Files
fz-lang/src/compiler/fzl_parser.c
T
2026-10-14 19:56:27 +03:00

773 lines
18 KiB
C

#include "compiler/fzl_token.h"
#include <compiler/fzl_parser.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
char plast_err[256];
fzl_parser_t *fzl_mkparser(fzl_reader_ll_t *first)
{
if (!first)
{
snprintf(plast_err, sizeof(plast_err), "token is null");
return NULL;
}
fzl_parser_t *parser = malloc(sizeof(fzl_parser_t));
if (!parser)
return NULL;
parser->root = NULL;
parser->cursor = first->next;
if (!parser->cursor)
return NULL;
return parser;
}
// --- Parse zone ---
// Make a new node
fzl_node_t *fzl_mknode(fzl_nodetype_t type, fzl_token_t *token)
{
fzl_node_t *node = malloc(sizeof(fzl_node_t));
if (!node)
return NULL;
node->type = type;
node->token = token;
node->parent = NULL;
node->next = NULL;
return node;
}
// Get token at cursor
static fzl_token_t *peek(fzl_parser_t *parser)
{
if (!parser || !parser->cursor)
return NULL;
return parser->cursor->token;
}
// Get token at (cursor + 1)
static fzl_token_t *next(fzl_parser_t *parser)
{
if (!parser || !parser->cursor)
return NULL;
fzl_reader_ll_t *next = parser->cursor->next;
if (!next)
return NULL;
return next->token;
}
// Consume token at cursor and move to the next
static fzl_token_t *consume(fzl_parser_t *parser)
{
if (!parser || !parser->cursor)
return NULL;
fzl_token_t *tok = parser->cursor->token;
parser->cursor = parser->cursor->next;
return tok;
}
// Check if current token have the same type as specified
static bool is(fzl_parser_t *parser, fzl_toktype_t type)
{
if (!parser)
return false;
fzl_token_t *tok = peek(parser);
if (!tok)
return false;
return tok->type == type;
}
// Convert arithmetic operator token type to node type
static fzl_nodetype_t math_op_to_type(fzl_toktype_t type)
{
switch (type)
{
case FZL_TOKTYPE_PLUS:
return FZL_NODETYPE_ADD;
case FZL_TOKTYPE_MINUS:
return FZL_NODETYPE_SUB;
case FZL_TOKTYPE_STAR:
return FZL_NODETYPE_MUL;
case FZL_TOKTYPE_SLASH:
return FZL_NODETYPE_DIV;
case FZL_TOKTYPE_CARET:
return FZL_NODETYPE_POW;
case FZL_TOKTYPE_PERCENT:
return FZL_NODETYPE_MOD;
default:
return 0;
}
}
static fzl_nodetype_t unary_op_to_type(fzl_toktype_t type)
{
switch (type)
{
case FZL_TOKTYPE_PLUS:
return FZL_NODETYPE_UNARY_PLUS;
case FZL_TOKTYPE_MINUS:
return FZL_NODETYPE_UNARY_MINUS;
case FZL_TOKTYPE_NOT:
return FZL_NODETYPE_NOT;
default:
return 0;
}
}
fzl_node_t *parse_expression(fzl_parser_t *parser);
// Parse primary
fzl_node_t *parse_primary(fzl_parser_t *parser)
{
if (!parser || !peek(parser))
return NULL;
if (is(parser, FZL_TOKTYPE_STR_LITERAL))
{
fzl_token_t *tok = consume(parser);
return fzl_mknode(FZL_NODETYPE_LITERAL, tok);
}
else if (is(parser, FZL_TOKTYPE_NUM_LITERAL))
{
fzl_token_t *tok = consume(parser);
return fzl_mknode(FZL_NODETYPE_LITERAL, tok);
}
else if (is(parser, FZL_TOKTYPE_BOOL_LITERAL))
{
fzl_token_t *tok = consume(parser);
return fzl_mknode(FZL_NODETYPE_LITERAL, tok);
}
else if (is(parser, FZL_TOKTYPE_LPAREN))
{
consume(parser);
fzl_node_t *node = parse_expression(parser);
if (!is(parser, FZL_TOKTYPE_RPAREN))
{
snprintf(plast_err, sizeof(plast_err), "expected ')' at %ld:%ld",
parser->cursor->token->ln, parser->cursor->token->cn);
return NULL;
}
consume(parser);
return node;
}
else if (is(parser, FZL_TOKTYPE_NAME))
{
fzl_token_t *name = consume(parser);
return fzl_mknode(FZL_NODETYPE_IDENTIFIER, name);
}
snprintf(plast_err, sizeof(plast_err), "failed to parse primary at %ld:%ld",
peek(parser)->ln, peek(parser)->cn);
return NULL;
}
// Parse postfix expressions (parens, index access, dots)
fzl_node_t *parse_postfix(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_primary(parser);
fzl_node_t *right = NULL;
