list of csv headers as values / variables

[fadingNA]

Hi,

I'm trying to get column name on CSV to create variable on the dashboard but I tried couple solution its did not work as. I expected.

| limit 1

I thought it should return columns header, but instead return list of value on the first row.

Could you provide more insight how to get column header name please.

Best,
Non

[yesoreyeram]

What does your CSV looks like and what column your are expecting see as variable? Also the uql you tried...

[yesoreyeram]

I haven't encountered this use case before to return column names. But with the limit 1 query you are just limiting the number of values rows which doesn't include headers.

I can't think of any straightforward solution. But you can hack enough this with JSONata inside the uql.

BTW, out of curiosity...why you needed column names? May be a wrong solution for your use case

[yesoreyeram]

Loving these sort of challanges and use cases. 😄

[fadingNA]

So my use cases might be a bit complex because I have different data sources:
• SQL
• CSV

I’m working with mixed data.
1. Joining them together through transformations — works.
2. Creating variables to filter dynamically based on user selections — not working.
Two variables:
1. Column name (selected by the user)
2. Value of the selected column

I have another approach in mind, but I want to use the full functionality of the Grafana dashboard and its plugins without manipulating the table schema or creating new tables.

So the idea is:
Once the user selects a column_name, then selects a column_value, I want to pass that column_value into the query to filter and display the correct data in the table.

@yesoreyeram

[yesoreyeram]

Challanges accepted 😀

[yesoreyeram]

right.. found a hack https://play.grafana.org/goto/KvJSQjoNR?orgId=1

[yesoreyeram]

right.. found a hack https://play.grafana.org/goto/KvJSQjoNR?orgId=1. Idea is to pretend the csv doesn't have headers and injecting a header. By using fake delimiter, entire first line will become single field.

For reference the UQL query is

parse-csv --delimiter ";;;" --columns 'headers' ## prepending fake headers using dummy delimiter
| limit 1                                       ## limit it to first row which contain headers
| project "header"=split("headers",',')         ## split the csv values into array
| mv-expand "server"="header"                   ## expand array item into rows
| where "server" != 'metric'                    ## removing unnecessary headers if required
# | echo "foo"                                  ## You can comment out line by prepending # to debug

and as a variable

[fadingNA]

@yesoreyeram this is absolute wonderful, I really appreciate it let me try it out.