How to get root keys and key types using jq

Question

Let's take this simple data file: http://data.cdc.gov/data.json

I know how to get the root key names:

jq keys_unsorted[] -r data.json

which produces:

@context
@id
@type
conformsTo
describedBy
dataset

And I know how to get the key types:

jq 'map(type)' data.json

Which produces:

[
  "string",
  "string",
  "string",
  "string",
  "string",
  "array"
]

Is there not some way of combining this in returning pairs? (what I am really trying to do is find the key name of the first root level array, if any). I can write a routine to figure it out, but this seem inelegant.

Bonus side question: How to you determine the type of a key (e.g., I would send "dataset" to jq in some form and would get "array" in return)?

Answer 1

jq 'first(path(.[] | select(type == "array"))[0])' < data.json

The top-level items .[] are filtered out with select(type == "array") which selects only items of array type; path() returns array representation of the path in ., i.e. the key names of the array items; first() extracts the first path.

So the result of the command is key name of the first top-level array item.

Sample Output

"dataset"

---

How to you determine the type of a key (e.g., I would send "dataset" to jq in some form and would get "array" in return).

You probably mean "the type of a value", because the keys must be strings in JSON. If the path is known (e.g. .dataset), then the type of the object can be fetched with the type function:

jq '.dataset | type' < data.json

Sample Output

"array"

Answer 2

The simplest approach to writing queries that depend on both key names and values is to use one of the "*_entries" family of filters. In your case:

$ jq -c 'to_entries[] | [.key, (.value|type)]' data.json

["@context","string"]
["@id","string"]
["@type","string"]
["conformsTo","string"]
["describedBy","string"]
["dataset","array"]    

If you wanted this presented in a more human-readable fashion, consider using @csv or @tsv, e.g.

$ jq -r 'to_entries[] | [.key, (.value|type)] | @csv' data.json
"@context","string"
"@id","string"
"@type","string"
"conformsTo","string"
"describedBy","string"
"dataset","array"

Or with less noise:

$ jq -r 'to_entries[] | "\(.key) \(.value|type)"' data.json
@context string
@id string
@type string
conformsTo string
describedBy string
dataset array

Bonus question

Here's a parametric approach to the second question. Let the file query.jq contain:

.[$key]|type

Then:

$ jq -r --arg key dataset -f query.jq data.json
array