Merge pull request #1778 from lcharlick/fetchart-google

Fetchart plugin: new google custom search engine backend

beetsplug/fetchart.py

@@ -96,8 +96,9 @@ class RequestMixin(object):
 # ART SOURCES ################################################################
 
 class ArtSource(RequestMixin):
-    def __init__(self, log):
+    def __init__(self, log, config):
         self._log = log
+        self._config = config
 
     def get(self, album):
         raise NotImplementedError()
@@ -157,6 +158,41 @@ class AlbumArtOrg(ArtSource):
             self._log.debug(u'no image found on page')
 
 
+class GoogleImages(ArtSource):
+    URL = u'https://www.googleapis.com/customsearch/v1'
+
+    def get(self, album):
+        """Return art URL from google custom search engine
+        given an album title and interpreter.
+        """
+        if not (album.albumartist and album.album):
+            return
+        search_string = (album.albumartist + ',' + album.album).encode('utf-8')
+        response = self.request(self.URL, params={
+            'key': self._config['google_key'].get(),
+            'cx': self._config['google_engine'].get(),
+            'q': search_string,
+            'searchType': 'image'
+        })
+
+        # Get results using JSON.
+        try:
+            data = response.json()
+        except ValueError:
+            self._log.debug(u'google: error loading response: {}'
+                            .format(response.text))
+            return
+
+        if 'error' in data:
+            reason = data['error']['errors'][0]['reason']
+            self._log.debug(u'google fetchart error: {0}', reason)
+            return
+
+        if 'items' in data.keys():
+            for item in data['items']:
+                yield item['link']
+
+
 class ITunesStore(ArtSource):
     # Art from the iTunes Store.
     def get(self, album):
@@ -361,7 +397,7 @@ class FileSystem(ArtSource):
 # Try each source in turn.
 
 SOURCES_ALL = [u'coverart', u'itunes', u'amazon', u'albumart',
-               u'wikipedia']
+               u'wikipedia', u'google']
 
 ART_SOURCES = {
     u'coverart': CoverArtArchive,
@@ -369,6 +405,7 @@ ART_SOURCES = {
     u'albumart': AlbumArtOrg,
     u'amazon': Amazon,
     u'wikipedia': Wikipedia,
+    u'google': GoogleImages,
 }
 
 # PLUGIN LOGIC ###############################################################
@@ -387,7 +424,10 @@ class FetchArtPlugin(plugins.BeetsPlugin, RequestMixin):
             'cautious': False,
             'cover_names': ['cover', 'front', 'art', 'album', 'folder'],
             'sources': ['coverart', 'itunes', 'amazon', 'albumart'],
+            'google_key': None,
+            'google_engine': u'001442825323518660753:hrh5ch1gjzm',
         })
+        self.config['google_key'].redact = True
 
         # Holds paths to downloaded images between fetching them and
         # placing them in the filesystem.
@@ -405,10 +445,14 @@ class FetchArtPlugin(plugins.BeetsPlugin, RequestMixin):
         available_sources = list(SOURCES_ALL)
         if not HAVE_ITUNES and u'itunes' in available_sources:
             available_sources.remove(u'itunes')
+        if not self.config['google_key'].get() and \
+                u'google' in available_sources:
+            available_sources.remove(u'google')
         sources_name = plugins.sanitize_choices(
             self.config['sources'].as_str_seq(), available_sources)
-        self.sources = [ART_SOURCES[s](self._log) for s in sources_name]
-        self.fs_source = FileSystem(self._log)
+        self.sources = [ART_SOURCES[s](self._log, self.config)
+                        for s in sources_name]
+        self.fs_source = FileSystem(self._log, self.config)
 
     # Asynchronous; after music is added to the library.
     def fetch_art(self, session, task):

docs/plugins/fetchart.rst

@@ -50,12 +50,18 @@ file. The available options are:
 - **sources**: List of sources to search for images. An asterisk `*` expands
   to all available sources.
   Default: ``coverart itunes amazon albumart``, i.e., everything but
-  ``wikipedia``. Enable those two sources for more matches at
+  ``wikipedia`` and ``google``. Enable those two sources for more matches at
   the cost of some speed.
+- **google_key**: Your Google API key (to enable the Google Custom Search
+  backend).
+  Default: None.
+- **google_engine**: The custom search engine to use.
+  Default: The `beets custom search engine`_, which searches the entire web.
 
