jq操作json文件

Question

Having as input below sample json:

{
    "name_A": {
        "logfile_one": [{
            "issue_desc": "desiredvalueone"
        }, {
            "issuetime": "desiredvaluetwo"
        }],
        "output": ["46312"]
    },
    "name_B": {
        "logfile_two": [{
            "issue_desc": "desiredvaluethirtyfour"
        }, {
            "issuetime": "desiredvaluetwo"
        }],
        "output": ["1", "2"]
    }
}

We would like the following output:

{
    "desiredvalueone": [{
        "name_A": "logfile_one"
    }],
    "desiredvaluetwo": [{
        "name_A": "logfile_one",
        "name_B": "logfile_two"
    }],
    ...
}

using t o_entries[] | .key o_entries[] | .key for instance, we manage to get " name_A " " name_B ", but very far from what we try to achieve. Or with

[to_entries[] | {"key": .key, "value": .value[]}] | from_entries

output was

{
  "name_A": [
    "46312"
  ],
  "name_B": [
    "1",
    "2"
  ]
}

Answer 1

The .... in the given expected output introduces several uncertainties beyond those in the problem statement itself, but the following program does produce results that are consistent with the expected output as shown.

The key to the solution given here is the helper function for combining a stream of JSON objects:

# Combine a stream of JSON objects into a single object
# by converting the value v to [v] at each
# key, and then appending the arrays at each key.
def combine(s):
  reduce s as $o ({};
    reduce ($o|keys[]) as $k (.; .[$k] += [$o[$k]]  ) );

The problem as I understand it can now be solved using to_entries/0 three times; if you're not familiar with that filter, the following might seem fairly incomprehensible:

combine( to_entries[]
   | .key as $k
   | .value
   | to_entries[]
   | select( .key | startswith("logfile_"))
   | .key as $innerkey
   | .value[]
   | to_entries[]
   | { (.value): { ($k): $innerkey }}
 ) | map_values(add)

The output with the given input is:

{
  "desiredvalueone": {
    "name_A": "logfile_one"
  },
  "desiredvaluetwo": {
    "name_A": "logfile_one",
    "name_B": "logfile_two"
  },
  "desiredvaluethirtyfour": {
    "name_B": "logfile_two"
  }
}

Answer 2

Here is a solution which only assumes the input has the following structure

{ 
  "k": {
     "k2": [{
        "k3": "v"

It works by generating a temporary array of objects of the format

{"v":"desiredvalueone","k":"name_A","k2":"logfile_one"}

then grouping by .v and combining groups and wrapping values in an array as requested.

[
  . as $d
| [
      keys[] as $k                           # e.g "name_A"...
    | ($d[$k] | keys[]
       | select($d[$k][.][0]|type=="object")
       ) as $k2                              # k2:  "logfile_one"
    | $d[$k][$k2][] as $v                    # v:  {"issue_desc":"desiredvalueone"},
    | ($v|keys[]) as $k3                     # k3: "issue_desc"
    | {v:$v[$k3], k:$k, k2:$k2}              # {"v":"desiredvalueone","k":"name_A","k2":"logfile_one"}
  ]
| group_by(.v)[]
| reduce .[] as $t (
    {}
  ; .[$t.v] += {($t.k): $t.k2}
  )
]
| add
| map_values([.])

The final output is

{
  "desiredvalueone": [
    {
      "name_A": "logfile_one"
    }
  ],
  "desiredvaluethirtyfour": [
    {
      "name_B": "logfile_two"
    }
  ],
  "desiredvaluetwo": [
    {
      "name_A": "logfile_one",
      "name_B": "logfile_two"
    }
  ]
}