corelang/new_ast.cpp

// Lexer::interns::map::allocator - array allocator, resizing
// Lexer::tokens                  - array allocator, resizing
//
// Parser::ast_arena                - arena for asts
// Lexer::interns::string_allocator - arena for interns
//
Intern_String keyword_struct;
Intern_String keyword_union;
Intern_String keyword_cast;
Intern_String keyword_enum;

Intern_String intern_void;
Intern_String intern_int;
Intern_String intern_str;
Intern_String intern_unsigned;

const U64 Parse_Ctx_ID = 115151;
struct Parse_Ctx:Lexer{
  Allocator *perm; // Stores: AST, tokens, interns
  Allocator *heap;

  U64 unique_ids;
  Map global_syms;
  Map type_map;

  Token empty_token;
  S64 indent;

  Intern_String intern(String string){
    return intern_string(&interns, string);
  }

  void init(Allocator *perm_allocator, Allocator *heap_allocator){
    perm = perm_allocator;
    heap = heap_allocator;

    global_syms = {heap};
    type_map    = {heap};

    lex_init(perm, heap, this);
    keyword_struct= intern_string(&interns, "struct"_s);
    keyword_union = intern_string(&interns, "union"_s);
    keyword_cast = intern_string(&interns, "cast"_s);
    keyword_enum  = intern_string(&interns, "enum"_s);
    interns.first_keyword = keyword_struct.str;
    interns.last_keyword  = keyword_enum.str;

    intern_void = intern_string(&interns, "void"_s);
    intern_int = intern_string(&interns, "int"_s);
    intern_str = intern_string(&interns, "String"_s);
    intern_unsigned = intern_string(&interns, "unsigned"_s);
  }
};

thread_local Parse_Ctx *pctx;
//-----------------------------------------------------------------------------
// AST
//-----------------------------------------------------------------------------
enum Ast_Kind{
  AST_NONE,

  AST_PACKAGE,

  AST_STR,
  AST_INT,
  AST_CAST,
  AST_IDENT,
  AST_INDEX,
  AST_UNARY,
  AST_BINARY,
  AST_COMPOUND_ITEM,
  AST_COMPOUND,

  AST_LAMBDA,
  AST_LAMBDA_PARAM,
  AST_CONST,
  AST_VAR,

  AST_TYPESPEC_IDENT,
  AST_TYPESPEC_POINTER,
  AST_TYPESPEC_ARRAY,
  AST_TYPESPEC_LAMBDA
};

struct Ast{
  U64 id;
  Ast_Kind kind;
  Token *pos;
};

struct Ast_Resolved_Type;
struct Ast_Typespec;
struct Ast_Expr:Ast{
  union{
    Intern_String intern_val;
    U64 int_val;
    struct{
      Token_Kind op;
      Ast_Expr *expr;
    }unary;
    struct{
      Token_Kind op;
      Ast_Expr *left;
      Ast_Expr *right;
    } binary;
    struct{
      Ast_Resolved_Type *type;
      Ast_Typespec *typespec;
      Array<Ast_Expr *> exprs;
    }compound;
    struct{
      Ast_Expr *name; // index | name
      Ast_Expr *index;
      Ast_Expr *item;
    }compound_item;
    struct{
      Ast_Expr *expr;
      Ast_Typespec *typespec;
    }cast;
    struct{
      Ast_Expr *expr;
      Ast_Expr *index;
    }index;
    struct{
      Intern_String name;
      Ast_Typespec *typespec;
    }lambda_param;
  };
};

struct Ast_Block : Ast {
  Array<Ast *> stmts;
};

struct Ast_Lambda : Ast_Expr {
  Array<Ast_Expr *> params;
  Ast_Typespec     *ret;
  Ast_Block        *block;
};

struct Ast_Resolved_Type;
struct Ast_Typespec:Ast{
  Ast_Resolved_Type *resolved_type;
  union{
    Ast_Typespec  *base;
    Intern_String  name;
    struct{
      Ast_Typespec *base;
      Ast_Expr     *expr;
    }arr;
    Ast_Lambda     *lambda;
  };
};

struct Ast_Decl:Ast{
  Intern_String name;
  union{
    struct{
      Ast_Typespec *typespec;
      Ast_Expr     *expr;
    }var;
  };
};

struct Ast_Package:Ast{
  Intern_String name;
  Array<Ast_Decl *> decls;
};

