595 lines
15 KiB
C
595 lines
15 KiB
C
typedef struct Parser_Error Parser_Error;
|
|
|
|
struct Parser_Error{
|
|
Parser_Error *next;
|
|
String message;
|
|
Token *token;
|
|
};
|
|
|
|
typedef struct Parser{
|
|
Token_Array tokens;
|
|
Arena *arena;
|
|
|
|
Parser_Error *first_error;
|
|
Parser_Error *last_error;
|
|
}Parser;
|
|
|
|
function void
|
|
parser_push_error(Parser *p, Token *token, char *str, ...){
|
|
String string;
|
|
{
|
|
va_list args1, args2;
|
|
va_start(args1, str);
|
|
va_copy(args2, args1);
|
|
string.len = vsnprintf(0, 0, str, args2);
|
|
va_end(args2);
|
|
|
|
string.str = arena_push_size(p->arena, string.len + 1);
|
|
vsnprintf((char*)string.str, string.len + 1, str, args1);
|
|
va_end(args1);
|
|
}
|
|
|
|
printf("Error: %s %s:%d\n", string.str, token->file.str, (S32)token->line);
|
|
Parser_Error *error = arena_push_struct(p->arena, Parser_Error);
|
|
error->message = string;
|
|
error->next = 0;
|
|
error->token = token;
|
|
SLLQueuePush(p->first_error, p->last_error, error);
|
|
|
|
__debugbreak();
|
|
}
|
|
|
|
//-----------------------------------------------------------------------------
|
|
// Parsing helpers
|
|
//-----------------------------------------------------------------------------
|
|
function Token *
|
|
token_get(Parser *p){
|
|
Token *result = token_array_iter_peek(&p->tokens, 0);
|
|
return result;
|
|
}
|
|
|
|
function Token *
|
|
token_is(Parser *p, Token_Kind kind){
|
|
Token *result = token_get(p);
|
|
if(result->kind == kind)
|
|
return result;
|
|
return 0;
|
|
}
|
|
|
|
function Token *
|
|
token_next(Parser *p){
|
|
Token *result = token_array_iter_next(&p->tokens);
|
|
return result;
|
|
}
|
|
|
|
function Token *
|
|
token_match(Parser *p, Token_Kind kind){
|
|
Token *token = token_get(p);
|
|
if(token->kind == kind){
|
|
token = token_next(p);
|
|
return token;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
function Token *
|
|
token_match_keyword(Parser *p, Intern_String string){
|
|
Token *token = token_get(p);
|
|
if(token->kind == TK_Keyword){
|
|
if(intern_compare(token->intern_val, string)){
|
|
token = token_next(p);
|
|
return token;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
function Token *
|
|
token_expect(Parser *p, Token_Kind kind){
|
|
Token *token = token_get(p);
|
|
if(token->kind == kind){
|
|
token = token_next(p);
|
|
return token;
|
|
}
|
|
parser_push_error(p, token, "Expected token of kind: %s, got instead token of kind: %s.", token_kind_string[kind], token_kind_string[token->kind]);
|
|
return 0;
|
|
}
|
|
|
|
//-----------------------------------------------------------------------------
|
|
// Expression parsing
|
|
//-----------------------------------------------------------------------------
|
|
/*
|
|
add = [+-]
|
|
mul = [/%*]
|
|
compare = == | != | >= | > | <= | <
|
|
logical = [&|^] | && | ||
|
|
unary = [&*-!~+] | ++ | --
|
|
|
|
atom_expr = Int
|
|
| Float
|
|
| String
|
|
| Identifier
|
|
| 'cast' '(' typespec ',' expr ')'
|
|
| 'size_type' '(' typespec ')'
|
|
| 'size_expr' '(' expr ')'
|
|
postfix_expr = atom_expr ('[' expr ']' | '.' Identifier | ++ | -- | '(' expr_list ')')*
|
|
unary_expr = unary ? unary_expr : atom_expr
|
|
mul_expr = atom_expr (mul atom_expr)*
|
|
add_expr = mul_expr (add mul_expr)*
|
|
logical_expr = add_expr (logical add_expr)*
|
|
compare_expr = logical_expr (compare logical_expr)*
|
|
ternary_expr = compare_expr ('?' ternary_expr ':' ternary_expr)?
