Fetching contributors…
Cannot retrieve contributors at this time
511 lines (457 sloc) 27.2 KB
''' Provide a pivot chart maker example app. Similar to Excel pivot charts,
but with additonal ability to explode into multiple charts.
See for more information.
from __future__ import division
import os
import math
import json
import pandas as pd
import collections
import as bio
import bokeh.layouts as bl
import bokeh.models.widgets as bmw
import bokeh.models.sources as bms
import as bmt
import bokeh.plotting as bp
import datetime
import six.moves.urllib.parse as urlp
#Defaults to configure:
COLORS = ['#5e4fa2', '#3288bd', '#66c2a5', '#abdda4', '#e6f598', '#fee08b', '#fdae61', '#f46d43', '#d53e4f', '#9e0142']*1000
C_NORM = "#31AADE"
CHARTTYPES = ['Dot', 'Line', 'Bar', 'Area']
STACKEDTYPES = ['Bar', 'Area']
AGGREGATIONS = ['None', 'Sum']
def get_data(data_source):
Read a csv into a pandas dataframe, and determine which columns of the dataframe
are discrete (strings), continuous (numbers), able to be filtered (aka filterable),
and able to be used as a series (aka seriesable). NA values are filled based on the type of column,
and the dataframe and columns are returned.
data_source (string): Path to csv file.
df_source (pandas dataframe): A dataframe of the csv source, with filled NA values.
cols (dict): Keys are categories of columns of df_source, and values are a list of columns of that category.
df_source = pd.read_csv(data_source)
cols = {}
cols['all'] = df_source.columns.values.tolist()
cols['discrete'] = [x for x in cols['all'] if df_source[x].dtype == object]
cols['continuous'] = [x for x in cols['all'] if x not in cols['discrete']]
cols['filterable'] = cols['discrete']+[x for x in cols['continuous'] if len(df_source[x].unique()) < 100]
cols['seriesable'] = cols['discrete']+[x for x in cols['continuous'] if len(df_source[x].unique()) < 60]
df_source[cols['discrete']] = df_source[cols['discrete']].fillna('{BLANK}')
df_source[cols['continuous']] = df_source[cols['continuous']].fillna(0)
return (df_source, cols)
def build_widgets(df_source, cols, defaults, init_load=False, init_config={}):
Use a dataframe and its columns to set widget options. Widget values may
be set by URL parameters via init_config.
df_source (pandas dataframe): Dataframe of the csv source.
cols (dict): Keys are categories of columns of df_source, and values are a list of columns of that category.
defaults (dict): Keys correspond to widgets, and values (str) are the default values of those widgets.
init_load (boolean, optional): If this is the initial page load, then this will be True, else False.
init_config (dict): Initial widget configuration passed via URL.
wdg (ordered dict): Dictionary of bokeh.model.widgets.
#Add widgets
wdg = collections.OrderedDict()
wdg['data'] = bmw.TextInput(title='Data Source (required)', value=defaults['data_source'], css_classes=['wdgkey-data'])
wdg['x_dropdown'] = bmw.Div(text='X-Axis (required)', css_classes=['x-dropdown'])
wdg['x'] = bmw.Select(title='X-Axis (required)', value=defaults['x'], options=['None'] + cols['all'], css_classes=['wdgkey-x', 'x-drop'])
wdg['x_group'] = bmw.Select(title='Group X-Axis By', value=defaults['x_group'], options=['None'] + cols['seriesable'], css_classes=['wdgkey-x_group', 'x-drop'])
wdg['y_dropdown'] = bmw.Div(text='Y-Axis (required)', css_classes=['y-dropdown'])
wdg['y'] = bmw.Select(title='Y-Axis (required)', value=defaults['y'], options=['None'] + cols['all'], css_classes=['wdgkey-y', 'y-drop'])
wdg['y_agg'] = bmw.Select(title='Y-Axis Aggregation', value='Sum', options=AGGREGATIONS, css_classes=['wdgkey-y_agg', 'y-drop'])
wdg['series_dropdown'] = bmw.Div(text='Series', css_classes=['series-dropdown'])
