Not able to use where in dataset filter

[aborruso]

Hi,
I’m trying to filter rows in a dataset, but when I run

curl "https://datasets-server.huggingface.co/filter?dataset=aborruso/open_cup_complessivo&config=default&split=train&where=CUP=B76G16005640001&length=2" \
-X GET -H "Authorization: Bearer ${API_TOKEN}"

I get "error":"Parameter 'where' is invalid"}.

It’s strange because here where is the main parameter.

What am I doing wrong?

Thank you

[severo]

Hi! It might seem counter-intuitive, but the string has to be quoted:

https://datasets-server.huggingface.co/filter?dataset=aborruso/open_cup_complessivo&config=default&split=train&where=CUP=‘B76G16005640001’&length=2

We even highlighted it in the docs

[aborruso]

@severo

Hi! It might seem counter-intuitive, but the string has to be quoted:

It’s not counter-intuitive, I am the one who is stupid

Thank you very much

[aborruso]

@severo I ask you a related question here.

In the data source I have 1586 record in which COMUNE='VILLAROSA'.

If I apply the same filter rows I have "num_rows_total": 177.

I think on the fact that it depends that the database is big and there is a limit. The search is probably done only on the first xxx records.
Is that the case? Where can I find something in the documentation to help me understand?

Thank you very much

[severo]

Exactly, this is the reason, see the field partial: true

As mentioned at the end of the /filter doc page, we run the filter on the first 5GB of data…