 Note: ``minwidth`` and ``enforce_ratio`` options require either `ImageMagick`_
 or `Pillow`_.
 
+.. _beets custom search engine: https://cse.google.com.au:443/cse/publicurl?cx=001442825323518660753:hrh5ch1gjzm
 .. _Pillow: https://github.com/python-pillow/Pillow
 .. _ImageMagick: http://www.imagemagick.org/
 
@@ -137,6 +143,24 @@ Once the library is installed, the plugin will use it to search automatically.
 .. _python-itunes: https://github.com/ocelma/python-itunes
 .. _pip: http://pip.openplans.org/
 
+Google custom search
+''''''''''''''''''''
+
+To use the google image search backend you need to
+`register for a Google API key`_. Set the ``google_key`` configuration
+option to your key, then add ``google`` to the list of sources in your
+configuration.
+
+.. _register for a Google API key: https://code.google.com/apis/console.
+
+Optionally, you can `define a custom search engine`_. Get your search engine's
+token and use it for your ``google_engine`` configuration option. The
+default engine searches the entire web for cover art.
+
+.. _define a custom search engine: http://www.google.com/cse/all
+
+Note that the Google custom search API is limited to 100 queries per day.
+After that, the fetchart plugin will fall back on other declared data sources.
 
 Embedding Album Art
 -------------------

test/test_art.py

@@ -65,13 +65,13 @@ class FetchImageTest(UseThePlugin):
         self.assertNotEqual(artpath, None)
 
 
-class FSArtTest(_common.TestCase):
+class FSArtTest(UseThePlugin):
     def setUp(self):
         super(FSArtTest, self).setUp()
         self.dpath = os.path.join(self.temp_dir, 'arttest')
         os.mkdir(self.dpath)
 
-        self.source = fetchart.FileSystem(logger)
+        self.source = fetchart.FileSystem(logger, self.plugin.config)
 
     def test_finds_jpg_in_directory(self):
         _common.touch(os.path.join(self.dpath, 'a.jpg'))
@@ -190,13 +190,13 @@ class CombinedTest(UseThePlugin):
         self.assertEqual(len(responses.calls), 0)
 
 
-class AAOTest(_common.TestCase):
+class AAOTest(UseThePlugin):
     ASIN = 'xxxx'
     AAO_URL = 'http://www.albumart.org/index_detail.php?asin={0}'.format(ASIN)
 
     def setUp(self):
         super(AAOTest, self).setUp()
-        self.source = fetchart.AlbumArtOrg(logger)
+        self.source = fetchart.AlbumArtOrg(logger, self.plugin.config)
 
     @responses.activate
     def run(self, *args, **kwargs):
@@ -226,6 +226,41 @@ class AAOTest(_common.TestCase):
         self.assertEqual(list(res), [])
 
 
+class GoogleImageTest(UseThePlugin):
+    def setUp(self):
+        super(GoogleImageTest, self).setUp()
+        self.source = fetchart.GoogleImages(logger, self.plugin.config)
+
+    @responses.activate
+    def run(self, *args, **kwargs):
+        super(GoogleImageTest, self).run(*args, **kwargs)
+
+    def mock_response(self, url, json):
+        responses.add(responses.GET, url, body=json,
+                      content_type='application/json')
+
+    def test_google_art_finds_image(self):
+        album = _common.Bag(albumartist="some artist", album="some album")
+        json = b'{"items": [{"link": "url_to_the_image"}]}'
+        self.mock_response(fetchart.GoogleImages.URL, json)
+        result_url = self.source.get(album)
+        self.assertEqual(list(result_url)[0], 'url_to_the_image')
+
+    def test_google_art_returns_no_result_when_error_received(self):
+        album = _common.Bag(albumartist="some artist", album="some album")
+        json = b'{"error": {"errors": [{"reason": "some reason"}]}}'
+        self.mock_response(fetchart.GoogleImages.URL, json)
+        result_url = self.source.get(album)
+        self.assertEqual(list(result_url), [])
+
+    def test_google_art_returns_no_result_with_malformed_response(self):
+        album = _common.Bag(albumartist="some artist", album="some album")
+        json = b"""bla blup"""
+        self.mock_response(fetchart.GoogleImages.URL, json)
+        result_url = self.source.get(album)
+        self.assertEqual(list(result_url), [])
+
+
 class ArtImporterTest(UseThePlugin):
     def setUp(self):
         super(ArtImporterTest, self).setUp()