def create_entries_by_unique_artist(billboard_df, start_year, end_year):
billboard_df = billboard_df[(billboard_df["Year"] >= start_year) & (billboard_df["Year"] <= end_year)]
billboard_df_temp = pd.DataFrame.copy(billboard_df)
billboard_unique_artists = []
billboard_songs = []
billboard_years = []
billboard_rank = []
for index_artist, row in billboard_df_temp.iterrows():
artist = row["Artist(s)"]
title = row["Title"]
year = row["Year"]
rank = row["Num"]
if artist == "Earth, Wind & Fire & The Emotions":
billboard_unique_artists.append("Earth, Wind & Fire")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("The Emotions")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Grover Washington, Jr. & Bill Withers":
billboard_unique_artists.append("Grover Washington, Jr.")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Bill Withers")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Dionne and Friends (Dionne Warwick, Gladys Knight, Elton John and Stevie Wonder)":
billboard_unique_artists.append("Dionne Warwick")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Gladys Knight")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Elton John")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Stevie Wonder")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Allen, KrisKris Allen":
billboard_unique_artists.append("Kris Allen")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Crosby, Stills, Nash & Young":
billboard_unique_artists.append("Crosby, Stills & Nash")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Neil Sedaka & Elton John":
billboard_unique_artists.append("Neil Sedaka")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Elton John")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif artist == "Neil & Dara Sedaka":
billboard_unique_artists.append("Neil Sedaka")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Dara Sedaka")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
elif artist == "Donna Summer and Brooklyn Dreams":
billboard_unique_artists.append("Donna Summer")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Brooklyn Dreams")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
elif artist == "Donny and Marie Osmond":
billboard_unique_artists.append("Donny Osmond")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
billboard_unique_artists.append("Marie Osmond")
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
elif artist in MULTIPLE_ARTIST_LIST:
billboard_unique_artists.append(artist)
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
else:
artist_comma_splitted = artist.split(", ")
for item in artist_comma_splitted:
if (item == "") or (item == " "):
continue
elif item in MULTIPLE_ARTIST_LIST:
billboard_unique_artists.append(item)
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
item_featuring_splitted = item.split(" featuring ")
for item2 in item_featuring_splitted:
if (item2 == "") or (item2 == " "):
continue
if item2 in MULTIPLE_ARTIST_LIST:
billboard_unique_artists.append(item2)
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
continue
elif (year >= 1982) & ((" and " in item2) or (item2.startswith("and "))):
if item2.startswith("and "):
item_and_splitted = item2.split("and ")
else:
item_and_splitted = item2.split(" and ")
for item3 in item_and_splitted:
if (item3 == "") or (item3 == " "):
continue
if item3 in ARTIST_DICTIONARY:
billboard_unique_artists.append(ARTIST_DICTIONARY[item3])
else:
billboard_unique_artists.append(item3)
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
else:
if item2 in ARTIST_DICTIONARY:
billboard_unique_artists.append(ARTIST_DICTIONARY[item2])
else:
billboard_unique_artists.append(item2)
billboard_rank.append(rank)
billboard_songs.append(title)
billboard_years.append(year)
data = {"Rank": billboard_rank, "Artist(s)": billboard_unique_artists, "Title": billboard_songs, "Year": billboard_years}
unique_artist_df = pd.DataFrame(data, columns = ["Rank", "Artist(s)", "Title", "Year"])
return unique_artist_df
def getArtists():
songs = request.json['songs']
### API Key Configuration
from pyechonest import config
config.ECHO_NEST_API_KEY="YZZS9XI0IMOLQRKQ6"
###################### Getting Artists + Song Info ########################
