I have a file like this
{"100":["0.00001","10","0.01"],
"101":["0.0001","100","0.1"],
"102":["1","1000","1"],
....
"103":["0.01","10000","10"]}
I want to swap the second number with the third, inside each array, converting it into
{"100":["0.00001","0.01","10"],
"101":["0.0001","0.1","100"],
"102":["1","1","1000"],
....
"103":["0.01","10","10000"]}
How do I do that with sed, awk, or whatever?
Notice the existence of { and } at the first and last records.
thanks
I would advise to avoid parsing JSON data using text oriented tool like awk or sed, etc...
Instead use a JSON parser or a language that relies on a JSON library like perl, python, etc..
Here is an example with jq:
<file jq '.[] |= (.[2] as $t | .[2] = .[1] | .[1] = $t)'
All objects are replaced with the permutation of the 3rd and 2nd array element. Note that the arrays in jq start with index 0.
More verbose than jq would be to use a scripting language that has JSON libraries available:
ruby -rjson -e '
data = JSON.parse File.read ARGV.shift
data.transform_values! {|v| v[1], v[2] = v[2], v[1]; v}
puts JSON.pretty_generate data
' file.json
perl -E '
use Path::Tiny;
use JSON;
$json = JSON->new->utf8;
$data = $json->decode( path(shift)->slurp_utf8 );
while (($key, $val) = each %$data) {
($val->[1], $val->[2]) = ($val->[2], $val->[1]);
$data->{$key} = $val;
}
$json->canonical->indent;
say $json->encode($data);
' file.json
Using jq and its reverse function:
jq '.[][1:] |= reverse' file.json
or
jq '.[][-2:] |= reverse' file.json
The expression .[][1:] selects slices containing all but the first element of each array in the document (.[][-2:] explicitly selects the last two elements in each array). These are reversed using reverse.
A possibly simpler approach:
Explicitly rewriting the arrays in the order that we want them by picking the elements in the correct order:
jq 'map_values([ .[0], .[2], .[1] ])' file.json
This recreates each array, using the elements from the original array in the order they are picked out.