path: root/src/parser.c

#include "parser.h"
#include "darray.h"

Node *
alloc_node(NodeType type) {
    // TODO: Use a bump allocator?
    // TODO: Free memory!
    Node *node = malloc(sizeof(Node));
    node->type = type;
    return node;
}

Token
next_token(Parser *parser) {
    return parser->tokens[parser->current++];
}

Token
peek_token(Parser *parser) {
    return parser->tokens[parser->current];
}

bool
has_next(Parser *parser) {
    return parser->current < array_size(parser->tokens);
}

Node *
parse_number(Parser *parser) {
    Token tok = next_token(parser);

    bool negative = false;
    int base = 10;
    char c = sv_next(&tok.value);
    if (c == '-') {
        negative = true;
        c = sv_next(&tok.value);
    }
    if (c == '+') {
        c = sv_next(&tok.value);
    }
    if (c == '0') {
        c = sv_next(&tok.value);
        if (c == 'x') {
            base = 16;
            c = sv_next(&tok.value);
        } else if (c == 'b') {
            base = 2;
            c = sv_next(&tok.value);
        } else if (!(c >= '0' && c <= '9')){
            push_error(ERR_TYPE_PARSER, ERR_MALFORMED_NUMBER, tok.line, tok.col);
            return NULL;
        }
    }

    // Integral part.
    u64 integral = 0;
    while (c != '\0') {
        ssize_t current = 0;
        if (c >= 'a' && c <= 'z' && base == 16) {
            current = (c - 'a') + 10;
        } else if (c >= 'A' && c <= 'Z' && base == 16) {
            current = (c - 'A') + 10;
        } else if (c >= '0' && c <= '9') {
            current = (c - '0');
        } else if (c == '.') {
            c = sv_next(&tok.value);
            break;
        } else {
            push_error(ERR_TYPE_PARSER, ERR_MALFORMED_NUMBER, tok.line, tok.col);
            return NULL;
        }
        integral = integral * base + current;
        c = sv_next(&tok.value);
    }

    // Fractional part.
    u64 fractional = 0;
    while (c != '\0') {
        ssize_t current = 0;
        if (c >= 'a' && c <= 'z' && base == 16) {
            current = (c - 'a') + 10;
        } else if (c >= 'A' && c <= 'Z' && base == 16) {
            current = (c - 'A') + 10;
        } else if (c >= '0' && c <= '9') {
            current = (c - '0');
        } else {
            push_error(ERR_TYPE_PARSER, ERR_MALFORMED_NUMBER, tok.line, tok.col);
            return NULL;
        }
        fractional = fractional * base + current;
        c = sv_next(&tok.value);
    }

    Node * node = alloc_node(NODE_NUMBER);
    node->number.negative = negative;
    node->number.integral = integral;
    node->number.fractional = fractional;
    return node;
}

Node *
parse_string(Parser *parser) {
    Token tok = next_token(parser);
    Node *node = alloc_node(NODE_STRING);
    node->string = tok.value;
    return node;
}

Node *
parse_symbol(Parser *parser) {
    Token tok = next_token(parser);
    Node *node = alloc_node(NODE_SYMBOL);
    node->string = tok.value;
    return node;
}

Node *
parse_bool(Parser *parser) {
    Token tok = next_token(parser);
    Node *node = alloc_node(NODE_BOOL);
    node->boolean = tok.type == TOKEN_TRUE;
    return node;
}

Node *
parse_builtin(Parser *parser) {
    Token op = next_token(parser);
    Node *node = alloc_node(NODE_BUILTIN);
    node->builtin.type = op.type;
    array_init(node->builtin.args, 0);
    while (has_next(parser)) {
        Token next = peek_token(parser);
        if (next.type == TOKEN_RPAREN) {
            next_token(parser);
            return node;
        }
        Node *arg = parse_next(parser);
        if (arg == NULL) {
            break;
        }
        array_push(node->builtin.args, arg);
    }
    push_error(ERR_TYPE_PARSER, ERR_UNMATCHED_PAREN, op.line, op.col);
    return NULL;
}

Node *
parse_def(Parser *parser) {
    Token op = next_token(parser);
    Node *symbol = parse_next(parser);
    if (symbol == NULL || symbol->type != NODE_SYMBOL) {
        push_error(ERR_TYPE_PARSER, ERR_MALFORMED_EXPR, op.line, op.col);
        return NULL;
    }

    // TODO: Check if it has type annotation.

