Ignore files with a badly encoded path

Closes #85

supysonic/cli.py

@@ -40,12 +40,12 @@ def __init__(self, name, stdout, interval = 5):
         self.__last_display = 0
         self.__last_len = 0
 
-    def __call__(self, scanned, total):
-        if time.time() - self.__last_display > self.__interval or scanned == total:
+    def __call__(self, scanned):
+        if time.time() - self.__last_display > self.__interval:
             if not self.__last_len:
                 self.__stdout.write("Scanning '{0}': ".format(self.__name))
 
-            progress = "{0}% ({1}/{2})".format((scanned * 100) / total, scanned, total)
+            progress = '{0} files scanned'.format(scanned)
             self.__stdout.write('\b' * self.__last_len)
             self.__stdout.write(progress)
             self.__stdout.flush()
@@ -190,11 +190,15 @@ def folder_scan(self, folders, force):
                 self.write_line()
 
         scanner.finish()
-        added, deleted = scanner.stats()
-
-        self.write_line("Scanning done")
-        self.write_line('Added: %i artists, %i albums, %i tracks' % (added[0], added[1], added[2]))
-        self.write_line('Deleted: %i artists, %i albums, %i tracks' % (deleted[0], deleted[1], deleted[2]))
+        stats = scanner.stats()
+
+        self.write_line('Scanning done')
+        self.write_line('Added: {0.artists} artists, {0.albums} albums, {0.tracks} tracks'.format(stats.added))
+        self.write_line('Deleted: {0.artists} artists, {0.albums} albums, {0.tracks} tracks'.format(stats.deleted))
+        if stats.errors:
+            self.write_line('Errors in:')
+            for err in stats.errors:
+                self.write_line('- ' + err)
 
     user_parser = CLIParser(prog = 'user', add_help = False)
     user_subparsers = user_parser.add_subparsers(dest = 'action')

supysonic/frontend/folder.py

@@ -104,9 +104,13 @@ def scan_folder(id = None):
         scanner.scan(folder)
 
     scanner.finish()
-    added, deleted = scanner.stats()
-
-    flash('Added: %i artists, %i albums, %i tracks' % (added[0], added[1], added[2]))
-    flash('Deleted: %i artists, %i albums, %i tracks' % (deleted[0], deleted[1], deleted[2]))
+    stats = scanner.stats()
+
+    flash('Added: {0.artists} artists, {0.albums} albums, {0.tracks} tracks'.format(stats.added))
+    flash('Deleted: {0.artists} artists, {0.albums} albums, {0.tracks} tracks'.format(stats.deleted))
+    if stats.errors:
+        flash('Errors in:')
+        for err in stats.errors:
+            flash('- ' + err)
     return redirect(url_for('folder_index'))
 

supysonic/scanner.py

@@ -30,20 +30,26 @@
 from .db import RatingFolder, RatingTrack
 from .py23 import strtype
 
+class StatsDetails(object):
+    def __init__(self):
+        self.artists = 0
+        self.albums = 0
+        self.tracks = 0
+
+class Stats(object):
+    def __init__(self):
+        self.added = StatsDetails()
+        self.deleted = StatsDetails()
+        self.errors = []
+
 class Scanner:
     def __init__(self, force = False, extensions = None):
         if extensions is not None and not isinstance(extensions, list):
             raise TypeError('Invalid extensions type')
 
         self.__force = force
 
-        self.__added_artists = 0
-        self.__added_albums  = 0
-        self.__added_tracks  = 0
-        self.__deleted_artists = 0
-        self.__deleted_albums  = 0
-        self.__deleted_tracks  = 0
-
+        self.__stats = Stats()
         self.__extensions = extensions
 
         self.__folders_to_check = set()
@@ -59,15 +65,28 @@ def scan(self, folder, progress_callback = None):
             raise TypeError('Expecting Folder instance, got ' + str(type(folder)))
 
