add support for rebuilding metadata of songs with partial tags

server/djrandom/metadata_fixer/metadata_fixer.py

@@ -63,15 +63,26 @@ class MetadataFixer(object):
         mp3.title = metadata.normalize_string(info['title'])
         mp3.artist = metadata.normalize_string(info['artist_name'])
 
-    def scan(self):
+    def scan(self, dry_run, partial):
         """Scan the database for new files."""
         n_bad = n_ok = n_err = 0
-        for mp3 in MP3.get_with_bad_metadata():
+        if partial:
+            entries = MP3.query.filter(
+                (MP3.state == MP3.READY)
+                 & (MP3.has_fingerprint == True)
+                 & ((MP3.artist == None)
+                    | (MP3.artist == '')
+                    | (MP3.title == '')
+                    | (MP3.title == None)))
+        else:
+            entries = MP3.get_with_bad_metadata()
+        for mp3 in entries:
             n_bad += 1
             log.info('searching metadata for %s' % mp3.sha1)
             try:
                 self.process(mp3)
                 mp3.state = MP3.READY
+                if not dry_run:
                     self.idx.add_mp3(mp3)
                 log.info('found: %s / %s' % (mp3.artist, mp3.title))
                 n_ok += 1
@@ -82,34 +93,38 @@ class MetadataFixer(object):
                 log.error(traceback.format_exc())
                 n_err += 1
                 mp3.state = MP3.ERROR
+            if not dry_run:
                 Session.add(mp3)
+        if not dry_run:
             Session.commit()
             self.idx.commit()
         log.debug('total: %d songs, found: %d' % (n_bad, n_ok))
 
-    def run(self, run_once):
+    def run(self, run_once, dry_run, partial):
         while True:
-            self.scan()
+            self.scan(dry_run, partial)
             if run_once:
                 break
             Session.remove()
             time.sleep(600)
 
 
-def run_fixer(solr_url, echonest_api_key, db_url, run_once):
+def run_fixer(solr_url, echonest_api_key, db_url, dry_run, partial, run_once):
     socket.setdefaulttimeout(300)
 
     init_db(db_url)
     fixer = MetadataFixer(solr_url, echonest_api_key)
-    fixer.run(run_once)
+    fixer.run(run_once, dry_run, partial)
 
 
 def main():
     parser = optparse.OptionParser()
     parser.add_option('--once', action='store_true')
+    parser.add_option('--dry_run', action='store_true')
     parser.add_option('--solr_url', default='http://localhost:8080/solr')
     parser.add_option('--echonest_api_key')
     parser.add_option('--db_url')
+    parser.add_option('--partial', action='store_true')
     daemonize.add_standard_options(parser)
     utils.read_config_defaults(
         parser, os.getenv('DJRANDOM_CONF', '/etc/djrandom.conf'))
@@ -125,8 +140,8 @@ def main():
         opts.foreground = True
 
     daemonize.daemonize(opts, run_fixer,
-                        (opts.solr_url, opts.echonest_api_key,
-                         opts.db_url, opts.once))
+                        (opts.solr_url, opts.echonest_api_key, opts.db_url,
+                         opts.dry_run, opts.partial, opts.once))
 
 
 if __name__ == '__main__':