Callbacks with dcc.Store in a loop

Dash Python
1

Hi,

is it possible to send data to dcc.Store in a loop?

I am trying to do something like:

metric_names = {
    "dbName_01":"metric_01",
    "dbName_02":"metric_02",
}

for m,n in metric_names.items():
    @app.callback(
        output=Output(f'intermediate-{n}', 'data'),
        inputs=Input('url', 'search'))
    def get_value(n):
        data = pd.DataFrame(list(db.collection.find({},{
            "months_12":f"${m}.12Months",
            "months_24":f"${m}.24Months",
            "months_36":f"${m}.36Months",
            "_id":0
        })))
        return json.dumps(data.to_dict(), default=str)

But when I get the data on another page with:

for m in metric_names.values():
    @callback(
        output=Output(f'{m}', 'children'),
        inputs=[
            Input(f'intermediate-{m}', 'data'),
            State(f'intermediate-{m}', 'data'),
            ])
    def get_value(data, _):
        if data is not None:
            data = pd.DataFrame(json.loads(data))
            graph = dcc.Graph(
                figure=get_growth_histogram(data, lower_clip=-100, upper_clip=1_000),
                style={"width": "100%"})
            return graph
        else:
            return no_update

I get the same graph in each iteration…

Getting the graphs in a loop works, when I don’t have the dcc.Store loop. That is why I think it is due to the dcc.Store loop.

Cheers
Nils

2

As long as the data is different in the stores, then it should be different.

3

Well, that’s the point.
The problem is just the stupid way I tried to pass the m into my callback function.
To pass the m correctly, I added some html.Data to my Layout:

*[html.Data(id=f"{m}-tmp", key=f"{m}") for m,_ in metric_names.items()],

And I grabbed the key from there to pass it into the callbacks.
Feels very hacky though :smiley:

for m,n in metric_names.items():
    @app.callback(
        output=Output(f'intermediate-{n}', 'data'),
        inputs=Input(f"{m}-tmp", 'key'))
    def get_value(metric):
        data = pd.DataFrame(list(db.collection.find({},{
            "months_12":f"${metric}.12Months",
            "months_24":f"${metric}.24Months",
            "months_36":f"${metric}.36Months",
            "_id":0
        })))
        return json.dumps(data.to_dict(), default=str)

Cheers
Nils

4

I mean, another less hacky way to go about it would be to use pattern-matching MATCH.

5

That is interesting, thanks for the advice - I’ll have a look!

Meanwhile I also found a less hacky way, by just passing the dcc.Store id as input to the callback. That way everything is way more readable.

Cheers
Nils

1 Like