Merge pull request #3519 from xhocquet/tekstowo-lyrics

Implement tekstowo lyrics provider

beetsplug/lyrics.py

@@ -401,6 +401,56 @@ class Genius(Backend):
         return lyrics_div.get_text()
 
 
+class Tekstowo(Backend):
+    # Fetch lyrics from Tekstowo.pl.
+
+    BASE_URL = 'http://www.tekstowo.pl'
+    URL_PATTERN = BASE_URL + '/wyszukaj.html?search-title=%s&search-artist=%s'
+
+    def fetch(self, artist, title):
+        url = self.build_url(title, artist)
+        search_results = self.fetch_url(url)
+        song_page_url = self.parse_search_results(search_results)
+
+        if not song_page_url:
+            return None
+
+        song_page_html = self.fetch_url(song_page_url)
+        return self.extract_lyrics(song_page_html)
+
+    def parse_search_results(self, html):
+        if not HAS_BEAUTIFUL_SOUP:
+            return None
+
+        html = _scrape_strip_cruft(html)
+        html = _scrape_merge_paragraphs(html)
+
+        try:
+            html = BeautifulSoup(html, "html.parser")
+        except HTMLParseError:
+            return None
+
+        song_row = html.find("div", class_="content"). \
+            find_all("div", class_="box-przeboje")[0]
+
+        if not song_row:
+            return None
+
+        href = song_row.find('a').get('href')
+        return self.BASE_URL + href
+
+    def extract_lyrics(self, html):
+        html = _scrape_strip_cruft(html)
+        html = _scrape_merge_paragraphs(html)
+
+        try:
+            html = BeautifulSoup(html, "html.parser")
+        except HTMLParseError:
+            return None
+
+        return html.find("div", class_="song-text").get_text()
+
+
 def remove_credits(text):
     """Remove first/last line of text if it contains the word 'lyrics'
     eg 'Lyrics by songsdatabase.com'
@@ -593,11 +643,13 @@ class Google(Backend):
 
 
 class LyricsPlugin(plugins.BeetsPlugin):
-    SOURCES = ['google', 'musixmatch', 'genius']
+    SOURCES = ['google', 'musixmatch', 'genius', 'tekstowo']
+    BS_SOURCES = ['google', 'genius', 'tekstowo']
     SOURCE_BACKENDS = {
         'google': Google,
         'musixmatch': MusiXmatch,
         'genius': Genius,
+        'tekstowo': Tekstowo,
     }
 
     def __init__(self):
@@ -636,6 +688,9 @@ class LyricsPlugin(plugins.BeetsPlugin):
         sources = plugins.sanitize_choices(
             self.config['sources'].as_str_seq(), available_sources)
 
+        if not HAS_BEAUTIFUL_SOUP:
+            sources = self.sanitize_bs_sources(sources)
+
         if 'google' in sources:
             if not self.config['google_API_key'].get():
                 # We log a *debug* message here because the default
@@ -645,18 +700,6 @@ class LyricsPlugin(plugins.BeetsPlugin):
                 self._log.debug(u'Disabling google source: '
                                 u'no API key configured.')
                 sources.remove('google')
-            elif not HAS_BEAUTIFUL_SOUP:
-                self._log.warning(u'To use the google lyrics source, you must '
-                                  u'install the beautifulsoup4 module. See '
-                                  u'the documentation for further details.')
-                sources.remove('google')
-
-        if 'genius' in sources and not HAS_BEAUTIFUL_SOUP:
-            self._log.debug(
-                u'The Genius backend requires BeautifulSoup, which is not '
-                u'installed, so the source is disabled.'
-            )
-            sources.remove('genius')
 
         self.config['bing_lang_from'] = [
             x.lower() for x in self.config['bing_lang_from'].as_str_seq()]
@@ -670,6 +713,17 @@ class LyricsPlugin(plugins.BeetsPlugin):
         self.backends = [self.SOURCE_BACKENDS[source](self.config, self._log)
                          for source in sources]
 
+    def sanitize_bs_sources(self, sources):
+        for source in self.BS_SOURCES:
+            if source in sources:
+                self._log.debug(u'To use the %s lyrics source, you must '
+                                u'install the beautifulsoup4 module. See '
+                                u'the documentation for further details.'
+                                % source)
+                sources.remove(source)
+
+        return sources
+
     def get_bing_access_token(self):
         params = {
             'client_id': 'beets',

docs/changelog.rst

@@ -288,6 +288,8 @@ Fixes:
 * Removed ``@classmethod`` decorator from dbcore.query.NoneQuery.match method
   failing with AttributeError when called. It is now an instance method.
   :bug:`3516` :bug:`3517`
+* :doc:`/plugins/lyrics`: Added Tekstowo.pl lyrics provider
+  :bug:`3344`
 * :doc:`/plugins/lyrics`: Tolerate missing lyrics div in Genius scraper.
   Thanks to :user:`thejli21`.
   :bug:`3535` :bug:`3554`

docs/plugins/lyrics.rst

@@ -3,10 +3,11 @@ Lyrics Plugin
 
 The ``lyrics`` plugin fetches and stores song lyrics from databases on the Web.
 Namely, the current version of the plugin uses `Musixmatch`_, `Genius.com`_,
-and, optionally, the Google custom search API.
+`Tekstowo.pl`_, and, optionally, the Google custom search API.
 
 .. _Musixmatch: https://www.musixmatch.com/
 .. _Genius.com: https://genius.com/
+.. _Tekstowo.pl: https://www.tekstowo.pl/
 
 
 Fetch Lyrics During Import
@@ -58,11 +59,11 @@ configuration file. The available options are:
   sources known to be scrapeable.
 - **sources**: List of sources to search for lyrics. An asterisk ``*`` expands
   to all available sources.
-  Default: ``google musixmatch genius``, i.e., all the
+  Default: ``google musixmatch genius tekstowo``, i.e., all the
   available sources. The ``google`` source will be automatically
   deactivated if no ``google_API_key`` is setup.
-  Both it and the ``genius`` source will only be enabled if BeautifulSoup is
-  installed.
+  The ``google``, ``genius``, and ``tekstowo`` sources will only be enabled if
+  BeautifulSoup is installed.
 
 Here's an example of ``config.yaml``::
 
@@ -155,15 +156,15 @@ After that, the lyrics plugin will fall back on other declared data sources.
 .. _pip: https://pip.pypa.io
 .. _BeautifulSoup: https://www.crummy.com/software/BeautifulSoup/bs4/doc/
 
-Activate Genius Lyrics
-----------------------
+Activate Genius and Tekstowo.pl Lyrics
+--------------------------------------------------------------------
 
-Like the Google backend, the Genius backend requires the `BeautifulSoup`_
-library. Install it by typing::
+Using the Genius or Tekstowo.pl backends requires `BeautifulSoup`_, which
+you can install using `pip`_ by typing::
 
     pip install beautifulsoup4
 
-The backend is enabled by default.
+These backends are enabled by default.
 
 .. _lyrics-translation:
 

test/test_lyrics.py

@@ -274,6 +274,8 @@ class LyricsPluginSourcesTest(LyricsGoogleBaseTest):
         dict(DEFAULT_SONG, backend=lyrics.Genius,
              # GitHub actions is on some form of Cloudflare blacklist.
              skip=os.environ.get('GITHUB_ACTIONS') == 'true'),
+        dict(artist=u'Boy In Space', title=u'u n eye',
+             backend=lyrics.Tekstowo),
     ]
 
     GOOGLE_SOURCES = [