How can I make jq assign values to environment variables in one command?

Question

Say, I provide jq with the following JSON body as input:

{"person1": {"name": "foo"}, "person2": {"name": "bar"}}

Is it possible to have jq assign person1.name to environment variable PERSON_1_NAME and person2.name to environment variable PERSON_2_NAME, so that I don't have to run jq multiple times with the same input, e.g.:

INPUT='{"person1": {"name": "foo"}, "person2": {"name": "bar"}}'
export PERSON_1_NAME=$(echo $INPUT | jq -r .person1)

?

Answer 1

You can do this like so:

INPUT='{"person1": {"name": "foo"}, "person2": {"name": "bar"}}'
eval "$(echo "$INPUT" | jq -r '[
  "export PERSON_1_NAME="+.person1.name,
  "export PERSON_2_NAME="+.person2.name
] | .[]')"

Output:

$ INPUT='{"person1": {"name": "foo"}, "person2": {"name": "bar"}}'
eval "$(echo "$INPUT" | jq -r '[
  "export PERSON_1_NAME="+.person1.name,
  "export PERSON_2_NAME="+.person2.name
] | .[]')"
$ echo $PERSON_1_NAME
foo
$ echo $PERSON_2_NAME
bar

If you have security concerns about the type of data you'll be handling, you can also avoid using eval to prevent any issues:

mapfile -t names < <(echo "$INPUT" | jq -r '.person1.name, .person2.name')
export PERSON_1_NAME="${names[0]}"
export PERSON_2_NAME="${names[1]}"

Output:

$ INPUT='{"person1": {"name": "foo"}, "person2": {"name": "bar"}}'
$ mapfile -t names < <(echo "$INPUT" | jq -r '.person1.name, .person2.name')
export PERSON_1_NAME="${names[0]}"
export PERSON_2_NAME="${names[1]}"
$ echo $PERSON_1_NAME
foo
$ echo $PERSON_2_NAME
bar

Answer 2

Assuming you want to use the key as part of the variable name, and that the key always has a sane value that can be used as part of a shell variable's name (we will not be verifying this), and that the name key holds a scalar that can be used as the value in a shell variable:

$ input='{"person1": {"name": "foo"}, "person2": {"name": "bar"}}'
$ eval "$(jq -n -r --argjson data "$input" '$data | to_entries[] | @sh "export \(.key)_name=\(.value.name)"')"
$ printf '%s\n' "$person1_name" "$person2_name"
foo
bar

This uses jq to create output like

export 'person1'_name='foo'
export 'person2'_name='bar'

... given the keys and the name sub-keys document passed as the jq variable $data.

The @sh output operator does some basic handling of shell escapes, which means that this also handles the cases where a name contains tabs, newlines, and quotes:

$ input='{"person1": {"name": "foo\tbar\nthe third"}, "person2": {"name": "A'\''Tuin"}}'
$ eval "$(jq -n -r --argjson data "$input" '$data | to_entries[] | @sh "export \(.key)_name=\(.value.name)"')"
$ printf '%s\n' "$person1_name" "$person2_name"
foo     bar
the third
A'Tuin

---

A jq command that creates the variable names in the format shown in the question, with the number counting up with each person:

$ jq -n -r --argjson data "$input" 'reduce ($data|to_entries[]) as $e ([]; . += [@sh "export PERSON_\(1+length)_NAME=\($e.value.name)"]) | .[]'
export PERSON_1_NAME='foo'
export PERSON_2_NAME='bar'

Answer 3

Some care is required with the quoting here:

eval "$(jq -r '@sh "export PERSON_1_NAME=\(.person1.name) PERSON_2_NAME=\(.person2.name)"' <<< '{"person1": {"name": "foo"}, "person2": {"name": "bar"}}')"

jq's @sh format/escape string does its best to escape output properly for shell ingestion.