Retry a few times if Genius returns an invalid response

In FFmpeg, a given encoding may not always point to the same format
string.

CHANGES.md

@@ -8,6 +8,18 @@ The release dates mentioned follow the format `DD-MM-YYYY`.
 
 ## [Unreleased]
 
+## [2.0.3] (Hotfix Release) - 18-05-2020
+### Fixed
+- Genius would sometimes return invalid lyrics. Retry a few times in such a case.
+
+## [2.0.2] (Hotfix Release) - 18-05-2020
+### Fixed
+- Skipping tracks with `-m` would crash.
+
+## [2.0.1] (Hotfix Release) - 18-05-2020
+### Fixed
+- `-o m4a` would always fail.
+
 ## [2.0.0] - 18-05-2020
 ### Migrating from v1.2.6 to v2.0.0
 For v2.0.0 to work correctly, you need to remove your previous `config.yml` due to
@@ -45,15 +57,14 @@ All the below changes were made as a part of #690.
   Such as `-o .mp3` is now written as `-o mp3`.
 - **[Breaking]** Search format now uses hyphen for word break instead of underscore. Such as
   `-sf "{artist} - {track_name}"` is now written as `-sf "{artist} - {track-name}"`.
-- **[Breaking]** `--write-sucessful` and `--skip` is renamed to `--write-succesful-file` and
+- **[Breaking]** `--write-successful` and `--skip` is renamed to `--write-successful-file` and
   `--skip-file` respectively.
-  Such as `-o .mp3` is now written as `-o mp3`.
 - Partial re-write and internal API refactor.
 - Enhance debug log output readability.
 - Internally adapt to latest changes made in Spotipy library.
 - Switch to `logging` + `coloredlogs` instead of `logzero`. Our loggers weren't being
   setup properly with `logzero`.
-- Simplify checking for an already track. Previously it also analyzed metadata
+- Simplify checking for an downloaded already track. Previously it also analyzed metadata
   for the already downloaded track to determine whether to overwrite the already downloaded
   track, which caused unexpected behvaiours at times.
 - Codebase is now more modular making it easier to use spotdl in python scripts.

spotdl/command_line/core.py

@@ -224,6 +224,11 @@ class Spotdl:
             quality=self.arguments["quality"],
             preftype=self.arguments["input_ext"],
         )
+        if stream is None:
+            logger.error('No matching streams found for given input format: "{}".'.format(
+                self.arguments["input_ext"]
+            ))
+            return
 
         if self.arguments["no_encode"]:
             output_extension = stream["encoding"]

spotdl/encode/encode_base.py

@@ -21,6 +21,13 @@ from spotdl.encode.exceptions import EncoderNotFoundError
 
 """
 
+_TARGET_FORMATS_FROM_ENCODING = {
+    "m4a": "mp4",
+    "mp3": "mp3",
+    "opus": "opus",
+    "flac": "flac"
+}
+
 
 class EncoderBase(ABC):
     """
@@ -44,6 +51,7 @@ class EncoderBase(ABC):
         self.encoder_path = encoder_path
         self._loglevel = loglevel
         self._additional_arguments = additional_arguments
+        self._target_formats_from_encoding = _TARGET_FORMATS_FROM_ENCODING
 
     def set_argument(self, argument):
         """
@@ -94,6 +102,14 @@ class EncoderBase(ABC):
         """
         pass
 
+    def target_format_from_encoding(self, encoding):
+        """
+        This method generates the target stream format from given
+        input encoding.
+        """
+        target_format = self._target_formats_from_encoding[encoding]
+        return target_format
+
     def re_encode_from_stdin(self, input_encoding, target_path):
         """
         This method must invoke the encoder to encode stdin to a

spotdl/encode/encoders/ffmpeg.py

@@ -76,7 +76,7 @@ class EncoderFFmpeg(EncoderBase):
             + ["-i", input_path] \
             + arguments.split() \
             + self._additional_arguments \
-            + ["-f", target_encoding] \
+            + ["-f", self.target_format_from_encoding(target_encoding)] \
             + [target_file]
 
         return command

spotdl/encode/encoders/tests/test_ffmpeg.py

@@ -48,7 +48,7 @@ class TestEncodingDefaults:
             '-acodec', 'copy',
             '-b:a', '192k',
             '-vn',
-            '-f', 'm4a',
+            '-f', 'mp4',
              target_path
         ]
         return command
@@ -112,7 +112,7 @@ class TestEncodingInDebugMode:
             '-acodec', 'copy',
             '-b:a', '192k',
             '-vn',
-            '-f', 'm4a',
+            '-f', 'mp4',
              target_path
         ]
         return command
@@ -180,7 +180,7 @@ class TestEncodingAndTrimSilence:
             '-b:a', '192k',
             '-vn',
             '-af', 'silenceremove=start_periods=1',
-            '-f', 'm4a',
+            '-f', 'mp4',
             target_path
         ]
         return command

spotdl/encode/tests/test_encode_base.py

@@ -86,3 +86,12 @@ class TestMethods:
     def test_encoder_not_found_error(self):
         with pytest.raises(EncoderNotFoundError):
             self.EncoderKid("/a/nonexistent/path", "0", [])
+
+    @pytest.mark.parametrize("encoding, target_format", [
+        ("m4a", "mp4"),
+        ("mp3", "mp3"),
+        ("opus", "opus"),
+        ("flac", "flac"),
+    ])
+    def test_target_format_from_encoding(self, encoderkid, encoding, target_format):
+        assert encoderkid.target_format_from_encoding(encoding) == target_format

spotdl/lyrics/providers/genius.py

@@ -48,14 +48,12 @@ class Genius(LyricBase):
         else:
             return response.read()
 
