Use YouTube as fallback for track metadata if not found on Spotify

spotdl/internals.py

@@ -1,6 +1,7 @@
 from logzero import logger as log
 import os
 import sys
+import urllib.request
 
 from spotdl import const
 
@@ -253,3 +254,12 @@ def remove_duplicates(tracks):
     local_set = set()
     local_set_add = local_set.add
     return [x for x in tracks if not (x in local_set or local_set_add(x))]
+
+
+def content_available(url):
+    try:
+        response = urllib.request.urlopen(url)
+    except HTTPError:
+        return False
+    else:
+        return response.getcode() < 300

spotdl/metadata.py

@@ -46,6 +46,8 @@ class EmbedMetadata:
     def __init__(self, music_file, meta_tags):
         self.music_file = music_file
         self.meta_tags = meta_tags
+        self.spotify_metadata = meta_tags["spotify_metadata"]
+        self.provider = "spotify" if meta_tags["spotify_metadata"] else "youtube"
 
     def as_mp3(self):
         """ Embed metadata to MP3 files. """
@@ -62,7 +64,7 @@ class EmbedMetadata:
         audiofile["lyricist"] = meta_tags["artists"][0]["name"]
         audiofile["arranger"] = meta_tags["artists"][0]["name"]
         audiofile["performer"] = meta_tags["artists"][0]["name"]
-        audiofile["website"] = meta_tags["external_urls"]["spotify"]
+        audiofile["website"] = meta_tags["external_urls"][self.provider]
         audiofile["length"] = str(meta_tags["duration"])
         if meta_tags["publisher"]:
             audiofile["encodedby"] = meta_tags["publisher"]
@@ -78,7 +80,7 @@ class EmbedMetadata:
         audiofile["TYER"] = TYER(encoding=3, text=meta_tags["year"])
         if meta_tags["publisher"]:
             audiofile["TPUB"] = TPUB(encoding=3, text=meta_tags["publisher"])
-        audiofile["COMM"] = COMM(encoding=3, text=meta_tags["external_urls"]["spotify"])
+        audiofile["COMM"] = COMM(encoding=3, text=meta_tags["external_urls"][self.provider])
         if meta_tags["lyrics"]:
             audiofile["USLT"] = USLT(
                 encoding=3, desc=u"Lyrics", text=meta_tags["lyrics"]
@@ -106,7 +108,7 @@ class EmbedMetadata:
         audiofile = MP4(music_file)
         self._embed_basic_metadata(audiofile, preset=M4A_TAG_PRESET)
         audiofile[M4A_TAG_PRESET["year"]] = meta_tags["year"]
-        audiofile[M4A_TAG_PRESET["comment"]] = meta_tags["external_urls"]["spotify"]
+        audiofile[M4A_TAG_PRESET["comment"]] = meta_tags["external_urls"][self.provider]
         if meta_tags["lyrics"]:
             audiofile[M4A_TAG_PRESET["lyrics"]] = meta_tags["lyrics"]
         try:
@@ -127,7 +129,7 @@ class EmbedMetadata:
         audiofile = FLAC(music_file)
         self._embed_basic_metadata(audiofile)
         audiofile["year"] = meta_tags["year"]
-        audiofile["comment"] = meta_tags["external_urls"]["spotify"]
+        audiofile["comment"] = meta_tags["external_urls"][self.provider]
         if meta_tags["lyrics"]:
             audiofile["lyrics"] = meta_tags["lyrics"]
 
