Files
fz-lang/src/compiler/fzl_reader.c
T
2026-10-15 21:41:45 +03:00

388 lines
10 KiB
C

#include "compiler/fzl_token.h"
#include "compiler/utils/fzl_linked.h"
#include <compiler/fzl_reader.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
// Last error from reader
static char last_err[256] = {0};
fzl_reader_t *fzl_mkreader(fzl_linked_t *linked)
{
fzl_reader_ll_t *first_node = malloc(sizeof(fzl_reader_ll_t));
first_node->token = NULL;
first_node->next = NULL;
fzl_reader_t *reader = malloc(sizeof(fzl_reader_t));
reader->first = first_node;
reader->last = first_node;
reader->linked = linked;
reader->source = linked->source;
reader->size = linked->total_sz;
reader->position = 0;
return reader;
}
// --- Read zone ---
// Move cursor forward
// @deprecated
static size_t move(fzl_reader_t *reader, size_t len)
{
if (!reader)
return 0;
return reader->position += len;
}
// Try parse name
fzl_token_t *parse_name(fzl_reader_t *reader)
{
size_t lpos = reader->position;
size_t size = 0;
do
{
lpos++;
size++;
} while ((reader->source[lpos] >= 'A' && reader->source[lpos] <= 'Z') ||
(reader->source[lpos] >= 'a' && reader->source[lpos] <= 'z') ||
(reader->source[lpos] >= '0' && reader->source[lpos] <= '9') ||
(reader->source[lpos] == '_'));
char *value = malloc(size + 1);
memcpy(value, reader->source + reader->position, size);
value[size] = '\0';
size_t start = reader->position;
move(reader, size);
fzl_token_t *tok;
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, start, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return NULL;
}
if (strcmp(value, "class") == 0)
tok =
fzl_mktoken(FZL_TOKTYPE_CLASS, l.line, l.col, value, size, l.filename);
else if (strcmp(value, "new") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_NEW, l.line, l.col, value, size, l.filename);
else if (strcmp(value, "fn") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_FUNCTION, l.line, l.col, value, size,
l.filename);
else if (strcmp(value, "return") == 0)
tok =
fzl_mktoken(FZL_TOKTYPE_RETURN, l.line, l.col, value, size, l.filename);
else if (strcmp(value, "if") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_IF, l.line, l.col, value, size, l.filename);
else if (strcmp(value, "else") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_ELSE, l.line, l.col, value, size, l.filename);
else if (strcmp(value, "true") == 0 || strcmp(value, "false") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_BOOL_LITERAL, l.line, l.col, value, size,
l.filename);
else if (strcmp(value, "str") == 0 || strcmp(value, "char") == 0 ||
strcmp(value, "bool") == 0 || strcmp(value, "uint") == 0 ||
strcmp(value, "int") == 0 || strcmp(value, "i8") == 0 ||
strcmp(value, "i16") == 0 || strcmp(value, "i32") == 0 ||
strcmp(value, "i64") == 0 || strcmp(value, "u8") == 0 ||
strcmp(value, "u16") == 0 || strcmp(value, "u32") == 0 ||
strcmp(value, "u64") == 0 || strcmp(value, "null") == 0)
tok = fzl_mktoken(FZL_TOKTYPE_TYPE, l.line, l.col, value, size, l.filename);
else
tok = fzl_mktoken(FZL_TOKTYPE_NAME, l.line, l.col, value, size, l.filename);
free(value);
return tok;
}
// Try parse string literal
fzl_token_t *parse_string_literal(fzl_reader_t *reader)
{
size_t lpos = reader->position;
size_t size = 0;
if (reader->source[lpos] != '\'' && reader->source[lpos] != '"')
return NULL;
char open_symbol = reader->source[lpos];
char close_symbol = '\0';
do
{
lpos++;
size++;
if (reader->source[lpos] == open_symbol)
close_symbol = open_symbol;
} while (reader->source[lpos] != open_symbol && lpos < reader->size);
size++;
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, reader->position, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return NULL;
}
if (close_symbol != open_symbol)
{
snprintf(last_err, sizeof(last_err), "'%s': unterminated string at %ld:%ld",
l.filename, l.line, l.col);
return NULL;
}
if (size == 2)
{
fzl_token_t *tok =
fzl_mktoken(FZL_TOKTYPE_STR_LITERAL, l.line, l.col, "", 0, l.filename);
move(reader, 2);
return tok;
}
char *value = malloc(size - 2);
memcpy(value, reader->source + reader->position + 1, size - 2);
fzl_token_t *tok = fzl_mktoken(FZL_TOKTYPE_STR_LITERAL, l.line, l.col, value,
size - 2, l.filename);
free(value);
move(reader, size);
return tok;
}
// Parse numeric literal
fzl_token_t *parse_number(fzl_reader_t *reader)
{
if (!reader)
return NULL;
size_t size = 0;
size_t lpos = reader->position;
do
{
size++;
lpos++;
} while ((reader->source[lpos] >= '0' && reader->source[lpos] <= '9') &&
lpos < reader->size);
char *value = malloc(size * sizeof(char) + 1);
memcpy(value, reader->source + reader->position, size);
value[size] = '\0';