|
|
expr = logical_expr
|
|
*/
|
|
function Expr *parse_expr(Parser *p);
|
|
|
|
function Expr *
|
|
parse_expr_atom(Parser *p){
|
|
Token *token = 0;
|
|
if((token = token_match(p, TK_StringLit))){
|
|
Expr *result = expr_str(p->arena, token);
|
|
return result;
|
|
}
|
|
else if((token = token_match(p, TK_Identifier))){
|
|
Expr *result = expr_identifier(p->arena, token);
|
|
return result;
|
|
}
|
|
else if((token = token_match(p, TK_Int))){
|
|
Expr *result = expr_int(p->arena, token);
|
|
return result;
|
|
}
|
|
else if((token = token_match(p, TK_OpenParen))){
|
|
Expr *expr = parse_expr(p);
|
|
token_expect(p, TK_CloseParen);
|
|
Expr *result = expr_paren(p->arena, token, expr);
|
|
return result;
|
|
}
|
|
else if((token = token_match_keyword(p, keyword_cast))){
|
|
token_expect(p, TK_OpenParen);
|
|
token_expect(p, TK_Identifier);
|
|
token_expect(p, TK_Comma);
|
|
token_expect(p, TK_Identifier);
|
|
token_expect(p, TK_CloseParen);
|
|
return 0;
|
|
}
|
|
else{
|
|
parser_push_error(p, token_get(p), "Failed to parse expression");
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
function B32
|
|
token_is_postfix(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind == TK_OpenBracket
|
|
|| token->kind == TK_OpenParen
|
|
|| token->kind == TK_Dot
|
|
|| token->kind == TK_Increment
|
|
|| token->kind == TK_Decrement;
|
|
return result;
|
|
}
|
|
|
|
function Expr_Compound_Field *
|
|
parse_expr_compound(Parser *p){
|
|
Token *token = token_get(p);
|
|
Expr_Compound_Field *result = 0;
|
|
if(token_match(p, TK_OpenBracket)){
|
|
Expr *index = parse_expr(p);
|
|
token_expect(p, TK_CloseBracket);
|
|
token_expect(p, TK_Assign);
|
|
Expr *expr = parse_expr(p);
|
|
result = expr_compound_index(p->arena, token, index, expr);
|
|
}
|
|
else{
|
|
Expr *expr = parse_expr(p);
|
|
if((token = token_match(p, TK_Assign))){
|
|
if(expr->kind != EK_Identifier){
|
|
parser_push_error(p, token, "Failed to parse compound literal, required identifier as left value");
|
|
}
|
|
result = expr_compound_named(p->arena, token, expr->intern_val, parse_expr(p));
|
|
|
|
}
|
|
else{
|
|
result = expr_compound_default(p->arena, token, expr);
|
|
}
|
|
}
|
|
return result;
|
|
}
|
|
|
|
function Expr_Compound_Field *
|
|
parse_expr_function_argument(Parser *p){
|
|
Token *token = token_get(p);
|
|
Expr_Compound_Field *result = 0;
|
|
Expr *expr1 = parse_expr(p);
|
|
if(token_match(p, TK_Assign)){
|
|
if(expr1->kind != EK_Identifier){
|
|
parser_push_error(p, token, "Failed to parse named function argument, required identifier as left value");
|
|
}
|
|
Expr *expr2 = parse_expr(p);
|
|
result = expr_compound_named(p->arena, token, expr1->intern_val, expr2);
|
|
}
|
|
else{
|
|
result = expr_compound_default(p->arena, token, expr1);
|
|
}
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_postfix(Parser *p){
|
|
Expr *left = parse_expr_atom(p);
|
|
while(token_is_postfix(p)){
|
|
Token *token = 0;
|
|
if((token = token_match(p, TK_OpenBracket))){
|
|
Expr *size = parse_expr(p);
|
|
token_expect(p, TK_CloseBracket);
|
|
left = expr_index(p->arena, token, left, size);
|
|
}
|
|
