42sh/src/lexer/lexer.c

141 lines
2.6 KiB
C
Raw Normal View History

2026-01-08 16:32:48 +01:00
#include "lexer.h"
2026-01-12 21:31:15 +01:00
#include <ctype.h>
2026-01-08 16:32:48 +01:00
#include <stdbool.h>
2026-01-10 19:57:36 +01:00
#include <stddef.h>
#include <stdlib.h>
#include <string.h>
2026-01-08 16:32:48 +01:00
#include "io_backend/io_backend.h"
2026-01-12 21:31:15 +01:00
#include "utils/string_utils/string_utils.h"
2026-01-08 16:32:48 +01:00
static char *end_last_token;
static ssize_t remaining_chars;
2026-01-12 21:31:15 +01:00
/* @brief: saves state for the next call the the lexer.
*
*/
static void save_state(char *stream, ssize_t i)
{
remaining_chars -= i;
end_last_token = stream + i;
return;
}
/* @return: true if a special character from the grammar was found,
* false otherwise.
*
*/
static bool is_special_char(char c)
{
return c == '\'' || c == '\n' || c == ';' || c == 'EOF';
2026-01-12 21:31:15 +01:00
}
static void new_token_keyword(struct token *tok, char *begin, ssize_t size)
2026-01-12 21:31:15 +01:00
{
if (strncmp(begin, "if", size) == 0)
{
tok->type = TOKEN_IF;
}
if (strncmp(begin, "fi", size) == 0)
2026-01-12 21:31:15 +01:00
{
tok->type = TOKEN_FI;
2026-01-12 21:31:15 +01:00
}
if (strncmp(begin, "then", size) == 0)
2026-01-12 21:31:15 +01:00
{
tok->type = TOKEN_THEN;
}
if (strncmp(begin, "else", size) == 0)
{
tok->type = TOKEN_ELSE;
2026-01-12 21:31:15 +01:00
}
}
struct token *new_token(char *begin, ssize_t size)
2026-01-08 16:32:48 +01:00
{
struct token *res = calloc(1, sizeof(struct token));
2026-01-08 16:32:48 +01:00
if (res == NULL)
return NULL;
// checks which type of token
// is special char
// is keyword
// otherwise -> WORD
char *token_data = calloc(size + 1, sizeof(char));
2026-01-08 16:32:48 +01:00
strncpy(res, begin, size);
return res;
}
char *stream_init(void)
{
char *stream;
if (remaining_chars == 0)
{
remaining_chars = stream_read(&stream);
}
else
{
stream = end_last_token;
}
2026-01-12 21:41:13 +01:00
char *trimed_stream = trim_blank_left(stream);
2026-01-12 21:31:15 +01:00
remaining_chars -= trimed_stream - stream;
stream = trimed_stream;
2026-01-08 16:32:48 +01:00
return stream;
}
struct token *peek_token(void)
2026-01-08 16:32:48 +01:00
{
char *stream = stream_init();
ssize_t i = 0;
while (i < remaining_chars)
{
2026-01-12 21:31:15 +01:00
if (is_special_char(stream[i]))
2026-01-08 16:32:48 +01:00
{
2026-01-12 21:31:15 +01:00
if (i == 0) // where we create spe_char token
i++;
2026-01-08 16:32:48 +01:00
break;
2026-01-12 21:31:15 +01:00
}
if (isblank(stream[i]))
{
break;
}
2026-01-08 16:32:48 +01:00
i++;
}
return new_token(stream, i);
}
struct token *pop_token(void)
{
2026-01-13 17:56:30 +01:00
char *stream = stream_init();
ssize_t i = 0;
while (i < remaining_chars)
{
if (is_special_char(stream[i]))
{
if (i == 0) // where we create spe_char token
i++;
break;
}
if (isblank(stream[i]))
{
break;
}
i++;
}
2026-01-12 21:31:15 +01:00
save_state(stream, i);
2026-01-13 17:56:30 +01:00
return new_token(stream, i);
2026-01-08 16:32:48 +01:00
}