0

I recently discovered that the new bokeh implemented more detailed hover options for stacked bar charts. Giddy, I updated bokeh right away and incorporated into a nested stacked vbar. Unbeknownst to me though, an unrelated callback that I had implemented under the previous release no longer works, and I do not know how to fix it.

I have a data table, its rows feature several columns with numbers. Selecting rows will sum the total for each selected row's columns, and display it in a Label. This aspect no longer works in 0.13.0. Please see a simplified recreation below:

Dataframe(df):
NAME  |  Sales  |  Returns | YEAR
John      33        3        2018
Mike      12        3        2018
Jim       19        9        2018
Tim        3        1        2017
Tom       20        0        2017

import pandas as pd
from bokeh.models import (ColumnDataSource, LabelSet, TapTool, OpenURL,CustomJS, NumberFormatter,
                              TableColumn, DataTable, Slider, CDSView,HoverTool,PreText, BoxSelectTool,
                              Label, Text, Button)
from bokeh.plotting import figure
from bokeh.io import curdoc
from bokeh.layouts import layout, column
from datetime import date
from bokeh.core.properties import expr
from bokeh.palettes import Spectral10

source=ColumnDataSource(data=df)

columns=[TableColumn(field="NAME", title="NAME"),
TableColumn(field="Sales", title= "Sales"),
TableColumn(field="Returns",title="Returns")]

data_table=DataTable(source=source, columns=columns)

def statsline(attr,old,new):
    selection=source.selected.indices
    if selection:
        sales=sum(source.data['Sales'][selection])
        returns=sum(source.data['Returns'][selection])

        income.text="SELECTION | Sales: " + str(sales) + " | Returns: " + str(returns)
p_stat=figure(plot_width=1400, plot_height=60, toolbar_location=None)

income=Label(x_units='screen',y_units='screen',x=90, y=5,
             text="Sales: " + str(df["Sales"][df["YEAR"]==date.today().year].sum()) +"   |   
             Returns: " +str(df["Returns"][df["YEAR"]==date.today().year].sum()),
             text_font_size='20pt')
p_stat.add_layout(income)
source.on_change('selected', statsline)

layout=layout([p_stat],[data_table])
curdoc().add_root(layout)

Thanks in advance. Also, is there a less hacky way to display KPIs in bokeh? I'm running out of ideas.

4

1 回答 1

0

Bokeh 中有几个与 DataTable 相关的回归0.13.0，它们将在下一个版本中解决。目前（但也将继续），这种模式将起作用：

source.selected.on_change('indices', ...)
于 2018-07-25T06:19:32.600 回答