thecookiess / rust-compiler

First attempt at making a compiler

an attempt at making a compiler in rust!

Future "Valid Syntax", negative whitespace significant indentation!

          hhh {
        eee {
      aaa {
   ddd {
aaa }
  ccc } 
    hhh }
      eee }

Core Functionality Required:

• variables WITH DATA TYPES
• operations: (boolean, logical//bitwise, binary)
• control flow branches: (if/else)
• loops: (while/for)
• types
• functions: (recursion << should be free?)
• array
• user defined struct

Currently working on:

• if

  • rework lexer to handle multi-symbol keywords (i.e "==" or "<=")
  • parse boolean comparison
  • else & else if parsing
  • code generation
    • invert 'jump' conditions
    • unsigned vs signed comparison (diff jump instructions) (no types, all signed)
    • binary expr conditions
      • conditional expr, either has explicity bool comparison or implicit, expr = lhs: (lhs > 0)
      • cmp reg1, reg2 ; compare arguments
      • set(EQUIVALENCE e.g e, le) al; sets register 'al' (8 bit) to 1,0 depending on cmp flag
      • movzx output_reg, al ; Move Zero Xtend. copies 'al' into 'output_reg' && zero init reg1 bits.
  • types of scope
    • inherits variables from parent scope (if, else if, else)
    • doesn't (new function, UNLESS class, inherits 'self')

• (kinda done) split 'TokenKind': 'Symbol' .. 'LogicalOp' .. 'BinaryOp' .. etc

• comments

• re-design multi-symbol in Lexer: match against longest multi-symbol to shortest, until finds match (or illegal token.)

• update grammar to match code.

• variable reassignment (mutability)

• Operators

  • modulus '%' op
  • Associativity
  • binary
  • logical
    • i.e: if (5) | 5 != 0 so -> if (true)
    • i.e: if (!5) | 5 == 0 so -> if (false)
    • eval expr, jump if zero. << un-optimal, uses 'al' reg when not necessary
  • bitwise
  • unary
    • LogicalNot
    • BitwiseNot

• dynamically place variables on stack if they are(nt) used immediately.

  • on expr generation.

• Testing infrastructure.

• data types

  • bools: al register (8 bit)
  • u8: al
  • u16: ax
  • u32: eax
  • u64: rax

• functions

  • return type
  • arguments
  • body
  • program entry point ("main")

• Implement C equivalent operators:

  • comma: ','
  • assignment: '=' | '+=' | '/=' | ..
  • modulus: %
  • logical not: !
  • unary minus: i.e -10
  • function call: '()'
  • array subscript: '[]'
  • struct member: '.' | '->'
  • Increment/Decrement (OPTIONAL)
  • cast: (type)

• (Joke mode:) negative whitespace significance, the most nested piece of code has 0 indentations, everything out has an indentation.

  • compiler error on uppercase