Merge pull request #111 from metabrainz/mode-ranges

Improve tag radio, add troi.py, use similar tags, other tag improvements

README.md

@@ -39,7 +39,7 @@ below.
 virtualenv -p python3 .ve
 source .ve/bin/activate
 pip3 install -r requirements.txt -r requirements_test.txt
-python3 -m troi.cli --help
+python3 troi.py --help
 ```
 
 **Windows**
@@ -48,22 +48,22 @@ python3 -m troi.cli --help
 virtualenv -p python .ve
 .ve\Scripts\activate.bat
 pip install -r requirements.txt -r requirements_test.txt
-python -m troi.cli --help
+python troi.py --help
 ```
 
 ## Basic commands
 
 List available patches:
 
-    python -m troi.cli list
+    python troi.py list
 
 Generate a playlist using a patch:
 
-    python -m troi.cli playlist --print [patch-name]
+    python troi.py playlist --print [patch-name]
 
 If the patch requires arguments, running it with no arguments will print a usage statement, e.g.
 
-    $ python -m troi.cli playlist --print area-random-recordings
+    $ python troi.py playlist --print area-random-recordings
     Usage: area-random-recordings [OPTIONS] AREA START_YEAR END_YEAR
 
       Generate a list of random recordings from a given area.
@@ -78,9 +78,9 @@ If the patch requires arguments, running it with no arguments will print a usage
 ## Running tests
 
 ```
-python3 -m troi.cli test
-python3 -m troi.cli test -v
-python3 -m troi.cli test -v <file to test>
+python3 troi.py test
+python3 troi.py test -v
+python3 troi.py test -v <file to test>
 ```
 
 ## Building Documentation

docs/lb_radio.rst

@@ -39,14 +39,15 @@ All terms have the following options:
 #. **medium**: Use medium mode for this term.
 #. **hard**: Use hard mode for this term.
 
-For artist terms, the following option applies:
+For artist and tag terms, the following option applies:
 
-#. **nosim**: Do not add similar artists, only output recordings from the given artist.
+#. **nosim**: Do not add similar artists/tags, only output recordings from the given artist/tag.
 
 For tag queries, the following options exist:
 
 #. **and**: For a tag query, if "and" is specified (the default) recordings will be chosen if all the given tags are applied to that recording.
 #. **or**: For a tag query, if "or" is specified, then recordings will be chosen if any of the tags are applied to the recording.
+#. **nosim**: Tag queries on medium and hard mode may include similar tags. Specifying nosim for a tag query ensures that no similar tags are used.
 
 For the stats term, the following options apply:
 

troi.py

@@ -0,0 +1,8 @@
+#!/usr/bin/env python3
+
+import sys
+from troi.cli import cli
+
+if __name__ == "__main__":
+    cli()
+    sys.exit(0)

troi/patches/lb_radio.py

@@ -19,6 +19,7 @@
 from troi.patches.lb_radio_classes.stats import LBRadioStatsRecordingElement
 from troi.patches.lb_radio_classes.recs import LBRadioRecommendationRecordingElement
 from troi import TARGET_NUMBER_OF_RECORDINGS, Playlist
+from troi.utils import interleave
 
 
 class LBRadioPatch(troi.patch.Patch):
@@ -130,12 +131,22 @@ def create(self, inputs):
             if "hard" in element["opts"]:
                 mode = "hard"
 
+            # Determine percent ranges based on mode -- this will likely need further tweaking
+            if mode == "easy":
+                start, stop = 0, 33
+            elif self.mode == "medium":
+                start, stop = 33, 66
+            else:
+                start, stop = 66, 100
+            self.local_storage["modes"] = {"easy": (0, 33), "medium": (33, 66), "hard": (66, 100)}
+
             if element["entity"] == "artist":
                 include_sim = False if "nosim" in element["opts"] else True
                 source = LBRadioArtistRecordingElement(element["values"][0], mode=mode, include_similar_artists=include_sim)
 
             if element["entity"] == "tag":
-                source = LBRadioTagRecordingElement(element["values"], mode=mode, operator="and")
+                include_sim = False if "nosim" in element["opts"] else True
+                source = LBRadioTagRecordingElement(element["values"], mode=mode, operator="and", include_similar_tags=include_sim)
 
             if element["entity"] == "tag-or":
                 source = LBRadioTagRecordingElement(element["values"], mode=mode, operator="or")

troi/patches/lb_radio_classes/artist.py

@@ -4,6 +4,7 @@
 from troi import Recording, Artist
 from troi.splitter import plist
 from troi import TARGET_NUMBER_OF_RECORDINGS
+from troi.utils import interleave
 
