mythmusic: various tweaks to the lyrics grabbers

* tweak the priorities of the grabbers
* fix the debugging on a couple of the grabbers
* attempt to fix the Baidu grabber
* Update the README

mythtv/programs/scripts/metadata/Music/lyrics/README

@@ -87,16 +87,18 @@ Options:
 
 
 
-Current Grabbers And Their Priority
-===================================
-
-EmbeddedLyrics 100
-FileLyrics     105
-TTPlayer       110
-Baidu          120
-GomAudio       135
-Lyrdb          150
-LyricsWiki     200
-Genius         210
-LyricsMode     220
-DarkLyrics     230
+Current Grabbers, Their Priority And Whether Synchronized
+=========================================================
+
+EmbeddedLyrics 100  Yes/No
+FileLyrics     105  Yes/No
+TTPlayer       110  Yes
+Alsong         120  Yes
+LetsSingIt     130  No
+LyricsCom      140  No
+LyricsWiki     150  No
+Genius         160  No
+LyricsMode     170  No
+DarkLyrics     180  No
+GomAudio       200  Yes
+Baidu          210  Yes

mythtv/programs/scripts/metadata/Music/lyrics/baidu.py

@@ -2,7 +2,7 @@
 """
 Scraper for http://www.baidu.com
 
-taxigps
+ronie
 """
 
 import sys
@@ -14,11 +14,11 @@
 from optparse import OptionParser
 from common import utilities
 
-__author__      = "Paul Harrison and 'taxigps'"
+__author__      = "Paul Harrison and 'ronie'"
 __title__       = "Baidu"
 __description__ = "Search http://www.baidu.com for lyrics"
 __version__     = "0.1"
-__priority__    = "120"
+__priority__    = "210"
 __syncronized__ = True
 
 debug = False
@@ -27,28 +27,27 @@
 
 class LyricsFetcher:
     def __init__( self ):
-        self.BASE_URL = 'http://box.zhangmen.baidu.com/x?op=12&count=1&title=%s$$%s$$$$'
-        self.LRC_URL = 'http://box.zhangmen.baidu.com/bdlrc/%d/%d.lrc'
+        self.BASE_URL = 'http://music.baidu.com/search/lrc?key=%s-%s'
+        self.LRC_URL = 'http://music.baidu.com%s'
 
     def get_lyrics(self, lyrics):
         utilities.log(debug, "%s: searching lyrics for %s - %s - %s" % (__title__, lyrics.artist, lyrics.album, lyrics.title))
 
         try:
-            url = self.BASE_URL % (urllib.quote(lyrics.title), urllib.quote((lyrics.artist)))
-            xml_str = urllib.urlopen(url).read()
-            lrcid_pattern = re.compile(r'<lrcid>(.+?)</lrcid>')
-            lrcid = int(re.search(lrcid_pattern, xml_str).group(1))
-            if lrcid == 0:
+            url = self.BASE_URL % (lyrics.title, lyrics.artist)
+            utilities.log(debug, "%s: searching url %s" % (__title__, url))
+            data = urllib.urlopen(url).read()
+            songmatch = re.search('song-title.*?<em>(.*?)</em>', data, flags=re.DOTALL)
+            track = songmatch.group(1)
+            artistmatch = re.search('artist-title.*?<em>(.*?)</em>', data, flags=re.DOTALL)
+            name = artistmatch.group(1)
+            urlmatch = re.search("down-lrc-btn.*?':'(.*?)'", data, flags=re.DOTALL)
+            found_url = urlmatch.group(1)
+            if (difflib.SequenceMatcher(None, lyrics.artist.lower(), name.lower()).ratio() > 0.8) and (difflib.SequenceMatcher(None, lyrics.title.lower(), track.lower()).ratio() > 0.8):
+                lyr = urllib.urlopen(self.LRC_URL % found_url).read()
+            else:
                 return False
-            lrc_url = self.LRC_URL % (lrcid/100, lrcid)
-            lyr = urllib.urlopen(lrc_url).read()
         except:
-            utilities.log(True, "%s: %s::%s (%d) [%s]" % (
-                   __title__, self.__class__.__name__,
-                   sys.exc_info()[ 2 ].tb_frame.f_code.co_name,
-                   sys.exc_info()[ 2 ].tb_lineno,
-                   sys.exc_info()[ 1 ]
-                   ))
             return False
 