    Node *value = parse_next(parser);
    if (value == NULL) {
        push_error(ERR_TYPE_PARSER, ERR_MALFORMED_EXPR, op.line, op.col);
        return NULL;
    }

    Token end = next_token(parser);
    if (end.type != TOKEN_RPAREN) {
        push_error(ERR_TYPE_PARSER, ERR_MALFORMED_EXPR, op.line, op.col);
        return NULL;
    }

    Node *node = alloc_node(NODE_DEF);
    node->def.symbol = symbol;
    node->def.value = value;

    // TODO: Register variable in symbol table.

    return node;
}

Node *
parse_paren(Parser *parser) {
    next_token(parser); // Skip paren.
    Token tok = peek_token(parser);

    switch (tok.type) {
        // Builtin functions.
        case TOKEN_ADD:
        case TOKEN_SUB:
        case TOKEN_MUL:
        case TOKEN_DIV:
        case TOKEN_MOD:
        case TOKEN_NOT:
        case TOKEN_AND:
        case TOKEN_OR: { return parse_builtin(parser); } break;
        // Special functions.
        case TOKEN_DEF: { return parse_def(parser); } break;
        default: break;
    }

    // TODO: Lookup value on symbol table.
    push_error(ERR_TYPE_PARSER, ERR_UNIMPLEMENTED, tok.line, tok.col);
    return NULL;
}

Node *
parse_next(Parser *parser) {
    Token tok = peek_token(parser);
    switch (tok.type) {
        case TOKEN_NUMBER: { return parse_number(parser); } break;
        case TOKEN_STRING: { return parse_string(parser); } break;
        case TOKEN_SYMBOL: { return parse_symbol(parser); } break;
        case TOKEN_TRUE:
        case TOKEN_FALSE: { return parse_bool(parser); } break;
        case TOKEN_LPAREN: { return parse_paren(parser); } break;
        case TOKEN_EOF: { return NULL; } break;
        default: {
            push_error(ERR_TYPE_PARSER, ERR_UNKNOWN_TOK_TYPE, tok.line, tok.col);
            return NULL;
        } break;
    }
}

void
print_node(Node *node) {
    switch (node->type) {
        case NODE_NUMBER: {
            if (node->number.negative) {
                printf("-");
            }
            if (node->number.fractional != 0) {
                printf("%zu.%zu", node->number.integral, node->number.fractional);
            } else {
                printf("%zu", node->number.integral);
            }
        } break;
        case NODE_SYMBOL:
        case NODE_STRING: {
            sv_write(&node->string);
        } break;
        case NODE_BOOL: {
            if (node->boolean) {
                printf("true");
            } else {
                printf("false");
            }
        } break;
        case NODE_BUILTIN: {
            printf("({%s}", token_str[node->builtin.type]);
            size_t n_args = array_size(node->builtin.args);
            if (n_args != 0) {
                printf(" ");
            }
            for (size_t i = 0; i < n_args; ++i) {
                print_node(node->builtin.args[i]);
                if (i < n_args - 1) {
                    printf(" ");
                }
            }
            printf(")");
        } break;
        case NODE_DEF: {
            printf("(def ");
            print_node(node->def.symbol);
            printf(" ");
            print_node(node->def.value);
            printf(")");
        } break;
        default: { printf("{#unknown#}"); } break;
    }
}

void
parse(Token *tokens) {
    Parser parser = {
        .tokens = tokens,
        .current = 0,
    };

    // DEBUG: TOKENS
    printf("-- tokens --\n");
    for (size_t i = 0; i < array_size(tokens); i++) {
        print_token(tokens[i]);
    }
    printf("------------\n");

    while (has_next(&parser)) {
        Node *node = parse_next(&parser);
        if (node == NULL) {
            return;
        }
        print_node(node);
        printf("\n");
    }
}