Files
fz-lang/src/compiler/fzl_parser.c
T

1518 lines
37 KiB
C

#include <compiler/fzl_parser.h>
#include <compiler/fzl_token.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
char plast_err[256];
fzl_token_t *last_known_tok = NULL;
fzl_token_t null_tok = {
.type = FZL_TOKTYPE_TYPE,
.filename = "",
.ln = 0,
.cn = 0,
.store_value = false,
.value = "null",
.value_sz = 4,
};
fzl_token_t this_tok = {.type = FZL_TOKTYPE_NAME,
.filename = "",
.ln = 0,
.cn = 0,
.store_value = false,
.value = "this",
.value_sz = 4};
fzl_parser_t *fzl_mkparser(fzl_reader_ll_t *first)
{
if (!first)
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));
memset(node, 0, 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;
last_known_tok = parser->cursor->token;
return parser->cursor->token;
}
// Consume token at cursor and move to the next
static fzl_token_t *consume(fzl_parser_t *parser)
{
if (!parser || !parser->cursor || !parser->cursor->token)
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) ||
is(parser, FZL_TOKTYPE_NUM_LITERAL) ||
is(parser, FZL_TOKTYPE_BOOL_LITERAL) ||
is(parser, FZL_TOKTYPE_NULL_LITERAL))
{
fzl_token_t *tok = consume(parser);
return fzl_mknode(FZL_NODETYPE_LITERAL, tok);
}
else if (is(parser, FZL_TOKTYPE_STR_FORMAT_START))
{
consume(parser);
size_t remainder = 2;
size_t count = 0;
fzl_node_t **entries = malloc(remainder * sizeof(fzl_node_t *));
if (!entries)
{
snprintf(plast_err, sizeof(plast_err), "out of memory");
return NULL;
}
while (peek(parser) && peek(parser)->type != FZL_TOKTYPE_STR_FORMAT_END)
{
if (is(parser, FZL_TOKTYPE_STR_FORMAT_END))
break;
if (remainder <= count)
{
remainder *= 2;
fzl_node_t **newentries =
realloc(entries, remainder * sizeof(fzl_node_t *));
if (!newentries)
{
snprintf(plast_err, sizeof(plast_err), "out of memory");
for (size_t i = 0; i < count; i++)
fzl_freeast(entries[i]);
free(entries);
return NULL;
}
entries = newentries;
}
fzl_node_t *expr = parse_expression(parser);
if (expr == NULL)
{
for (size_t i = 0; i < count; i++)
fzl_freeast(entries[i]);
free(entries);
return NULL;
}
entries[count++] = expr;
}
fzl_token_t *end = consume(parser);
if (end->type != FZL_TOKTYPE_STR_FORMAT_END)
{
snprintf(plast_err, sizeof(plast_err),
"'%s':%zu:%zu => unterminated format string",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
for (size_t i = 0; i < count; i++)
fzl_freeast(entries[i]);
free(entries);
}
if (count < 1)
{
snprintf(plast_err, sizeof(plast_err),
"'%s':%zu:%zu => do you really need empty format string?",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
for (size_t i = 0; i < count; i++)
fzl_freeast(entries[i]);
free(entries);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_FMT_STRING, entries[0]->token);
if (!node)
{
snprintf(plast_err, sizeof(plast_err), "out of memory");
for (size_t i = 0; i < count; i++)
fzl_freeast(entries[i]);
free(entries);
return NULL;
}
node->list.entries = entries;
node->list.entry_count = count;
return node;
}
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), "'%s': expected ')' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
fzl_freeast(node);
return NULL;
}
consume(parser);
return node;
}
else if (is(parser, FZL_TOKTYPE_NAME))
{
fzl_token_t *name = consume(parser);
if (!is(parser, FZL_TOKTYPE_NAME))
return fzl_mknode(FZL_NODETYPE_IDENTIFIER, name);
fzl_token_t *realname = consume(parser); // :D
fzl_node_t *decl = fzl_mknode(FZL_NODETYPE_DECLARATION, name);
decl->var.type = name;
decl->var.identifier = fzl_mknode(FZL_NODETYPE_IDENTIFIER, realname);
return decl;
}
else if (is(parser, FZL_TOKTYPE_TYPE))
{
fzl_token_t *type = consume(parser);
fzl_token_t *name = consume(parser);
if (!name || name->type != FZL_TOKTYPE_NAME)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected name after type at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
