clean-up parsing a bit

inc/minishell.h

@@ -207,4 +207,5 @@ 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);
+t_token	*ft_parse_token(t_lexer *lexer);
 #endif

src/tokenizer/lexer.c

@@ -6,7 +6,7 @@
 /*   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       */
+/*   Updated: 2025/02/04 20:40:27 by qmennen          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -43,31 +43,38 @@ void ft_lexer_readchar(t_lexer *lexer)
 	lexer->n_pos++;
 }
 
-char *ft_lexer_readword(t_lexer *lexer)
+static char	*ft_parse_quotes(t_lexer *lexer)
+{
+	int		start;
+	int		len;
+	char	qc;
+	char	*word;
+
+	qc = lexer->current_char;
+	word = NULL;
+	ft_lexer_readchar(lexer);
+	start = lexer->pos;
+	while (lexer->current_char != '\0' && lexer->current_char != qc)
+		ft_lexer_readchar(lexer);
+	len = lexer->pos - start;
+	word = malloc(sizeof(char) * len);
+	ft_strlcpy(word, lexer->input + start, len + 1);
+	if (lexer->current_char == qc)
+		ft_lexer_readchar(lexer);
+	return (word);
+}
+
+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;
+		return (ft_parse_quotes(lexer));
-		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;

src/tokenizer/tokenizer.c

@@ -6,59 +6,40 @@
 /*   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       */
+/*   Updated: 2025/02/04 20:28:00 by qmennen          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
-# include "minishell.h"
+#include "minishell.h"
 
-t_token	*ft_token_next(t_lexer *lexer)
+t_token *ft_token_next(t_lexer *lexer)
 {
-	t_token	*token;
+	t_token *token;
-	char	*word;
+	char *word;
-	int		current_pos;
+	int current_pos;
 
 	token = NULL;
 	while (ft_isspace(lexer->current_char))
 		ft_lexer_readchar(lexer);
 	current_pos = lexer->pos;
-	switch(lexer->current_char)
+	if (lexer->current_char == '\0')
+		token = ft_token_new(T_EOF, NULL, current_pos);
+	else if (lexer->current_char == '<' || lexer->current_char == '>' || lexer->current_char == '|')
+		token = ft_parse_token(lexer);
+	else if (ft_isprint(lexer->current_char))
 	{
-		case '\0':
+		word = ft_lexer_readword(lexer);
-            token = ft_token_new(T_EOF, NULL, current_pos);
+		token = ft_token_new(T_WORD, word, current_pos);
-            break;
+		free(word);
-        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);
-			}
 	}
+	else
+		token = ft_token_new(T_ERROR, NULL, current_pos);
 	return (token);
 }
 
-void	ft_clear_tokenlist(void *content)
+void ft_clear_tokenlist(void *content)
 {
-	t_token	*token;
+	t_token *token;
 
 	token = (t_token *)content;
 	ft_token_free(token);

src/tokenizer/tokens.c

@@ -6,12 +6,39 @@
 /*   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       */
+/*   Updated: 2025/02/04 20:29:06 by qmennen          ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
 # include "minishell.h"
 
+t_token	*ft_parse_token(t_lexer *lexer)
+{
+	t_token	*token;
+	
+	token = NULL;
+	if (lexer->current_char == '|')
+	{
+		token = ft_token_new(T_PIPE, "|", lexer->pos);
+	}
+	else if (lexer->current_char == '<')
+	{
+		token = ft_token_new(T_REDIRECT_IN, "<", lexer->pos);
+	}
+	else if (lexer->current_char == '>' && lexer->input[lexer->pos + 1] == '>')
+	{	
+		token = ft_token_new(T_APPEND_OUT, ">>", lexer->pos);	
+		ft_lexer_readchar(lexer);
+	}
+	else if (lexer->current_char == '>')
+	{
+		token = ft_token_new(T_REDIRECT_OUT, ">", lexer->pos);	
+	}
+	
+	ft_lexer_readchar(lexer);
+	return (token);
+}
+
 t_token	*ft_token_new(TokenType type, char *c, int pos)
 {
 	t_token	*token;