minishell/tests/test_word_splitting.c

/* ************************************************************************** */
/*                                                                            */
/*                                                        :::      ::::::::   */
/*   word_splitting.c                                   :+:      :+:    :+:   */
/*                                                    +:+ +:+         +:+     */
/*   By: khais <marvin@42.fr>                       +#+  +:+       +#+        */
/*                                                +#+#+#+#+#+   +#+           */
/*   Created: 2025/02/13 15:17:56 by khais             #+#    #+#             */
/*   Updated: 2025/03/06 16:50:56 by khais            ###   ########.fr       */
/*                                                                            */
/* ************************************************************************** */

#include <assert.h>
#include "testutil.h"
#include "../src/parser/wordsplit/wordsplit.h"
#include <unistd.h>
#include "libft.h"
#include <stdlib.h>

/*
** https://bash-hackers.gabe565.com/syntax/words/
*/
static void	test_wordsplit_singleword(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("echo");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert(NULL == wordlist_get(words, 1));
	wordlist_destroy(words);
}

static void	test_wordsplit_singleword_with_blanks(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("\t  \t echo  \t\t ");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert(NULL == wordlist_get(words, 1));
	wordlist_destroy(words);
}

static void	test_wordsplit_multiword(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("\t  echo\tThe  file is named    $MYFILE  \t");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert_strequal("The", wordlist_get(words, 1)->word);
	assert_strequal("   ", wordlist_get(words, 1)->marker);
	assert_strequal("file", wordlist_get(words, 2)->word);
	assert_strequal("    ", wordlist_get(words, 2)->marker);
	assert_strequal("is", wordlist_get(words, 3)->word);
	assert_strequal("  ", wordlist_get(words, 3)->marker);
	assert_strequal("named", wordlist_get(words, 4)->word);
	assert_strequal("     ", wordlist_get(words, 4)->marker);
	assert_strequal("$MYFILE", wordlist_get(words, 5)->word);
	assert_strequal("       ", wordlist_get(words, 5)->marker);
	assert(NULL == wordlist_get(words, 6));
	wordlist_destroy(words);
}

static void	test_wordsplit_multiword_with_single_quotes(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("\t  echo\t' \t The  file is named    $MYFILE  ' \t");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert_strequal("' \t The  file is named    $MYFILE  '", wordlist_get(words, 1)->word);
	// This marker is one char shorter because the tab character above is
	// represented here with two chars
	ft_dprintf(STDERR_FILENO, "the marker below is shorter than the string above because the string above contains a tab, it is normal\n");
	assert_strequal(" '''''''''''''''''''''''''''''''''' ", wordlist_get(words, 1)->marker);
	assert(NULL == wordlist_get(words, 2));
	wordlist_destroy(words);
}

static void	test_wordsplit_multiword_with_double_quotes(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("\t  echo\t\" \t The  file is named    $MYFILE  \" \t");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert_strequal("\" \t The  file is named    $MYFILE  \"", wordlist_get(words, 1)->word);
	// This marker is one char shorter because the tab character above is
	// represented here with two chars
	ft_dprintf(STDERR_FILENO, "the marker below is shorter than the string above because the string above contains a tab, it is normal\n");
	assert_strequal(" \"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\" ", wordlist_get(words, 1)->marker);
	assert(NULL == wordlist_get(words, 2));
	wordlist_destroy(words);
}

static void	test_wordsplit_mixed_single_in_double(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("hello \"mixed ' \tquotes \t'\" there");
	assert_strequal("hello", wordlist_get(words, 0)->word);
	assert_strequal("     ", wordlist_get(words, 0)->marker);
	assert_strequal("\"mixed ' \tquotes \t'\"", wordlist_get(words, 1)->word);
	assert_strequal(" \"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\" ", wordlist_get(words, 1)->marker);
	assert_strequal("there", wordlist_get(words, 2)->word);
	assert_strequal("     ", wordlist_get(words, 2)->marker);
	assert(NULL == wordlist_get(words, 3));
	wordlist_destroy(words);
}

static void	test_wordsplit_mixed_double_in_single(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("hello 'mixed \" quotes \"' there");
	assert_strequal("hello", wordlist_get(words, 0)->word);
	assert_strequal("     ", wordlist_get(words, 0)->marker);
	assert_strequal("'mixed \" quotes \"'", wordlist_get(words, 1)->word);
	assert_strequal(" '''''''''''''''' ", wordlist_get(words, 1)->marker);
	assert_strequal("there", wordlist_get(words, 2)->word);
	assert_strequal("     ", wordlist_get(words, 2)->marker);
	assert(NULL == wordlist_get(words, 3));
	wordlist_destroy(words);
}

