Lyrics: rewrite the system of fetchers to be more flexible.

also use Qt's networking module instead of urllib - which removes the
need for explicit multithreading here.

2 files changed, 165 insertions, 90 deletions

diff --git a/TODO b/TODO
index 379a66e..8271283 100644
--- a/TODO
+++ b/TODO
@@ -13,3 +13,4 @@ other TODO:
 - split library by letters (like Amarok)
 - icons could use some improving
 - waste less memory (library and playlist are main suspects)
+- Lyrics - should use only Qt's xml functions and allow user to select lyrics
diff --git a/nephilim/plugins/Lyrics.py b/nephilim/plugins/Lyrics.py
index 381009a..960c400 100644
--- a/nephilim/plugins/Lyrics.py
+++ b/nephilim/plugins/Lyrics.py
@@ -15,22 +15,23 @@
 #    along with Nephilim.  If not, see <http://www.gnu.org/licenses/>.
 #
 
-from PyQt4        import QtGui, QtCore
+from PyQt4        import QtGui, QtCore, QtNetwork
 from PyQt4.QtCore import QVariant
 
-import socket
 import os
 import re
-import urllib
 from   lxml import etree
 
 from ..plugin import Plugin
 from ..       import misc
 
 class LyricsWidget(QtGui.QWidget):
+    #public
+    lyrics_loaded = None
+
     # public, read-only
-    plugin = None  # plugin
-    logger = None
+    plugin        = None  # plugin
+    logger        = None
 
     # private
     __text_view = None  # text-object
@@ -68,8 +69,6 @@ class LyricsWidget(QtGui.QWidget):
         self.layout().addWidget(self.__label, 0, 1)
         self.layout().addWidget(self.__text_view, 1, 1)
 
-        self.connect(self.plugin, QtCore.SIGNAL('new_lyrics_fetched'), self.set_lyrics)
-
     def set_lyrics(self, song, lyrics, flags = 0):
         """Set currently displayed lyrics for song. flags parameter is
            unused now."""
@@ -87,6 +86,7 @@ class LyricsWidget(QtGui.QWidget):
         if lyrics:
             self.logger.info('Setting new lyrics.')
             self.__text_view.insertPlainText(lyrics.decode('utf-8'))
+            self.lyrics_loaded = True
         else:
             self.logger.info('Lyrics not found.')
             self.__text_view.insertPlainText('Lyrics not found.')
@@ -100,32 +100,37 @@ class LyricsWidget(QtGui.QWidget):
 class Lyrics(Plugin):
     # public, read-only
     o           = None
-    """A dict of { site name : function }. Function takes a song and returns lyrics
-       as a python string or None if not found."""
-    sites       = {}
+    """A dict of { site name : fetcher object }. The fetcher object provides a
+       fetch(song)Function takes a song and emits finished(song, lyrics) signal
+       when finished. Lyrics is either a python unicode string, QString
+       or None if not found."""
 
     # private
     DEFAULTS          = {'sites' : QtCore.QStringList(['lyricwiki', 'animelyrics']), 'lyricdir' : '$musicdir/$songdir',
                          'lyricname' : '.lyrics_nephilim_$artist_$album_$title', 'store' : True}
     __available_sites = {}
+    __fetchers        = {}
+    __results         = 0
     __lyrics_dir      = None
     __lyrics_path     = None
 
     def __init__(self, parent, mpclient, name):
         Plugin.__init__(self, parent, mpclient, name)
 
-        self.__available_sites['lyricwiki']   = self.__fetch_lyricwiki
-        self.__available_sites['animelyrics'] = self.__fetch_animelyrics
+        self.__available_sites['lyricwiki']   = self.FetchLyricwiki
+        self.__available_sites['animelyrics'] = self.FetchAnimelyrics
 
     def _load(self):
-        self.o = LyricsWidget(self)
         for site in self.__available_sites:
             if site in self.settings().value('%s/sites'%self.name).toStringList():
-                self.sites[site] = self.__available_sites[site]
+                self.__fetchers[site] = self.__available_sites[site](self)
+        self.o = LyricsWidget(self)
+        for fetcher in self.__fetchers:
+            self.connect(self.__fetchers[fetcher], QtCore.SIGNAL('finished'), self.__new_lyrics_fetched)
         self.connect(self.mpclient, QtCore.SIGNAL('song_changed'), self.refresh)
     def _unload(self):
         self.o     = None
-        self.sites = []
+        self.sites = {}
         self.disconnect(self.mpclient, QtCore.SIGNAL('song_changed'), self.refresh)
     def info(self):
         return "Show (and fetch) the lyrics of the currently playing song."
@@ -133,18 +138,11 @@ class Lyrics(Plugin):
     def _get_dock_widget(self):
         return self._create_dock(self.o)
 
