proof of concept on lexing

Makefile

@@ -1,12 +1,12 @@
 # **************************************************************************** #
 #                                                                              #
-#                                                   :::  o_   ::::::    :::    #
+#                                                         :::      ::::::::    #
-#    Makefile                                      :+:  /    :+::+:    :+:     #
+#    Makefile                                           :+:      :+:    :+:    #
-#                                                 +:+   >   +:++:+    +:+      #
+#                                                     +:+ +:+         +:+      #
-#    By: whaffman <whaffman@student.codam.nl>    +#+  +:+  +#++#++:++#++       #
+#    By: qmennen <qmennen@student.codam.nl>         +#+  +:+       +#+         #
-#                                               +#+ +#+#+ +#++#+    +#+ \o/    #
+#                                                 +#+#+#+#+#+   +#+            #
-#    Created: 2024/10/15 11:48:46 by whaffman   #+#+# #+#+# #+#    #+#   |     #
+#    Created: 2024/10/15 11:48:46 by whaffman          #+#    #+#              #
-#    Updated: 2024/11/07 15:28:08 by whaffman   ###   ###  ###    ###   / \    #
+#    Updated: 2025/02/04 16:57:05 by qmennen          ###   ########.fr        #
 #                                                                              #
 # **************************************************************************** #
 
@@ -22,7 +22,7 @@ LIBFT = $(LIBFT_PATH)/libft.a
 
 OBJ_PATH = obj
 
-VPATH = src
+VPATH = src:src/tokenizer
 SOURCES = $(shell basename -a $(shell find $(SRC_PATH) -type f -name "*.c"))
 
 OBJECTS = $(addprefix $(OBJ_PATH)/, $(SOURCES:.c=.o))
@@ -32,7 +32,7 @@ CC = cc
 RM = rm -rf
 
 INCLUDES = -I./$(INC_PATH) -I./$(LIBFT_INC_PATH) 
-CFLAGS = -Wall -Wextra -Werror -MMD
+CFLAGS = -Wall -Wextra -Werror -fsanitize=address,undefined -MMD
 
 UNAME_S	:= $(shell uname -s)
 ifeq ($(UNAME_S),Linux)

inc/minishell.h

@@ -147,6 +147,8 @@
 #ifndef MINISHELL_H
 # define MINISHELL_H
 
+# define TRUE 1
+
 # include <stdio.h>
 # include <stdlib.h>
 # include <unistd.h>
@@ -161,5 +163,48 @@
 # include <sys/ioctl.h>
 # include <termios.h>
 # include <term.h>
+# include "libft.h"
 
-#endif // MINISHELL_H
+typedef enum
+{
+	T_WORD,
+	T_PIPE,
+	T_REDIRECT_IN,
+	T_REDIRECT_OUT,
+	T_APPEND_OUT,
+	T_EOF,
+	T_ERROR
+}	TokenType;
+
+typedef struct s_token
+{
+	TokenType	type;
+	char		*value;
+	int			position;
+}	t_token;
+
+typedef struct s_lexer
+{
+	char	*input;
+	int		pos;
+	int		n_pos;
+	char	current_char;
+}	t_lexer;
+
+int		ft_isspace(const char c);
+
+/**
+ * Lexer
+ */
+t_lexer	*ft_lexer_new(const char *input);
+void	ft_lexer_free(t_lexer *lexer);
+void	ft_lexer_readchar(t_lexer *lexer);
+char	*ft_lexer_readword(t_lexer *lexer);
+/**
+ * Token
+ */
+t_token	*ft_token_next(t_lexer *lexer);
+t_token	*ft_token_new(TokenType type, char *c, int pos);
+void	ft_token_free(t_token *token);
+void	ft_clear_tokenlist(void *content);
+#endif

src/prompt.c

@@ -4,6 +4,7 @@
 #include <unistd.h>
 #include <string.h>
 #include "libft.h"
+#include "minishell.h"
 
 
 // /**
@@ -69,20 +70,43 @@ void print_prompt(void)
     free(cwd);
 }
 
+void	print_list(void *content)
+{
+	t_token *token;
+	token = (t_token *)content;
+	ft_printf("%s\n", token->value);
+}
+
 int main(int argc, char **argv, char **envp)
 {
     (void)argc;
-    (void)argv;
+	(void)envp;
-    char **env;
+    // char **env;
-    t_enviroment *enviroment = NULL;
+    // t_enviroment	*enviroment = NULL;
+	t_lexer			*lexer;
+	t_token			*token;
+	t_list			*list;
 
-    while (*envp != NULL)
+    // while (*envp != NULL)
-    {
+    // {
-        env = ft_split(*envp, '=');
+    //     env = ft_split(*envp, '=');
-        add_enviroment(&enviroment, env[0], env[1]);
+    //     add_enviroment(&enviroment, env[0], env[1]);
-        envp++;
+    //     envp++;
-    }
+    // }
 
