def request_to_parameters(self, criticize_type, track, request_parameters):
if criticize_type != TrackCriticizeType.Parameter and not track:
raise Exception("Target track is None")
parameters = []
if criticize_type == TrackCriticizeType.Parameter:
if "bpm" in request_parameters:
parameters.append(Parameter("bpm", request_parameters["bpm"]))
elif criticize_type == TrackCriticizeType.Pattern:
criticize = CriticizePattern(pattern=request_parameters["pattern"])
direction = criticize.get_direction()
parameters.append(Parameter("pattern", criticize.pattern))
for c in criticize.get_targets():
parameters.append(Parameter(c.name, vector_utils.to_value(getattr(track, c.name)), direction))
elif criticize_type == TrackCriticizeType.Like:
for c in self.__get_criticizables():
parameters.append(Parameter(c, vector_utils.to_value(getattr(track, c))))
if "q" in request_parameters and request_parameters["q"]:
parameters.append(Parameter("q", request_parameters["q"]))
if Parameter.first_or_default("genre", parameters) and not Parameter.first_or_default("genre_score", parameters):
genre = Parameter.first_or_default("genre", parameters)
if Track.genre_to_score(genre):
g_score = Track.genre_to_score(genre)
parameters.append(Parameter("genre_score", g_score))
return parameters
def __get_track(cls, track_id, tracks):
if not track_id:
Exception("Track id is not defined")
track = None
track = next(iter(filter(lambda t: t.id == track_id, tracks)), None)
if track is None:
track = Track.find_by_id(track_id)
return track
def get_scored_tracks(cls, parameters, track, initial_tracks):
tracks = []
trial_count = 0
finder = Track()
base_track = track
pa = ParameterAdapter()
conditions = pa.parameters_to_conditions(parameters)
while trial_count < RecommendApi.TRACK_TRIAL_LIMIT and len(tracks) <= RecommendApi.TRACK_COUNT_BASE:
try:
# get tracks by criticizes
if len(tracks) > 0:
conditions["offset"] = len(tracks)
if trial_count == 0:
tracks += initial_tracks
new_tracks = finder.find(conditions)
tracks += list(filter(lambda t: t.id not in [t.id for t in tracks], new_tracks))
# filter by inputed parameters
if track:
tracks = list(filter(lambda t: pa.filter_by_parameters(parameters, track, t), tracks))
except HTTPError as ex:
pass
trial_count += 1
sleep(0.5)
scored = tracks
if len(tracks) > 0:
if track is None:
base_track = tracks[0]
evaluator = Track.make_evaluator(TrackCriticizePattern)
scored = evaluator.calc_score(tracks, base_track)
scored = scored[:RecommendApi.TRACK_COUNT_BASE]
return scored
def get_default_conditions(cls):
default_condition = {}
base_date = datetime.now() - timedelta(days=730)
default_condition["created_at"] = {"from": base_date.strftime("%Y-%m-%d %H:%M:%S")}
random_score = random.uniform(-1, 1)
random_genres = Track.score_to_genres(random_score, 1)
default_condition["genres"] = ",".join(random_genres)
default_condition["filter"] = "streamable"
return default_condition
def parameters_to_conditions(self, parameters):
default_condition = self.get_default_conditions()
c_dict = {}
is_condition_set = False
for key in self.__get_conditionables():
p = Parameter.first_or_default(key, parameters)
if p is not None and p.value is not None:
is_condition_set = True
if key == "bpm":
value = int(p.value)
if p.direction == CriticizeDirection.Up:
c_dict[key] = {"from": self.adjust_parameter(key, value, True)}
elif p.direction == CriticizeDirection.Down:
c_dict[key] = {"to": self.adjust_parameter(key, value, False)}
else:
c_dict[key] = {"from": self.adjust_parameter(key, value, False),
"to": self.adjust_parameter(key, value, True)}
elif key == "genre_score":
if p.direction == CriticizeDirection.Around:
c_dict[key] = p.value
else:
c_dict[key] = self.adjust_parameter(key, p.value, p.direction == CriticizeDirection.Up)
elif key == "created_at":
value = self.adjust_parameter(key, p.value, p.direction == CriticizeDirection.Up)
c_dict[key] = {"from": value.strftime("%Y-%m-%d %H:%M:%S")}
else:
c_dict[key] = p.value
if "genre_score" in c_dict:
genre = Track.score_to_genre(c_dict["genre_score"])
c_dict["genres"] = genre
c_dict.pop("genre_score")
elif "genre" in c_dict:
c_dict["genres"] = c_dict["genre"]
if "genre" in c_dict:
c_dict.pop("genre")
if is_condition_set:
c_dict["filter"] = default_condition["filter"]
else:
c_dict = default_condition
return c_dict
def get_criticize_pattern(cls, request):
if request.method == "POST":
posted = json.loads(request.body.decode("utf-8"))
track_id = posted["track_id"]
tracks = cls.__get_session_tracks(request)
track = cls.__get_track(track_id, tracks)
evaluator = Track.make_evaluator()
criticize_patterns = evaluator.make_pattern(tracks, track)
questions = TrackCriticizePattern.patterns_to_questions(criticize_patterns, track, tracks)
serialized_question = json.dumps(questions)
return HttpResponse(serialized_question, content_type="application/json")
else:
raise Exception("You have to use POST method when access the get_criticize_pattern.")
def test_criticize_by_pattern(self):
pa = ParameterAdapter()
selected_track = random.sample(self.tracks, 1)[0]
# by parameter
criticize_type = TrackCriticizeType.Pattern
evaluator = Track.make_evaluator()
criticize_patterns = evaluator.make_pattern(self.tracks, selected_track)
pattern = random.sample(criticize_patterns, 1)[0]
post_parameters = {"pattern": pattern.pattern}
parameters = pa.request_to_parameters(criticize_type, selected_track, post_parameters)
print(map(lambda p: p.__str__(), parameters))
scored = RecommendApi.get_scored_tracks(parameters, selected_track, self.tracks)
print("tracks: {0}".format(len(scored)))
self.print_tracks(map(lambda s: s.item, scored[:10]))
def get_favorite_tracks(cls, parameters, track, initial_tracks):
if track is None:
Exception("If getting favorite, you have to set track parameter")
tracks = []
favoriters = []
trial_count = 0
pa = ParameterAdapter()
user_evaluator = User.make_evaluator()
while trial_count < RecommendApi.TRACK_TRIAL_LIMIT and len(tracks) <= RecommendApi.TRACK_COUNT_BASE:
try:
# get tracks by criticizes
if trial_count == 0:
favoriters = track.get_favoriters()
if len(favoriters) > 0:
favoriters = user_evaluator.calc_score(favoriters, favoriters[0])
else:
break
if len(favoriters) > trial_count:
new_tracks = favoriters[trial_count].item.get_favorites()
tracks += list(filter(lambda t: t.id not in [t.id for t in tracks], new_tracks))
tracks = list(filter(lambda t: pa.filter_by_parameters(parameters, track, t), tracks))
except HTTPError as ex:
pass
trial_count += 1
sleep(0.5)
scored = []
if len(tracks) > 0:
evaluator = Track.make_evaluator(TrackCriticizePattern)
scored = evaluator.calc_score(tracks, track)
else:
scored = cls.get_scored_tracks(parameters, track, tracks)
scored = scored[:RecommendApi.TRACK_COUNT_BASE]
return scored