changing the json using jq in bash

Question

i have the following json

{
    "name" : "qwerty",
    "values" :[
    {
        "field1" : [
            "val1"
            ],
        "field2" : [
            "val2"
            ],
        "name1" : [["a", "b"], ["c", "d"]]
    },
    {
        "field1" : [
            "val3"
            ],
        "field2" : [
            "val4"
            ],
        "name1" : [["a", "b"], ["c", "d"]]
    },
    {
        "field1" : [
            "val5"
            ],
        "field2" : [
            "val6"
            ],
         "name1" : [["a", "b"], ["c", "d"]]
    }
    ]
}

I need to change the above json to the following using jq in bash

{
    "name" : "qwerty",
    "values" :[
    {
        "field1" :  "val1",
        "field2" :  "val2",
        "new_name" : [["a", "b"], ["c", "d"]]
    },
    {
        "field1" : "val3",
        "field2" : "val4",
        "new_name" : [["a", "b"], ["c", "d"]]
    },
    {
        "field1" : "val5",
        "field2" : "val6",
        "new_name" : [["a", "b"], ["c", "d"]]
    }
    ]

}

Here i am facing the following issues :

I tried parsing the inner json with tag values and replace the '[' ']' with spaces, however, when i try to put the "values" in a variable in the form of list, jq is prettifying and then showing each new line as a element of an array.

The number of inner jsons in the values array is not fixed.

Can some one please help me with framing the jq statement to be run in bash to make the required changes.

@fedorqui thnx for pointing that out, corrected the question — vamsi, Aug 18 '16 at 10:12
Do you know that it is specifically `.values[].field1` and `.values[].field2` you want to change, or is it any singleton list anywhere that should be collapsed to a single value? — chepner, Aug 18 '16 at 11:50
Yes "field1, field2, name" are fixed in each json. But the number of jsons is not fixed. — vamsi, Aug 18 '16 at 12:02
sorry for reopening the question. There is a alight change in my requirement, i would like to know if this can be done if we dont have the number of fields also fixed. That is "field1, field2, name" are also not fixed in each json. — vamsi, Aug 24 '16 at 06:25

Tiago Lopo · Answer 1 · 2016-08-18T12:10:06.097

The snippet below should do what you want:

jq '{
    "name": .name, 
    "values": [ 
        {
            "field1" : .values[0].field1[0], 
            "field2" : .values[0].field2[0],
            "New_name": .values[0].name1  
        },
        {
            "field1" : .values[1].field1[0],
            "field2" : .values[1].field2[0],
            "new_name" : .values[1].name1
        },
        {
            "field1" : .values[2].field1[0],
            "field2" : .values[2].field2[0],
            "new_name" : .values[2].name1
        }
    ]  
}' < /tmp/input.json

EDIT

Since the number of objects are not fixed the snippet below will do:

jq '{ 
        "name" : .name,
        "values" : [
                .values[] as $in | 
                { 
                        "field1" : $in.field1[0],
                        "field2" : $in.field2[0],
                        "new_name" : $in.name1 
                }
        ]

}' < /tmp/input.json

thnx for the quick response. But, the number of inner jsons in the values array is not fixed. — vamsi, Aug 18 '16 at 10:54

chepner · Accepted Answer · 2016-08-18T17:54:34.980

2

This should work; I'm not sure if there is a way to refactor the assignments to field1 and field2:

jq '.values[] |= (.field1=.field1[0] | .field2=.field2[0])' tmp1.json

edited Aug 18 '16 at 17:54

answered Aug 18 '16 at 12:59

chepner

497,756
71
530
681

Another option is to create an object with the new values and merge it. But I think that's uglier and potentially more destructive than the assignments. – Jeff Mercado Aug 18 '16 at 16:01
Based on the name alone, I thought `foreach` might be relevant, but I can't seem to wrap my mind around how it works. It was all I could do to come up with something like `jq '.values[] |= (foreach ["field1", "field2"] as $f (empty;.[$f]=.[$f][0];empty))' tmp1.json` that at least complies; it produces no output. – chepner Aug 18 '16 at 16:05
In my head, the way I see it is `foreach` is like `reduce` but with an additional extraction part. So in your example in your comment, you initialized with an empty value, and try to assign values to it. But empty is empty so it yields no results for every iteration. Then for the extraction part, you extracted empty so nothing comes from it. With that approach, you could just stick with `reduce` (with a slight adjustment): `.values[] |= reduce ("field1", "field2") as $f (.;.[$f]=.[$f][0])` – Jeff Mercado Aug 18 '16 at 16:16
OK, nice. I think I understand `reduce` a little better now. – chepner Aug 18 '16 at 16:43
@JeffMercado But the output json being created is not the one i want, though it has the changes required. Being new to jq, I tried playing around with reduce (jq '. | reduce("field1","field2") as $f (.values[];.[$f]=.[$f][0])' tmp1.json) , but it didnt work. – vamsi Aug 18 '16 at 17:47
@vamsi Sorry, I was eyeballing the output and though I saw that it was the same. – chepner Aug 18 '16 at 17:55
1

Did some thinking and here's another way you could modify the multiple values: `.values[] |= (.["field1", "field2"] |= .[0])`. – Jeff Mercado Aug 19 '16 at 17:39
@JeffMercado sorry for reopening the question, but i would like to know if this can be done if we dont have the number of fields also fixed. That is "field1, field2, name" are also not fixed in each json. – vamsi Aug 24 '16 at 06:24

score 0 · Answer 3 · answered Aug 30 '17 at 21:27

0

Here is another solution.

  .values |= map({
     field1:   .field1[0],
     field2:   .field2[0],
     new_name: .name1
  })

answered Aug 30 '17 at 21:27

jq170727

13,159
3
46
56

changing the json using jq in bash

3 Answers3

Linked