jq reformatting decimals in scientific notation —

Question

I found difference between json-data created by JavaScipt and via jq with bash (and other programming languages). With JavaScript I can create decimal numbers with up to six digits after the point, even when I use float(). But with jq its different, because adding a decimal value takes four digits after the decimal point only.

My problem is that I need decimal numbers to store in SQL, with up to six digits after the point.

Example:

$ JSON='{"decimal":0.00001}'
$ echo "$JSON" | jq .
{
  "decimal": 1e-05
}

My goal is to validate the decimal with this line ...

if [[ "$TMP_DECIMAL" =~ ^[0-9]+([.][0-9]+)?$ ]] ; then

Any tip / suggsestion ?

Answer 1

You can't change jq's behavior -- at present date, relevant feature requests are still open -- but you can reformat your numbers after they've been retrieved. For example:

json='{"decimal":0.00001}'
decimal=$(jq '.decimal' <<<"$json")
decimal_f=$(awk -v decimal="$decimal" 'BEGIN { printf("%f\n", decimal) }' </dev/null)

echo "JQ emitted $decimal; reformatted as $decimal_f"

Answer 2

Also, you can reformat your JSON using perl module JSON::PP.

perl -0777 -MJSON::PP -E '$s=<>; $j=JSON::PP->new->ascii->pretty->allow_nonref->allow_bignum;$p=$j->decode($s);say $j->encode($p)'

or nicer:

perl -0777 -MJSON::PP -E '
    $j=JSON::PP->new->ascii->pretty->allow_nonref->allow_bignum;
    $p=$j->decode(<>);
    say $j->encode($p)'

The crucial is the allow_bignum.

Example:

echo '{"decimal":0.00000001}' | perl ....

prints

{
   "decimal" : 0.00000001
}

but without the allow_bignum prints

{
   "decimal" : 1e-08
}

Ps: ... and also, is possible to validate the whole json using perl... :)