parse: include location of rule ref in AST

[anderseknert]

AST information from a rule head currently only includes the location of the value, not the rule ref var itself.

package location

foo := "bar"

opa parse location.rego --format json --json-include locations

{
  "rules": [
    {
      "head": {
        "name": "foo",
        "value": {
          "location": {
            "file": "location.rego",
            "row": 3,
            "col": 8
          },
          "type": "string",
          "value": "bar"
        },
        "assign": true,
        "ref": [
          {
            "type": "var",
            "value": "foo"
          }
        ]
      }
    }
  ]
}

For tooling that works with the AST, this location information would be useful.

[charlieegan3]

Hmm, interesting. I think that we might be able to dig into where NewHead and RefHead are used and make sure we're setting the location on the Head structs as they're initially created. It supports the JSON opts, but I don't think the location is getting set during parsing.

[Trolloldem]

I have taken a look into the code of NewHead and RefHead as suggested by @charlieegan3, and noticed that when they are called in parseHead, the defer function should handle the process of setting the Location of the Head of a Rule.

While investigating the motivation that causes the output to have missing location, I tried to use the following code as an example:

package test

default allow = false
allow = true {
   input.method == "GET"
   input.path = ["getSalary", user]
   input.user == user
}

In this case the output relative to the line allow = true is the following:

      "head": {
        "name": "allow",
        "value": {
          "location": {
            "file": "t2.rego",
            "row": 4,
            "col": 9
          },
          "type": "boolean",
          "value": true
        },
        "ref": [
          {
            "type": "var",
            "value": "allow"
          }
        ],
        "location": {
          "file": "t2.rego",
          "row": 4,
          "col": 1
        }
      },
      "location": {
        "file": "t2.rego",
        "row": 4,
        "col": 1
      }

There is a similar part of the output for the default allow = false.

What I noticed after this, is that in parseModule, the statement used in the example of @anderseknert is handled by the Body case insted of the *Rule one. This causes the method ParseRuleFromBody to create a new Rule struct. The problem with this approach is that the code used to configure the correct JSON setting in Parse is never called on this new struct.

I have tried to add similar code in the switch statement mentioned above and this is the output for the example of @anderseknert:

"rules": [
    {
      "body": [
        {
          "index": 0,
          "location": {
            "file": "test.rego",
            "row": 3,
            "col": 8
          },
          "terms": {
            "location": {
              "file": "test.rego",
              "row": 3,
              "col": 8
            },
            "type": "boolean",
            "value": true
          }
        }
      ],
      "head": {
        "name": "foo",
        "value": {
          "location": {
            "file": "test.rego",
            "row": 3,
            "col": 8
          },
          "type": "string",
          "value": "bar"
        },
        "assign": true,
        "ref": [
          {
            "type": "var",
            "value": "foo"
          }
        ],
        "location": {
          "file": "test.rego",
          "row": 3,
          "col": 1
        }
      },
      "location": {
        "file": "test.rego",
        "row": 3,
        "col": 1
      }
    }
  ]

I can work on this to further investigate if only this addition fixes the problem and add some tests.

[anderseknert]

That'd be amazing! Thanks @Trolloldem 😃

[anderseknert]

Seems to be missing still in some cases: #6860