-    print_enviroment(enviroment);
+	lexer = ft_lexer_new(argv[1]);
+	list = NULL;	
+	while (TRUE)
+	{
+		token = ft_token_next(lexer);
+		if (token->type == T_EOF)
+			break;
+		ft_lstadd_back(&list, ft_lstnew(token));
+	}
+	ft_lstiter(list, print_list);
+	ft_lstclear(&list, ft_clear_tokenlist);
+	ft_token_free(token);
+	ft_lexer_free(lexer);
+    // print_enviroment(enviroment);
     return 0;
 }

src/tokenizer/lexer.c

@@ -0,0 +1,84 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   lexer.c                                            :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: qmennen <qmennen@student.codam.nl>         +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2025/02/04 18:04:07 by qmennen           #+#    #+#             */
+/*   Updated: 2025/02/04 20:11:48 by qmennen          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+#include "minishell.h"
+
+t_lexer *ft_lexer_new(const char *input)
+{
+	t_lexer *lexer;
+
+	lexer = malloc(sizeof(t_lexer));
+	if (!lexer)
+	{
+		perror("failed assigning lexer memory");
+		exit(EXIT_FAILURE);
+	}
+	lexer->input = ft_strdup(input);
+	lexer->pos = 0;
+	lexer->n_pos = 1;
+	lexer->current_char = '\0';
+	if (ft_strlen(input) > 0)
+		lexer->current_char = *input;
+	return (lexer);
+}
+
+void ft_lexer_readchar(t_lexer *lexer)
+{
+	if ((size_t)lexer->n_pos > ft_strlen(lexer->input))
+	{
+		lexer->current_char = '\0';
+		return;
+	}
+	lexer->current_char = lexer->input[lexer->n_pos];
+	lexer->pos = lexer->n_pos;
+	lexer->n_pos++;
+}
+
+char *ft_lexer_readword(t_lexer *lexer)
+{
+	int start;
+	int len;
+	char quote;
+	char *word;
+
+	start = lexer->pos;
+	if (lexer->current_char == '"' || lexer->current_char == '\'')
+	{
+		quote = lexer->current_char;
+		ft_lexer_readchar(lexer);
+		start = lexer->pos;
+		while (lexer->current_char != '\0' && lexer->current_char != quote)
+			ft_lexer_readchar(lexer);
+		if (lexer->current_char == quote)
+		{
+			len = lexer->pos - start;
+			word = malloc(sizeof(char) * len);
+			ft_strlcpy(word, lexer->input + start, len + 1);
+			ft_lexer_readchar(lexer);
+			return (word);
+		}
+	}
+	start = lexer->pos;
+	while (ft_isprint(lexer->current_char) && lexer->current_char != '<' && lexer->current_char != '>' && lexer->current_char != '|' && lexer->current_char != '\0')
+		ft_lexer_readchar(lexer);
+	len = lexer->pos - start;
+	word = malloc(sizeof(char) * len);
+	ft_strlcpy(word, lexer->input + start, len + 1);
+	return (word);
+}
+
+void ft_lexer_free(t_lexer *lexer)
+{
+	if (lexer->input)
+		free(lexer->input);
+	free(lexer);
+}

src/tokenizer/tokenizer.c

@@ -0,0 +1,65 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   tokenizer.c                                        :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: qmennen <qmennen@student.codam.nl>         +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2025/02/04 16:07:58 by qmennen           #+#    #+#             */
+/*   Updated: 2025/02/04 20:14:50 by qmennen          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+# include "minishell.h"
+
+t_token	*ft_token_next(t_lexer *lexer)
+{
+	t_token	*token;
+	char	*word;
+	int		current_pos;
+
+	token = NULL;
+	while (ft_isspace(lexer->current_char))
+		ft_lexer_readchar(lexer);
+	current_pos = lexer->pos;
+	switch(lexer->current_char)
+	{
+		case '\0':
+            token = ft_token_new(T_EOF, NULL, current_pos);
+            break;
+        case '|':
+            token = ft_token_new(T_PIPE, "|", current_pos);
+            ft_lexer_readchar(lexer);
+            break;
+		case '<':
+			token = ft_token_new(T_REDIRECT_IN, "<", current_pos);
+            ft_lexer_readchar(lexer);
+            break;
+		case '>':
+            ft_lexer_readchar(lexer);
+			if (lexer->current_char == '>')
+			{
+				token = ft_token_new(T_APPEND_OUT, ">>", current_pos);
+				ft_lexer_readchar(lexer);
+			}
+			else
+				token = ft_token_new(T_REDIRECT_OUT, ">", current_pos);
+            break;
+		default:
+			if (ft_isprint(lexer->current_char))
+			{
+				word = ft_lexer_readword(lexer);
+				token = ft_token_new(T_WORD, word, current_pos);
+				free(word);
+			}
+	}
+	return (token);
+}
+
+void	ft_clear_tokenlist(void *content)
+{
+	t_token	*token;
+
+	token = (t_token *)content;
+	ft_token_free(token);
+}

src/tokenizer/tokens.c

@@ -0,0 +1,39 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   tokens.c                                           :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: qmennen <qmennen@student.codam.nl>         +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2025/02/04 18:02:56 by qmennen           #+#    #+#             */
+/*   Updated: 2025/02/04 18:22:11 by qmennen          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+# include "minishell.h"
+
+t_token	*ft_token_new(TokenType type, char *c, int pos)
+{
+	t_token	*token;
+
+	token = malloc(sizeof(t_token));
+	if (!token)
+	{
+		perror("failed assigning token memory");
+		exit(EXIT_FAILURE);
+	}
+	token->type = type;
+	token->position = pos;
+	if (c)
+		token->value = ft_strdup(c);
+	else
+		token->value = NULL;
+	return (token);
+}
+
+void	ft_token_free(t_token *token)
+{
+	if (token->value)
+		free(token->value);
+	free(token);
+}

src/util.c

@@ -0,0 +1,18 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   util.c                                             :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: qmennen <qmennen@student.codam.nl>         +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2025/02/04 16:31:29 by qmennen           #+#    #+#             */
+/*   Updated: 2025/02/04 16:32:08 by qmennen          ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+# include "minishell.h"
+
+int	ft_isspace(const char c)
+{
+	return (c == ' ' || c == '\t' || c == '\v' || c == '\n' || c == '\v');
+}