fzl_node_t *decl = fzl_mknode(FZL_NODETYPE_DECLARATION, type);
decl->var.type = type;
decl->var.identifier = fzl_mknode(FZL_NODETYPE_IDENTIFIER, name);
return decl;
}
if (!plast_err[0])
snprintf(plast_err, sizeof(plast_err),
"'%s': expected expression at %ld:%ld", last_known_tok->filename,
last_known_tok->ln, last_known_tok->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);
if (!left)
return NULL;
if (is(parser, FZL_TOKTYPE_LPAREN) || is(parser, FZL_TOKTYPE_DOT))
{
while (is(parser, FZL_TOKTYPE_LPAREN) || is(parser, FZL_TOKTYPE_DOT))
{
fzl_token_t *post = consume(parser);
if (post->type == FZL_TOKTYPE_LPAREN)
{
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)
{
for (size_t i = 0; i < args_count; i++)
fzl_freeast(args[i]);
free(args);
fzl_freeast(left);
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)
{
for (size_t i = 0; i < args_count; i++)
fzl_freeast(args[i]);
free(args);
fzl_freeast(left);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected ')' after argument list at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->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;
}
else if (post->type == FZL_TOKTYPE_DOT)
{
if (!is(parser, FZL_TOKTYPE_NAME))
{
fzl_freeast(left);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected name after '.' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
fzl_token_t *name = consume(parser);
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_FIELD, name);
node->memb.parent = left;
node->memb.field = fzl_mknode(FZL_NODETYPE_IDENTIFIER, name);
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)
return NULL;
if (is(parser, FZL_TOKTYPE_CARET))
{
fzl_token_t *caret = consume(parser);
right = parse_unary(parser);
if (!right)
{
fzl_freeast(left);
if (!plast_err[0])
snprintf(plast_err, sizeof(plast_err),
"'%s': expected expression at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->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)
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)
{
fzl_freeast(left);
if (!plast_err[0])
snprintf(plast_err, sizeof(plast_err),
"'%s': expected expression at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->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)
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)
{
fzl_freeast(left);
if (!plast_err[0])
snprintf(plast_err, sizeof(plast_err),
"'%s': expected expression at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->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)
{
fzl_freeast(left);
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)
{
fzl_freeast(left);
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)
{
fzl_freeast(left);
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;
}
// Object construct parsing
fzl_node_t *parse_new(fzl_parser_t *parser)
{
if (!is(parser, FZL_TOKTYPE_NEW))
return parse_or(parser);
fzl_token_t *new = consume(parser);
fzl_node_t *objname = parse_or(parser);
if (!objname || (objname->type != FZL_NODETYPE_IDENTIFIER &&
objname->type != FZL_NODETYPE_FIELD))
{
fzl_freeast(objname);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected class name after 'new' keyword at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_CONSTRUCT, new);
node->construct.name = objname;
return node;
}
// Assignments parsing
fzl_node_t *parse_assignment(fzl_parser_t *parser)
{
fzl_node_t *left = parse_new(parser);
if (!left)
return NULL;
if (is(parser, FZL_TOKTYPE_EQ))
{
fzl_token_t *asn = consume(parser);
if (!left || (left->type != FZL_NODETYPE_IDENTIFIER &&
left->type != FZL_NODETYPE_DECLARATION &&
left->type != FZL_NODETYPE_FIELD))
{
fzl_freeast(left);
snprintf(plast_err, sizeof(plast_err),
"'%s': cannot assign to '%s' at %ld:%ld",
last_known_tok->filename, (char *)left->token->value,
