Rava/parser/parser.c

#define _POSIX_C_SOURCE 200809L
#include "parser.h"
#include <stdlib.h>
#include <string.h>
#include <stdio.h>

RavaASTNode_t* rava_ast_node_create(RavaASTNodeType_e type, int line, int column) {
    RavaASTNode_t *node = calloc(1, sizeof(RavaASTNode_t));
    node->type = type;
    node->line = line;
    node->column = column;
    node->parent = NULL;
    node->children = NULL;
    node->children_count = 0;
    node->children_capacity = 0;
    return node;
}

void rava_ast_node_destroy(RavaASTNode_t *node) {
    if (!node) return;

    for (size_t i = 0; i < node->children_count; i++) {
        rava_ast_node_destroy(node->children[i]);
    }
    free(node->children);

    switch (node->type) {
        case RAVA_AST_IDENTIFIER_EXPR:
            free(node->data.identifier.name);
            break;
        case RAVA_AST_CLASS_DECL:
            free(node->data.class_decl.name);
            free(node->data.class_decl.modifiers);
            rava_ast_node_destroy(node->data.class_decl.type);
            break;
        case RAVA_AST_METHOD_DECL:
            free(node->data.method_decl.name);
            free(node->data.method_decl.modifiers);
            rava_ast_node_destroy(node->data.method_decl.return_type);
            break;
        case RAVA_AST_VAR_DECL:
            free(node->data.var_decl.name);
            rava_ast_node_destroy(node->data.var_decl.type);
            rava_ast_node_destroy(node->data.var_decl.initializer);
            break;
        case RAVA_AST_TYPE:
            free(node->data.type.type_name);
            break;
        case RAVA_AST_MEMBER_ACCESS_EXPR:
            free(node->data.member_access.member);
            rava_ast_node_destroy(node->data.member_access.object);
            break;
        case RAVA_AST_CALL_EXPR:
            rava_ast_node_destroy(node->data.call.callee);
            for (size_t i = 0; i < node->data.call.arguments_count; i++) {
                rava_ast_node_destroy(node->data.call.arguments[i]);
            }
            free(node->data.call.arguments);
            break;
        case RAVA_AST_NEW_EXPR:
            rava_ast_node_destroy(node->data.new_expr.type);
            for (size_t i = 0; i < node->data.new_expr.arguments_count; i++) {
                rava_ast_node_destroy(node->data.new_expr.arguments[i]);
            }
            free(node->data.new_expr.arguments);
            break;
        case RAVA_AST_LITERAL_EXPR:
            if (node->data.literal.literal_type == RAVA_TOKEN_LITERAL_STRING) {
                free(node->data.literal.value.string_value);
            }
            break;
        default:
            break;
    }

    free(node);
}

void rava_ast_node_add_child(RavaASTNode_t *parent, RavaASTNode_t *child) {
    if (!parent || !child) return;

    if (parent->children_count >= parent->children_capacity) {
        size_t new_capacity = parent->children_capacity == 0 ? 4 : parent->children_capacity * 2;
        parent->children = realloc(parent->children, sizeof(RavaASTNode_t*) * new_capacity);
        parent->children_capacity = new_capacity;
    }

    parent->children[parent->children_count++] = child;
    child->parent = parent;
}

RavaParser_t* rava_parser_create(RavaLexer_t *lexer) {
    RavaParser_t *parser = malloc(sizeof(RavaParser_t));
    parser->lexer = lexer;
    parser->current_token = rava_lexer_next_token(lexer);
    parser->peek_token = rava_lexer_next_token(lexer);
    parser->error_message = NULL;
    parser->had_error = false;
    return parser;
}

void rava_parser_destroy(RavaParser_t *parser) {
    if (!parser) return;

    rava_token_destroy(parser->current_token);
    rava_token_destroy(parser->peek_token);
    free(parser->error_message);
    free(parser);
}

void _rava_parser_advance(RavaParser_t *parser) {
    rava_token_destroy(parser->current_token);
    parser->current_token = parser->peek_token;
    parser->peek_token = rava_lexer_next_token(parser->lexer);
}

bool _rava_parser_check(RavaParser_t *parser, RavaTokenType_e type) {
    return parser->current_token->type == type;
}

bool _rava_parser_match(RavaParser_t *parser, RavaTokenType_e type) {
    if (_rava_parser_check(parser, type)) {
        _rava_parser_advance(parser);
        return true;
    }
    return false;
}

bool _rava_parser_expect(RavaParser_t *parser, RavaTokenType_e type, const char *message) {
    if (_rava_parser_check(parser, type)) {
        _rava_parser_advance(parser);
        return true;
    }

    parser->had_error = true;
    if (parser->error_message) free(parser->error_message);
    parser->error_message = malloc(256);
    snprintf(parser->error_message, 256, "%s at line %d, column %d",
             message, parser->current_token->line, parser->current_token->column);
    return false;
}