wdg['series'] = bmw.Select(title='Separate Series By', value=defaults['series'], options=['None'] + cols['seriesable'],
css_classes=['wdgkey-series', 'series-drop'])
wdg['series_legend'] = bmw.Div(text='', css_classes=['series-drop'])
wdg['explode_dropdown'] = bmw.Div(text='Explode', css_classes=['explode-dropdown'])
wdg['explode'] = bmw.Select(title='Explode By', value=defaults['explode'], options=['None'] + cols['seriesable'], css_classes=['wdgkey-explode', 'explode-drop'])
wdg['explode_group'] = bmw.Select(title='Group Exploded Charts By', value=defaults['explode_group'], options=['None'] + cols['seriesable'],
css_classes=['wdgkey-explode_group', 'explode-drop'])
wdg['filters'] = bmw.Div(text='Filters', css_classes=['filters-dropdown'])
for j, col in enumerate(cols['filterable']):
val_list = [str(i) for i in sorted(df_source[col].unique().tolist())]
wdg['heading_filter_'+str(j)] = bmw.Div(text=col, css_classes=['filter-head'])
wdg['filter_'+str(j)] = bmw.CheckboxGroup(labels=val_list, active=list(range(len(val_list))), css_classes=['wdgkey-filter_'+str(j), 'filter'])
wdg['update'] = bmw.Button(label='Update Filters', button_type='success', css_classes=['filters-update'])
wdg['adjustments'] = bmw.Div(text='Plot Adjustments', css_classes=['adjust-dropdown'])
wdg['chart_type'] = bmw.Select(title='Chart Type', value=defaults['chart_type'], options=CHARTTYPES, css_classes=['wdgkey-chart_type', 'adjust-drop'])
wdg['plot_width'] = bmw.TextInput(title='Plot Width (px)', value=str(PLOT_WIDTH), css_classes=['wdgkey-plot_width', 'adjust-drop'])
wdg['plot_height'] = bmw.TextInput(title='Plot Height (px)', value=str(PLOT_HEIGHT), css_classes=['wdgkey-plot_height', 'adjust-drop'])
wdg['plot_title'] = bmw.TextInput(title='Plot Title', value='', css_classes=['wdgkey-plot_title', 'adjust-drop'])
wdg['plot_title_size'] = bmw.TextInput(title='Plot Title Font Size', value=str(PLOT_FONT_SIZE), css_classes=['wdgkey-plot_title_size', 'adjust-drop'])
wdg['opacity'] = bmw.TextInput(title='Opacity (0-1)', value=str(OPACITY), css_classes=['wdgkey-opacity', 'adjust-drop'])
wdg['x_scale'] = bmw.TextInput(title='X Scale', value=str(X_SCALE), css_classes=['wdgkey-x_scale', 'adjust-drop'])
wdg['x_min'] = bmw.TextInput(title='X Min', value='', css_classes=['wdgkey-x_min', 'adjust-drop'])
wdg['x_max'] = bmw.TextInput(title='X Max', value='', css_classes=['wdgkey-x_max', 'adjust-drop'])
wdg['x_title'] = bmw.TextInput(title='X Title', value='', css_classes=['wdgkey-x_title', 'adjust-drop'])
wdg['x_title_size'] = bmw.TextInput(title='X Title Font Size', value=str(PLOT_FONT_SIZE), css_classes=['wdgkey-x_title_size', 'adjust-drop'])
wdg['x_major_label_size'] = bmw.TextInput(title='X Labels Font Size', value=str(PLOT_AXIS_LABEL_SIZE), css_classes=['wdgkey-x_major_label_size', 'adjust-drop'])
wdg['x_major_label_orientation'] = bmw.TextInput(title='X Labels Degrees', value=str(PLOT_LABEL_ORIENTATION),
css_classes=['wdgkey-x_major_label_orientation', 'adjust-drop'])
wdg['y_scale'] = bmw.TextInput(title='Y Scale', value=str(Y_SCALE), css_classes=['wdgkey-y_scale', 'adjust-drop'])
wdg['y_min'] = bmw.TextInput(title='Y Min', value='', css_classes=['wdgkey-y_min', 'adjust-drop'])
wdg['y_max'] = bmw.TextInput(title='Y Max', value='', css_classes=['wdgkey-y_max', 'adjust-drop'])
wdg['y_title'] = bmw.TextInput(title='Y Title', value='', css_classes=['wdgkey-y_title', 'adjust-drop'])
wdg['y_title_size'] = bmw.TextInput(title='Y Title Font Size', value=str(PLOT_FONT_SIZE), css_classes=['wdgkey-y_title_size', 'adjust-drop'])
wdg['y_major_label_size'] = bmw.TextInput(title='Y Labels Font Size', value=str(PLOT_AXIS_LABEL_SIZE), css_classes=['wdgkey-y_major_label_size', 'adjust-drop'])
