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_return;
Intern_String keyword_if;
Intern_String keyword_else;
Intern_String keyword_for;
Intern_String keyword_cast;
Intern_String keyword_enum;

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

struct Ast_Package;
struct Sym;
struct Parse_Ctx:Lexer{
  Allocator *perm; // Stores: AST, tokens, interns
  Allocator *heap;

  U64 unique_ids;
  Map type_map;

  Ast_Package *resolving_package;
  Map resolved;
  Map syms;
  S32 scope;
  Array<Sym *> local_syms;

  Token empty_token;
  S64 indent;

  String_Builder gen;

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

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

    gen        = {perm};
    resolved   = {heap};
    syms       = {heap};
    type_map   = {heap};
    local_syms = {heap};

    lex_init(perm, heap, this);
    keyword_struct= intern("struct"_s);
    keyword_union = intern("union"_s);
    keyword_cast = intern("cast"_s);
    keyword_return = intern("return"_s);
    keyword_if = intern("if"_s);
    keyword_else = intern("else"_s);
    keyword_for = intern("for"_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_INIT,
  AST_IF,
  AST_IF_NODE,
  AST_RETURN,
  AST_BLOCK,
  AST_LAMBDA,
  AST_LAMBDA_ARG,
  AST_CONST,
  AST_VAR,

  AST_TYPESPEC_IDENT,
  AST_TYPESPEC_POINTER,
  AST_TYPESPEC_ARRAY,
  AST_TYPESPEC_LAMBDA
};

struct Ast{
  U64 id;
  Token *pos;

  Ast_Kind kind;
  // @todo?
  // bool is_atom: 1;
  // bool is_stmt: 1;
  // bool is_expr: 1;
  // bool is_decl: 1;
  // bool is_named: 1;
};

struct Ast_Resolved_Type;
struct Ast_Typespec;
struct Ast_Expr:Ast{};

struct Ast_Atom: Ast_Expr{
  union{
    Intern_String intern_val;
    U64 int_val;
  };
};

struct Ast_Compound_Item: Ast_Expr{
  Ast_Atom *name; // index | name
  Ast_Expr *index;
  Ast_Expr *item;
};

struct Ast_Compound: Ast_Expr{
  Ast_Resolved_Type *type;
  Ast_Typespec *typespec;
  Array<Ast_Compound_Item *> exprs;
};

struct Ast_Unary: Ast_Expr{
  Token_Kind op;
  Ast_Expr *expr;
};

struct Ast_Cast: Ast_Expr{
  Ast_Expr *expr;
  Ast_Typespec *typespec;
};

struct Ast_Index: Ast_Expr{
  Ast_Expr *expr;
  Ast_Expr *index;
};

struct Ast_Binary: Ast_Expr{
  Token_Kind op;
  Ast_Expr *left;
  Ast_Expr *right;
};

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

struct Ast_Return: Ast{
  Ast_Expr *expr;
};

struct Ast_Init: Ast{
  Token_Kind op;
  Ast_Atom *ident;
  Ast_Expr *expr;
};

struct Ast_If_Node: Ast{
  Ast_Expr  *expr ;
  Ast_Block *block;
  Ast_Init  *init;
};

struct Ast_If: Ast{
  Array<Ast_If_Node *> ifs;
};

struct Ast_Lambda_Arg: Ast_Expr{
  Intern_String name;
  Ast_Typespec *typespec;
};

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

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

struct Ast_Named:Ast{
  Intern_String name;
};

struct Ast_Var: Ast_Named{
  Ast_Typespec *typespec;
  Ast_Expr *expr;
};

struct Ast_Const: Ast_Named{
  Ast_Expr *expr;
};