## Basic info on the artists and song from the Echonest API based on a spotify ID sent to the
## app. This part also collects any available information on tempo, danceability and energy
## from the Echonest API for use in subsetting the recommended playlist
from pyechonest import track
artists = []
EchoNestSongId = []
EchoNestArtistId = []
energy = []
dance = []
tempo = []
for i in range(len(songs)):
songs[i] = songs[i].replace("spotify:track:", "spotify-WW:track:")
try:
t = track.track_from_id(songs[i])
EchoNestSongId.append(t.song_id)
EchoNestArtistId.append(t.id)
artists.append(t.artist)
if "energy" in vars(t):
energy.append(t.energy)
if "danceability" in vars(t):
dance.append(t.danceability)
if "tempo" in vars(t):
tempo.append(t.tempo)
except:
continue
###################### Terms Per Artist // All Terms ########################
### Building Terms from Artists
artist_dict = {}
terms = []
from pyechonest import artist
##Creates a dictionary with artists and terms associated with that artist, also creates a vector of all terms
for i in range(len(artists)):
artistTerm = artist.Artist(artists[i]).terms
find_term = []
limitTerms = 4
for artistT in artistTerm:
find_term.append(artistT['name'])
limitTerms -= 1
if limitTerms == 0:
break
artist_dict[artists[i]] = find_term;
terms.extend(find_term)
###################### Weights for Terms ########################
##Creates weights for each term based on the number of times it appeared in artist list
terms_dict = {}
for i in range(len(terms)):
if terms[i] not in terms_dict.items():
find_count = terms.count(terms[i])
terms_dict[terms[i]] = find_count
###################### Weights for Artists ########################
## Adds all of the weights per term per artist to get the overall weight for an artist
## 5 subtracted from total weight (number of terms used per artist, each guarenteed a count of 5)
weights = {}
for key in artist_dict:
getWeight=0
for value in artist_dict[key]:
if value in terms:
getWeight = getWeight+terms_dict[value]
weights[key] = getWeight
#checking weights
#for key in weights:
# print key +":" + str(weights[key])
#NOTE::: If we ever want to add more than 5 songs, we have to order the dictionary and select top 5 artists of terms.
###################### Finding Similar Artists w/Weights ########################
createURL = "http://developer.echonest.com/api/v4/artist/similar?api_key=YZZS9XI0IMOLQRKQ6"
for key in weights:
createURL += "&id=" +artist.Artist(key).id +"^" + str(weights[key])
simMake = urllib2.urlopen(createURL);
echonestSim = simMake.read();
findSimilarArtist = []
for artist in echonestSim.split(":"):
if "name" in artist:
artistShrink = artist.split(",")[0]
artistShrink= artistShrink.translate(None, string.punctuation)
artistShrink = artistShrink.strip()
findSimilarArtist.append(artistShrink)
findSimilarArtists = findSimilarArtist[1:30]
###################### Setting Limits on Audio for Song Rec ########################
### Used to find songs in range of input songs on danceability, enerygy, and tempo, and return full range
### values if there is not enough data to calculate an adjusted range
### NOTE: Currently - tempo and energy not used
if len(dance) > 1:
min_danceability = dance[dance.index(min(dance))]
max_danceability = dance[dance.index(max(dance))]
else:
min_danceability = 0
max_danceability = 1
if len(tempo) > 1:
min_tempo = tempo[tempo.index(min(tempo))]
max_tempo = tempo[tempo.index(max(tempo))]
else:
min_tempo = 0
max_tempo = 500
if len(energy) > 1:
min_energy = energy[energy.index(min(energy))]
max_energy = energy[energy.index(max(energy))]
else:
min_energy = 0
max_energy = 1
###################### Selecting Top Songs for Recommended Artists + Creates Playlist ########################
playlist = []
for i in findSimilarArtists:
createURL="http://developer.echonest.com/api/v4/song/search?api_key=YZZS9XI0IMOLQRKQ6&artist_id=" +i+ "&bucket=id:spotify-WW&bucket=tracks&sort=song_hotttnesss-desc&min_danceability=" + str(min_danceability) + "&max_danceability=" + str(max_danceability) + "&results=5"
getURL = urllib2.urlopen(createURL);
clean_page = getURL.read();
if "spotify-WW:track" in clean_page:
get_track = clean_page[clean_page.index("spotify-WW:track")-1:clean_page.index("spotify-WW:track")+60].replace(":", ",").split(",")[2].translate(None, string.punctuation)
playlist.append("spotify:track:" + get_track)
playlist = jsonify( { 'playlist': playlist} )
return(playlist)