feat: Merge branch 'lexer' into dev

This commit is contained in:
Gu://em_ 2026-01-10 19:28:59 +01:00
commit 03c35d5366
6 changed files with 280 additions and 0 deletions

View file

@ -0,0 +1,67 @@
#include "lexer.h"
#include <stdbool.h>
#include "io_backend/io_backend.h"
static char *end_last_token;
static ssize_t remaining_chars;
char *new_token(char *begin, size_t size)
{
char *res = calloc(size + 1, sizeof(char));
if (res == NULL)
return NULL;
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;
}
return stream;
}
char *get_token(void)
{
char *stream = stream_init();
bool inquotes = false;
ssize_t i = 0;
while (i < remaining_chars)
{
switch (stream[i])
{
case '\'':
inquotes = !inquotes;
break;
case ' ' | '\n' | '\t':
if (inquotes)
break;
else
{
// token creation
// skip blank char
// exit from loop
char *token = new_token(stream, i);
}
default:
break;
}
i++;
}
remaining_chars -= i;
}

View file

@ -1,4 +1,38 @@
#ifndef LEXER_H
#define LEXER_H
#include <sys/types.h>
/* @return: char*, the next token
*
*/
char *get_token(void);
/*
* @warning: NOT IMPLEMENTED.
*
* @note: maybe usefull for subshells.
*/
char *get_token_str(void);
/*
* @brief: return a newly allocated token.
* This token contains [size] chars, starting from [begin].
*
* @return: NULL on error, null-terminated char* otherwise.
*
*/
char *new_token(char *begin, ssize_t size);
/*
* @brief: checks if the stream used for the last token creation is empty.
* If it is, it calls stream_read() from IO_backend,
* and sets [remaing_chars].
* If not, it starts from the end of the last token.
*
* @return: char* stream from which we tokenise.
*/
char *stream_init(void);
#endif /* ! LEXER_H */

View file

@ -0,0 +1,16 @@
#include "utils/string_utils/string_utils.h"
ssize_t skip_blanks(char **str)
{
if (str == NULL || *str == NULL)
{
return 0;
}
ssize_t skipped = 0;
while (str[skipped] != '\0' && !isblank(str[skipped]))
{
skipped++;
}
*str += skipped;
return skipped;
}

View file

@ -0,0 +1,10 @@
#ifndef STRING_UTILS_H
#define STRING_UTILS_H
/*
* @brief: skips blank characters at the beginning of [str].
* @return: number of characters skipped.
*/
ssize_t skip_blanks(char **str);
#endif /* STRING_UTILS_H */

View file

@ -0,0 +1,56 @@
#include <criterion/criterion.h>
#include <criterion/new/assert.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include "lexer/lexer.h"
TestSuite(token_creation);
Test(token_creation, basic)
{
char input[] = "Hello World";
char actual[] = new_token(input, 5);
char expected[] = "Hello";
cr_expect(eq(str, actual, expected));
free(actual);
}
Test(token_creation, nul)
{
char input[] = NULL;
char actual[] = new_token(input, 5);
cr_expect(actual == NULL);
}
Test(token_creation, too_large)
{
char input[] = "Hel";
char actual[] = new_token(input, 5);
cr_expect(actual == NULL);
}
Test(token_creation, empty)
{
char input[] = "";
char actual[] = new_token(input, 5);
cr_expect(actual == NULL);
}
Test(token_creation, basic_long)
{
char input[] = "Hello World! This project is a mini shell, I love BIG G.";
char actual[] = new_token(input, 42);
char expected[] = calloc(42 + 1, sizeof(char));
strncpy(input, expected, 42);
cr_expect(eq(str, actual, expected));
free(actual);
free(expected);
}

View file

@ -0,0 +1,97 @@
#include <criterion/criterion.h>
#include <criterion/new/assert.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include "utils/string_utils/string_utils.h"
TestSuite(string_utils);
Test(string_utils, skipblank_basic)
{
char input[] = " Hello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 2;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_noblank)
{
char input[] = "Hello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 0;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_tab)
{
char input[] = "\tHello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 1;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_space_tab)
{
char input[] = " \tHello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 2;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_2tab_1space)
{
char input[] = "\t \tHello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 3;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_a_lot)
{
char input[] = "\t \t \tHello World";
char expected_str[] = "Hello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 8;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_newline)
{
char input[] = "\nHello World";
char expected_str[] = "\nHello World";
ssize_t actual = skip_blanks(input);
ssize_t expected = 0;
cr_expect(eq(str, input, expected_str));
cr_expect(actual == expected);
}
Test(string_utils, skipblank_nul)
{
char *input = NULL;
ssize_t actual = skip_blanks(input);
ssize_t expected = 0;
cr_expect(input == NULL);
cr_expect(actual == expected);
}