-    class FetchThread(QtCore.QThread):
-        def __init__(self, parent, fetch_func, song):
-            QtCore.QThread.__init__(self)
-            self.setParent(parent)
-            self.fetch_func = fetch_func
-            self.song = song
-        def run(self):
-            self.fetch_func(self.song)
-
     def refresh(self):
         """Attempt to automatically get lyrics first from a file, then from the internet."""
         self.logger.info('Autorefreshing lyrics.')
+        self.__results = 0
+        self.o.lyrics_loaded = False
         song = self.mpclient.current_song()
         if not song:
             self.__lyrics_dir  = ''
@@ -160,13 +158,12 @@ class Lyrics(Plugin):
             lyrics = file.read()
             file.close()
             if lyrics:
-                return self.emit(QtCore.SIGNAL('new_lyrics_fetched'), song, lyrics)
+                return self.o.set_lyrics(song, lyrics)
         except IOError, e:
             self.logger.info('Error reading lyrics file: %s.'%e)
 
-
-        thread = self.FetchThread(self, self.__fetch_lyrics, song)
-        thread.start()
+        for fetcher in self.__fetchers.values():
+            fetcher.fetch(song)
 
     def save_lyrics_file(self, lyrics, path = None):
         """Save lyrics to a file specified in path.
@@ -197,68 +194,14 @@ class Lyrics(Plugin):
         except IOError, e:
             self.logger.error('Error removing lyrics file %s: %s'%(path, e))
 
-    def __fetch_lyrics(self, song):
-        self.logger.info('Trying to download lyrics from internet.')
-        lyrics = None
-        for site in self.sites:
-            self.logger.info('Trying %s.'%site)
-            lyrics = self.sites[site](song)
-            if lyrics:
-                if self.settings().value(self.name + '/store').toBool():
-                    self.save_lyrics_file(lyrics)
-                return self.emit(QtCore.SIGNAL('new_lyrics_fetched'), song, lyrics)
-
-        self.emit(QtCore.SIGNAL('new_lyrics_fetched'), song, None)
-
-    def __fetch_lyricwiki(self, song):
-        url  = 'http://lyricwiki.org/api.php?%s' %urllib.urlencode({'func':'getSong',
-                   'artist':song.artist().encode('utf-8'), 'song':song.title().encode('utf-8'),
-                   'fmt':'xml'})
-        try:
-            # get url for lyrics
-            tree = etree.HTML(urllib.urlopen(url).read())
-            if tree.find('.//lyrics').text == 'Not found':
-                return None
-            #get page with lyrics and change <br> tags to newlines
-            url    = tree.find('.//url').text
-            page   = re.sub('<br>|<br/>|<br />', '\n', urllib.urlopen(url).read())
-            html   = etree.HTML(page)
-            lyrics = ''
-            for elem in html.iterfind('.//div'):
-                if elem.get('class') == 'lyricbox':
-                    lyrics += etree.tostring(elem, method = 'text', encoding = 'utf-8')
-            return lyrics
-        except (socket.error, IOError), e:
-            self.logger.error('Error downloading lyrics from LyricWiki: %s.'%e)
-            return None
-
-    def __fetch_animelyrics(self, song):
-        url = 'http://www.animelyrics.com/search.php?%s'%urllib.urlencode({'q':song.artist().encode('utf-8'),
-                                                                           't':'performer'})
-        try:
-            #get url for lyrics
-            self.logger.info('Searching Animelyrics: %s.'%url)
-            tree   = etree.HTML(urllib.urlopen(url).read())
-            url    = None
-            for elem in tree.iterfind('.//a'):
-                if ('href' in elem.attrib) and elem.text and (song.title() in elem.text):
-                    url = 'http://www.animelyrics.com/%s'%elem.get('href')
-            if not url:
-                return None
-            #get lyrics
-            self.logger.info('Found song URL: %s.'%url)
-            tree = etree.HTML(urllib.urlopen(url).read())
-            ret = ''
-            for elem in tree.iterfind('.//pre'):
-                if elem.get('class') == 'lyrics':
-                    ret += '%s\n\n'%etree.tostring(elem, method = 'text', encoding = 'utf-8')
-            return ret
-        except socket.error, e:
-            self.logger.error('Error downloading lyrics from Animelyrics: %s.'%e)
-            return None
-        except AttributeError:
-            # lyrics not found
-            return None
+    def __new_lyrics_fetched(self, song, lyrics):
+        self.logger.info('Got new lyrics.')
+        self.__results += 1
+        if lyrics and self.settings().value(self.name + '/store').toBool():
+            self.save_lyrics_file(lyrics)
+            return self.o.set_lyrics(song, lyrics)
+        elif self.__results >= len(self.__fetchers) and not self.o.lyrics_loaded:
+            self.o.set_lyrics(song, None)
 
     class SettingsWidgetLyrics(Plugin.SettingsWidget):
         lyricdir  = None
@@ -309,3 +252,134 @@ class Lyrics(Plugin):
 