-    def _get_lyrics_text(self, html):
+    def _get_lyrics_text(self, paragraph):
         """
         Extracts and returns the lyric content from the provided HTML.
         """
-        soup = BeautifulSoup(html, "html.parser")
-        lyrics_paragraph = soup.find("p")
-        if lyrics_paragraph:
-            return lyrics_paragraph.get_text()
+        if paragraph:
+            return paragraph.get_text()
         else:
             raise LyricsNotFoundError(
                 "The lyrics for this track are yet to be released on Genius."
@@ -81,6 +79,7 @@ class Genius(LyricBase):
         """
         encoded_query = urllib.request.quote(query.replace(" ", "+"))
         search_url = self.base_search_url + encoded_query
+        logger.debug('Fetching Genius search results from "{}".'.format(search_url))
         metadata = self._fetch_search_page(search_url)
 
         lyric_url = None
@@ -105,6 +104,7 @@ class Genius(LyricBase):
         Returns the lyric string for the track best matching the
         given query.
         """
+        logger.debug('Fetching lyrics for the search query on "{}".'.format(query))
         try:
             lyric_url = self.best_matching_lyric_url_from_query(query)
         except LyricsNotFoundError:
@@ -123,11 +123,29 @@ class Genius(LyricBase):
         lyric_url = self.guess_lyric_url_from_artist_and_track(artist, track)
         return self.from_url(lyric_url, linesep, timeout)
 
-    def from_url(self, url, linesep="\n", timeout=None):
+    def from_url(self, url, linesep="\n", retries=5, timeout=None):
         """
         Returns the lyric string for the given URL.
         """
+        logger.debug('Fetching lyric text from "{}".'.format(url))
         lyric_html_page = self._fetch_url_page(url, timeout=timeout)
-        lyrics = self._get_lyrics_text(lyric_html_page)
+        soup = BeautifulSoup(lyric_html_page, "html.parser")
+        paragraph = soup.find("p")
+        # If <p> has a class (like <p class="bla">), then we got an invalid
+        # response. Retry in such a case.
+        invalid_response = paragraph.get("class") is not None
+        to_retry = retries > 0 and invalid_response
+        if to_retry:
+            logger.debug(
+                "Retrying since Genius returned invalid response for search "
+                "results. Retries left: {retries}.".format(retries=retries)
+            )
+            return self.from_url(url, linesep=linesep, retries=retries-1, timeout=timeout)
+
+        if invalid_response:
+            raise LyricsNotFoundError(
+                'Genius returned invalid response for the search URL "{}".'.format(url)
+            )
+        lyrics = self._get_lyrics_text(paragraph)
         return lyrics.replace("\n", linesep)
 

spotdl/metadata/providers/youtube.py

@@ -53,7 +53,7 @@ class YouTubeSearch:
         quoted_query = urllib.request.quote(query)
         return self.base_search_url.format(quoted_query)
 
-    def _fetch_response_html(self, url, retries=5):
+    def _fetch_response_html(self, url):
         response = urllib.request.urlopen(url)
         soup = BeautifulSoup(response.read(), "html.parser")
         return soup
@@ -119,12 +119,11 @@ class YouTubeSearch:
         videos = self._fetch_search_results(html, limit=limit)
         to_retry = retries > 0 and self._is_server_side_invalid_response(videos, html)
         if to_retry:
-            retries -= 1
             logger.debug(
                 "Retrying since YouTube returned invalid response for search "
                 "results. Retries left: {retries}.".format(retries=retries)
             )
-            return self.search(query, limit=limit, retries=retries)
+            return self.search(query, limit=limit, retries=retries-1)
         return YouTubeVideos(videos)
 
 
@@ -138,6 +137,7 @@ class YouTubeStreams(StreamsBase):
         self.all = []
 
         for stream in audiostreams:
+            encoding = "m4a" if "mp4a" in stream.audio_codec else stream.audio_codec
             standard_stream = {
                 # Store only the integer part for bitrate. For example
                 # the given bitrate would be "192kbps", we store only
@@ -145,7 +145,7 @@ class YouTubeStreams(StreamsBase):
                 "bitrate": int(stream.abr[:-4]),
                 "connection": None,
                 "download_url": stream.url,
-                "encoding": stream.audio_codec,
+                "encoding": encoding,
                 "filesize": None,
             }
             establish_connection = threading.Thread(

spotdl/metadata_search.py

@@ -163,7 +163,7 @@ class MetadataSearch:
         if video is None:
             raise NoYouTubeVideoMatchError(
                 'No matching videos found on YouTube for the search query "{}".'.format(
-                    search_query
+                    query
                 )
             )
         return video

spotdl/version.py

@@ -1,2 +1,2 @@
-__version__ = "2.0.0"
+__version__ = "2.0.3"