wdg['circle_size'] = bmw.TextInput(title='Circle Size (Dot Only)', value=str(CIRCLE_SIZE), css_classes=['wdgkey-circle_size', 'adjust-drop'])
wdg['bar_width'] = bmw.TextInput(title='Bar Width (Bar Only)', value=str(BAR_WIDTH), css_classes=['wdgkey-bar_width', 'adjust-drop'])
wdg['line_width'] = bmw.TextInput(title='Line Width (Line Only)', value=str(LINE_WIDTH), css_classes=['wdgkey-line_width', 'adjust-drop'])
wdg['download'] = bmw.Button(label='Download csv', button_type='success')
wdg['export_config'] = bmw.Div(text='Export Config to URL', css_classes=['export-config', 'bk-bs-btn', 'bk-bs-btn-success'])
#use init_config (from 'widgets' parameter in URL query string) to configure widgets.
if init_load:
for key in init_config:
if key in wdg:
if hasattr(wdg[key], 'value'):
wdg[key].value = str(init_config[key])
elif hasattr(wdg[key], 'active'):
wdg[key].active = init_config[key]
#Add update functions for widgets
wdg['data'].on_change('value', update_data)
for name in wdg_col:
wdg[name].on_change('value', update_wdg_col)
for name in wdg_non_col:
wdg[name].on_change('value', update_wdg)
return wdg
def set_df_plots(df_source, cols, wdg):
Apply filters, scaling, aggregation, and sorting to source dataframe, and return the result.
df_source (pandas dataframe): Dataframe of the csv source.
cols (dict): Keys are categories of columns of df_source, and values are a list of columns of that category.
wdg (ordered dict): Dictionary of bokeh model widgets.
df_plots (pandas dataframe): df_source after having been filtered, scaled, aggregated, and sorted.
df_plots = df_source.copy()
#Apply filters
for j, col in enumerate(cols['filterable']):
active = [wdg['filter_'+str(j)].labels[i] for i in wdg['filter_'+str(j)].active]
if col in cols['continuous']:
active = [float(i) for i in active]
df_plots = df_plots[df_plots[col].isin(active)]
#Scale Axes
if wdg['x_scale'].value != '' and wdg['x'].value in cols['continuous']:
df_plots[wdg['x'].value] = df_plots[wdg['x'].value] * float(wdg['x_scale'].value)
if wdg['y_scale'].value != '' and wdg['y'].value in cols['continuous']:
df_plots[wdg['y'].value] = df_plots[wdg['y'].value] * float(wdg['y_scale'].value)
#Apply Aggregation
if wdg['y_agg'].value == 'Sum' and wdg['y'].value in cols['continuous']:
groupby_cols = [wdg['x'].value]
if wdg['x_group'].value != 'None': groupby_cols = [wdg['x_group'].value] + groupby_cols
if wdg['series'].value != 'None': groupby_cols = [wdg['series'].value] + groupby_cols
if wdg['explode'].value != 'None': groupby_cols = [wdg['explode'].value] + groupby_cols
if wdg['explode_group'].value != 'None': groupby_cols = [wdg['explode_group'].value] + groupby_cols
df_plots = df_plots.groupby(groupby_cols, as_index=False, sort=False)[wdg['y'].value].sum()
#Sort Dataframe
sortby_cols = [wdg['x'].value]
if wdg['x_group'].value != 'None': sortby_cols = [wdg['x_group'].value] + sortby_cols
if wdg['series'].value != 'None': sortby_cols = [wdg['series'].value] + sortby_cols
if wdg['explode'].value != 'None': sortby_cols = [wdg['explode'].value] + sortby_cols
if wdg['explode_group'].value != 'None': sortby_cols = [wdg['explode_group'].value] + sortby_cols
df_plots = df_plots.sort_values(sortby_cols).reset_index(drop=True)
#Rearrange column order for csv download
unsorted_columns = [col for col in df_plots.columns if col not in sortby_cols + [wdg['y'].value]]
df_plots = df_plots[sortby_cols + unsorted_columns + [wdg['y'].value]]
return df_plots
def create_figures(df_plots, wdg, cols):
Create figures based on the data in a dataframe and widget configuration, and return figures in a list.