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

function Ast_Typespec *ast_typespec_name(Token *pos, Intern_String name);
//-----------------------------------------------------------------------------
// AST Constructors beginning with expressions
//-----------------------------------------------------------------------------
#define AST_NEW(T,ikind,ipos)                                          \
  Ast_##T *result = exp_alloc_type(pctx->perm, Ast_##T, AF_ZeroMemory);\
  result->kind    = AST_##ikind;                                       \
  result->pos     = ipos;                                              \
  result->id      = ++pctx->unique_ids

function Ast_Atom *
ast_expr_string(Token *pos, Intern_String string){
  AST_NEW(Atom, STR, pos);
  result->intern_val = string;
  return result;
}

function Ast_Atom *
ast_expr_identifier(Token *pos, Intern_String string){
  AST_NEW(Atom, IDENT, pos);
  result->intern_val = string;
  return result;
}

function Ast_Atom *
ast_expr_integer(Token *pos, S64 integer){
  AST_NEW(Atom, INT, pos);
  result->int_val = integer;
  return result;
}

function Ast_Expr *
ast_expr_binary(Ast_Expr *left, Ast_Expr *right, Token *op){
  AST_NEW(Binary, BINARY, op);
  result->op    = op->kind;
  result->left  = left;
  result->right = right;
  return result;
}

function Ast_Compound *
ast_expr_compound(Token *pos, Ast_Typespec *typespec, Array<Ast_Compound_Item *> exprs){
  AST_NEW(Compound, COMPOUND, pos);
  result->typespec = typespec;
  result->exprs    = exprs.tight_copy(pctx->perm);
  return result;
}

function Ast_Compound_Item *
ast_expr_compound_item(Token *pos, Ast_Expr *index, Ast_Atom *name, Ast_Expr *item){
  AST_NEW(Compound_Item, COMPOUND_ITEM, pos);
  result->name  = name;
  result->index = index;
  result->item  = item;
  return result;
}

function Ast_Expr *
ast_expr_cast(Token *pos, Ast_Expr *expr, Ast_Typespec *typespec){
  AST_NEW(Cast, CAST, pos);
  result->expr     = expr;
  result->typespec = typespec;
  return result;
}

function Ast_Expr *
ast_expr_unary(Token *pos, Token_Kind op, Ast_Expr *expr){
  AST_NEW(Unary, UNARY, pos);
  result->expr = expr;
  result->op   = op;
  return result;
}

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

function Ast_Lambda *
ast_lambda(Token *pos, Array<Ast_Lambda_Arg *> params, Ast_Typespec *ret, Ast_Block *block){
  AST_NEW(Lambda, LAMBDA, pos);
  result->args  = params.tight_copy(pctx->perm);
  result->ret   = ret;
  result->block = block;
  if(!ret){
    result->ret = ast_typespec_name(0, intern_void);
  }
  return result;
}

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

function Ast_Lambda_Arg *
ast_expr_lambda_arg(Token *pos, Intern_String name, Ast_Typespec *typespec){
  AST_NEW(Lambda_Arg, LAMBDA_ARG, pos);
  result->name = name;
  result->typespec = typespec;
  return result;
}

function Ast_Block *
ast_block(Token *pos, Array<Ast *> stmts){
  AST_NEW(Block, BLOCK, pos);
  result->stmts = stmts.tight_copy(pctx->perm);
  return result;
}

function Ast_If *
ast_if(Token *pos, Array<Ast_If_Node *> ifs){
  AST_NEW(If, IF, pos);
  result->ifs = ifs.tight_copy(pctx->perm);
  return result;
}

function Ast_If_Node *
ast_if_node(Token *pos, Ast_Init *init, Ast_Expr *expr, Ast_Block *block){
  AST_NEW(If_Node, IF_NODE, pos);
  result->block    = block;
  result->expr     = expr;
  result->init = init;
  return result;
}

function Ast_Init *
ast_init(Token *pos, Token_Kind op, Ast_Atom *ident, Ast_Expr *expr){
  AST_NEW(Init, INIT, pos);
  result->op    = op;
  result->ident = ident;
  result->expr  = expr;
  return result;
}

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

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

//-----------------------------------------------------------------------------
// Declarations
//-----------------------------------------------------------------------------

function Ast_Var *
ast_var(Token *pos, Ast_Typespec *typespec, Intern_String name, Ast_Expr *expr){
  AST_NEW(Var, VAR, pos);
  result->expr     = expr;
  result->typespec = typespec;
  result->name     = name;
  return result;
}

function Ast_Const *
ast_const(Token *pos, Intern_String name, Ast_Expr *expr){
  AST_NEW(Const, CONST, pos);
  result->expr = expr;
  result->name = name;
  return result;
}

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