How can I programmatically select rows in a table?

I’m trying to figure out how to select various rows in a Table programmatically. The attached notebook is a small example that I thought might work based on the documentation for the selected attr for a DataSource but unfortunately it doesn’t. (I tried both 0d and 1d thinking that might be the problem but neither seemed to do anything, and since I wanted entire rows selected it didn’t seem like 2d would apply).

I’ve looked everywhere I can think of for an example of how to do this or additional documentation but have come up short. Could someone please describe the correct way to specify which rows in a table should be selected?

table_selection_test.ipynb (13 KB)

Does anyone have any ideas on this? I have a very similar need.

Hi Jim,

I don’t have any help for you unfortunately, but I’m interested in the problem. I looked at the notebook you provided. Does that method of setting the selected property of the datasource work for other chart types? If so, maybe the events aren’t propagating in data_table.coffee they way they do in other objects?

For others, here’s the code from the notebook Jim attached:

import bokeh

from bokeh.plotting import figure, output_notebook, show

from bokeh.models import DataTable, ColumnDataSource, TableColumn

from bokeh.io import push_notebook

import pandas as pd

print(‘bokeh version: %s’ % bokeh.version)

output_notebook()

df = pd.DataFrame( {‘AAA’ : [4,5,6,7], ‘BBB’ : [10,20,30,40],‘CCC’ : [100,50,-30,-50]})

data = df.to_dict(orient=‘list’)

tableSource = ColumnDataSource(df)

columns = [TableColumn(field=“AAA”, title=“AAA”),

TableColumn(field=“BBB”, title=“BBB”),

TableColumn(field=“CCC”, title=“CCC”)]

bokehTable = DataTable(columns=columns, source=tableSource,

editable=False, fit_columns=True, height=300,

row_headers=True, selectable=True, sortable=True)

tableSource.selected = {‘1d’:{‘True’:(0,1)}}

show(bokehTable)

I looked through the examples for anything similar to this but didn’t find anything. Maybe others will have ideas.

cheers,

Dennis

Hi Dennis,

I looked around a bit to see how selection is handled with the different chart type and found this example (scroll down to the Selected and Unselected Glyphs section). On the surface it seems like that kind of approach might also work for tables if there were some way to tell a table to render a row differently (i.e. selected). Obviously the table can do that since the rows display differently when you click on them. But It’s not obvious to me how to trigger that rendering behavior externally.

I was just trying to do this. I couldn’t figure how to get anywhere with Jim’s suggestion, and the solution offered here Redirecting to Google Groups didn’t trigger a visible selection.

I think this would be a useful feature, as I’ve found that DataTable selections are useful to set document state parameters and trigger events. Having the default parameter rows visibly selected on load would the document more readable, and make it possible to incrementally change the parameters from their default state (rather than having them reset when the first selection is made).

I also want to do this, as I think its the only way to force the table to scroll to the last row in the table. I found this here, which implies it is possible to do.

SlickGrid also has APIs of its own for scrolling, they are not currently exposed via any Bokeh methods, but but they could be used from a CustomJS callback:

Slick.Grid · 6pac/SlickGrid Wiki · GitHub

Regarding, selections, since this is an old pot that has been revived, it should be mentioned that the "0d", "1d" syntax is long-deprecated (and has only been read only for that time) and will be removed in the next few releases. To set a selection on a table, update the selection on the data source that drives it: source.selected.indices = [...]