extern RavaASTNode_t* _rava_parser_parse_type(RavaParser_t *parser);
extern RavaASTNode_t* _rava_parser_parse_expression(RavaParser_t *parser);
extern RavaASTNode_t* _rava_parser_parse_statement(RavaParser_t *parser);
extern RavaASTNode_t* _rava_parser_parse_block(RavaParser_t *parser);

static RavaModifier_e _rava_token_to_modifier(RavaTokenType_e type) {
    switch (type) {
        case RAVA_TOKEN_KEYWORD_PUBLIC: return RAVA_MODIFIER_PUBLIC;
        case RAVA_TOKEN_KEYWORD_PRIVATE: return RAVA_MODIFIER_PRIVATE;
        case RAVA_TOKEN_KEYWORD_PROTECTED: return RAVA_MODIFIER_PROTECTED;
        case RAVA_TOKEN_KEYWORD_STATIC: return RAVA_MODIFIER_STATIC;
        case RAVA_TOKEN_KEYWORD_FINAL: return RAVA_MODIFIER_FINAL;
        case RAVA_TOKEN_KEYWORD_ABSTRACT: return RAVA_MODIFIER_ABSTRACT;
        case RAVA_TOKEN_KEYWORD_NATIVE: return RAVA_MODIFIER_NATIVE;
        case RAVA_TOKEN_KEYWORD_SYNCHRONIZED: return RAVA_MODIFIER_SYNCHRONIZED;
        case RAVA_TOKEN_KEYWORD_TRANSIENT: return RAVA_MODIFIER_TRANSIENT;
        case RAVA_TOKEN_KEYWORD_VOLATILE: return RAVA_MODIFIER_VOLATILE;
        case RAVA_TOKEN_KEYWORD_STRICTFP: return RAVA_MODIFIER_STRICTFP;
        default: return RAVA_MODIFIER_PUBLIC;
    }
}

static bool _rava_is_modifier(RavaTokenType_e type) {
    return type == RAVA_TOKEN_KEYWORD_PUBLIC ||
           type == RAVA_TOKEN_KEYWORD_PRIVATE ||
           type == RAVA_TOKEN_KEYWORD_PROTECTED ||
           type == RAVA_TOKEN_KEYWORD_STATIC ||
           type == RAVA_TOKEN_KEYWORD_FINAL ||
           type == RAVA_TOKEN_KEYWORD_ABSTRACT ||
           type == RAVA_TOKEN_KEYWORD_NATIVE ||
           type == RAVA_TOKEN_KEYWORD_SYNCHRONIZED ||
           type == RAVA_TOKEN_KEYWORD_TRANSIENT ||
           type == RAVA_TOKEN_KEYWORD_VOLATILE ||
           type == RAVA_TOKEN_KEYWORD_STRICTFP;
}

void _rava_parser_parse_modifiers(RavaParser_t *parser, RavaModifier_e **modifiers, size_t *count) {
    size_t capacity = 4;
    *modifiers = malloc(sizeof(RavaModifier_e) * capacity);
    *count = 0;

    while (_rava_is_modifier(parser->current_token->type)) {
        if (*count >= capacity) {
            capacity *= 2;
            *modifiers = realloc(*modifiers, sizeof(RavaModifier_e) * capacity);
        }
        (*modifiers)[(*count)++] = _rava_token_to_modifier(parser->current_token->type);
        _rava_parser_advance(parser);
    }
}

RavaASTNode_t* _rava_parser_parse_type(RavaParser_t *parser) {
    RavaASTNode_t *type_node = rava_ast_node_create(RAVA_AST_TYPE,
                                                     parser->current_token->line,
                                                     parser->current_token->column);

    if (_rava_parser_check(parser, RAVA_TOKEN_IDENTIFIER) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_INT) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_BOOLEAN) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_CHAR) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_BYTE) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_SHORT) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_LONG) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_FLOAT) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_DOUBLE) ||
        _rava_parser_check(parser, RAVA_TOKEN_KEYWORD_VOID)) {

        type_node->data.type.type_name = strdup(parser->current_token->lexeme);
        type_node->data.type.is_array = false;
        type_node->data.type.array_dimensions = 0;
        _rava_parser_advance(parser);

        while (_rava_parser_check(parser, RAVA_TOKEN_LBRACKET)) {
            if (parser->peek_token && parser->peek_token->type == RAVA_TOKEN_RBRACKET) {
                _rava_parser_advance(parser);
                _rava_parser_advance(parser);
                type_node->data.type.is_array = true;
                type_node->data.type.array_dimensions++;
            } else {
                break;
            }
        }
    } else {
        parser->had_error = true;
        if (parser->error_message) free(parser->error_message);
        parser->error_message = strdup("Expected type name");
        rava_ast_node_destroy(type_node);
        return NULL;
    }

    return type_node;
}

RavaASTNode_t* rava_parser_parse(RavaParser_t *parser);