_time(metadata.get("~#length", 0)))
             item.setText(2, metadata["artist"])

picard/ui/mainwindow.py

@@ -410,7 +410,7 @@ class MainWindow(QtGui.QMainWindow):
                 if obj.linked_file:
                     orig_metadata = obj.linked_file.orig_metadata
                     metadata = obj.linked_file.metadata
-                    statusBar = obj.linked_file.filename
+                    statusBar = "%s (%d%%)" % (obj.linked_file.filename, obj.linked_file.similarity * 100)
                     file = obj.linked_file
                 else:
                     orig_metadata = obj.metadata