Merge pull request #12 from dpu-kaiser/tokenizer

Tokenizer

9 files changed, 174 insertions, 14 deletions

diff --git a/Makefile b/Makefile
index 41551ec..4ac27f7 100644
--- a/Makefile
+++ b/Makefile
@@ -12,7 +12,7 @@ HEADERS =  -I include -I $(LIB_DIR)/libft
 
 VPATH   := src
 SRC     := main.c debug_tools.c init.c signal_handling.c repl.c new_token.c \
-           free_token.c new_node.c free_node.c
+           free_token.c new_node.c free_node.c tokenizer.c
 
 OBJ_DIR := _obj
 OBJ     := $(addprefix $(OBJ_DIR)/, $(SRC:%.c=%.o))
diff --git a/include/token.h b/include/token.h
index 38e758f..d7ff9f9 100644
--- a/include/token.h
+++ b/include/token.h
@@ -3,10 +3,10 @@
 /*                                                        :::      ::::::::   */
 /*   token.h                                            :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: dkaiser <dkaiser@student.42heilbronn.de    +#+  +:+       +#+        */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2024/06/27 13:27:18 by dkaiser           #+#    #+#             */
-/*   Updated: 2024/06/28 14:59:19 by dkaiser          ###   ########.fr       */
+/*   Updated: 2024/08/05 13:23:27 by chuhlig          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -45,5 +45,6 @@ t_token						*new_redir_token(int type, t_token *previous,
 								t_token *next);
 
 void						free_token(t_token *token);
+void						tokenizer(char *s, t_token **token_list);
 
-#endif
+#endif
\ No newline at end of file
diff --git a/lib/libft/Makefile b/lib/libft/Makefile
index 3c2fb91..6951c43 100644
--- a/lib/libft/Makefile
+++ b/lib/libft/Makefile
@@ -30,6 +30,7 @@ SRC =		ft_atoi.c \
 			ft_strlen.c \
 			ft_strmapi.c \
 			ft_strncmp.c \
+			ft_strncpy.c \
 			ft_strnstr.c \
 			ft_strrchr.c \
 			ft_strtrim.c \
diff --git a/lib/libft/ft_memset.c b/lib/libft/ft_memset.c
index 58c5e1e..085df1d 100644
--- a/lib/libft/ft_memset.c
+++ b/lib/libft/ft_memset.c
@@ -3,10 +3,10 @@
 /*                                                        :::      ::::::::   */
 /*   ft_memset.c                                        :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: dkaiser <dkaiser@student.42heilbronn.de    +#+  +:+       +#+        */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2024/03/05 09:58:19 by dkaiser           #+#    #+#             */
-/*   Updated: 2024/03/10 13:13:09 by dkaiser          ###   ########.fr       */
+/*   Updated: 2024/07/11 23:52:13 by chuhlig          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -35,3 +35,13 @@ void	*ft_memset(void *b, int c, size_t len)
 /* 	printf("ft_memset: %s\n", ft_memset((char *)str, 'A' + 255, 5)); */
 /* 	printf("memset: %s\n", memset((char *)str, 'A' + 255, 5)); */
 /* } */
+
+// void	*ft_memset(void *b, int c, size_t len)
+// {
+// 	void	*savearg;
+
+// 	savearg = b;
+// 	while (len--)
+// 		*(unsigned char *)b++ = (unsigned char)c;
+// 	return (savearg);
+// }
\ No newline at end of file
diff --git a/lib/libft/ft_strncpy.c b/lib/libft/ft_strncpy.c
new file mode 100644
index 0000000..a1a2293
--- /dev/null
+++ b/lib/libft/ft_strncpy.c
@@ -0,0 +1,21 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   ft_strncpy.c                                       :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2024/08/05 13:41:47 by chuhlig           #+#    #+#             */
+/*   Updated: 2024/08/09 15:26:40 by chuhlig          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+char	*ft_strncpy(char *s1, char *s2, int n)
+{
+	int	i;
+
+	i = -1;
+	while (++i < n && s2[i])
+		s1[i] = s2[i];
+	return (s1);
+}
diff --git a/lib/libft/libft.h b/lib/libft/libft.h
index 2de723b..67fbb0f 100644
--- a/lib/libft/libft.h
+++ b/lib/libft/libft.h
@@ -3,10 +3,10 @@
 /*                                                        :::      ::::::::   */
 /*   libft.h                                            :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: dkaiser <dkaiser@student.42heilbronn.de    +#+  +:+       +#+        */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2024/03/10 16:37:54 by dkaiser           #+#    #+#             */
-/*   Updated: 2024/06/24 16:44:44 by dkaiser          ###   ########.fr       */
+/*   Updated: 2024/08/05 13:55:13 by chuhlig          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -57,6 +57,7 @@ void				ft_putchar_fd(char c, int fd);
 void				ft_putstr_fd(char *s, int fd);
 void				ft_putendl_fd(char *s, int fd);
 void				ft_putnbr_fd(int n, int fd);
+char				*ft_strncpy(char *s1, char *s2, int n);
 
