Maybe add a NOT predicate about mumps-parser HOT 2 OPEN

On the one hand, adding the NOT and AND predicates makes this an actual PEG parser.
On the other hand, adding those predicates would make it much, much harder to use the Grammar Arrays to generate sample MUMPS code for fuzz testing.

Additionally, automatically creating a railroad track syntax diagram using the grammar array is pretty trivial with the current state, but it's not possible to generate proper diagrams if we have AND and NOT predicates. This is because an AND predicate conceptually means that an input string has to follow two separate "railroad tracks" at the same time.

from mumps-parser.

Augmenting the code to handle the AND and NOT predicates is actually pretty easy. It would just be a change to the ParseSubtree code to check for certain nodes in the map that tell us what we need to check for and exclude. Adding support for the the reasonable subnodes to the grammar array is pretty straight-forward.

For example, if these predicates get added, the grammar map for tokens would need to be augmented to hold data like this:
; @subtree = "token"
; @subtree@("value") = the name of the token. This should match some name occurring in the first coordinate of the @Map
; @subtree@("force") = the behavior of the parser to force when parsing this or deeper subtrees.
; @subtree@("AlsoParsableAs") = a subtree that the input stream must be parsable as. This is the AND predicate
; @subtree@("NotParsableAs") = a subtree that the input stream must not be parsable as. This is the NOT predicate

Adding these predicates wouldn't require that we change the structure of the parse tree output.

from mumps-parser.