The explode widget determines if there will be multiple figures.
df_plots (pandas dataframe): Dataframe of csv source after being filtered, scaled, aggregated, and sorted.
wdg (ordered dict): Dictionary of bokeh model widgets.
cols (dict): Keys are categories of columns of df_source, and values are a list of columns of that category.
plot_list (list): List of bokeh.model.figures.
plot_list = []
df_plots_cp = df_plots.copy()
if wdg['explode'].value == 'None':
plot_list.append(create_figure(df_plots_cp, df_plots, wdg, cols))
if wdg['explode_group'].value == 'None':
for explode_val in df_plots_cp[wdg['explode'].value].unique().tolist():
df_exploded = df_plots_cp[df_plots_cp[wdg['explode'].value].isin([explode_val])]
plot_list.append(create_figure(df_exploded, df_plots, wdg, cols, explode_val))
for explode_group in df_plots_cp[wdg['explode_group'].value].unique().tolist():
df_exploded_group = df_plots_cp[df_plots_cp[wdg['explode_group'].value].isin([explode_group])]
for explode_val in df_exploded_group[wdg['explode'].value].unique().tolist():
df_exploded = df_exploded_group[df_exploded_group[wdg['explode'].value].isin([explode_val])]
plot_list.append(create_figure(df_exploded, df_plots, wdg, cols, explode_val, explode_group))
return plot_list
def create_figure(df_exploded, df_plots, wdg, cols, explode_val=None, explode_group=None):
Create and return a figure based on the data in a dataframe and widget configuration.
df_exploded (pandas dataframe): Dataframe of just the data that will be plotted in this figure.
df_plots (pandas dataframe): Dataframe of all plots data, used only for maintaining consistent series colors.
wdg (ordered dict): Dictionary of bokeh model widgets.
cols (dict): Keys are categories of columns of df_source, and values are a list of columns of that category.
explode_val (string, optional): The value in the column designated by wdg['explode'] that applies to this figure.
explode_group (string, optional): The value in the wdg['explode_group'] column that applies to this figure.
p (bokeh.model.figure): A figure, with all glyphs added by the add_glyph() function.