last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *right = parse_assignment(parser);
if (!right)
{
fzl_freeast(left);
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;
}
// Expression parsing
fzl_node_t *parse_expression(fzl_parser_t *parser)
{
return parse_assignment(parser);
}
fzl_node_t *parse_statement(fzl_parser_t *parser);
// Parse class definition
fzl_node_t *parse_class(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_token_t *class = consume(parser);
fzl_token_t *name = consume(parser);
if (!name || name->type != FZL_TOKTYPE_NAME)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected class name at %ld:%ld", last_known_tok->filename,
last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_token_t *lbrace = consume(parser);
if (!lbrace || lbrace->type != FZL_TOKTYPE_LBRACE)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected '{' after class name at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
size_t fremainder = 2;
size_t fcount = 0;
fzl_node_t **fields = malloc(fremainder * sizeof(fzl_node_t *));
if (!fields)
{
snprintf(plast_err, sizeof(plast_err),
"failed to allocate space for class fields");
return NULL;
}
size_t mremainder = 2;
size_t mcount = 0;
fzl_node_t **methods = malloc(mremainder * sizeof(fzl_node_t *));
if (!methods)
{
snprintf(plast_err, sizeof(plast_err),
"failed to allocate space for class fields");
free(fields);
return NULL;
}
while (!is(parser, FZL_TOKTYPE_RBRACE))
{
fzl_node_t *node = parse_statement(parser);
if (!node)
return NULL;
if (node->type == FZL_NODETYPE_DECLARATION)
{
if (fremainder <= fcount)
{
fremainder *= 2;
fields = realloc(fields, fremainder * sizeof(fzl_node_t *));
if (!fields)
{
snprintf(plast_err, sizeof(plast_err),
"failed to resize fields array");
fzl_freeast(node);
free(methods);
return NULL;
}
}
fields[fcount++] = node;
}
else if (node->type == FZL_NODETYPE_FUNCTION)
{
if (mremainder <= mcount)
{
mremainder *= 2;
methods = realloc(methods, mremainder * sizeof(fzl_node_t *));
if (!methods)
{
snprintf(plast_err, sizeof(plast_err),
"failed to resize methods array");
fzl_freeast(node);
free(fields);
return NULL;
}
}
fzl_node_t **new_args = realloc(node->fn.args, (++node->fn.args_count) *
sizeof(fzl_node_t *));
if (!new_args)
{
snprintf(plast_err, sizeof(plast_err),
"failed to reallocate method args list");
free(fields);
fzl_freeast(node);
return NULL;
}
if (node->fn.args_count > 1)
memmove(&new_args[1], new_args,
(node->fn.args_count - 1) * sizeof(fzl_node_t *));
fzl_node_t *this_id = fzl_mknode(FZL_NODETYPE_IDENTIFIER, &this_tok);
fzl_node_t *this_dec = fzl_mknode(FZL_NODETYPE_DECLARATION, name);
this_dec->var.type = name;
this_dec->var.identifier = this_id;
new_args[0] = this_dec;
node->fn.args = new_args;
methods[mcount++] = node;
}
else
{
fzl_freeast(node);
snprintf(plast_err, sizeof(plast_err),
"'%s': unknown type of class member at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
free(fields);
free(methods);
return NULL;
}
}
fzl_token_t *rbra = consume(parser);
if (rbra->type != FZL_TOKTYPE_RBRACE)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected '}' after class definition at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
free(fields);
free(methods);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_CLASS, class);
node->class.name = name;
node->class.fields = fields;
node->class.fields_count = fcount;
node->class.methods = methods;
node->class.methods_count = mcount;
return node;
}
// Functions parsing
fzl_node_t *parse_function(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_token_t *fn = consume(parser);
if (!fn || fn->type != FZL_TOKTYPE_FUNCTION)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected keyword 'fn' at %ld:%ld", last_known_tok->filename,