if (!left)
{
snprintf(plast_err, sizeof(plast_err), "failed to parse postfix at %ld:%ld",
peek(parser)->ln, peek(parser)->cn);
return NULL;
}
while (is(parser, FZL_TOKTYPE_LPAREN))
{
consume(parser);
size_t args_capacity = 2;
size_t args_count = 0;
fzl_node_t **args = malloc(args_capacity * sizeof(fzl_node_t *));
while (!is(parser, FZL_TOKTYPE_RPAREN))
{
fzl_node_t *arg = parse_expression(parser);
if (!arg)
{
free(args);
return NULL;
}
if (args_capacity <= args_count)
{
args_capacity *= 2;
args = realloc(args, args_capacity * sizeof(fzl_node_t));
}
args[args_count] = arg;
args_count++;
if (is(parser, FZL_TOKTYPE_COMMA))
consume(parser);
}
fzl_token_t *rparen = consume(parser);
if (!rparen || rparen->type != FZL_TOKTYPE_RPAREN)
{
free(args);
snprintf(plast_err, sizeof(plast_err),
"expected ')' after argument list at %ld:%ld", peek(parser)->ln,
peek(parser)->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_CALL, left->token);
node->call.callee = left;
node->call.args = args;
node->call.args_count = args_count;
left = node;
}
return left;
}
// Parse unary operations
fzl_node_t *parse_unary(fzl_parser_t *parser)
{
if (!parser)
return NULL;
if (!is(parser, FZL_TOKTYPE_MINUS) && !is(parser, FZL_TOKTYPE_PLUS) &&
!is(parser, FZL_TOKTYPE_NOT))
return parse_postfix(parser);
fzl_token_t *op = consume(parser);
fzl_node_t *operand = parse_unary(parser);
if (!operand)
return NULL;
fzl_node_t *result = fzl_mknode(unary_op_to_type(op->type), op);
result->unary_op.operand = operand;
return result;
}
fzl_node_t *parse_power(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_unary(parser);
fzl_node_t *right = NULL;
if (!left)
{
snprintf(plast_err, sizeof(plast_err), "failed to parse power at %ld:%ld",
peek(parser)->ln, peek(parser)->cn);
return NULL;
}
if (is(parser, FZL_TOKTYPE_CARET))
{
fzl_token_t *caret = consume(parser);
right = parse_unary(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err),
"failed to parse right side of power at %ld:%ld",
peek(parser)->ln, peek(parser)->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_POW, caret);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
// Parse FACTOR: multiplication and division
fzl_node_t *parse_factor(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_power(parser);
fzl_node_t *right = NULL;
if (!left)
{
if (!plast_err[0])
snprintf(plast_err, sizeof(plast_err), "invalid expression at %ld:%ld",
parser->cursor->token->ln, parser->cursor->token->cn);
return NULL;
}
while (is(parser, FZL_TOKTYPE_STAR) || is(parser, FZL_TOKTYPE_SLASH) ||
is(parser, FZL_TOKTYPE_PERCENT))
{
fzl_token_t *op = consume(parser);
right = parse_factor(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err),
"unfinished factor (mul/div) expression at %ld:%ld",
left->token->ln, left->token->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(math_op_to_type(op->type), op);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
// Parse TERM: addition and subtraction
fzl_node_t *parse_term(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_factor(parser);
fzl_node_t *right = NULL;
if (!left)
{
snprintf(plast_err, sizeof(plast_err), "invalid expression at %ld:%ld",
parser->cursor->token->ln, parser->cursor->token->cn);
return NULL;
}
while (is(parser, FZL_TOKTYPE_PLUS) || is(parser, FZL_TOKTYPE_MINUS))
{
fzl_token_t *op = consume(parser);
right = parse_factor(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err),
"unfinished term (add/sub) expression at %ld:%ld",
left->token->ln, left->token->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(math_op_to_type(op->type), op);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
// Parse conditional operations
