%%ruby --out socrata_data

# with --out, data written to the stdout in this ruby cell 
# will be mapped to a Python variable (socrata_data) after execution.

require 'soda/client'
require 'json'      

# Set up client object with domain and application token
client = SODA::Client.new({:domain => "opendata.socrata.com", 
                           :app_token => "eqZC5q2iEmFXdIu2qEbtZkWgP"})

# Get data with dataset identifier
response = client.get("ed74-c6ni")

# Print dataset to stdout as a JSON 
puts response.to_json

import pandas as pd

# Read the retrieved JSON dataset (df stands for dataframe)
df = pd.read_json(socrata_data)

df.head()  # print the first 5 lines of the dataframe

df.shape  # print the dataframe's size

song_by_artist = df.groupby('artist').size().to_dict()

song_by_artist

song_by_artist_4plus = {k:v for k,v in song_by_artist.items() if v>=4}

song_by_artist_4plus

import numpy as np  

# Lists of keys and values
my_keys = song_by_artist_4plus.keys()
my_vals = song_by_artist_4plus.values()

# Find indices of sorted values (first converted to a numpy array)
i_sorted = np.argsort(np.array(my_vals))[::-1]

# Sort both the keys and value list 
my_keys_sorted = [my_keys[i] for i in i_sorted]
my_vals_sorted = [my_vals[i] for i in i_sorted]

import plotly.plotly as py  

from plotly.graph_objs import Figure, Data, Layout
from plotly.graph_objs import Bar
from plotly.graph_objs import XAxis, YAxis, Marker, Font, Margin

my_bar = Bar(x=my_keys_sorted,  # labels of the x-axis
             y=my_vals_sorted,  # values of the y-axis
             marker= Marker(color='#2ca02c'))  # a nice green color

my_data = Data([my_bar])  # make data object, (Data accepts only list)

my_title = 'Number of songs listed in the Guardian\'s<br>\
<em>Top 1,000 Songs to Hear Before You Die</em> per artist with 4 or more songs'
my_ytitle = 'Number of songs per artist'

my_layout = Layout(title=my_title,   # set plot title
                   showlegend=False, # remove legend 
                   font= Font(family='Georgia, serif', # set global font family
                              color='#635F5D'),        #   and color 
                   plot_bgcolor='#EFECEA',   # set plot color to grey
                   xaxis= XAxis(title='',             # no x-axis title
                                tickangle=45,         # tick labels' angle
                                ticks='outside',      # draw ticks outside axes 
                                ticklen=8,            # tick length
                                tickwidth=1.5,),       #   and width, 
                   yaxis= YAxis(title=my_ytitle,      # y-axis title
                                gridcolor='#FFFFFF',  # white grid lines
                                ticks='outside',     
                                ticklen=8,           
                                tickwidth=1.5),
                   autosize=False,  # manual figure size
                   width=700,       
                   height=500,
                   margin= Margin(b=140)  # increase bottom margin, 
                  )                       # to fit long x-axis tick labels

my_fig = Figure(data=my_data, layout=my_layout)

py.iplot(my_fig, filename='socrata1')

# Rows which have 'artist' name in song_by_artist_4plus
i_good = (df['artist'].isin(song_by_artist_4plus))  

df_good = df[i_good]  # a new dataframe

df_good.shape   # a much smaller dataframe than the original

my_text = []  # init. the hover-text list

# Loop through the sorted artist names, so that my_text
# will have to same ordering as the values linked to 'x' and 'y' in my_data
for k in my_keys_sorted:
    
    # Slice dataframe to artist name and sort songs by year
    i_artist = (df['artist']==k)
    df_tmp = df_good[i_artist].sort(columns='year')
    
    my_text_tmp = ''  # init. string 
    cnt_song = 0                   # song counter for given artist
    N_song = len(df_tmp['title'])  # total number of song for given artist
    
    # Loop through songs
    for i_song, song in df_tmp.iterrows():
        
        # Add to string and counter
        my_text_tmp += song['title']+' ('+str(song['year'])+')<br>'
        cnt_song += 1
        
        # Skip if song list is too long to fit on figure
        if cnt_song>12:
            diff = N_song - cnt_song
            my_text_tmp += ' and '+str(diff)+' more ...'
            break
    
    # Append hover-text list
    my_text += [my_text_tmp]
    
# Update figure object 
my_fig['data'][0].update(text=my_text)

from plotly.graph_objs import Annotation

my_anno_text = '<em>Open Data by Socrata</em><br>\
Hover over the bars to see list of songs'

my_anno = Annotation(text=my_anno_text,  # annotation text
                     x=0.95,  # position's x-coord
                     y=0.95,  #   and y-coord
                     xref='paper',  # use paper coords
                     yref='paper',  #  for both coordinates
                     font= Font(size=14),  # increase font size (default is 12)
                     showarrow=False,      # remove arrow 
                     bgcolor='#FFFFFF',      # white background
                     borderpad=4)       # space bt. border and text (in px)

# Update figure object
my_fig['layout'].update(annotations=[my_anno])

py.iplot(my_fig, filename='socrata1-hover')

# CSS styling within IPython notebook
from IPython.core.display import HTML
import urllib2
def css_styling():
    url = 'https://raw.githubusercontent.com/plotly/python-user-guide/master/custom.css'
    styles = urllib2.urlopen(url).read()
    return HTML(styles)

css_styling()