Suffix match order bug

[ChristopherChudzicki]

Note: I came across this bug while rewriting the parser/evaluator. I'm creating the issue mostly so I can reference it in a future PR. (A PR which would also fix this bug.)

from mitxgraders.helpers.calc import evaluator

variables = {}
functions = {}
suffixes_ok = {  'ab': 2 }
suffixes_error = { 'a': 10, 'ab': 2 }

expr = '5ab'

evaluator(expr, variables, functions, suffixes_ok) # works as expected
evaluator(expr, variables, functions, suffixes_error) # throws an error

The issue is that a is a front substring of ab. I think that the grammar was originally intended to handle this case because in defining suffixes, MatchFirst is used, which can be used to specify the order in which to check. (So above, MatchFirst(['ab', 'a']) would have been fine.) But we give MatchFirst a bunch of dictionary keys, which aren't ordered:

# Define our suffixes
suffix = MatchFirst(Literal(k) for k in self.suffixes.keys())

[ChristopherChudzicki]

Incidentally, it is also possible to do some pretty silly things with suffixes:

from mitxgraders.helpers.calc import evaluator

variables = {}
functions = {}
suffixes_ok = { '-1': 10 }

expr = '5-1'

result, _ = evaluator(expr, variables, functions, suffixes_ok)
print(result) # 10.0

[jolyonb]

You sure that doesn't give you 50?

I was aware the suffixes had these issues, but since we were restricting them to % and hard-coded metric suffixes, it seemed safe enough. Feel free to fix if you like. I suspect that it should probably restrict to single character suffixes that cannot include certain characters.

[jolyonb]

Resolved by #134