Dynamically add json object into array with jq - json

Question

Below is the template of my employee.json file
{
"orgConfig": {
"departments": []
}
}
where departments will have array of departments like below
{
"name" : "physics",
"id" : "1234",
"head" : "abcd"
}
similarly
{
"name" : "chemistry",
"id" : "3421",
"head" : "xyz"
}
so the final array structure i want to construct is as below
{
"orgConfig": {
"departments": [
{
"name" : "physics",
"id" : "1234",
"head" : "abcd"
},
{
"name" : "chemistry",
"id" : "3421",
"head" : "xyz"
},
{
"name" : "Maths",
"id" : "4634",
"head" : "jklm"
}
]
}
}
Below is the code where i am adding the json elements to an departments array dynamically
#!/bin/bash
source department.properties # will have departments=physiscs,chemistry,Maths,computers .. etc
IFS=',' read -ra NAMES <<< "$departmentsToImport"
position=0
for i in "${NAMES[#]}"; do
#./jsonfiles will chemistry.json, physics.json, Maths.json etc
value=`cat ./jsonfiles/$i.json`
if [ $position -eq 0 ]
then
cat employee.json | jq --arg value "$value" '.orgConfig.departments[0] |= .+ $value' > tmp.json && mv tmp.json employee.json
else
cat employee.json | jq --arg position "$position" value "$value" '.orgConfig.departments[$position] |= .+ $value' > tmp.json && mv tmp.json employee.json
fi
((position++))
rm -rf tmp.json
done
exit $?
but program throws below error
jq: error (at <stdin>:51): Cannot index array with string "1"
But if use direct index instead of variable position then it works fine.
cat employee.json | jq --argjson value "$value" '.orgConfig.departments[1] |= .+ $value' > tmp.json && mv tmp.json employee.json
I do not know how many key value maps of departments i have. I cannot hard code the index. Any help to above problem and Dynamically add json object into array?
Thanks

Answer 1

This can be done by turning off jq's automatic input reading, then piping the first explicit input through a filter that modifies its input using the remaining explicit inputs. Make sense? No :) But the code itself is simple:
jq -n 'input | .orgConfig.departments += [inputs]' \
employee.json chemistry.json physics.json math.json
The first input reads from the first argument (employee.json).
The += gets its input from the input filter. Its left operand
selects the field to update; the right operand provides the value
to update it with.
inputs reads from the remaining command-line arguments, and puts their contents in an array, each file in a separate element.
Combining this with your shell code to select the correct course files yields
source department.properties
IFS=, read -ra NAMES <<< "$departmentsToImport"
for c in "${NAMES[#]}"; do
courses+=("./jsonfiles/$c.json")
done
jq -n 'input | .orgConfig.departments += [inputs]' employee.json "${courses[#]}"

Answer 2

The task can be accomplished without invoking jq more than once.
Something very much like the following should suffice:
jq -s '{orgConfig: {departments: . }}' jsonfiles/*.json
This solution of course assumes the .json files all hold valid JSON.
The trick is to use the -s (aka --slurp) option, as this converts the input into an array for you. You'll probably find that using -s results in better run-times than some other approaches.