Lyrics: fix Lyricwiki. again.

i hope they stop doing that
also switch to QXmlStreamReader for parsing lyrics.

1 files changed, 25 insertions, 18 deletions

diff --git a/nephilim/plugins/Lyrics.py b/nephilim/plugins/Lyrics.py
index 99eff72..2bd3ea4 100644
--- a/nephilim/plugins/Lyrics.py
+++ b/nephilim/plugins/Lyrics.py
@@ -19,7 +19,6 @@ from PyQt4        import QtGui, QtCore, QtNetwork
 from PyQt4.QtCore import QVariant
 
 import os
-import re
 from   lxml import etree
 
 from ..plugin import Plugin
@@ -80,7 +79,7 @@ class LyricsWidget(QtGui.QWidget):
 
     #### public ####
     def set_lyrics(self, song, lyrics, flags = 0):
-        """Set currently displayed lyrics for song. flags parameter is
+        """Set currently displayed lyrics (unicode string) for song. flags parameter is
            unused now."""
         if not song:
             self.__label.clear()
@@ -95,7 +94,7 @@ class LyricsWidget(QtGui.QWidget):
                              %(song['title'], song['artist'], song['album']))
         if lyrics:
             self.logger.info('Setting new lyrics.')
-            self.__text_view.insertPlainText(lyrics.decode('utf-8'))
+            self.__text_view.insertPlainText(lyrics)
             self.lyrics_loaded = True
         else:
             self.logger.info('Lyrics not found.')
@@ -195,23 +194,31 @@ class Lyrics(Plugin):
                 return self.finish()
             self.logger.info('Found Lyricwiki song URL: %s.'%url)
 
-            self.rep = self.nam.get(QtNetwork.QNetworkRequest(url))
+            # XXX temporary hack to work around lyricwiki.org -> lyrics.wikia.org transition
+            url.setHost('lyrics.wikia.com')
+            url.setPath('/lyrics%s'%url.path())
+            req = QtNetwork.QNetworkRequest(url)
+            self.rep = self.nam.get(req)
             self.rep.finished.connect(self.__handle_lyrics)
 
         def __handle_lyrics(self):
-            #TODO this should use Qt xml functions too
             lyrics = ''
-            page = unicode(self.rep.readAll(), encoding = 'utf-8')
-            page = re.sub('<br>|<br/>|<br />', '\n', page)
-            try:
-                html   = etree.HTML(page)
-            except etree.XMLSyntaxError, e:
-                self.logger.error('Error parsing lyrics: %s' %e)
-                return self.finish()
+            xml    = QtCore.QXmlStreamReader(self.rep)
+            while not xml.atEnd():
+                token = xml.readNext()
+                if token == QtCore.QXmlStreamReader.StartElement:
+                    if xml.name() == 'div' and xml.attributes().value('class') == 'lyricbox':
+                        while not xml.atEnd():
+                            token = xml.readNext()
+                            if token == QtCore.QXmlStreamReader.EndElement and xml.name() == 'div':
+                                break
+                            elif token == QtCore.QXmlStreamReader.StartElement and xml.name() == 'br':
+                                lyrics += '\n'
+                            elif token == QtCore.QXmlStreamReader.Characters:
+                                lyrics += xml.text()
+            if xml.hasError():
+                self.logger.warning('Error parsing lyrics: %s'%xml.errorString())
 
-            for elem in html.iterfind('.//div'):
-                if elem.get('class') == 'lyricbox':
-                    lyrics += etree.tostring(elem, method = 'text', encoding = 'utf-8')
             self.finish(lyrics)
 
     class FetchAnimelyrics(common.MetadataFetcher):
@@ -363,7 +370,7 @@ class Lyrics(Plugin):
         try:
             self.logger.info('Trying to read lyrics from file %s.'%self.__lyrics_path)
             file   = open(self.__lyrics_path, 'r')
-            lyrics = file.read()
+            lyrics = file.read().decode('utf-8')
             file.close()
             if lyrics:
                 return self.o.set_lyrics(song, lyrics)
@@ -374,7 +381,7 @@ class Lyrics(Plugin):
             fetcher.fetch(song)
 
     def save_lyrics_file(self, lyrics, path = None):
-        """Save lyrics to a file specified in path.
+        """Save lyrics (unicode string) to a file specified in path.
            If path is None, then a default value is used."""
         self.logger.info('Saving lyrics...')
         try:
@@ -382,7 +389,7 @@ class Lyrics(Plugin):
                 file = open(path, 'w')
             else:
                 file = open(self.__lyrics_path, 'w')
-            file.write(lyrics)
+            file.write(lyrics.encode('utf-8'))
             file.close()
             self.logger.info('Lyrics successfully saved.')
         except IOError, e: