Remove comma after last value inside an array(Linux)

I have this type of data. Values inside "percents" can be multiple one after another.

I want to remove comma after last value(Inside an array), but no luck.

          "percentiles_value": {
             "percentiles": {
                 "field": "value",
                 "percents": [
                      25,
                      50,
                  ],
                 "keyed": true,
                 "tdigest": {
                     "compression": 100
                 }
             }
        }
    }

Can anybody suggest how to achieve the same with sed or awk.

The larger question is "why?"

That comma should not cause any JSON parser an error; so why spend time on such a task?

Is this homework or related to some school assignment?

This is neither homework nor school assignment.

This is opensearch transform job to handle huge bank data for your information.

And yes, it is simple as that the comma with last value creates issue.

image1917×402 33.4 KB

Job executed successfully without comma at last value.

image1920×732 67.1 KB

Here is a sed "solution":

sed '$!N; s/,\( *\n *\]\)/\1/; P; D'

It puts two lines in the input buffer, looks for a comma in the first line and a ] in the second line.

N put a second line in the input buffer; the embedded newline becomes \n
P print the first line
D delete the first line
\( \) capture group
\1 reference to the match of the 1st capture group
The $! is needed by a non-GNU sed where in the last line N exits without default-printing.

Of course sed has no idea about the structure.
A real solution is to fix your json parser.

That is interesting.

Normally, based on my experience with JSON with Javascript, Ruby and PHP is that the trailing comma in a array such as the one you described so well is ignored and does not cause an exception condition.

Thank you for the excellent reply and explanation @akshay.kulkarni

I guess the opensearch parser is "too restrictive" or "very restrictive" for reasons unknown to us since other parsers just ignore the trailing comma in an array such as you have provided.

Thanks @MadeInGermany

Suggested solution worked well with my transform_job_generator_script.

And yes elasticsearch/opensearch are very restrictive.

Here is the generated json output:

image1397×967 11 KB

Looks like in this case, @akshay.kulkarni is not in a position to change the (overly restrictive) behavior of the parser, so thanks to @MadeInGermany for his excellent suggestion, "saving the day" for Akshay.

Well done.

Note, it seems there should be a configuration variable in that parser (could not find one) which sets the JSON parser mode to "less restrictive" so this comma is not an issue.

Thank you people for this useful information.