@@ -147,7 +149,8 @@ class EmbedMetadata:
         meta_tags = self.meta_tags
         audiofile[preset["artist"]] = meta_tags["artists"][0]["name"]
         audiofile[preset["albumartist"]] = meta_tags["album"]["artists"][0]["name"]
-        audiofile[preset["album"]] = meta_tags["album"]["name"]
+        if meta_tags["album"]["name"]:
+            audiofile[preset["album"]] = meta_tags["album"]["name"]
         audiofile[preset["title"]] = meta_tags["name"]
         audiofile[preset["date"]] = meta_tags["release_date"]
         audiofile[preset["originaldate"]] = meta_tags["release_date"]

spotdl/patcher.py

@@ -0,0 +1,31 @@
+import pafy
+
+from spotdl import internals
+
+
+def _getbestthumb(self):
+    url = self._ydl_info["thumbnails"][0]["url"]
+    if url:
+        return url
+
+    part_url = "http://i.ytimg.com/vi/%s/" % self.videoid
+    # Thumbnail resolution sorted in descending order
+    thumbs = ("maxresdefault.jpg",
+              "sddefault.jpg",
+              "hqdefault.jpg",
+              "mqdefault.jpg",
+              "default.jpg")
+    for thumb in thumbs:
+        url = part_url + thumb
+        if self._content_available(url):
+            return url
+
+
+@classmethod
+def _content_available(cls, url):
+    return internals.content_available(url)
+
+def patch_pafy():
+    pafy.backend_shared.BasePafy._bestthumb = None
+    pafy.backend_shared.BasePafy._content_available = _content_available
+    pafy.backend_shared.BasePafy.getbestthumb = _getbestthumb

spotdl/spotify_tools.py

@@ -81,6 +81,7 @@ def generate_metadata(raw_song):
     # Some sugar
     meta_tags["year"], *_ = meta_tags["release_date"].split("-")
     meta_tags["duration"] = meta_tags["duration_ms"] / 1000.0
+    meta_tags["spotify_metadata"] = True
     # Remove unwanted parameters
     del meta_tags["duration_ms"]
     del meta_tags["available_markets"]

spotdl/youtube_tools.py

@@ -14,6 +14,12 @@ from spotdl import const
 # Read more on mps-youtube/pafy#199
 pafy.g.opener.addheaders.append(("Range", "bytes=0-"))
 
+# Implement unreleased methods on Pafy object
+# More info: https://github.com/mps-youtube/pafy/pull/211
+if pafy.__version__ <= "0.5.4":
+    from spotdl import patcher
+    patcher.patch_pafy()
+
 
 def set_api_key():
     if const.args.youtube_api_key:
@@ -49,6 +55,8 @@ def match_video_and_metadata(track, force_pafy=True):
         track = slugify(content.title).replace("-", " ")
         if not const.args.no_metadata:
             meta_tags = spotify_tools.generate_metadata(track)
+            if meta_tags is None: # and const.args.allow_youtube:
+                meta_tags = generate_metadata(content)
     else:
         # Let it generate metadata, youtube doesn't know spotify slang
         if not const.args.no_metadata or internals.is_spotify(track):
@@ -56,11 +64,37 @@ def match_video_and_metadata(track, force_pafy=True):
 
         if force_pafy:
             content = go_pafy(track, meta_tags)
+            if meta_tags is None: # and const.args.allow_youtube:
+                meta_tags = generate_metadata(content)
         else:
             content = None
     return content, meta_tags
 
 
+def generate_metadata(content):
+    """ Fetch a song's metadata from YouTube. """
+    meta_tags = {"spotify_metadata": False,
+                 "name": content.title,
+                 "artists": [{"name": content.author}],
+                 "duration": content.length,
+                 "external_urls": {"youtube": content.watchv_url},
+                 "album": {"images" : [{"url": content.getbestthumb()}], "name": None},
+                 "year": content.published.split("-")[0],
+                 "release_date": content.published.split(" ")[0],
+                 "type": "track",
+                 "disc_number": 1,
+                 "track_number": 1,
+                 "total_tracks": 1,
+                 "publisher": None,
+                 "external_ids": {"isrc": None},
+                 "lyrics": None,
+                 "copyright": None,
+                 "genre": None,
+                 }
+
+    return meta_tags
+
+
 def get_youtube_title(content, number=None):
     """ Get the YouTube video's title. """
     title = content.title