fzl_node_t *parse_cond(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_term(parser);
fzl_node_t *right = NULL;
if (!left)
return NULL;
while (is(parser, FZL_TOKTYPE_EQEQ) || is(parser, FZL_TOKTYPE_NEQ) ||
is(parser, FZL_TOKTYPE_LANGLE) || is(parser, FZL_TOKTYPE_LEQ) ||
is(parser, FZL_TOKTYPE_RANGLE) || is(parser, FZL_TOKTYPE_REQ))
{
fzl_token_t *op = consume(parser);
right = parse_term(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err), "unfinished expression at %ld:%ld",
op->ln, op->cn);
return NULL;
}
fzl_node_t *result;
switch (op->type)
{
case FZL_TOKTYPE_EQEQ:
result = fzl_mknode(FZL_NODETYPE_EQUAL, op);
break;
case FZL_TOKTYPE_NEQ:
result = fzl_mknode(FZL_NODETYPE_NEQUAL, op);
break;
case FZL_TOKTYPE_LANGLE:
result = fzl_mknode(FZL_NODETYPE_LESS, op);
break;
case FZL_TOKTYPE_LEQ:
result = fzl_mknode(FZL_NODETYPE_LESS_EQUAL, op);
break;
case FZL_TOKTYPE_RANGLE:
result = fzl_mknode(FZL_NODETYPE_GREATER, op);
break;
case FZL_TOKTYPE_REQ:
result = fzl_mknode(FZL_NODETYPE_GREATER_EQUAL, op);
break;
default:
return NULL;
}
result->binary_op.left = left;
result->binary_op.right = right;
left = result;
}
return left;
}
// Parse conditional &&
fzl_node_t *parse_and(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_cond(parser);
fzl_node_t *right = NULL;
if (!left)
return NULL;
while (is(parser, FZL_TOKTYPE_DAMPERSAND))
{
fzl_token_t * or = consume(parser);
right = parse_cond(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err), "unfinished expression at %ld:%ld",
or->ln, or->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_AND, or);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
// Parse conditional ||
fzl_node_t *parse_or(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_node_t *left = parse_and(parser);
fzl_node_t *right = NULL;
if (!left)
return NULL;
while (is(parser, FZL_TOKTYPE_DVBAR))
{
fzl_token_t * or = consume(parser);
right = parse_and(parser);
if (!right)
{
snprintf(plast_err, sizeof(plast_err), "unfinished expression at %ld:%ld",
or->ln, or->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_OR, or);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
// Expression parsing
fzl_node_t *parse_expression(fzl_parser_t *parser) { return parse_and(parser); }
// Statement parsing
fzl_node_t *parse_statement(fzl_parser_t *parser)
{
if (is(parser, FZL_TOKTYPE_TYPE))
{
fzl_token_t *type = peek(parser);
if (type->type == FZL_TOKTYPE_TYPE)
consume(parser);
if (type->type == FZL_TOKTYPE_TYPE || next(parser)->type == FZL_TOKTYPE_EQ)
{
fzl_node_t *identifier = parse_primary(parser);
if (!identifier || identifier->type != FZL_NODETYPE_IDENTIFIER)
{
snprintf(plast_err, sizeof(plast_err), "expected identifier at %ld:%ld",
type->ln, type->cn);
return NULL;
}
fzl_node_t *left = fzl_mknode(FZL_NODETYPE_DECLARATION, type);
left->var.type = type;
left->var.identifier = identifier;
if (!is(parser, FZL_TOKTYPE_EQ))
return left;
fzl_node_t *right;
while (is(parser, FZL_TOKTYPE_EQ))
{
fzl_token_t *asn = consume(parser);
right = parse_expression(parser);
if (!right)
return NULL;
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_ASSIGN, asn);
node->binary_op.left = left;
node->binary_op.right = right;
left = node;
}
return left;
}
}
if (is(parser, FZL_TOKTYPE_NAME) && next(parser)->type == FZL_TOKTYPE_EQ)
{
fzl_token_t *name = consume(parser);
fzl_node_t *left = fzl_mknode(FZL_NODETYPE_IDENTIFIER, name);
while (is(parser, FZL_TOKTYPE_EQ))
{
fzl_token_t *eq = consume(parser);
fzl_node_t *right = parse_expression(parser);
if (!right)
return NULL;
fzl_node_t *assn = fzl_mknode(FZL_NODETYPE_ASSIGN, eq);
assn->binary_op.left = left;
assn->binary_op.right = right;
left = assn;
}
