I have multiple JSON files like these:

{
  "object1": {
    "tags": ["A"],

    "something": "else",
    "other": "data"
  },
  "object2": {
    "tags": ["A", "B"]
  }
}

and

{
  "object3": {
    "tags": ["C"],

    "something": "else",
    "other": "data"
  },
  "object4": {
    "tags": ["A"]
  }
}

It is guaranteed that keys of all the objects (object1 - object4) are unique, across all files.

I need to generate a different json file, that would be an array of used tags, and each tag would have extra information which objects use it:

[
  { 
    "tag": "A",
    "objects": ["object1", "object2", "object4"]
  },
  { 
    "tag": "B",
    "objects": ["object2"]
  },
  { 
    "tag": "C",
    "objects": ["object3"]
  }
]

Order of tags in this output array is irrelevant.

So far I have: cat *.json | jq -s add | jq '[.[].tags[]] | unique' which gives me array of tags used across all files, but I don't quite know how to get the list of objects for those tags. I suspect that this is not a right approach, because I am loosing some information (source of the tag) during this transformation.

Any help with jq expressions would be appreciated. Thank you.

CodePudding user response：

One way would be to reduce the input using to_entries

jq -s '
  add | reduce to_entries[] as $e ({}; .[$e.value.tags[]]  = [$e.key])
' *.json

which would give you a structure like this

{
  "A": [
    "object1",
    "object2",
    "object4"
  ],
  "B": [
    "object2"
  ],
  "C": [
    "object3"
  ]
}

Demo

To then convert this into your desired structure, append another to_entries

jq -s '
  add | reduce to_entries[] as $e ({}; .[$e.value.tags[]]  = [$e.key])
  | to_entries | map({tag:.key, objects:.value})
' *.json

[
  {
    "tag": "A",
    "objects": [
      "object1",
      "object2",
      "object4"
    ]
  },
  {
    "tag": "B",
    "objects": [
      "object2"
    ]
  },
  {
    "tag": "C",
    "objects": [
      "object3"
    ]
  }
]

Demo