Support deep evaluation in evaluate_keyed_by

[hneiva]

Add deep=False argument to evaluate_keyed_by, and if deep is true, then try to evaluate all attributes of provided value, traversing objects and lists.

As an alternative, a "wrapper" function (ie deep_evaluate_keyed_by) that traverses an object/list and calls the original evaluate_keyed_by.

[RamitPandey22]

hey I am interested to contribute on this if no one is already maintaining this?

[hneiva]

Hello @RamitPandey22 !
I checked with the team and no one is currently working on this.

Here is a quick code example of this request:

from taskgraph.util.keyed_by import evaluate_keyed_by
import json

data = {
    "list-type": [
        {
            "name": {
                "by-foo": {
                    "bar": "name is foobar0",
                    "baz": "name is foobaz0",
                }
            }
        },
        {
            "name": {
                "by-foo": {
                    "bar": "name is foobar1",
                    "baz": "name is foobaz1",
                }
            }
        },
    ],
    "dict-type": {
        "one": {
            "name": {
                "by-foo": {
                    "bar": "name is foobar0",
                    "baz": "name is foobaz0",
                }
            }
        },
        "two": {
            "name": {
                "by-foo": {
                    "bar": "name is foobar1",
                    "baz": "name is foobaz1",
                }
            }
        },
    },
}

# Evaluating a value as expected
evaluated_value = evaluate_keyed_by(data["dict-type"]["one"]["name"], "example", {"foo": "bar"})
print(evaluated_value)
assert evaluated_value == "name is foobar0"

# This would be nice to have (passing in the entire data structure)
evaluated_structure = evaluate_keyed_by(data, "example", {"foo": "bar"})
print(json.dumps(evaluated_structure, indent=4))
assert evaluated_structure["dict-type"]["one"]["name"] == "name is foobar0"

If you have any questions on how to get started, feel free to ask here or in the Matrix Firefox CI room. (My handle is @hneiva there as well)