return left;
}
if (is(parser, FZL_TOKTYPE_IF))
{
fzl_token_t *if_tok = consume(parser);
fzl_token_t *lparen = consume(parser);
if (lparen->type != FZL_TOKTYPE_LPAREN)
{
snprintf(plast_err, sizeof(plast_err),
"expected '(' before condition at %ld:%ld", lparen->ln,
lparen->cn);
return NULL;
}
fzl_node_t *condition = parse_expression(parser);
if (!condition)
return NULL;
fzl_token_t *rparen = consume(parser);
if (rparen->type != FZL_TOKTYPE_RPAREN)
{
snprintf(plast_err, sizeof(plast_err),
"expected ')' after condition at %ld:%ld", rparen->ln,
rparen->cn);
return NULL;
}
fzl_node_t *else_block = NULL;
fzl_node_t *true_block = parse_statement(parser);
if (!true_block)
return NULL;
if (is(parser, FZL_TOKTYPE_ELSE))
{
consume(parser); // else
else_block = parse_statement(parser);
if (!else_block)
return NULL;
}
fzl_node_t *stmt_node = fzl_mknode(FZL_NODETYPE_CONDITIONAL, if_tok);
stmt_node->cond.condition = condition;
stmt_node->cond.true_block = true_block;
stmt_node->cond.else_block = else_block;
return stmt_node;
}
if (is(parser, FZL_TOKTYPE_LBRACE))
{
fzl_token_t *block_tok = consume(parser);
size_t remainder = 2;
size_t count = 0;
fzl_node_t **stmts = malloc(remainder * sizeof(fzl_node_t *));
while (peek(parser) && !is(parser, FZL_TOKTYPE_RBRACE))
{
fzl_node_t *node = parse_statement(parser);
if (!node)
return NULL;
if (remainder <= count)
{
remainder *= 2;
stmts = realloc(stmts, remainder);
}
stmts[count++] = node;
}
fzl_token_t *rbrace = consume(parser);
if (!rbrace || rbrace->type != FZL_TOKTYPE_RBRACE)
{
snprintf(plast_err, sizeof(plast_err),
"missing '}' after block at %ld:%ld", rbrace->ln, rbrace->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_BLOCK, block_tok);
node->list.entries = stmts;
node->list.entry_count = count;
return node;
}
return parse_expression(parser);
}
// Internal parsing entry point
fzl_node_t *parse_program(fzl_parser_t *parser)
{
fzl_node_t *first = NULL;
fzl_node_t *last = NULL;
fzl_node_t *node = NULL;
while (peek(parser) != NULL)
{
node = parse_statement(parser);
if (!node)
return NULL;
if (!last)
{
first = node;
last = node;
}
else
{
last->next = node;
last = node;
}
}
if (!first)
{
snprintf(plast_err, sizeof(plast_err), "nothing to parse >:(");
return NULL;
}
return first;
}
// Parse tokens and build AST
int fzl_parse(fzl_parser_t *parser)
{
fzl_node_t *ast = parse_program(parser);
if (!ast)
return -1;
parser->root = ast;
return 0;
}
// --- Parse zone end ---
char *fzl_parser_geterror() { return plast_err; }
void fzl_freenode(fzl_node_t *node)
{
if (!node)
return;
free(node);
}
void fzl_freeast(fzl_node_t *root)
{
if (!root)
return;
while (root)
{
if (root->type == FZL_NODETYPE_BLOCK)
{
if (root->list.entries)
{
for (size_t i = 0; i < root->list.entry_count; i++)
if (root->list.entries[i])
fzl_freeast(root->list.entries[i]);
free(root->list.entries);
}
}
else if (root->type == FZL_NODETYPE_CALL)
{
if (root->call.args_count)
{
for (size_t i = 0; i < root->call.args_count; i++)
if (root->call.args[i])
fzl_freeast(root->call.args[i]);
free(root->call.args);
}
}
else if (root->type == FZL_NODETYPE_ADD || root->type == FZL_NODETYPE_SUB ||
root->type == FZL_NODETYPE_MUL || root->type == FZL_NODETYPE_DIV ||
root->type == FZL_NODETYPE_POW || root->type == FZL_NODETYPE_MOD)
{
if (root->binary_op.left)
fzl_freeast(root->binary_op.left);
if (root->binary_op.right)
fzl_freeast(root->binary_op.right);
}
else if (root->type == FZL_NODETYPE_UNARY_PLUS ||
root->type == FZL_NODETYPE_UNARY_MINUS)
{
if (root->unary_op.operand)
fzl_freeast(root->unary_op.operand);
}
fzl_freenode(root);
root = root->next;
}
}
void fzl_freeparser(fzl_parser_t *parser)
{
if (!parser)
return;
free(parser);
}