static void	test_wordsplit_mixed_broken(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("hello '\"mixed 'quotes'\"' there");
	assert_strequal("hello", wordlist_get(words, 0)->word);
	assert_strequal("     ", wordlist_get(words, 0)->marker);
	assert_strequal("'\"mixed 'quotes'\"'", wordlist_get(words, 1)->word);
	assert_strequal(" '''''''        ' ", wordlist_get(words, 1)->marker);
	assert_strequal("there", wordlist_get(words, 2)->word);
	assert_strequal("     ", wordlist_get(words, 2)->marker);
	assert(NULL == wordlist_get(words, 3));
	wordlist_destroy(words);
}

static void	test_wordsplit_unclosed_single(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("'hello");
	assert(words == NULL);
}

static void	test_wordsplit_unclosed_double(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("\"hello");
	assert(words == NULL);
}

static void	test_wordsplit_operator_word(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit(">test");
	assert_strequal(">", wordlist_get(words, 0)->word);
	assert_strequal(" ", wordlist_get(words, 0)->marker);
	assert_strequal("test", wordlist_get(words, 1)->word);
	assert_strequal("    ", wordlist_get(words, 1)->marker);
	assert(NULL == wordlist_get(words, 2));
	wordlist_destroy(words);
}

static void	test_wordsplit_all_operators(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("|&&||()<>><<>");
	assert_strequal("|", wordlist_get(words, 0)->word);
	assert_strequal(" ", wordlist_get(words, 0)->marker);
	assert_strequal("&&", wordlist_get(words, 1)->word);
	assert_strequal("  ", wordlist_get(words, 1)->marker);
	assert_strequal("||", wordlist_get(words, 2)->word);
	assert_strequal("  ", wordlist_get(words, 2)->marker);
	assert_strequal("(", wordlist_get(words, 3)->word);
	assert_strequal(" ", wordlist_get(words, 3)->marker);
	assert_strequal(")", wordlist_get(words, 4)->word);
	assert_strequal(" ", wordlist_get(words, 4)->marker);
	assert_strequal("<", wordlist_get(words, 5)->word);
	assert_strequal(" ", wordlist_get(words, 5)->marker);
	assert_strequal(">>", wordlist_get(words, 6)->word);
	assert_strequal("  ", wordlist_get(words, 6)->marker);
	assert_strequal("<<", wordlist_get(words, 7)->word);
	assert_strequal("  ", wordlist_get(words, 7)->marker);
	assert_strequal(">", wordlist_get(words, 8)->word);
	assert_strequal(" ", wordlist_get(words, 8)->marker);
	assert(NULL == wordlist_get(words, 9));
	wordlist_destroy(words);
}

static void	test_wordsplit_operator_combining(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("|||>>><<<&&&");
	assert_strequal("||", wordlist_get(words, 0)->word);
	assert_strequal("  ", wordlist_get(words, 0)->marker);
	assert_strequal("|", wordlist_get(words, 1)->word);
	assert_strequal(" ", wordlist_get(words, 1)->marker);
	assert_strequal(">>", wordlist_get(words, 2)->word);
	assert_strequal("  ", wordlist_get(words, 2)->marker);
	assert_strequal(">", wordlist_get(words, 3)->word);
	assert_strequal(" ", wordlist_get(words, 3)->marker);
	assert_strequal("<<", wordlist_get(words, 4)->word);
	assert_strequal("  ", wordlist_get(words, 4)->marker);
	assert_strequal("<", wordlist_get(words, 5)->word);
	assert_strequal(" ", wordlist_get(words, 5)->marker);
	assert_strequal("&&", wordlist_get(words, 6)->word);
	assert_strequal("  ", wordlist_get(words, 6)->marker);
	assert_strequal("&", wordlist_get(words, 7)->word);
	assert_strequal(" ", wordlist_get(words, 7)->marker);
	assert(NULL == wordlist_get(words, 8));
	wordlist_destroy(words);
}

static void	test_wordsplit_var_substitution(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("echo VAR=$VAR here");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert(0 == wordlist_get(words, 0)->flags);
	assert_strequal("VAR=$VAR", wordlist_get(words, 1)->word);
	assert_strequal("        ", wordlist_get(words, 1)->marker);
	assert(W_HASDOLLAR == wordlist_get(words, 1)->flags);
	assert_strequal("here", wordlist_get(words, 2)->word);
	assert_strequal("    ", wordlist_get(words, 2)->marker);
	assert(0 == wordlist_get(words, 2)->flags);
	wordlist_destroy(words);
}