     def get_settings_widget(self):
         return self.SettingsWidgetLyrics(self)
+
+    class Fetcher(QtCore.QObject):
+        """A basic class for lyrics fetchers. Provides a fetch(song) function,
+           emits a finished(song, lyrics) signal when done; lyrics is either a QString,
+           Python unicode string or None if not found."""
+        #public, read-only
+        logger = None
+
+        #private
+        nam  = None  # NetworkAccessManager
+        srep = None  # search results NetworkReply
+        lrep = None  # lyrics page NetworkReply
+        song = None  # current song
+
+        def __init__(self, plugin):
+            QtCore.QObject.__init__(self, plugin)
+
+            self.nam = QtNetwork.QNetworkAccessManager()
+            self.logger = plugin.logger
+        def fetch(self, song):
+            """Reimplement this in subclasses."""
+            pass
+
+        def finish(self, lyrics = None):
+            """A private convenience function to clean up and emit finished().
+               Feel free to reimplement/not use it."""
+            self.srep = None
+            self.lrep = None
+            self.emit(QtCore.SIGNAL('finished'), self.song, lyrics)
+            self.song = None
+
+    class FetchLyricwiki(Fetcher):
+
+        def fetch(self, song):
+            # abort any existing connections
+            if self.srep:
+                self.srep.abort()
+                self.srep = None
+            if self.lrep:
+                self.lrep.abort()
+                self.lrep = None
+            self.song = song
+
+            url = QtCore.QUrl('http://lyricwiki.org/api.php')
+            url.setQueryItems([('func', 'getSong'), ('artist', song.artist()),
+                               ('song', song.title()), ('fmt', 'xml')])
+
+            self.logger.info('Searching Lyricwiki: %s.'%url)
+            self.srep = self.nam.get(QtNetwork.QNetworkRequest(url))
+            self.connect(self.srep, QtCore.SIGNAL('finished()'), self.__handle_search_res)
+
+        def __handle_search_res(self):
+            url = None
+            xml = QtCore.QXmlStreamReader(self.srep)
+            while not xml.atEnd():
+                token = xml.readNext()
+                if token == QtCore.QXmlStreamReader.StartElement:
+                    if xml.name() == 'url':
+                        url = QtCore.QUrl() # the url is already percent-encoded
+                        url.setEncodedUrl(xml.readElementText().toLatin1())
+                    elif xml.name() == 'lyrics' and xml.readElementText() == 'Not found':
+                        xml.clear()
+                        return self.finish()
+            if xml.hasError():
+                self.logger.error('Error parsing seach results.%s'%xml.errorString())
+
+            if not url:
+                self.logger.error('Didn\'t find the URL in Lyricwiki search results.')
+                return self.finish()
+            self.logger.info('Found Lyricwiki song URL: %s.'%url)
+
+            self.lrep = self.nam.get(QtNetwork.QNetworkRequest(url))
+            self.connect(self.lrep, QtCore.SIGNAL('finished()'), self.__handle_lyrics)
+
+        def __handle_lyrics(self):
+            #TODO this should use Qt xml functions too
+            lyrics = ''
+            page = unicode(self.lrep.readAll(), encoding = 'utf-8')
+            page = re.sub('<br>|<br/>|<br />', '\n', page)
+            html   = etree.HTML(page)
+            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(Fetcher):
+
+        def fetch(self, song):
+            # abort any existing connections
+            if self.srep:
+                self.srep.abort()
+                self.srep = None
+            if self.lrep:
+                self.lrep.abort()
+                self.lrep = None
+            self.song = song
+
+            url = QtCore.QUrl('http://www.animelyrics.com/search.php')
+            url.setQueryItems([('t', 'performer'), ('q', self.song.artist())])
+
+            self.logger.info('Searching Animelyrics: %s.'%url)
+            self.srep = self.nam.get(QtNetwork.QNetworkRequest(url))
+            self.connect(self.srep, QtCore.SIGNAL('finished()'), self.__handle_search_res)
+
+        def __handle_search_res(self):
+            # TODO use Qt xml functions
+            tree = etree.HTML(unicode(self.srep.readAll(), encoding = 'utf-8', errors='ignore'))
+            self.srep = None
+
+            url = None
+            for elem in tree.iterfind('.//a'):
+                if ('href' in elem.attrib) and elem.text and (self.song.title() in elem.text):
+                    url = QtCore.QUrl('http://www.animelyrics.com/%s'%elem.get('href'))
+
+            if not url:
+                self.logger.info('Didn\'t find the URL in Animelyrics search results.')
+                return self.finish()
+            self.logger.info('Found Animelyrics song URL: %s.'%url)
+
+            self.lrep = self.nam.get(QtNetwork.QNetworkRequest(url))
+            self.connect(self.lrep, QtCore.SIGNAL('finished()'), self.__handle_lyrics)
+
+        def __handle_lyrics(self):
+            lyrics = ''
+            tree   = etree.HTML(unicode(self.lrep.readAll(), encoding = 'utf-8'))
+            for elem in tree.iterfind('.//pre'):
+                if elem.get('class') == 'lyrics':
+                    lyrics += '%s\n\n'%etree.tostring(elem, method = 'text', encoding = 'utf-8')
+
+            self.finish(lyrics)
+