feat: Add basic url filter parser

[AdityaHegde]

For human readable url params we need a parser for filter param.
Possible expressions,

country IN ('US','IN') and state = 'ABC'
country NIN ('US','IN') and (state = 'ABC' or lat >= 12.56)
country NIN ('US','IN') and state having (lat >= 12.56)

• Write a grammer and parser for the expression.
• Convert the parsed object to V1Expressioin

[ericpgreen2]

Overall this looks good, but left various comments/questions throughout. Additionally, for posterity, a README.md file in the url-state/filters directory would be helpful to explain our filter language, nearly, and how to generate the parser.

[ericpgreen2]

What happens when the string cannot be properly parsed? Can we add some test cases for the failure cases to show what happens? (Given user-provided strings will certainly fail occasionally)

[AdityaHegde]

Added a couple to demonstrate the error. We should get a UX pass eventually.

[ericpgreen2]

Might be clearer to have a discrete andPostprocessor and an orPostprocessor

[AdityaHegde]

They have the exact same implementation. This is the suggested usage of post post processors, being able to reuse them. Feel like it would be redundant.

[ericpgreen2]

Where are these referenced? I don't see any explicit references, but I'm guessing maybe the primitive types at the bottom (e.g. sqstring, int, decimal) are coming from here?

[AdityaHegde]

Yes, the docs arent here for these right now. They point to the source code of these files: https://github.com/kach/nearley/tree/master/builtin

[ericpgreen2]

I'm not totally following these. At first glance, it looks like adding these conditions would prevent multiple "ANDs" and multiple "ORs" in the same statement. But I see in the test cases that "X and Y and Z" is possible. So, mind just clarifying in the comment what these do?

[AdityaHegde]

These are used in expr (boolean_expr _ "AND"i _):+ non_and_expr for AND. It matches continuous AND ended by a non AND expr.

Added some comments around this.