jq how to filter by field value in list

Question

having an input like this:

[
    {
        "foo": "aaa",
        "bar": 111
    },
    {
        "foo": "bbb",
        "bar": 111
    },
    {
        "foo": "ccc",
        "bar": 222
    },
    {
        "foo": "aaa",
        "bar": 333
    },
    {
        "foo": "ddd",
        "bar": 444
    }
]

i would like to select all objects with "foo" key equal to "aaa" or "bbb". so the solution is obvious:

.[] | select (.foo=="aaa" or.foo=="bbb" ) .[] | select (.foo=="aaa" or.foo=="bbb" ) ( https://jqplay.org/s/x7FGo1uQNW )

but i would like to enhance it and replace that x=y or x=z to sql'ish styled x in (y,z) . and i got stuck, as the natural try:

.[] | select (.foo in (["aaa", "bbb"]) )

results in an error:

jq: error: syntax error, unexpected IDENT, expecting ';' or ')' (Unix shell quoting issues?) at, line 1:

i also tried this:

.[] | select (.foo | in (["aaa", "bbb"]) )

but also not great...

jq: error (at:21): Cannot check whether array has a string key

is this even posible?

Answer 1

hmm, i managed to do it with this:

.[] | select(.foo as $tmpvar | ["aaa", "bbb"] | index ($tmpvar ) )

https://jqplay.org/s/g7AyRgARdU

According to this answer: https://stackoverflow.com/a/46470951/2244766 in versions above 1.5 there is a new IN operator that makes life a bit easier:

.[] | select(.foo|IN("aaa","bbb"))

Answer 2

The SQL-style operators don't work well for me as a straight-forward selection mechanism; I believe they have a very specific use-case, for which they are uniquely suitable for, and for anything else they're (at best) clunky. At least that's been my experience. And I haven't really figured out what that specific use-case is, either.

With all of that as a backdrop, my recommendation is to use a simple regex test:

map(select(.foo | test("aaa|bbb")))

Given the example JSON:

<~> $ jq . /tmp/so4229.json
[
  {
    "foo": "aaa",
    "bar": 111
  },
  {
    "foo": "bbb",
    "bar": 111
  },
  {
    "foo": "ccc",
    "bar": 222
  },
  {
    "foo": "aaa",
    "bar": 333
  },
  {
    "foo": "ddd",
    "bar": 444
  }
]

the above filter would result in:

<~> $ jq 'map(select(.foo | test("aaa|bbb")))' /tmp/so4229.json
[
  {
    "foo": "aaa",
    "bar": 111
  },
  {
    "foo": "bbb",
    "bar": 111
  },
  {
    "foo": "aaa",
    "bar": 333
  }
]

If you need to generate the regex based on other data within the JSON, you can do that, too:

. as $data | map(select(.bar==111) | .foo) | join("|") as $regex | . = $data | map(select(.foo | test($regex)))

which would result in:

<~> $ jq '. as $data | map(select(.bar==111) | .foo) | join("|") as $regex | . = $data | map(select(.foo | test($regex)))' /tmp/so4229.json
[
  {
    "foo": "aaa",
    "bar": 111
  },
  {
    "foo": "bbb",
    "bar": 111
  },
  {
    "foo": "aaa",
    "bar": 333
  }
]

There may be a better way to run through the JSON twice (once to get the regex values, once to use it).