Improved Grammar, Implemented Parsing to IR

[Nikil-Shyamsunder]

Grammar and Lexing Changes:

• Removed all operators more complex than bitwise.
• Eliminated else if statements and other complexities for now.
• Introduced struct syntax in the grammar.
• Functions now require a struct type parameter, specified using angle bracket syntax (<T>).
• Added assert_eq as a new statement type.

IR Changes:

• Separated Binary and Unary operations for better structure and clarity.
• Introduced BoxedExpr:
  • Serves as an intermediate representation before full parsing into Expr.
  • Ensures compatibility with pest’s PrattParser, avoiding Rust’s compilation issues when mutating the same structure within multiple closures.
  • Temporarily stores start and end characters until they can be added to the transaction during parsing.
• Added Helper Functions symbol_id_from_name and struct_id_from_name; required for correct parsing of symbols and struct definitions.

Serialize & Typecheck Updates:

• Minor adjustments to accommodate the new IR structure.
• Operators (Ops) are now independent enumerations.

Parser Changes:

• Implemented the entire parsing stage within parser.rs.
• Test Suite Overhaul:
  • Added new tests for all updated files.
  • Removed obsolete files that no longer define structs.
  • Introduced new struct definitions where necessary.
  • Tests now parse and then reserialize.

[fpham0701]

For the function symbol_id_from_name and other getter functions that you made, isn't it possible to just index utilizing the symbol table itself? I.e. st[(name)]?

[fpham0701]

I saw one of the calyx_go_done tests still named calyx_go_down so I just changed it. woops!

[Nikil-Shyamsunder]

For the function symbol_id_from_name and other getter functions that you made, isn't it possible to just index utilizing the symbol table itself? I.e. st[(name)]?

the field by_name of SymbolTable is private, so I think this is the best way to do it.

[ekiwi]

This is what @fpham0701 is refering to: https://github.com/ekiwi/protocols/blob/6ad45f0259494569b5c3b7d679df4da26565a7d9/src/ir.rs#L330

[Nikil-Shyamsunder]

This required changing the implementation of the indexing and fixing the test cases that use it accordingly. I then removed the helper function I was using and replaced it with this more simple indexing.

The only thing to consider as a downside is that it is harder to capture a parsing failure (e.g. someone attempts to assign to an undeclared variable, an expression refers to a non-existent symbol) and return an expressive error message now because this panics with "no entry found for key"

[Nikil-Shyamsunder]

Okay, I've reverted for now after deciding that the panic issue causes problems for proper error handling in the parsing stage.