# If x_group has a value, create a combined column in the dataframe for x and x_group
x_col = wdg['x'].value
if wdg['x_group'].value != 'None':
x_col = str(wdg['x_group'].value) + '_' + str(wdg['x'].value)
df_exploded[x_col] = df_exploded[wdg['x_group'].value].map(str) + ' ' + df_exploded[wdg['x'].value].map(str)
#Build x and y ranges and figure title
kw = dict()
#Set x and y ranges. When x is grouped, there is added complication of separating the groups
xs = df_exploded[x_col].values.tolist()
ys = df_exploded[wdg['y'].value].values.tolist()
if wdg['x_group'].value != 'None':
kw['x_range'] = []
unique_groups = df_exploded[wdg['x_group'].value].unique().tolist()
unique_xs = df_exploded[wdg['x'].value].unique().tolist()
for i, ugr in enumerate(unique_groups):
for uxs in unique_xs:
kw['x_range'].append(str(ugr) + ' ' + str(uxs))
#Between groups, add entries that consist of spaces. Increase number of spaces from
#one break to the next so that each entry is unique
kw['x_range'].append(' ' * (i + 1))
elif wdg['x'].value in cols['discrete']:
kw['x_range'] = sorted(set(xs))
if wdg['y'].value in cols['discrete']:
kw['y_range'] = sorted(set(ys))
#Set figure title
kw['title'] = wdg['plot_title'].value
seperator = '' if kw['title'] == '' else ', '
if explode_val is not None:
if explode_group is not None:
kw['title'] = kw['title'] + seperator + "%s = %s" % (wdg['explode_group'].value, str(explode_group))
seperator = '' if kw['title'] == '' else ', '
kw['title'] = kw['title'] + seperator + "%s = %s" % (wdg['explode'].value, str(explode_val))
#Add figure tools
hover = bmt.HoverTool(
("ser", "@ser_legend"),
("x", "@x_legend"),
("y", "@y_legend"),
TOOLS = [bmt.BoxZoomTool(), bmt.PanTool(), hover, bmt.ResetTool(), bmt.SaveTool()]
#Create figure with the ranges, titles, and tools, and adjust formatting and labels
p = bp.figure(plot_height=int(wdg['plot_height'].value), plot_width=int(wdg['plot_width'].value), tools=TOOLS, **kw)
p.toolbar.active_drag = TOOLS[0]
p.title.text_font_size = wdg['plot_title_size'].value + 'pt'
p.xaxis.axis_label = wdg['x_title'].value
p.yaxis.axis_label = wdg['y_title'].value
p.xaxis.axis_label_text_font_size = wdg['x_title_size'].value + 'pt'
p.yaxis.axis_label_text_font_size = wdg['y_title_size'].value + 'pt'
p.xaxis.major_label_text_font_size = wdg['x_major_label_size'].value + 'pt'
p.yaxis.major_label_text_font_size = wdg['y_major_label_size'].value + 'pt'
p.xaxis.major_label_orientation = 'horizontal' if wdg['x_major_label_orientation'].value == '0' else math.radians(float(wdg['x_major_label_orientation'].value))
if wdg['x'].value in cols['continuous']:
if wdg['x_min'].value != '': p.x_range.start = float(wdg['x_min'].value)
if wdg['x_max'].value != '': p.x_range.end = float(wdg['x_max'].value)
if wdg['y'].value in cols['continuous']:
if wdg['y_min'].value != '': p.y_range.start = float(wdg['y_min'].value)
if wdg['y_max'].value != '': p.y_range.end = float(wdg['y_max'].value)
#Add glyphs to figure
c = C_NORM
if wdg['series'].value == 'None':
if wdg['y_agg'].value != 'None' and wdg['y'].value in cols['continuous']:
xs = df_exploded[x_col].values.tolist()
ys = df_exploded[wdg['y'].value].values.tolist()
add_glyph(wdg, p, xs, ys, c)
full_series = df_plots[wdg['series'].value].unique().tolist() #for colors only
if wdg['chart_type'].value in STACKEDTYPES: #We are stacking the series
xs_full = sorted(df_exploded[x_col].unique().tolist())
y_bases_pos = [0]*len(xs_full)
y_bases_neg = [0]*len(xs_full)
for i, ser in enumerate(df_exploded[wdg['series'].value].unique().tolist()):
c = COLORS[full_series.index(ser)]
df_series = df_exploded[df_exploded[wdg['series'].value].isin([ser])]
xs_ser = df_series[x_col].values.tolist()
ys_ser = df_series[wdg['y'].value].values.tolist()
if wdg['chart_type'].value not in STACKEDTYPES: #The series will not be stacked
add_glyph(wdg, p, xs_ser, ys_ser, c, series=ser)
else: #We are stacking the series
ys_pos = [ys_ser[xs_ser.index(x)] if x in xs_ser and ys_ser[xs_ser.index(x)] > 0 else 0 for i, x in enumerate(xs_full)]
ys_neg = [ys_ser[xs_ser.index(x)] if x in xs_ser and ys_ser[xs_ser.index(x)] < 0 else 0 for i, x in enumerate(xs_full)]
ys_stacked_pos = [ys_pos[i] + y_bases_pos[i] for i in range(len(xs_full))]
ys_stacked_neg = [ys_neg[i] + y_bases_neg[i] for i in range(len(xs_full))]
add_glyph(wdg, p, xs_full, ys_stacked_pos, c, y_bases=y_bases_pos, series=ser)
add_glyph(wdg, p, xs_full, ys_stacked_neg, c, y_bases=y_bases_neg, series=ser)
y_bases_pos = ys_stacked_pos
y_bases_neg = ys_stacked_neg
return p
def add_glyph(wdg, p, xs, ys, c, y_bases=None, series=None):
Add a glyph to a Bokeh figure, depending on the chosen chart type.