last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_token_t *name = consume(parser);
if (!name || name->type != FZL_TOKTYPE_NAME)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected function name at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_token_t *lpar = consume(parser);
if (!lpar || lpar->type != FZL_TOKTYPE_LPAREN)
{
snprintf(plast_err, sizeof(plast_err), "'%s': expected '(' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
size_t remainder = 1;
size_t count = 0;
fzl_node_t **args = malloc(remainder * sizeof(fzl_node_t *));
while (peek(parser) && !is(parser, FZL_TOKTYPE_RPAREN))
{
fzl_node_t *arg = parse_expression(parser);
if (!arg)
return NULL;
if (remainder <= count)
{
remainder *= 2;
args = realloc(args, remainder * sizeof(fzl_node_t *));
}
args[count++] = arg;
if (is(parser, FZL_TOKTYPE_COMMA))
consume(parser);
}
fzl_token_t *rpar = consume(parser);
if (!rpar || rpar->type != FZL_TOKTYPE_RPAREN)
{
free(args);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected ')' after argument list at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_token_t *type = NULL;
if (is(parser, FZL_TOKTYPE_TYPE) || is(parser, FZL_TOKTYPE_NAME))
type = consume(parser);
if (!type)
type = &null_tok;
if (!is(parser, FZL_TOKTYPE_LBRACE))
{
free(args);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected '{' after function declaration at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *body = parse_statement(parser);
if (!body)
{
free(args);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_FUNCTION, name);
node->fn.args = args;
node->fn.args_count = count;
node->fn.body = body;
node->fn.name = name;
node->fn.type = type;
return node;
}
// Return parsing
fzl_node_t *parse_return(fzl_parser_t *parser)
{
if (!parser)
return NULL;
fzl_token_t *ret = consume(parser);
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_RETURN, ret);
if (!is(parser, FZL_TOKTYPE_SEMICOLON))
{
fzl_node_t *retval = parse_expression(parser);
if (!retval)
{
fzl_freeast(node);
return NULL;
}
node->ret.value = retval;
}
return node;
}
fzl_node_t *parse_if(fzl_parser_t *parser)
{
if (!parser)
return NULL;
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),
"'%s': expected '(' before condition at %ld:%ld", lparen->filename,
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)
{
fzl_freeast(condition);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected ')' after condition at %ld:%ld", rparen->filename,
rparen->ln, rparen->cn);
return NULL;
}
fzl_node_t *else_block = NULL;
fzl_node_t *true_block = parse_statement(parser);
if (!true_block)
{
fzl_freeast(condition);
return NULL;
}
if (is(parser, FZL_TOKTYPE_ELSE))
{
consume(parser); // else
else_block = parse_statement(parser);
if (!else_block)
{
fzl_freeast(condition);
fzl_freeast(true_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;
}
// Block statement parsing
fzl_node_t *parse_block(fzl_parser_t *parser)
{
if (!parser)
return NULL;
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)
{
free(stmts);
return NULL;
}
if (remainder <= count)
{
remainder *= 2;
stmts = realloc(stmts, remainder * sizeof(fzl_node_t *));
}
stmts[count++] = node;
}
fzl_token_t *rbrace = consume(parser);
if (!rbrace || rbrace->type != FZL_TOKTYPE_RBRACE)
{
for (size_t i = 0; i < count; i++)
fzl_freeast(stmts[i]);
free(stmts);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing '}' after block at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->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;
}
// For-loop parsing
fzl_node_t *parse_for(fzl_parser_t *parser)
{
fzl_token_t *tok = consume(parser);
fzl_token_t *lpar = consume(parser);
if (!lpar || lpar->type != FZL_TOKTYPE_LPAREN)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected '(' after for keyword at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *start = NULL;