//-----------------------------------------------------------------------------
// AST Constructors beginning with expressions
//-----------------------------------------------------------------------------
#define AST_NEW(T,ikind,ipos)                                 \
  Ast_##T *result = exp_alloc_type(pctx->perm, Ast_##T);      \
  result->kind    = ikind;                                    \
  result->pos     = ipos;                                     \
  result->id      = ++pctx->unique_ids

function Ast_Expr *
ast_expr_string(Token *pos, Intern_String string){
  AST_NEW(Expr, AST_STR, pos);
  result->intern_val = string;
  return result;
}

function Ast_Expr *
ast_expr_identifier(Token *pos, Intern_String string){
  AST_NEW(Expr, AST_IDENT, pos);
  result->intern_val = string;
  return result;
}

function Ast_Expr *
ast_expr_integer(Token *pos, S64 integer){
  AST_NEW(Expr, AST_INT, pos);
  result->int_val = integer;
  return result;
}

function Ast_Expr *
ast_expr_binary(Ast_Expr *left, Ast_Expr *right, Token *op){
  AST_NEW(Expr, AST_BINARY, op);
  result->binary.op    = op->kind;
  result->binary.left  = left;
  result->binary.right = right;
  return result;
}

function Ast_Expr *
ast_expr_compound(Token *pos, Ast_Typespec *typespec, Array<Ast_Expr *> exprs){
  AST_NEW(Expr, AST_COMPOUND, pos);
  result->compound.typespec = typespec;
  result->compound.exprs    = exprs.tight_copy(pctx->perm);
  return result;
}

function Ast_Expr *
ast_expr_compound_item(Token *pos, Ast_Expr *index, Ast_Expr *name, Ast_Expr *item){
  AST_NEW(Expr, AST_COMPOUND_ITEM, pos);
  result->compound_item.name  = name;
  result->compound_item.index = index;
  result->compound_item.item  = item;
  return result;
}

function Ast_Expr *
ast_expr_cast(Token *pos, Ast_Expr *expr, Ast_Typespec *typespec){
  AST_NEW(Expr, AST_CAST, pos);
  result->cast.expr     = expr;
  result->cast.typespec = typespec;
  return result;
}

function Ast_Expr *
ast_expr_unary(Token *pos, Token_Kind op, Ast_Expr *expr){
  AST_NEW(Expr, AST_UNARY, pos);
  result->unary.expr = expr;
  result->unary.op   = op;
  return result;
}

function Ast_Expr *
ast_expr_index(Token *pos, Ast_Expr *expr, Ast_Expr *index){
  AST_NEW(Expr, AST_INDEX, pos);
  result->index.expr = expr;
  result->index.index = index;
  return result;
}


function Ast_Lambda *
ast_lambda(Token *pos, Array<Ast_Expr *> params, Ast_Typespec *ret){
  AST_NEW(Lambda, AST_LAMBDA, pos);
  result->params = params.tight_copy(pctx->perm);
  result->ret    = ret;
  return result;
}

function Ast_Expr *
ast_expr_lambda_empty(Token *pos){
  AST_NEW(Expr, AST_LAMBDA, pos);
  return result;
}

function Ast_Expr *
ast_expr_lambda_param(Token *pos, Intern_String name, Ast_Typespec *typespec){
  AST_NEW(Expr, AST_LAMBDA_PARAM, pos);
  result->lambda_param.name = name;
  result->lambda_param.typespec = typespec;
  return result;
}

//-----------------------------------------------------------------------------
// Typespecs
//-----------------------------------------------------------------------------
function Ast_Typespec *
ast_typespec_name(Token *pos, Intern_String name){
  AST_NEW(Typespec, AST_TYPESPEC_IDENT, pos);
  result->name = name;
  return result;
}

function Ast_Typespec *
ast_typespec_pointer(Token *pos, Ast_Typespec *base){
  AST_NEW(Typespec, AST_TYPESPEC_POINTER, pos);
  result->base = base;
  return result;
}

function Ast_Typespec *
ast_typespec_array(Token *pos, Ast_Typespec *base, Ast_Expr *expr){
  AST_NEW(Typespec, AST_TYPESPEC_ARRAY, pos);
  result->arr.base = base;
  result->arr.expr = expr;
  return result;
}

function Ast_Typespec *
ast_typespec_lambda(Token *pos, Ast_Lambda *lambda){
  AST_NEW(Typespec, AST_TYPESPEC_LAMBDA, pos);
  result->lambda = lambda;
  return result;
}

//-----------------------------------------------------------------------------
// Declarations
//-----------------------------------------------------------------------------
function Ast_Decl *
ast_decl_func(Token *pos, Intern_String name){
  AST_NEW(Decl, AST_LAMBDA, pos);
  result->name = name;
  return result;
}

function Ast_Decl *
ast_decl_var(Token *pos, Ast_Typespec *typespec, Intern_String name, Ast_Expr *expr){
  AST_NEW(Decl, AST_VAR, pos);
  result->var.expr     = expr;
  result->var.typespec = typespec;
  result->name         = name;
  return result;
}

function Ast_Decl *
ast_decl_const(Token *pos, Intern_String name, Ast_Expr *expr){
  Ast_Decl *result = ast_decl_var(pos, 0, name, expr);
  result->kind = AST_CONST;
  return result;
}

function Ast_Package *
ast_package(Token *pos, String name, Array<Ast_Decl *> decls){
  AST_NEW(Package, AST_PACKAGE, pos);
  result->decls = decls.tight_copy(pctx->perm);
  result->name = intern_string(&pctx->interns, name);
  return result;
}