static void	test_wordsplit_var_substitution_quotes(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("VAR=\"$VAR\"");
	assert_strequal("VAR=\"$VAR\"", wordlist_get(words, 0)->word);
	assert_strequal("     \"\"\"\" ", wordlist_get(words, 0)->marker);
	assert(W_HASDOLLAR & wordlist_get(words, 0)->flags);
	wordlist_destroy(words);
	words = minishell_wordsplit("VAR='$VAR'");
	assert_strequal("VAR='$VAR'", wordlist_get(words, 0)->word);
	assert((W_HASDOLLAR & wordlist_get(words, 0)->flags) == 0);
	assert_strequal("     '''' ", wordlist_get(words, 0)->marker);
	wordlist_destroy(words);
}

static void	test_wordsplit_quote_detection_nonnested(void)
{
	t_wordlist	*words;

	words = minishell_wordsplit("echo 'single quotes' here \"double quotes\" here");
	assert_strequal("echo", wordlist_get(words, 0)->word);
	assert_strequal("    ", wordlist_get(words, 0)->marker);
	assert(0 == wordlist_get(words, 0)->flags);
	assert_strequal("'single quotes'", wordlist_get(words, 1)->word);
	assert_strequal(" ''''''''''''' ", wordlist_get(words, 1)->marker);
	assert(W_QUOTED == wordlist_get(words, 1)->flags);
	assert_strequal("here", wordlist_get(words, 2)->word);
	assert_strequal("    ", wordlist_get(words, 2)->marker);
	assert(0 == wordlist_get(words, 2)->flags);
	assert_strequal("\"double quotes\"", wordlist_get(words, 3)->word);
	assert_strequal(" \"\"\"\"\"\"\"\"\"\"\"\"\" ", wordlist_get(words, 3)->marker);
	assert((W_QUOTED | W_DQUOTE) == wordlist_get(words, 3)->flags);
	assert_strequal("here", wordlist_get(words, 4)->word);
	assert_strequal("    ", wordlist_get(words, 4)->marker);
	assert(0 == wordlist_get(words, 4)->flags);
	wordlist_destroy(words);
}

static void	test_wordsplit_quote_detection_nested_double_in_simple(void)
{
	t_wordlist	*words;
	char		*str;

	str = "'these are single quotes \"with double\" inside'";
	words = minishell_wordsplit(str);
	assert_strequal(str, wordlist_get(words, 0)->word);
	assert_strequal(" '''''''''''''''''''''''''''''''''''''''''''' ", wordlist_get(words, 0)->marker);
	assert(W_QUOTED == wordlist_get(words, 0)->flags);
	wordlist_destroy(words);
	str = "'\"these are single quotes with double inside\"'";
	words = minishell_wordsplit(str);
	assert_strequal(str, wordlist_get(words, 0)->word);
	assert_strequal(" '''''''''''''''''''''''''''''''''''''''''''' ", wordlist_get(words, 0)->marker);
	assert(W_QUOTED == wordlist_get(words, 0)->flags);
	wordlist_destroy(words);
}

static void	test_wordsplit_quote_detection_nested_single_in_double(void)
{
	t_wordlist	*words;
	char		*str;

	str = "\"these are double quotes 'with single' inside\"";
	words = minishell_wordsplit(str);
	assert_strequal(str, wordlist_get(words, 0)->word);
	assert((W_QUOTED | W_DQUOTE) == wordlist_get(words, 0)->flags);
	assert_strequal(" \"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\" ", wordlist_get(words, 0)->marker);
	wordlist_destroy(words);
	str = "\"'these are double quotes with single inside'\"";
	words = minishell_wordsplit(str);
	assert_strequal(str, wordlist_get(words, 0)->word);
	assert_strequal(" \"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\" ", wordlist_get(words, 0)->marker);
	assert((W_QUOTED | W_DQUOTE) == wordlist_get(words, 0)->flags);
	wordlist_destroy(words);
}

int main(void) {
	test_wordsplit_singleword();
	test_wordsplit_singleword_with_blanks();
	test_wordsplit_multiword();
	test_wordsplit_multiword_with_single_quotes();
	test_wordsplit_multiword_with_double_quotes();
	test_wordsplit_mixed_single_in_double();
	test_wordsplit_mixed_double_in_single();
	test_wordsplit_mixed_broken();
	test_wordsplit_unclosed_single();
	test_wordsplit_unclosed_double();
	test_wordsplit_operator_word();
	test_wordsplit_all_operators();
	test_wordsplit_operator_combining();
	test_wordsplit_var_substitution();
	test_wordsplit_var_substitution_quotes();
	test_wordsplit_quote_detection_nonnested();
	test_wordsplit_quote_detection_nested_double_in_simple();
	test_wordsplit_quote_detection_nested_single_in_double();
	return (0);
}