if (!is(parser, FZL_TOKTYPE_SEMICOLON))
{
start = parse_expression(parser);
if (!start)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': expected loop init or ';' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
fzl_token_t *sc = consume(parser);
if (!sc || sc->type != FZL_TOKTYPE_SEMICOLON)
{
fzl_freeast(start);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing ';' after init section at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
}
else
consume(parser);
fzl_node_t *condition = NULL;
if (!is(parser, FZL_TOKTYPE_SEMICOLON))
{
condition = parse_expression(parser);
if (!condition)
{
fzl_freeast(start);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected loop condition or ';' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
fzl_token_t *sc = consume(parser);
if (!sc || sc->type != FZL_TOKTYPE_SEMICOLON)
{
fzl_freeast(start);
fzl_freeast(condition);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing ';' after condition at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
}
else
consume(parser);
fzl_node_t *step = NULL;
if (!is(parser, FZL_TOKTYPE_RPAREN))
{
step = parse_expression(parser);
if (!step)
{
fzl_freeast(start);
fzl_freeast(condition);
snprintf(plast_err, sizeof(plast_err),
"'%s': expected loop step end or ')' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln,
last_known_tok->cn);
return NULL;
}
}
fzl_token_t *rpar = consume(parser);
if (!rpar || rpar->type != FZL_TOKTYPE_RPAREN)
{
fzl_freeast(start);
fzl_freeast(condition);
fzl_freeast(step);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing ')' after loop setup at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *body = parse_statement(parser);
if (!body)
{
fzl_freeast(start);
fzl_freeast(condition);
fzl_freeast(step);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing body of for loop at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_LOOP, tok);
node->loop.init = start;
node->loop.cond = condition;
node->loop.step = step;
node->loop.body = body;
return node;
}
// While-loop parsing
fzl_node_t *parse_while(fzl_parser_t *parser)
{
fzl_token_t *tok = consume(parser);
fzl_token_t *lpar = consume(parser);
if (!lpar || lpar->type != FZL_TOKTYPE_LPAREN)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': missing '(' after while keyword at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *condition = parse_expression(parser);
if (!condition)
{
snprintf(plast_err, sizeof(plast_err),
"'%s': missing condition for 'while' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_token_t *rpar = consume(parser);
if (!rpar || rpar->type != FZL_TOKTYPE_RPAREN)
{
fzl_freeast(condition);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing ')' after while loop condition at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *body = parse_statement(parser);
if (!body)
{
fzl_freeast(condition);
snprintf(plast_err, sizeof(plast_err),
"'%s': missing while loop body at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
fzl_node_t *node = fzl_mknode(FZL_NODETYPE_LOOP, tok);
node->loop.cond = condition;
node->loop.body = body;
return node;
}
// `while` and `for` loops
fzl_node_t *parse_loop(fzl_parser_t *parser)
{
if (!peek(parser))
return NULL;
if (is(parser, FZL_TOKTYPE_FOR))
return parse_for(parser);
else if (is(parser, FZL_TOKTYPE_WHILE))
return parse_while(parser);
snprintf(plast_err, sizeof(plast_err),
"'%s': unknown type of loop at %ld:%ld", last_known_tok->filename,
last_known_tok->ln, last_known_tok->cn);
return NULL;
}
// `break` keyword parsing
fzl_node_t *parse_break(fzl_parser_t *parser)
{
fzl_token_t *tok = consume(parser);
if (!tok)
{
snprintf(plast_err, sizeof(plast_err), "'%s': EOF",
last_known_tok->filename);
