def get_state_name(state_code):
""" Return long name of the state, None if the code doesn't exist """
# codes - list of dicts with keys Code and Name
codes = reader.load_json_data("Data/state_codes.json")
result = [state["Name"] for state in codes if state["Code"] == state_code]
return None if len(result) == 0 else result[0]
def filter_states_without_language(states, language):
""" Return list of states without the language as official language """
states = [state['country'] for state in states]
states_with_langs = reader.load_json_data(reader.LANGUAGE_FILE_NAME)
states_by_language = alter_country_by_languages(states_with_langs)
return [
state for state in states if state in states_by_language
and language not in states_by_language[state]
]
def alter_top_artists_JSON(states):
""" Collect and return all data about top artists into one dictionary:
key - state, values - top artists of the state
"""
result = {}
for state in states:
top_artists = reader.load_json_data(reader.TOP_ARTISTS_DIR + state +
".json")
result[state] = top_artists
return result
def make_graph():
print("Making the graph")
data = reader.load_json_data(reader.FINAL_DATA)
values = [state[1] for state in data]
countries = [state[0] for state in data]
indices = [i for i in range(len(countries))]
width = 0.8
plt.bar(indices, values, width=width, align="center")
plt.ylabel('Percent of artists')
plt.title(
'The countries in which the most listened artists are from the country'
)
plt.xticks(indices, countries, rotation=30)
plt.tight_layout()
plt.savefig("output.png")
plt.show()
def analyze_data(states):
""" Find the states with the most listened artists that are from the country
(if there are more then 8 of them in the state)
and save the data into data.json as a list of lists: [country, number of artists]
"""
print("Analyzing data")
# Number of artists listened in their country among the top 100
number_of_artists_in_their_country = {}
for country in states:
artists_ids = reader.load_json_data(reader.TOP_ARTISTS_DIR + country +
".json")
count = data_query.count_artists_from_country(artists_ids, country)
number_of_artists_in_their_country[country] = count
sorted_countries = sorted(number_of_artists_in_their_country.items(),
reverse=True,
key=lambda x: x[1])
filtered_countries = [state for state in sorted_countries if state[1] > 8]
reader.save_json_data(filtered_countries, reader.FINAL_DATA)
def produce_artists_country(states):
""" Create Data/ArtistsCountry.json
representing the most popular artists and their country """
# the wanted dict of artists and their countries;
# in case there is some data stored already
artists_countries = reader.load_json_data("Data/ArtistsCountry.json")
top_artists = data_query.alter_top_artists_JSON(states)
for country in states:
for artist in top_artists[country]:
if artist not in artists_countries.keys():
print("Adding " + artist + " who is often listened in " +
country)
artist_info = api_query.MB_API_get_artist_info(artist)
if artist_info is not None:
# sometimes it does not corespondent even if it should
# e.g. http://musicbrainz.org/ws/2/artist/7d2f5f88-5da7-4d94-8be4-b4fadaf5b3f6
if artist_info['id'] != artist:
artist_info['id'] = artist
data_query.connect_artist_and_country(
artist_info, artists_countries)
reader.save_json_data(artists_countries,
"Data/ArtistsCountry.json")
def get_country_of_artist(artist):
artists_country = reader.load_json_data("Data/ArtistsCountry.json")
return artists_country[artist] if artist in artists_country else "Unknown"
def get_initial_states():
""" Return list of strings of wanted states """
states = reader.load_json_data("Data/country-by-population.json")
states = data_query.filter_states_by_population_limit(states, 5_000_000)
states = data_query.filter_states_without_language(states, "English")
return states