else if((token = token_match(p, TK_OpenParen))){
|
|
left = expr_call(p->arena, token, left);
|
|
if(!token_is(p, TK_CloseParen)){
|
|
do {
|
|
Expr_Compound_Field *field = parse_expr_function_argument(p);
|
|
expr_call_push(left, field);
|
|
} while(token_match(p, TK_Comma));
|
|
}
|
|
token_expect(p, TK_CloseParen);
|
|
}
|
|
else if(token_match(p, TK_Dot)){
|
|
token = token_expect(p, TK_Identifier);
|
|
left = expr_field(p->arena, token, left);
|
|
}
|
|
else{
|
|
token = token_next(p);
|
|
assert(token->kind == TK_Increment || token->kind == TK_Decrement);
|
|
left = expr_postfix_unary(p->arena, token, left);
|
|
}
|
|
}
|
|
return left;
|
|
}
|
|
|
|
function B32
|
|
token_is_unary(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind == TK_Add
|
|
|| token->kind == TK_Increment
|
|
|| token->kind == TK_Decrement
|
|
|| token->kind == TK_Sub
|
|
|| token->kind == TK_Mul
|
|
|| token->kind == TK_BitAnd
|
|
|| token->kind == TK_Neg
|
|
|| token->kind == TK_Not;
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_unary(Parser *p){
|
|
if(token_is_unary(p)){
|
|
Token *op = token_next(p);
|
|
Expr *right = parse_expr_unary(p);
|
|
Expr *result = expr_unary(p->arena, op, right);
|
|
return result;
|
|
}
|
|
else{
|
|
return parse_expr_postfix(p);
|
|
}
|
|
}
|
|
|
|
function B32
|
|
token_is_mul(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind >= TK_FirstMul && token->kind <= TK_LastMul;
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_mul(Parser *p){
|
|
Expr *left = parse_expr_unary(p);
|
|
while(token_is_mul(p)){
|
|
Token *op = token_next(p);
|
|
Expr *right = parse_expr_unary(p);
|
|
left = expr_binary(p->arena, op, left, right);
|
|
}
|
|
return left;
|
|
}
|
|
|
|
function B32
|
|
token_is_add(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind >= TK_FirstAdd && token->kind <= TK_LastAdd;
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_add(Parser *p){
|
|
Expr *left = parse_expr_mul(p);
|
|
while(token_is_add(p)){
|
|
Token *op = token_next(p);
|
|
Expr *right = parse_expr_mul(p);
|
|
left = expr_binary(p->arena, op, left, right);
|
|
}
|
|
return left;
|
|
}
|
|
|
|
function B32
|
|
token_is_logical(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind >= TK_FirstLogical && token->kind <= TK_LastLogical;
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_logical(Parser *p){
|
|
Expr *left = parse_expr_add(p);
|
|
while(token_is_logical(p)){
|
|
Token *op = token_next(p);
|
|
Expr *right = parse_expr_add(p);
|
|
left = expr_binary(p->arena, op, left, right);
|
|
}
|
|
return left;
|
|
}
|
|
|
|
function B32
|
|
token_is_compare(Parser *p){
|
|
Token *token = token_get(p);
|
|
B32 result = token->kind >= TK_FirstCompare && token->kind <= TK_LastCompare;
|
|
return result;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_compare(Parser *p){
|
|
Expr *left = parse_expr_logical(p);
|
|
while(token_is_compare(p)){
|
|
Token *op = token_next(p);
|
|
Expr *right = parse_expr_logical(p);
|
|
left = expr_binary(p->arena, op, left, right);
|
|
}
|
|
return left;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr_ternary(Parser *p){
