1
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 277 100 174 100 103 464 274
--:--:-- --:--:-- --:--:--
736{"result":{"progressId":"ES_7PBXiq5gg67u9Vr","percentComplete":0.0,"status":"inProg
ress"},"meta":{"requestId":"**********************","httpStatus":"200 -
OK"}}

I need to extract the highlighted part beginning with ES_

I tried using

curl <...> | sed '"progressId"\:"(ES_[A-Za-z0-9]{15})'

however it still stores the entire output in the environment variable.

AdminBee
  • 22,803
Viraj Deshpande
  • 25
  • 2
  • 6
  • 1
    Welcome to the site. When posting console output, please always ensure proper formatting. I tried to edit your post, but please verify that the sample output is exactly as would appear on your console so that contributors can understand the structure of the text and point you to a working regular expression. Btw, note that your sed command is malformed since it does not contain any command; I'm surprised you didn't get any error message ... – AdminBee May 15 '20 at 07:22
  • 1
    Please ensure that the output is properly formatted. As it stands now, it looks like you're getting broken JSON back, which I must assume means that you've botched the copy-and-paste somehow. Since you haven't copied the output properly, not even a sed solution can be trusted to work reliably. – Kusalananda May 15 '20 at 11:45

2 Answers2

5

Best to use a json parsing tool to parse json. jq is a popular one.

set -o pipefail # if supported by your shell
var=$(curl -s "$url" | jq -r .result.progressId) || exit

Or:

json=$(curl -s "$url") || exit
var=$(printf '%s\n' "$json" | jq -r .result.progressId) || exit

(which allows you to extract more information from that json data later on).

Stéphane Chazelas
  • 544,893
2

This should work:

curl <...> | sed -E -n 's/.*(ES[^"]+).+/\1/p'

Explanation:

  • .* means 'any char 0 or more times'
  • ES[^"]+ means 'ES followed by any char which is not " one or more times'
  • .+ means 'any char 1 or more times'
  • \1 means 'the first group', so what's inside ()
  • p means 'print only matching lines'

If you want to store it in a variable you can do it with backtick ` like this:

VAR="$(curl <...> | sed -E -n 's/.*(ES[^"]+).+/\1/p')"
Francesco
  • 836