wdg (ordered dict): Dictionary of bokeh model widgets.
p (bokeh.model.figure): Bokeh figure.
xs (list): List of x-values. These could be numeric or strings.
ys (list): List of y-values. These could be numeric or strings. If series data is stacked, these values include stacking.
c (string): Color to use for this series.
y_bases (list, optional): Only used when stacking series. This is the previous cumulative stacking level.
series (string): Name of current series for this glyph.
alpha = float(wdg['opacity'].value)
y_unstacked = list(ys) if y_bases is None else [ys[i] - y_bases[i] for i in range(len(ys))]
ser = ['None']*len(xs) if series is None else [series]*len(xs)
if wdg['chart_type'].value == 'Dot':
source = bms.ColumnDataSource({'x': xs, 'y': ys, 'x_legend': xs, 'y_legend': y_unstacked, 'ser_legend': ser})'x', 'y', source=source, color=c, size=int(wdg['circle_size'].value), fill_alpha=alpha, line_color=None, line_width=None)
elif wdg['chart_type'].value == 'Line':
source = bms.ColumnDataSource({'x': xs, 'y': ys, 'x_legend': xs, 'y_legend': y_unstacked, 'ser_legend': ser})
p.line('x', 'y', source=source, color=c, alpha=alpha, line_width=float(wdg['line_width'].value))
elif wdg['chart_type'].value == 'Bar':
if y_bases is None: y_bases = [0]*len(ys)
centers = [(ys[i] + y_bases[i])/2 for i in range(len(ys))]
heights = [abs(ys[i] - y_bases[i]) for i in range(len(ys))]
source = bms.ColumnDataSource({'x': xs, 'y': centers, 'x_legend': xs, 'y_legend': y_unstacked, 'h': heights, 'ser_legend': ser})
p.rect('x', 'y', source=source, height='h', color=c, fill_alpha=alpha, width=float(wdg['bar_width'].value), line_color=None, line_width=None)
elif wdg['chart_type'].value == 'Area':
if y_bases is None: y_bases = [0]*len(ys)
xs_around = xs + xs[::-1]
ys_around = y_bases + ys[::-1]
source = bms.ColumnDataSource({'x': xs_around, 'y': ys_around})
p.patch('x', 'y', source=source, alpha=alpha, fill_color=c, line_color=None, line_width=None)
def build_series_legend(df_plots, series_val):
Return html for series legend, based on values of column that was chosen for series, and global COLORS.
df_plots (pandas dataframe): Dataframe of all plots data.
series_val (string): Header for column chosen as series.
series_legend_string (string): html to be used as legend.
series_legend_string = '<div class="legend-header">Series Legend</div><div class="legend-body">'
if series_val != 'None':
active_list = df_plots[series_val].unique().tolist()
for i, txt in reversed(list(enumerate(active_list))):
series_legend_string += '<div class="legend-entry"><span class="legend-color" style="background-color:' + str(COLORS[i]) + ';"></span>'
series_legend_string += '<span class="legend-text">' + str(txt) +'</span></div>'
series_legend_string += '</div>'
return series_legend_string
def update_data(attr, old, new):
When data source is updated, rebuild widgets and plots.
defaults['data_source'] = gl['widgets']['data'].value
for w in wdg_col:
defaults[w] = 'None'
defaults['chart_type'] = 'Dot'
gl['df_source'], gl['columns'] = get_data(defaults['data_source'])
gl['widgets'] = build_widgets(gl['df_source'], gl['columns'], defaults)
gl['controls'].children = list(gl['widgets'].values())
gl['plots'].children = []
def update_wdg(attr, old, new):
When general widgets are updated (not in wdg_col), update plots only.
