Sorting partially missing arrays with jmespath

Question

I am trying to sort this source data outeractivities by outer_name and, then, activities by and inner_name:

Sourcedata.json:

[
    {
        "outeractivities": [
            {
                "outer_name": "2"
            },
            {
                "outer_name": "1",
                "activities": [
                    {
                        "inner_name": "B"
                    },
                    {
                        "inner_name": "A"
                    }
                ]
            }
        ]
    }
]

Expected output:

[
    {
        "outeractivities": [
            {
                "activities": [
                    {
                        "inner_name": "A"
                    },
                    {
                        "inner_name": "B"
                    }
                ],
                "outer_name": "1"
            },
            {
                "outer_name": "2"
            }
        ]
    }
]

By using

cat Sourcedata.json | jp "[].{outeractivities:sort_by(outeractivities[]. \
{outer_name:outer_name, activities:sort_by(activities[].{inner_name:inner_name},&inner_name) \
},&outer_name)  \
}"

I will get the exception:

Error evaluating JMESPath expression: Invalid type for: <nil>, expected: []jmespath.jpType{"array"}

If I change

{
          "outer_name": "2"
}

to

{
          "outer_name": "2",
          "activities": []
}

It does work.
How can I fix this?

β.εηοιτ.βε · Accepted Answer · 2022-01-26T09:43:44.870

Since the error is occurring due to the fact that sort_by is failing to process properly when the property to sort on does not exists, what you could do is to make a list of two lists:

one with all the outeractivities having the activities property, sorted on this field
one with all the outeractivities that does not have the activities property

And, then, flatten those two lists.

Given the query:

[].{ 
  outeractivities: sort_by(
    [
      outeractivities[?activities == null], 
      outeractivities[?activities != null].{ 
        outer_name: outer_name, 
        activities: sort_by(activities, &inner_name)
      }
    ] [], 
    &outer_name
  )
}

This yields the expected:

[
  {
    "outeractivities": [
      {
        "outer_name": "1",
        "activities": [
          {
            "inner_name": "A"
          },
          {
            "inner_name": "B"
          }
        ]
      },
      {
        "outer_name": "2"
      }
    ]
  }
]

Sorting partially missing arrays with jmespath

1 Answers1