return NULL;
}
return fzl_mknode(FZL_NODETYPE_BREAK, tok);
}
// Statement parsing
fzl_node_t *parse_statement(fzl_parser_t *parser)
{
if (!peek(parser))
return NULL;
fzl_node_t *res = NULL;
if (is(parser, FZL_TOKTYPE_CLASS))
return parse_class(parser);
else if (is(parser, FZL_TOKTYPE_FUNCTION))
return parse_function(parser);
else if (is(parser, FZL_TOKTYPE_RETURN))
res = parse_return(parser);
else if (is(parser, FZL_TOKTYPE_IF))
return parse_if(parser);
else if (is(parser, FZL_TOKTYPE_LBRACE))
return parse_block(parser);
else if (is(parser, FZL_TOKTYPE_FOR) || is(parser, FZL_TOKTYPE_WHILE))
return parse_loop(parser);
else if (is(parser, FZL_TOKTYPE_BREAK))
res = parse_break(parser);
if (!res)
res = parse_expression(parser);
if (!res)
{
fzl_freeast(res);
return NULL;
}
if (!is(parser, FZL_TOKTYPE_SEMICOLON))
{
fzl_freeast(res);
snprintf(plast_err, sizeof(plast_err), "'%s': missing ';' at %ld:%ld",
last_known_tok->filename, last_known_tok->ln, last_known_tok->cn);
return NULL;
}
while (is(parser, FZL_TOKTYPE_SEMICOLON))
consume(parser);
return res;
}
// Internal parsing entry point
fzl_node_t *parse_program(fzl_parser_t *parser)
{
if (!peek(parser))
return NULL;
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)
{
fzl_freeast(first);
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)
{
last_known_tok = NULL;
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)
{
switch (root->type)
{
case FZL_NODETYPE_BREAK:
case FZL_NODETYPE_IDENTIFIER:
case FZL_NODETYPE_LITERAL:
break;
case FZL_NODETYPE_FMT_STRING:
case FZL_NODETYPE_BLOCK:
if (root->list.entries)
{
for (size_t i = 0; i < root->list.entry_count; i++)
fzl_freeast(root->list.entries[i]);
free(root->list.entries);
}
break;
case FZL_NODETYPE_CALL:
fzl_freeast(root->call.callee);
if (root->call.args_count)
{
for (size_t i = 0; i < root->call.args_count; i++)
fzl_freeast(root->call.args[i]);
}
if (root->call.args)
free(root->call.args);
break;
case FZL_NODETYPE_CONDITIONAL:
fzl_freeast(root->cond.condition);
fzl_freeast(root->cond.true_block);
fzl_freeast(root->cond.else_block);
break;
case FZL_NODETYPE_LOOP:
fzl_freeast(root->loop.init);
fzl_freeast(root->loop.cond);
fzl_freeast(root->loop.step);
fzl_freeast(root->loop.body);
break;
case FZL_NODETYPE_ADD:
case FZL_NODETYPE_SUB:
case FZL_NODETYPE_MUL:
case FZL_NODETYPE_DIV:
case FZL_NODETYPE_POW:
case FZL_NODETYPE_MOD:
case FZL_NODETYPE_OR:
case FZL_NODETYPE_AND:
case FZL_NODETYPE_EQUAL:
case FZL_NODETYPE_NEQUAL:
case FZL_NODETYPE_LESS:
case FZL_NODETYPE_LESS_EQUAL:
case FZL_NODETYPE_GREATER:
case FZL_NODETYPE_GREATER_EQUAL:
case FZL_NODETYPE_ASSIGN:
fzl_freeast(root->binary_op.left);
fzl_freeast(root->binary_op.right);
break;
case FZL_NODETYPE_UNARY_PLUS:
case FZL_NODETYPE_UNARY_MINUS:
case FZL_NODETYPE_NOT:
fzl_freeast(root->unary_op.operand);
break;
case FZL_NODETYPE_FUNCTION:
if (root->fn.args_count)
for (size_t i = 0; i < root->fn.args_count; i++)
fzl_freeast(root->fn.args[i]);
free(root->fn.args);
fzl_freeast(root->fn.body);
break;
case FZL_NODETYPE_RETURN:
fzl_freeast(root->ret.value);
break;
case FZL_NODETYPE_CLASS:
if (root->class.fields_count)
for (size_t i = 0; i < root->class.fields_count; i++)
fzl_freeast(root->class.fields[i]);
free(root->class.fields);
if (root->class.methods_count)
for (size_t i = 0; i < root->class.methods_count; i++)
fzl_freeast(root->class.methods[i]);
free(root->class.methods);
break;
case FZL_NODETYPE_CONSTRUCT:
fzl_freeast(root->construct.name);
break;
case FZL_NODETYPE_FIELD:
fzl_freeast(root->memb.field);
fzl_freeast(root->memb.parent);
break;
case FZL_NODETYPE_DECLARATION:
fzl_freeast(root->var.identifier);
break;
}
fzl_node_t *next = root->next;
fzl_freenode(root);
root = next;
}
}
void fzl_freeparser(fzl_parser_t *parser)
{
if (!parser)
return;
free(parser);
}