|
|
Expr *cond = parse_expr_compare(p);
|
|
Token *token = 0;
|
|
if((token = token_match(p, TK_Question))){
|
|
Expr *on_true = parse_expr_ternary(p);
|
|
token_expect(p, TK_Colon);
|
|
Expr *on_false = parse_expr_ternary(p);
|
|
Expr *result = expr_ternary(p->arena, token, cond, on_true, on_false);
|
|
return result;
|
|
}
|
|
return cond;
|
|
}
|
|
|
|
function Expr *
|
|
parse_expr(Parser *p){
|
|
return parse_expr_ternary(p);
|
|
}
|
|
|
|
//-----------------------------------------------------------------------------
|
|
// Test code
|
|
//-----------------------------------------------------------------------------
|
|
function void expr_print(Expr *expr);
|
|
function void
|
|
token_print(Token *token){
|
|
printf("%.*s", (S32)token->len, token->str);
|
|
}
|
|
|
|
function void
|
|
expr_compound_print(Expr_Compound_Field *field){
|
|
switch(field->kind){
|
|
case COMPOUND_Default: {
|
|
expr_print(field->init);
|
|
}break;
|
|
case COMPOUND_Named: {
|
|
printf("[%s] = ", field->name.s.str);
|
|
expr_print(field->init);
|
|
}break;
|
|
case COMPOUND_Index: {
|
|
printf("[");
|
|
expr_print(field->index);
|
|
printf("] = ");
|
|
expr_print(field->init);
|
|
}break;
|
|
default: invalid_codepath;
|
|
}
|
|
}
|
|
|
|
function void
|
|
expr_print(Expr *expr){
|
|
switch(expr->kind) {
|
|
case EK_Int:case EK_String:case EK_Identifier: {
|
|
token_print(expr->token);
|
|
} break;
|
|
|
|
case EK_SizeExpr:{
|
|
printf("size_expr(");
|
|
expr_print(expr->size_expr.expr);
|
|
printf(")");
|
|
}break;
|
|
|
|
case EK_Compound:{
|
|
// typespec_print(p);
|
|
printf("{");
|
|
for(Expr_Compound_Field *n = expr->compound.first; n; n=n->next){
|
|
expr_compound_print(n);
|
|
if(n!=expr->compound.last) printf(",");
|
|
}
|
|
printf("}");
|
|
} break;
|
|
|
|
case EK_SizeType:{
|
|
printf("size_type(");
|
|
printf(")");
|
|
}break;
|
|
|
|
case EK_Paren:{
|
|
printf("(");
|
|
expr_print(expr->paren.expr);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_Field:{
|
|
expr_print(expr->field.expr);
|
|
printf(".%s", expr->field.name.s.str);
|
|
} break;
|
|
|
|
case EK_Binary:{
|
|
printf("(");
|
|
expr_print(expr->binary.left);
|
|
token_print(expr->token);
|
|
expr_print(expr->binary.right);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_PostfixUnary:{
|
|
printf("(");
|
|
expr_print(expr->unary.expr);
|
|
token_print(expr->token);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_Unary:{
|
|
printf("(");
|
|
token_print(expr->token);
|
|
expr_print(expr->unary.expr);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_Ternary:{
|
|
printf("(");
|
|
expr_print(expr->ternary.cond);
|
|
printf("?");
|
|
expr_print(expr->ternary.on_true);
|
|
printf(":");
|
|
expr_print(expr->ternary.on_false);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_Cast:{
|
|
printf("(");
|
|
printf("(");
|
|
//print_typespec(expr->cast.type);
|
|
printf(")");
|
|
expr_print(expr->cast.expr);
|
|
printf(")");
|
|
} break;
|
|
|
|
case EK_Index:{
|
|
expr_print(expr->index.atom);
|
|
printf("[");
|
|
expr_print(expr->index.index);
|
|
printf("]");
|
|
}break;
|
|
|
|
case EK_Call:{
|
|