 OVERHYPED_SIMILAR_ARTISTS = [
     "b10bbbfc-cf9e-42e0-be17-e2c3e1d2600d",  # The Beatles
@@ -16,13 +17,10 @@
     "5b11f4ce-a62d-471e-81fc-a69a8278c7da",  # Nirvana
     "f59c5520-5f46-4d2c-b2c4-822eabf53419",  # Linkin Park
     "cc0b7089-c08d-4c10-b6b0-873582c17fd6",  # System of a Down
+    "ebfc1398-8d96-47e3-82c3-f782abcdb13d",  # Beach boys
 ]
 
 
-def interleave(lists):
-    return [val for tup in zip(*lists) for val in tup]
-
-
 class LBRadioArtistRecordingElement(troi.Element):
     """
         Given an artist, find its similar artists and their popular tracks and return one 
@@ -109,22 +107,26 @@ def fetch_artist_names(self, artist_mbids):
 
     def read(self, entities):
 
+        # Fetch our mode ranges
+        start, stop = self.local_storage["modes"][self.mode]
+
         self.data_cache = self.local_storage["data_cache"]
         artists = [{"mbid": self.artist_mbid}]
 
         # First, fetch similar artists if the user didn't override that.
         if self.include_similar_artists:
             # Fetch similar artists for original artist
             similar_artists = self.get_similar_artists(self.artist_mbid)
-            if len(similar_artists) == 0:
-                raise RuntimeError("Not enough similar artist data available for artist %s. Please choose a different artist." %
-                                   self.artist_name)
+            #            if len(similar_artists) == 0:
+            #                raise RuntimeError(f"Not enough similar artist data available for artist {self.artist_name}. Please choose a different artist.")
 
-            # Verify and lookup artist mbids
-            for artist in similar_artists[:self.MAX_NUM_SIMILAR_ARTISTS]:
+            # select artists
+            for artist in similar_artists[start:stop]:
                 artists.append({"mbid": artist["artist_mbid"]})
+                if len(artists) >= self.MAX_NUM_SIMILAR_ARTISTS:
+                    break
 
-        # For all fetched artists, fetcht their names
+        # For all fetched artists, fetch their names
         artist_names = self.fetch_artist_names([i["mbid"] for i in artists])
         for artist in artists:
             if artist["mbid"] not in artist_names:
@@ -136,23 +138,23 @@ def read(self, entities):
             self.data_cache[artist["mbid"]] = artist["name"]
 
         # start crafting user feedback messages
-        msg = "artist: using seed artist %s" % artists[0]["name"]
-        if self.include_similar_artists:
-            msg += " and similar artists: " + ", ".join([a["name"] for a in artists[1:]])
+        msgs = []
+        if self.include_similar_artists and len(artists) == 1:
+            msgs.append(f"Seed artist {artist_names[self.artist_mbid]} no similar artists.")
         else:
-            msg += " only"
-
-        self.local_storage["user_feedback"].append(msg)
+            if self.include_similar_artists and len(artists) < 4:
+                msgs.append(f"Seed artist {artist_names[self.artist_mbid]} few similar artists.")
+            msg = "artist: using seed artist %s" % artists[0]["name"]
+            if self.include_similar_artists:
+                msg += " and similar artists: " + ", ".join([a["name"] for a in artists[1:]])
+            else:
+                msg += " only"
+            msgs.append(msg)
+
+        for msg in msgs:
+            self.local_storage["user_feedback"].append(msg)
         self.data_cache["element-descriptions"].append("artist %s" % artists[0]["name"])
 
-        # Determine percent ranges based on mode -- this will likely need further tweaking
-        if self.mode == "easy":
-            start, stop = 0, 50
-        elif self.mode == "medium":
-            start, stop = 25, 75
-        else:
-            start, stop = 50, 100
-
         # Now collect recordings from the artist and similar artists and return an interleaved
         # strem of recordings.
         for i, artist in enumerate(artists):
@@ -161,6 +163,10 @@ def read(self, entities):
                 continue
 
             recs_plist = plist(self.fetch_top_recordings(artist["mbid"]))
+            if len(recs_plist) < 20:
+                self.local_storage["user_feedback"].append(
+                    f"Artist {artist['name']} only has {'no' if len(recs_plist) == 0 else 'few'} top recordings.")
+
             recordings = []
             for recording in recs_plist.random_item(start, stop, self.max_top_recordings_per_artist):
                 recordings.append(recording)