Before anything else, english isn't my laguage. I'll try to be understandable :-) Second, I am not a regular Linux user, I'm just starting out.

I'm using curl and the API of a web chat platform to obtain the status of my team. Each member possess an ID. For each ID, I pipe the result in a text file agent-#.txt. So in my folder, if I have 5 members, I obtain 5 files: agent-1.txt, agent-2.txt and so on.

Each file contain this:

    "used": 0,
    "last": "2018-06-01 15:25:55",
    "presence": [],
    "away": false,
    "free": 3,
    "last_utc": "2018-06-01 13:25:55",
    "last_timezone": "Europe/Paris",
    "online": true,
    "offline": true

Next, using sed I delete the data I don't need:

sed -i -e '/{/d; /last/d; /presence/d; /last_utc/d; /last_timezone/d; /offline/d; /}/d; s/"//g; s/,//g; s/[ ]//g; /^\s*$/d' *

I obtain, for each file:


Where # is a number (total used and free slots for the agent).

Ok, but now I want to combine those data to obtain:

  1. The total number of members whose status is online:true
  2. The total number of free and used slot

Ideally I'm searching for a command that will output the total in another file. But I you have an idea (command, full script) to deal with the whole process, I'm also very interested!


Using your unmodified JSON files (before you run sed over them), and jq:

Number of people online:

jq -s 'map(select(.online))|length' agent-*.txt

Number of free slots:

jq -s '[.[].free]|add' agent-*.txt

Number of used slots:

jq -s '[.[].used]|add' agent-*.txt

All three as a single JSON document:

jq -s '{online: map(select(.online))|length, free: [.[].free]|add, used: [.[].used]|add}' agent-*.txt

Same as above, but just the numbers:

jq -s '(map(select(.online))|length), ([.[].free]|add), ([.[].used]|add)' agent-*.txt
  • Thanks for linking to jq. Looks like a useful binary. – maulinglawns Aug 3 '18 at 11:48
  • Hi ! Sorry for the lack of news. I'll try this solution soon and report ASAP :-) – Grishkaone Aug 11 '18 at 14:47

You could concatenate the agent-*.txt files and process them by awk.

Note the -F : to use full column as field separator.

An incomplete example:


awk -F : -f - agent-*.txt <<EOF
    printf("used: %d\n", used);
    printf("online: %d\n", online);

Probably, if you change carefully the matching expressions, you'll can skip the sed step.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.