Update pandas dataframe

Dash Python
1

How can I update my pandas dataframe variables by changes in the user data table on dash?
I was thinking of using global dataframes and changing them through callbacks… but not sure if that would be stable

2

I use hidden div’s to store my variables, but you can also use dcc.Store; I my case I convert to JSON (this may not be necessary but it works for me). So I use the following to convert my panda data_df to json.

This is the conversion needed to assign data via callback’s Output:
data_df.to_json(orient='split')

To convert data passed in via callback’s Input, I do the following:
data_df = pd.read_json(input_data_variable, orient='split')

3

Sweet thanks a lot - I appreciate it :slight_smile:

1 Like
4

Are you able to show your actual callbacks as well if possible

5 
All the `children` components are Hidden Div's in my layout.  

import pandas as pd
...

@app.callback( 
    output=Output('slider-container', 'children'), 
    inputs=[Input('hilo-data-container', 'children'), 
    Input('session-data-container', 'children')]) 
def show_slider(hilo_props, session_props):
    # hilo_props and session_props inputs are panda df's converted to JSON
    hilo_df = pd.read_json(hilo_props, orient='split')
    session_df = pd.read_json(session_props, orient='split')

    # Do some stuff
    return some_panda_df.to_json(orient='split')
6

This is what I used… It works if I just call update_sometable, in which case my table gets updated. However it won’t update my table when I click the button through the dash app. Any idea what could be missing?

@app.callback(Output('dccStore', 'data'),
             [Input('updateButton', 'n_clicks')],
             [State('my-dropdown', 'value')])
def update_sometable(n_clicks, dropdown_selection):
    def update_table(dropdown_selection, sometable):

        sometable.loc[0, 'column'] = 5

        return sometable
    
    if n_clicks > 0: 
        if not dropdown_selection:
            return {}
        
        update_table(dropdown_selection, sometable)

    return {}
7

I mean call update_sometable in a jupyter notebooks cell

8

Actually this is fine, it is saved in the app cache
But for some reason I cannot get the State of the data table which is odd
Are you able to show an example where the input is the state of a data table?

9

For example, this will not work:

@app.callback(Output('dccStore', 'data'),
             [Input('updateButton', 'n_clicks')],
             [State('my-dropdown', 'value'), State('mytable', 'data')])
def update_sometable(n_clicks, dropdown_selection, rows):
    def update_table(dropdown_selection, sometable, rows):

        sometable.loc[0, 'column'] = rows[0]['column']

        return sometable
    
    if n_clicks > 0: 
        if not dropdown_selection:
            return {}
        
        update_table(dropdown_selection, sometable, rows)

    return {}
10

Not sure what you mean by “where the input is the state of a data table”…But in your code above, you don’t have a return after your update_table call. You could tweak your code as follows:

if n_clicks > 0: 
    if not dropdown_selection:
        return {}
    else:    
        return update_table(dropdown_selection, sometable)

If you want to update a datatable from a callback, or to obtain the state of the data in the datatable in a callback, the following illustrate this.

1 - update datatable ( i.e. id=my-datatable) data parameter.

@app.callback([Output('dccStore', 'data'),
    [Output('my-datatable', 'data')],
    [Input('updateButton', 'n_clicks')],
    [State('my-dropdown', 'value')])
    def update_sometable(n_clicks, dropdown_selection):

2 - get the state of the datatable data parameter:

@app.callback(Output('dccStore', 'data'),
    [Input('updateButton', 'n_clicks')],
    [State('my-dropdown', 'value'),
    State('my-datatable', 'data')])
    def update_sometable(n_clicks, dropdown_selection, my-datatable_data):