gbuoy/lib/parser.pegjs

// AST factories
{
    const attr = (name, args) => ({ name, args })
    const stmt = (stmt_type, args) => ({ type: "stmt", stmt_type, args })
    const decl = (decl_type, name, attrs, args) => ({ type: "decl", decl_type, name, attrs: attrs || [], args })
    const uexpr = (op, arg) => ({ type: "unary", op, arg })
    const bexpr = (op, left, right) => ({ type: "binary", op, left, right })

    const assocl = (head, tail) => tail.reduce((left, arg) => {
        const [op, right] = arg
        return bexpr(op, left, right)
    }, head)
}

Program = WS @(Decl / Stmt)* WS

// Declarations
Decl = VarDecl
VarDecl = attrs:Attrs? type:Type name:Ident SEMI { return decl("var", name, attrs, { type }) }

// Statements
Stmt = @AssignStmt SEMI
AssignStmt = name:Ident ASSIGN expr:Expr { return stmt("assign", { name, expr }) }

// Expressions
Expr = AddExpr

AddExpr = head:UnaryExpr tail:(op:AddOp e:UnaryExpr)* { return assocl(head, tail || []) }
AddOp = PLUS { return "add" }

UnaryExpr = op:UnaryOp? e:BaseExpr { return op ? uexpr(op, e) : e }
UnaryOp = TILDE { return "bit_negate" }
    / MINUS { return "arith_negate" }

BaseExpr = Ident / Number / LPAREN @Expr RPAREN

// Attributes
Attrs = LATTR @AttrList RATTR
AttrList = h:Attr t:(COMMA @Attr)* { return [h].concat(t) }
    / '' { return [] }
Attr = name:Ident args:(LPAREN @AttrArgs RPAREN)? { return attr(name, args || []) }
AttrArgs = h:AttrArg t:(COMMA @AttrArg)* { return [h].concat(t) }
    / '' { return [] }
AttrArg = Ident / Number

// Types
Type = PrimitiveType
PrimitiveType = S8 / U8 / U16

// Terminals
Ident = h:[a-zA-Z_] t:[0-9a-zA-Z_]* WS { return h + t.join('') }
Number = digits:[0-9]+ WS { return parseInt(digits.join(''), 10) }
    / '0x' digits:[a-fA-F0-9]+ WS { return parseInt(digits.join(''), 16) }
    / '0b' digits:[01]+ WS { return parseInt(digits.join(''), 2) }

// Terminal keywords
S8 = @'s8' WS
U8 = @'u8' WS
U16 = @'u16' WS

// Terminal symbols
PLUS = '+' WS
TILDE = '~' WS
MINUS = '-' WS
SEMI = ';' WS
ASSIGN = '<-' WS
LPAREN = '(' WS
RPAREN = ')' WS
COMMA = ',' WS
LATTR = '[[' WS
RATTR = ']]' WS

// Misc
WS = [ \t\r\n]*