from IPython.display import Image
Image(url='http://i.imgur.com/h5GLIHc.gif')

import pandas as pd
import datetime

from IPython.html import widgets 
from IPython.display import display, clear_output

import plotly.plotly as py
from plotly.graph_objs import *
import plotly
from plotly.widgets import GraphWidget

df = pd.read_csv('https://raw.githubusercontent.com/plotly/widgets/master/ipython-examples/311_150k.csv', parse_dates=True, index_col=1)
df = df
df.head()

grouped = df.resample('24H', how='count')

slider = widgets.IntSliderWidget()
slider.min = 1
slider.value=24
slider.description = 'Window length (hours)'

column_headers_dropdown = widgets.DropdownWidget()
column_headers_dropdown.values  = {column: column for column in df.columns}
column_headers_dropdown.value = 'Complaint Type'
column_headers_dropdown.description = 'Select which column to graph (e.g. Tree)'


search_complaints_text_input = widgets.TextWidget()
search_complaints_text_input.description = 'Search complaint types'

graph_time_series_agg = GraphWidget('https://plot.ly/~chris/4103')
graph_time_series_agg.ranges = {'x': [1413648000000, 1416157200000], 'y': [0, 7744.210526315789]}
graph_complaints = GraphWidget('https://plot.ly/~chris/4103')

def replot(on_zoom=False):
    search_text = search_complaints_text_input.value
    slider_val = slider.value
    column_name = column_headers_dropdown.value

    # Time window that we're looking at
    ranges = graph_time_series_agg.ranges
    xr = ranges['x']
    x1 = datetime.datetime.fromtimestamp(int(xr[0]/1000.))
    x2 = datetime.datetime.fromtimestamp(int(xr[1]/1000.))

    # Text value that we're looking at
    search_idx = df['Complaint Type'].str.contains(search_complaints_text_input.value).fillna(False)
    windowed_idx = search_idx & (df.index>x1) & (df.index<x2)

    if not on_zoom:
        # Replot the time series aggregate on input change
        grouped = df[search_idx].resample('{}H'.format(slider_val), how='count')

        graph_time_series_agg.restyle({
            'x': [grouped['Complaint Type'].index],
            'y': [grouped['Complaint Type']],
            'type': 'bar'
        })

    
    # Replot the value counts aggregate
    vc = df[windowed_idx][column_name].value_counts()
    
    graph_complaints.restyle({'x': [vc.index], 'y': [vc.values], 'type': 'bar'})

def on_trait_change():
    replot()

search_complaints_text_input.on_trait_change(on_trait_change, 'value')
column_headers_dropdown.on_trait_change(on_trait_change, 'value')
slider.on_trait_change(on_trait_change, 'value')

def replot_complaints_on_zoom(_, ranges):
    graph_time_series_agg.ranges = ranges

    replot(True)

graph_time_series_agg.on_zoom(replot_complaints_on_zoom)

display(slider)
display(search_complaints_text_input)
display(column_headers_dropdown)

display(graph_time_series_agg)
display(graph_complaints)

replot()

graph_time_series_agg.relayout({'title': 'Number of complaints over time'})
graph_complaints.relayout({'yaxis.title': 'Number of events'})

# CSS styling within IPython notebook - feel free to re-use
from IPython.core.display import HTML
import urllib2

HTML(urllib2.urlopen('http://bit.ly/1Bf5Hft').read())