Compare json files but ignore values

Question

I would like to compare two json files and report differencies but I am interested in keys only and not values. So for example the "json-diff" between the following two files (of course they are much more complicated):

{
    "http": {
        "https": true,
        "swagger": {
            "enabled": false
        },
        "scalingFactors": [0.1, 0.2]
    }
}

{
    "http": {
        "https": true,
        "swagger": {
            "enabled": true
        },
        "scalingFactors": [0.1, 0.1],
        "test": true
    }
}

should report that there is missing key:

• http.test

but

should not report that the following keys have different values:

• http.swagger.enabled
• http.scalingFactors

I looked at the jq tool but I am not sure how to ignore values.

Answer 1

Ignoring potential complications having to do with arrays, looking at the "symmetric difference" of the sets of paths to scalars would make sense. As a starting point, you could thus consider:

jq -c '
 [paths(scalars)] as $f1
 | [input | paths(scalars)] as $f2
 | ($f1 - $f2) + ($f2 - $f1)' file1.json file2.json 

You might want to stringify the paths, but then again, it might be wise to avoid doing so if the mapping to the strings is not invertible.

---

If arrays are present, you might want to compare the paths while ignoring the array indices:

def p: [paths(scalars) | map(select(type=="string"))] | unique;
p as $f1
| (input | p) as $f2
| ($f1 - $f2) + ($f2 - $f1)
| .[]

The last line ensures that the result is a (possibly empty) stream, the point being that this makes it easy to check the return code to determine whether any difference was detected: simply use the -e command-line option. If there are no differences, the return code will then be 4.

One way to check if the stream is empty would be to use the -4