 typedef struct s_list
 {
diff --git a/src/main.c b/src/main.c
index 016ede6..8523b9e 100644
--- a/src/main.c
+++ b/src/main.c
@@ -3,14 +3,14 @@
 /*                                                        :::      ::::::::   */
 /*   main.c                                             :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: dkaiser <dkaiser@student.42heilbronn.de    +#+  +:+       +#+        */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2024/06/22 17:14:03 by dkaiser           #+#    #+#             */
-/*   Updated: 2024/06/25 13:58:24 by dkaiser          ###   ########.fr       */
+/*   Updated: 2024/07/18 16:44:14 by chuhlig          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
-#include "minishell.h"
+#include "../include/minishell.h"
 
 int	main(void)
 {
diff --git a/src/repl.c b/src/repl.c
index 85d227f..1fd7be7 100644
--- a/src/repl.c
+++ b/src/repl.c
@@ -3,18 +3,22 @@
 /*                                                        :::      ::::::::   */
 /*   repl.c                                             :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: dkaiser <dkaiser@student.42heilbronn.de    +#+  +:+       +#+        */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2024/06/24 16:07:04 by dkaiser           #+#    #+#             */
-/*   Updated: 2024/06/25 15:03:00 by dkaiser          ###   ########.fr       */
+/*   Updated: 2024/08/09 15:27:11 by chuhlig          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
-#include "minishell.h"
+#include "../include/minishell.h"
+#include "token.h"
 
 void	repl(const char *prompt)
 {
 	char	*input;
+	t_token	*token_list;
+	t_token	*current;
+	t_token	*next;
 
 	while (1)
 	{
@@ -22,6 +26,15 @@ void	repl(const char *prompt)
 		if (input == NULL)
 			return ;
 		add_history(input);
+		token_list = NULL;
+		tokenizer(input, &token_list);
+		current = token_list;
+		while (current != NULL)
+		{
+			next = current->next;
+			free_token(current);
+			current = next;
+		}
 		free(input);
 	}
 }
diff --git a/src/tokenizer.c b/src/tokenizer.c
new file mode 100644
index 0000000..34685ac
--- /dev/null
+++ b/src/tokenizer.c
@@ -0,0 +1,113 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   tokenizer.c                                        :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: chuhlig <chuhlig@student.42.fr>            +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2024/06/28 20:55:50 by chuhlig           #+#    #+#             */
+/*   Updated: 2024/08/09 15:40:00 by chuhlig          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+#include "minishell.h"
+#include "token.h"
+
+void	print_token(t_token *token)
+{
+	if (DEBUG)
+	{
+		if (token->type == STRING_TOKEN)
+		{
+			printf("STRING_TOKEN: %s\n", token->content.string);
+		}
+		else if (token->type == REDIR_TOKEN)
+		{
+			printf("REDIR_TOKEN: %d\n", token->content.redir_type);
+		}
+		else if (token->type == PIPE_TOKEN)
+		{
+			printf("PIPE_TOKEN\n");
+		}
+		else if (token->type == NEWLINE_TOKEN)
+		{
+			printf("NEWLINE_TOKEN\n");
+		}
+	}
+}
+
+void	conditional_print(char *string, int start_of_string, int i,
+	t_token **token_list)
+{
+	char	*line;
+	int		len;
+
+	len = i - start_of_string + 1;
+	if (len > 0)
+	{
+		line = (char *)malloc(len + 1);
+		if (!line)
+		{
+			exit(EXIT_FAILURE);
+		}
+		ft_strncpy(line, string + start_of_string, len);
+		line[len] = '\0';
+		while (*line == ' ' || *line == '\t')
+			line++;
+		if (*line != '\0')
+		{
+			*token_list = new_str_token(line, *token_list, NULL);
+			print_token(*token_list);
+		}
+	}
+}
+
+void	handle_special_chars(char *s, int *i, int *start, t_token **token_list)
+{
+	conditional_print(s, *start, *i - 1, token_list);
+	if (s[*i] == '<' && s[*i + 1] == '<')
+		*token_list = new_redir_token(INPUT_LIMITER, *token_list, NULL);
+	else if (s[*i] == '>' && s[*i + 1] == '>')
+		*token_list = new_redir_token(OUTPUT_APPEND, *token_list, NULL);
+	else if (s[*i] == '<')
+		*token_list = new_redir_token(INPUT_FILE, *token_list, NULL);
+	else if (s[*i] == '>')
+		*token_list = new_redir_token(OUTPUT_OVERRIDE, *token_list, NULL);
+	else if (s[*i] == '|')
+		*token_list = new_token(PIPE_TOKEN, *token_list, NULL);
+	else if (s[*i] == '\n')
+		*token_list = new_token(NEWLINE_TOKEN, *token_list, NULL);
+	print_token(*token_list);
+	if (s[*i + 1] == '<' || s[*i + 1] == '>')
+		(*i)++;
+	*start = *i + 1;
+}
+
+void	tokenizer(char *s, t_token **token_list)
+{
+	char	quote_check;
+	int		pos;
+	int		i;
+	int		f;
+
+	pos = 0;
+	i = -1;
+	f = 0;
+	while (s[++i])
+	{
+		if (!f && ft_strchr("|<>\n", s[i]))
+			handle_special_chars(s, &i, &pos, token_list);
+		else if (f && s[i] == quote_check)
+			f = 0;
+		else if (!f && ft_strchr("\'\"", s[i]))
+		{
+			f = 1;
+			quote_check = s[i];
+		}
+		if ((!f && (s[i] == ' ' || s[i] == '\t')) || i == ft_strlen(s) - 1)
+		{
+			conditional_print(s, pos, i, token_list);
+			pos = i + 1;
+		}
+	}
+}