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, reader->position, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return NULL;
}
fzl_token_t *tok = fzl_mktoken(FZL_TOKTYPE_NUM_LITERAL, l.line, l.col, value,
size, l.filename);
free(value);
if (!tok)
return NULL;
move(reader, size);
return tok;
}
// Add single-char token
fzl_token_t *add_single(fzl_reader_t *reader, fzl_toktype_t type)
{
size_t p = move(reader, 1);
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, p, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return NULL;
}
return fzl_mktoken(type, l.line, l.col, NULL, 0, l.filename);
}
// Add double-char token
fzl_token_t *add_double(fzl_reader_t *reader, fzl_toktype_t type)
{
size_t p = move(reader, 2);
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, p, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return NULL;
}
return fzl_mktoken(type, l.line, l.col, NULL, 0, l.filename);
}
// Skip symbols until '\n' or EOF
void skip_comment(fzl_reader_t *reader)
{
move(reader, 2);
while (reader->position < reader->size)
{
if (reader->source[reader->position++] == '\n')
break;
}
}
int fzl_read(fzl_reader_t *reader)
{
if (reader->position >= reader->size)
return 0;
do
{
size_t p = reader->position;
fzl_token_t *token;
char curr = reader->source[p];
if (curr == '\n' || curr == ' ' || curr == '\r')
{
move(reader, 1);
continue;
}
if (curr == '\'' || curr == '"')
token = parse_string_literal(reader);
else if (curr == '/' && reader->source[p + 1] == '/')
{
skip_comment(reader);
continue;
}
else if (curr == '=' && reader->source[p + 1] == '=')
token = add_double(reader, FZL_TOKTYPE_EQEQ);
else if (curr == '>' && reader->source[p + 1] == '=')
token = add_double(reader, FZL_TOKTYPE_REQ);
else if (curr == '<' && reader->source[p + 1] == '=')
token = add_double(reader, FZL_TOKTYPE_LEQ);
else if (curr == '!' && reader->source[p + 1] == '=')
token = add_double(reader, FZL_TOKTYPE_NEQ);
else if (curr == '|' && reader->source[p + 1] == '|')
token = add_double(reader, FZL_TOKTYPE_DVBAR);
else if (curr == '&' && reader->source[p + 1] == '&')
token = add_double(reader, FZL_TOKTYPE_DAMPERSAND);
else if (curr == '>')
token = add_single(reader, FZL_TOKTYPE_RANGLE);
else if (curr == '<')
token = add_single(reader, FZL_TOKTYPE_LANGLE);
else if (curr == '!')
token = add_single(reader, FZL_TOKTYPE_NOT);
else if (curr == ';')
token = add_single(reader, FZL_TOKTYPE_SEMICOLON);
else if (curr == '{')
token = add_single(reader, FZL_TOKTYPE_LBRACE);
else if (curr == '}')
token = add_single(reader, FZL_TOKTYPE_RBRACE);
else if (curr == '(')
token = add_single(reader, FZL_TOKTYPE_LPAREN);
else if (curr == ')')
token = add_single(reader, FZL_TOKTYPE_RPAREN);
else if (curr == ',')
token = add_single(reader, FZL_TOKTYPE_COMMA);
else if (curr == '.')
token = add_single(reader, FZL_TOKTYPE_DOT);
else if (curr == '*')
token = add_single(reader, FZL_TOKTYPE_STAR);
else if (curr == '/')
token = add_single(reader, FZL_TOKTYPE_SLASH);
else if (curr == '+')
token = add_single(reader, FZL_TOKTYPE_PLUS);
else if (curr == '-')
token = add_single(reader, FZL_TOKTYPE_MINUS);
else if (curr == '^')
token = add_single(reader, FZL_TOKTYPE_CARET);
else if (curr == '%')
token = add_single(reader, FZL_TOKTYPE_PERCENT);
else if (curr == '=')
token = add_single(reader, FZL_TOKTYPE_EQ);
else if (curr >= 'A' && curr <= 'Z' || curr >= 'a' && curr <= 'z' ||
curr == '_')
token = parse_name(reader);
else if (curr >= '0' && curr <= '9')
token = parse_number(reader);
else
{
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, p, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return -1;
}
sprintf(last_err, "'%s': undefined token '%c' at %ld:%ld", l.filename,
curr, l.line, l.col);
return -1;
}
if (!token)
{
if (!last_err[0])
{
fzl_file_loc l;
bool ok = fzl_linked_getpos(reader->linked, p, &l);
if (!ok)
{
snprintf(last_err, sizeof(last_err), "failed to locate line:col");
return -1;
}
sprintf(last_err, "'%s': cannot parse token at %ld:%ld", l.filename,
l.line, l.col);
}
return -1;
}
fzl_reader_ll_t *tok_node = malloc(sizeof(fzl_reader_ll_t));
tok_node->token = token;
reader->last->next = tok_node;
reader->last = tok_node;
} while (reader->position < reader->size);
return 0;
}
// --- Read zone end ---
char *fzl_reader_geterror() { return last_err; }
void free_node_and_next(fzl_reader_ll_t *node)
{
if (!node)
return;
if (node->next)
free_node_and_next(node->next);
if (node->token)
fzl_freetoken(node->token);
free(node);
}
void fzl_freereader(fzl_reader_t *reader)
{
if (!reader)
return;
if (reader->first)
free_node_and_next(reader->first);
free(reader);
}