Develop expression language for executing rules

[ubdbra001]

bidsschematools (BST) has tooling for parsing the Javascript-ish rules in the schema, but the validator currently doesn't parse rules and check these parsed rules.

This issue should address the following:

• Parse the rules (using or possibly reimplementing the BST tooling)
• Execute the selector rules on specific file contexts to identify which checks should be run for a specific file
• Execute the checks against that file (and report errors)?

Suggestion was to start by looking at the pyparsing package for examples.

Is this roughly correct @effigies? What did I miss?

[effigies]

Looks right

[ubdbra001]

Use schema.meta.expressiontests as starting point for parsing and executing, they don't need an evaluation context.

[effigies]

from types import SimpleNamespace
context = SimpleNamespace()

context.true = True
context.false = False
context.null = None

def intersects(a, b):
    return list(dict.fromkeys(a).keys() & b)

intersects(range(5), range(3, 8))

context.intersects = intersects

def evalexpr(ast, context):
    # match ast:

    pass

from bidsschematools.expressions import parse

parse("false || true")
parse('[1, 2, 3]')

[effigies]

Came up in the call, it might be elegant to create the expression language constants as a module:

true = True
false = False
null = None

def intersects(a, b):
    return list(set(a) & set(b))

__all__ = ['true', 'false', 'intersects']
constants = {key: globals()[key] for key in __all__}

Then:

from .expression_language import constants

result = evaluate(ast, constants | bids_context)