expr_print(expr->call.atom);
|
|
printf("(");
|
|
for(Expr_Compound_Field *n = expr->call.first; n; n=n->next){
|
|
expr_compound_print(n);
|
|
if(n!=expr->call.last) printf(",");
|
|
}
|
|
printf(")");
|
|
}break;
|
|
default: {invalid_codepath;} break;
|
|
}
|
|
}
|
|
|
|
|
|
function S64
|
|
eval_expr(Expr *expr){
|
|
switch(expr->kind){
|
|
case EK_Int: return expr->int_val; break;
|
|
case EK_Unary:{
|
|
S64 left = eval_expr(expr->unary.expr);
|
|
switch(expr->unary.op){
|
|
case TK_Not: return !left; break;
|
|
case TK_Neg: return ~left; break;
|
|
case TK_Sub: return -left; break;
|
|
case TK_Add: return +left; break;
|
|
default: invalid_codepath;
|
|
}
|
|
} break;
|
|
case EK_Ternary:{
|
|
S64 cond = eval_expr(expr->ternary.cond);
|
|
if(cond) return eval_expr(expr->ternary.on_true);
|
|
else return eval_expr(expr->ternary.on_false);
|
|
} break;
|
|
case EK_Paren: return eval_expr(expr->paren.expr); break;
|
|
case EK_Binary: {
|
|
S64 left = eval_expr(expr->binary.left);
|
|
S64 right = eval_expr(expr->binary.right);
|
|
switch(expr->binary.op){
|
|
case TK_Add: return left + right; break;
|
|
case TK_Sub: return left - right; break;
|
|
case TK_Mul: return left * right; break;
|
|
case TK_Div: return left / right; break;
|
|
case TK_Mod: return left % right; break;
|
|
case TK_Equals: return left == right; break;
|
|
case TK_NotEquals: return left != right; break;
|
|
case TK_GreaterThenOrEqual: return left >= right; break;
|
|
case TK_LesserThenOrEqual: return left <= right; break;
|
|
case TK_GreaterThen: return left > right; break;
|
|
case TK_LesserThen: return left < right; break;
|
|
case TK_BitAnd: return left & right; break;
|
|
case TK_BitOr: return left | right; break;
|
|
case TK_BitXor: return left ^ right; break;
|
|
case TK_And: return left && right; break;
|
|
case TK_Or: return left || right; break;
|
|
case TK_LeftShift: return left << right; break;
|
|
case TK_RightShift: return left >> right; break;
|
|
default: invalid_codepath;
|
|
}
|
|
} break;
|
|
default: invalid_codepath;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
function void
|
|
parse_test(){
|
|
Arena *scratch = arena_begin_scratch();
|
|
String test_case = lit("32+52-242*2/424%5-23"
|
|
" 1<<5>>6<<2 "
|
|
" 5*(4/3)*(2+5) "
|
|
" 0&1 == 1&0 "
|
|
" 1&&5*3 "
|
|
" 1&&5||0 "
|
|
" 1>5>=5==0 "
|
|
" 1>5 ? 1 : 2 "
|
|
" !!!!!1 "
|
|
" ~~1 + -!2 "
|
|
" 1 + ++Thing[12]++ + ++Thing[12].expr + --Not_Thing[156](Thing) + test_func(asd=func1, af=func2, gg=func3)"
|
|
);
|
|
Parser parser = {
|
|
.tokens = lex_stream(scratch, test_case, lit("expr_test")),
|
|
.arena = scratch,
|
|
};
|
|
Parser *p = &parser;
|
|
|
|
S64 t = 5;
|
|
S64 test_val[] = {
|
|
(32+52-242*2/424%5-23),
|
|
(((1<<5)>>6)<<2),
|
|
5*(4/3)*(2+5),
|
|
(0&1) == (1&0),
|
|
1&&(t*3),
|
|
(1&&t)||0,
|
|
1>t>=t==0,
|
|
1>t ? 1 : 2,
|
|
!!!!!1,
|
|
~~1 + -!2,
|
|
};
|
|
for(int i = 0; i < buff_cap(test_val); i++){
|
|
Expr *expr = parse_expr(p);
|
|
S64 val = eval_expr(expr);
|
|
assert(val == test_val[i]);
|
|
}
|
|
|
|
Expr *expr = parse_expr(p);
|
|
expr_print(expr);
|
|
|
|
arena_end_scratch();
|
|
}
|
|
|
|
|