Advanced JQ techniques-CodePudding

I'm looking for some help to increase my jq-foo, and could use a little guidance. I have some etl jobs that I script using JQ and I want to see if there is a way to make the runs more efficient in terms of CPU utilization and or speed.

Heres a sample payload:

{
  "timestamp": 1636601959,
  "uniqueId": "Foo",
  "value": 10
},
{
  "timestamp": 1636601859,
  "uniqueId": "Bar",
  "value": 13
}

and I want to do something like (pseudo-code):

if [ (epoch 15 minutes ago) -le timestamp ]; then 
   name=uniqueId; value_total=value(total); uniqueId_count=(uniqueId(count_total))
fi

Right now I do something like (pseudo-code):

for jq[timestamps] in $(json); do
  if [ (epoch 15 minutes ago)  -le timestamp ]; then 
    name=uniqueId; value_total=(value_total   value); uniqueId_count=(uniqueId_count   1). 
  fi
done

Is there a way to simplify this without using a for loop to iterate over each object by calling jq 'select() ...' over and over?

Thanks in advanced

CodePudding user response：

If you either get rid of the commas between the objects or enclose the whole input with array brackets (in order to get proper JSON), then you can do (use the -s option for {}{}{} style, or drop it for [{],{},{}] style):

jq -s --argjson delta $((15*60)) --argjson addvalue 10 '

 (now - $delta) as $pivot
 | map(select(.timestamp >= $pivot))
 | group_by(.uniqueId)
 | map({
    name: first.uniqueId,
    value_total: map(.value) | (add   $addvalue),
    uniqueId_count: length
  })

'