         # Scan new/updated files
-        files = [ os.path.join(root, f) for root, _, fs in os.walk(folder.path) for f in fs if self.__is_valid_path(os.path.join(root, f)) ]
-        total = len(files)
-        current = 0
-
-        for path in files:
-            self.scan_file(path)
-            current += 1
-            if progress_callback:
-                progress_callback(current, total)
+        to_scan = [ folder.path ]
+        scanned = 0
+        while to_scan:
+            path = to_scan.pop()
+            for f in os.listdir(path):
+                try: # test for badly encoded filenames
+                    f.encode('utf-8')
+                except UnicodeError:
+                    self.__stats.errors.append(path)
+                    continue
+
+                full_path = os.path.join(path, f)
+                if os.path.islink(full_path):
+                    continue
+                elif os.path.isdir(full_path):
+                    to_scan.append(full_path)
+                elif os.path.isfile(full_path) and self.__is_valid_path(full_path):
+                    self.scan_file(full_path)
+                    scanned += 1
+
+                    if progress_callback:
+                        progress_callback(scanned)
 
         # Remove files that have been deleted
         for track in Track.select(lambda t: t.root_folder == folder):
@@ -90,15 +109,15 @@ def finish(self):
                 continue
 
             self.__artists_to_check.add(album.artist.id)
-            self.__deleted_albums += 1
+            self.__stats.deleted.albums += 1
             album.delete()
         self.__albums_to_check.clear()
 
         for artist in Artist.select(lambda a: a.id in self.__artists_to_check):
             if not artist.albums.is_empty() or not artist.tracks.is_empty():
                 continue
 
-            self.__deleted_artists += 1
+            self.__stats.deleted.artists += 1
             artist.delete()
         self.__artists_to_check.clear()
 
@@ -168,7 +187,7 @@ def scan_file(self, path):
             trdict['artist'] = trartist
 
             Track(**trdict)
-            self.__added_tracks += 1
+            self.__stats.added.tracks += 1
         else:
             if tr.album.id != tralbum.id:
                 self.__albums_to_check.add(tr.album.id)
@@ -192,7 +211,7 @@ def remove_file(self, path):
         self.__folders_to_check.add(tr.folder.id)
         self.__albums_to_check.add(tr.album.id)
         self.__artists_to_check.add(tr.artist.id)
-        self.__deleted_tracks += 1
+        self.__stats.deleted.tracks += 1
         tr.delete()
 
     @db_session
@@ -231,7 +250,7 @@ def __find_album(self, artist, album):
             return al
 
         al = Album(name = album, artist = ar)
-        self.__added_albums += 1
+        self.__stats.added.albums += 1
 
         return al
 
@@ -241,7 +260,7 @@ def __find_artist(self, artist):
             return ar
 
         ar = Artist(name = artist)
-        self.__added_artists += 1
+        self.__stats.added.artists += 1
 
         return ar
 
@@ -289,5 +308,5 @@ def __try_read_tag(self, metadata, field, default = None, transform = lambda x:
             return default
 
     def stats(self):
-        return (self.__added_artists, self.__added_albums, self.__added_tracks), (self.__deleted_artists, self.__deleted_albums, self.__deleted_tracks)
+        return self.__stats
 

tests/base/test_scanner.py

@@ -56,10 +56,8 @@ def test_scan(self):
 
     @db_session
     def test_progress(self):
-        def progress(processed, total):
+        def progress(processed):
             self.assertIsInstance(processed, int)
-            self.assertIsInstance(total, int)
-            self.assertLessEqual(processed, total)
 
         self.scanner.scan(db.Folder[self.folderid], progress)
 
@@ -192,7 +190,13 @@ def test_scan_tag_change(self):
             self.assertIsNotNone(db.Album.get(name = 'Awesome album'))
 
     def test_stats(self):
-        self.assertEqual(self.scanner.stats(), ((1,1,1),(0,0,0)))
+        stats = self.scanner.stats()
+        self.assertEqual(stats.added.artists, 1)
+        self.assertEqual(stats.added.albums, 1)
+        self.assertEqual(stats.added.tracks, 1)
+        self.assertEqual(stats.deleted.artists, 0)
+        self.assertEqual(stats.deleted.albums, 0)
+        self.assertEqual(stats.deleted.tracks, 0)
 
 if __name__ == '__main__':
     unittest.main()