         enc = chardet.detect(lyr)

mythtv/programs/scripts/metadata/Music/lyrics/darklyrics.py

@@ -16,7 +16,7 @@
 __author__      = "Paul Harrison and smory'"
 __title__       = "DarkLyrics"
 __description__ = "Search http://www.darklyrics.com/ - the largest metal lyrics archive on the Web"
-__priority__    = "230";
+__priority__    = "180";
 __version__     = "0.1"
 __syncronized__ = False;
 

mythtv/programs/scripts/metadata/Music/lyrics/embedlrc.py

@@ -8,12 +8,12 @@
 import sys, os, re, chardet
 import xml.dom.minidom as xml
 from optparse import OptionParser
-from common import *
+from common import utilities
 
 __author__      = "Paul Harrison and 'ronin'"
 __title__       = "EmbeddedLyrics"
 __description__ = "Search tracks tag for embedded lyrics"
-__version__     = "0.1"
+__version__     = "0.2"
 __priority__    = "100"
 __syncronized__ = True
 

mythtv/programs/scripts/metadata/Music/lyrics/genius.py

@@ -25,7 +25,7 @@
 __author__      = "Paul Harrison and ronie'"
 __title__       = "Genius"
 __description__ = "Search http://www.genius.com for lyrics"
-__priority__    = "210"
+__priority__    = "160"
 __version__     = "0.1"
 __syncronized__ = False
 

mythtv/programs/scripts/metadata/Music/lyrics/gomaudio.py

@@ -11,6 +11,7 @@
 import hashlib
 import urllib
 import re
+import unicodedata
 from optparse import OptionParser
 from common import utilities
 from common import audiofile
@@ -28,6 +29,11 @@
 
 GOM_URL = "http://newlyrics.gomtv.com/cgi-bin/lyrics.cgi?cmd=find_get_lyrics&file_key=%s&title=%s&artist=%s&from=gomaudio_local"
 
+def remove_accents(data):
+    nfkd_data = unicodedata.normalize('NFKD', data)
+    return u"".join([c for c in nfkd_data if not unicodedata.combining(c)])
+
+
 class gomClient(object):
     '''
     privide Gom specific function, such as key from mp3
@@ -62,7 +68,7 @@ def get_lyrics(self, lyrics):
                 key = gomClient.GetKeyFromFile(lyrics.filename)
             if not key:
                 return False
-            url = GOM_URL %(key, urllib.quote(lyrics.title.decode("utf-8").encode("euc-kr")), urllib.quote(lyrics.artist.decode("utf-8").encode("euc-kr")))
+            url = GOM_URL %(key, urllib.quote(remove_accents(lyrics.title.decode('utf-8')).encode('euc-kr')), (remove_accents(lyrics.artist.decode('utf-8')).encode('euc-kr')))
             response = urllib.urlopen(url)
             Page = response.read()
         except:

mythtv/programs/scripts/metadata/Music/lyrics/lyricsmode.py

@@ -8,7 +8,7 @@
 __author__      = "Paul Harrison and ronie'"
 __title__       = "LyricsMode"
 __description__ = "Search http://www.lyricsmode.com for lyrics"
-__priority__    = "220"
+__priority__    = "170"
 __version__     = "0.1"
 __syncronized__ = False
 
@@ -141,6 +141,8 @@ def buildVersion():
     sys.exit(0)
 
 def main():
+    global debug
+
     parser = OptionParser()
 
     parser.add_option('-v', "--version", action="store_true", default=False,

mythtv/programs/scripts/metadata/Music/lyrics/lyricswiki.py

@@ -12,8 +12,8 @@
 __author__      = "Paul Harrison and ronie'"
 __title__       = "LyricsWiki"
 __description__ = "Search http://lyrics.wikia.com for lyrics"
-__priority__    = "200"
-__version__     = "0.1"
+__priority__    = "150"
+__version__     = "0.2"
 __syncronized__ = False
 
 

mythtv/programs/scripts/metadata/Music/lyrics/ttplayer.py

@@ -247,6 +247,8 @@ def buildVersion():
     sys.exit(0)
 
 def main():
+    global debug
+
     parser = OptionParser()
 
     parser.add_option('-v', "--version", action="store_true", default=False,