def update_wdg_col(attr, old, new):
When widgets in wdg_col are updated, set the options of all wdg_col widgets,
and update plots.
def set_wdg_col_options():
Limit available options for wdg_col widgets based on their selected values, so that users
cannot select the same value for two different wdg_col widgets.
cols = gl['columns']
wdg = gl['widgets']
#get list of selected values and use to reduce selection options.
sels = [str(wdg[w].value) for w in wdg_col if str(wdg[w].value) !='None']
all_reduced = [x for x in cols['all'] if x not in sels]
ser_reduced = [x for x in cols['seriesable'] if x not in sels]
for w in wdg_col:
val = str(wdg[w].value)
none_append = [] if val == 'None' else ['None']
opt_reduced = all_reduced if w in wdg_col_all else ser_reduced
wdg[w].options = [val] + opt_reduced + none_append
def update_plots():
Make sure x axis and y axis are set. If so, set the dataframe for the plots and build them.
if gl['widgets']['x'].value == 'None' or gl['widgets']['y'].value == 'None':
gl['plots'].children = []
gl['df_plots'] = set_df_plots(gl['df_source'], gl['columns'], gl['widgets'])
gl['widgets']['series_legend'].text = build_series_legend(gl['df_plots'], gl['widgets']['series'].value)
gl['plots'].children = create_figures(gl['df_plots'], gl['widgets'], gl['columns'])
def download():
Download a csv file of the currently viewed data to the downloads/ directory,
with the current timestamp.
gl['df_plots'].to_csv(os.path.dirname(os.path.realpath(__file__)) + '/downloads/out '+"%Y-%m-%d %H-%M-%S-%f")+'.csv', index=False)
#initialize globals dict
gl = {'df_source':None, 'df_plots':None, 'columns':None, 'widgets':None, 'controls': None, 'plots':None}
#List of widgets that use columns as their selectors
wdg_col_all = ['x', 'y'] #all columns available for these widgets
wdg_col_ser = ['x_group', 'series', 'explode', 'explode_group'] #seriesable columns available for these widgets
wdg_col = wdg_col_all + wdg_col_ser
#List of widgets that don't use columns as selector and share general widget update function
wdg_non_col = ['chart_type', 'y_agg', 'plot_title', 'plot_title_size',
'plot_width', 'plot_height', 'opacity', 'x_min', 'x_max', 'x_scale', 'x_title',
'x_title_size', 'x_major_label_size', 'x_major_label_orientation',
'y_min', 'y_max', 'y_scale', 'y_title', 'y_title_size', 'y_major_label_size',
'circle_size', 'bar_width', 'line_width']
#Specify default widget values
defaults = {}
defaults['data_source'] = os.path.dirname(os.path.realpath(__file__)) + '/csv/US_electric_power_generation.csv'
for w in wdg_col:
defaults[w] = 'None'
defaults['x'] = 'Year'
defaults['y'] = 'Electricity Generation (TWh)'
defaults['series'] = 'Technology'
defaults['explode'] = 'Case'
defaults['chart_type'] = 'Area'
#On initial load, read 'widgets' parameter from URL query string and use to set data source (data_source)
#and widget configuration object (wdg_config)
wdg_config = {}
args = bio.curdoc().session_context.request.arguments
wdg_arr = args.get('widgets')
if wdg_arr is not None:
wdg_config = json.loads(urlp.unquote(wdg_arr[0].decode('utf-8')))
if 'data' in wdg_config:
defaults['data_source'] = str(wdg_config['data'])
for w in wdg_col:
defaults[w] = 'None'
#build widgets and plots
gl['df_source'], gl['columns'] = get_data(defaults['data_source'])
gl['widgets'] = build_widgets(gl['df_source'], gl['columns'], defaults, init_load=True, init_config=wdg_config)
gl['controls'] = bl.widgetbox(list(gl['widgets'].values()), id='widgets_section')
gl['plots'] = bl.column([], id='plots_section')
layout = bl.row(gl['controls'], gl['plots'], id='layout')
bio.curdoc().title = "Exploding Pivot Chart Maker"