word splitting: a single word is not split

Sorry this commit is a bit big, lots of groudwork is being established here
This commit is contained in:
Khaïs COLIN 2025-02-13 15:17:30 +01:00
parent 871571e258
commit a083800506
Signed by: logistic-bot
SSH key fingerprint: SHA256:RlpiqKeXpcPFZZ4y9Ou4xi2M8OhRJovIwDlbCaMsuAo
9 changed files with 262 additions and 0 deletions

View file

@ -27,6 +27,9 @@ srcs = \
src/get_command.c \ src/get_command.c \
src/parser/matchers/identifier.c \ src/parser/matchers/identifier.c \
src/parser/matchers/metacharacter.c \ src/parser/matchers/metacharacter.c \
src/parser/worddesc/worddesc.c \
src/parser/wordlist/wordlist.c \
src/parser/wordsplit/wordsplit.c \
objs = $(srcs:.c=.o) objs = $(srcs:.c=.o)
export objs export objs

View file

@ -0,0 +1,33 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* worddesc.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 17:20:36 by khais #+# #+# */
/* Updated: 2025/02/13 17:23:11 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#include "worddesc.h"
#include "libft.h"
#include <stdlib.h>
/*
** allocate a new worddesc with zeroed flags and the given word as word.
**
** return null in case of error, or if word is null
*/
t_worddesc *worddesc_create(char *word)
{
t_worddesc *retvalue;
if (word == NULL)
return (NULL);
retvalue = ft_calloc(1, sizeof(t_worddesc));
if (retvalue == NULL)
return (NULL);
retvalue->word = word;
return (retvalue);
}

View file

@ -0,0 +1,33 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* worddesc.h :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 15:47:58 by khais #+# #+# */
/* Updated: 2025/02/13 17:20:25 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#ifndef WORDDESC_H
# define WORDDESC_H
/*
** A logical word for the parser.
**
** See section 3.2.4 https://aosabook.org/en/v1/bash.html for more details.
**
** We are basically duplicating the way bash does it.
*/
typedef struct s_worddesc
{
/*
** The word itself
*/
char *word;
} t_worddesc;
t_worddesc *worddesc_create(char *word);
#endif

View file

@ -0,0 +1,57 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* wordlist.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 17:07:01 by khais #+# #+# */
/* Updated: 2025/02/14 13:29:16 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#include "wordlist.h"
#include "libft.h"
#include <stdlib.h>
/*
** create a new wordlist element, with next set to null and word set to the
** passed worddesc.
**
** in case of error, return null
*/
t_wordlist *wordlist_create(t_worddesc *word)
{
t_wordlist *retvalue;
retvalue = ft_calloc(1, sizeof(t_wordlist));
if (retvalue == NULL)
return (NULL);
retvalue->next = NULL;
retvalue->word = word;
return (retvalue);
}
/*
** free all memory associated with this wordlist.
*/
void wordlist_destroy(t_wordlist *wordlist)
{
free(wordlist);
}
/*
** get the worddesc at position idx in the given wordlist.
**
** return null if out of range or wordlist is null
*/
t_worddesc *wordlist_get(t_wordlist *wordlist, int idx)
{
if (wordlist == NULL || idx < 0)
return (NULL);
while (idx > 0 && wordlist != NULL)
wordlist = wordlist->next;
if (wordlist == NULL)
return (NULL);
return (wordlist->word);
}

View file

@ -0,0 +1,41 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* wordlist.h :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 15:46:02 by khais #+# #+# */
/* Updated: 2025/02/14 13:29:05 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#ifndef WORDLIST_H
# define WORDLIST_H
# include "../worddesc/worddesc.h"
/*
** Linked-list of WORDS
**
** cf. section 3.2.4 of https://aosabook.org/en/v1/bash.html
**
** we are copying how bash does it.
*/
typedef struct s_wordlist
{
/*
** pointer to the next element in the list
*/
struct s_wordlist *next;
/*
** pointer to the word at this position in the list
*/
struct s_worddesc *word;
} t_wordlist;
t_wordlist *wordlist_create(t_worddesc *word);
void wordlist_destroy(t_wordlist *wordlist);
t_worddesc *wordlist_get(t_wordlist *wordlist, int idx);
#endif

View file

@ -0,0 +1,25 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* wordsplit.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 17:02:32 by khais #+# #+# */
/* Updated: 2025/02/14 13:28:38 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#include "wordsplit.h"
#include <stdlib.h>
/*
** split a string into words, respecting quotes etc.
**
** cf. Token Recognition section at
** https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html
*/
t_wordlist *minishell_wordsplit(char *original)
{
return (wordlist_create(worddesc_create(original)));
}

View file

@ -0,0 +1,20 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* wordsplit.h :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 15:52:48 by khais #+# #+# */
/* Updated: 2025/02/13 15:54:30 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#ifndef WORDSPLIT_H
# define WORDSPLIT_H
# include "../wordlist/wordlist.h"
t_wordlist *minishell_wordsplit(char *original);
#endif

View file

@ -3,6 +3,7 @@
rawtests = \ rawtests = \
test_env_manip \ test_env_manip \
metacharacters \ metacharacters \
word_splitting \
ifeq ($(CFLAGS),) ifeq ($(CFLAGS),)
CFLAGS = -Wall -Wextra -Werror -g CFLAGS = -Wall -Wextra -Werror -g

49
tests/word_splitting.c Normal file
View file

@ -0,0 +1,49 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* word_splitting.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: khais <marvin@42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2025/02/13 15:17:56 by khais #+# #+# */
/* Updated: 2025/02/13 17:24:33 by khais ### ########.fr */
/* */
/* ************************************************************************** */
#include <assert.h>
#include "testutil.h"
#include "../src/parser/wordsplit/wordsplit.h"
#include <stdlib.h>
/*
** https://bash-hackers.gabe565.com/syntax/words/
*/
void test_wordsplit_singleword(void)
{
t_wordlist *words;
words = minishell_wordsplit("echo");
assert_strequal("echo", wordlist_get(words, 0)->word);
assert(NULL == wordlist_get(words, 1));
wordlist_destroy(words);
}
void test_wordsplit_basic(void)
{
t_wordlist *words;
words = minishell_wordsplit("echo The file is named $MYFILE");
assert_strequal("echo", wordlist_get(words, 0)->word);
assert_strequal("The", wordlist_get(words, 1)->word);
assert_strequal("file", wordlist_get(words, 2)->word);
assert_strequal("is", wordlist_get(words, 3)->word);
assert_strequal("named", wordlist_get(words, 4)->word);
assert_strequal("$MYFILE", wordlist_get(words, 5)->word);
assert(NULL == wordlist_get(words, 6));
wordlist_destroy(words);
}
int main(void) {
test_wordsplit_singleword();
return (0);
}