parser as a pushdown automata, test

42sh/Makefile

@@ -170,6 +170,7 @@ main/shell_script.c\
 main/sig_handler.c\
 parser/parse.c\
 parser/ft_parse.c\
+parser/produce_prim_sym.c\
 parser/get_instruction.c\
 parser/get_sub_instruction.c\
 parser/parse_dgreat.c\
@@ -186,7 +187,8 @@ parser/parse_while.c\
 parser/parse_if.c\
 parser/parse_elif.c\
 parser/parse_else.c\
-parser/parse_word.c
+parser/parse_word.c\
+parser/read_stack.c
 
 SRCS		=	$(addprefix $(SRC_DIR), $(SRC_BASE))
 OBJS		=	$(addprefix $(OBJ_DIR), $(SRC_BASE:.c=.o))

42sh/includes/lexer.h

@@ -6,7 +6,7 @@
 /*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/12/01 12:15:50 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/06 20:35:43 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:26:01 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -24,8 +24,27 @@ typedef struct s_nest	t_nest;
 
 typedef long long		t_type;
 
-# define TK_LESS		(1 << 0)
+/*
-# define TK_GREAT		(1 << 1)
+ * Token need :
+ * DLESSDASH
+ * CLOBBER
+ * LESSGREAT
+ * IO_NUMBER
+ * NAME
+ * DSEMI
+ * ASSIGNEMENT_WORD
+ * Lbrace
+ * Rbrace
+ * Bang
+ * Case
+ * Esac
+ * for
+ * in
+ *
+*/
+
+# define TK_LESS		(1 << 0)	//transparent
+# define TK_GREAT		(1 << 1)	//transparent
 # define TK_DLESS		(1 << 2)
 # define TK_DGREAT		(1 << 3)
 # define TK_LESSAND		(1 << 4)
@@ -83,7 +102,6 @@ enum	e_lexstate
 	IF,
 	THEN,
 	FI,
-	LIST,
 	COMMENT,
 };
 

42sh/includes/parser.h

@@ -6,7 +6,7 @@
 /*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/12/01 12:15:54 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/06 20:35:28 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:58:13 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -15,8 +15,87 @@
 
 # include "minishell.h"
 
-#define INSTRUCTION			(TK_WHILE | TK_IF | TK_ELIF | TK_NEWLINE | TK_SEMI\
+/*
-							| TK_ELSE | TK_UNTIL)
+ * Parse POSIX grammar
+ *
+*/
+
+typedef unsigned long long int 		t_sym;
+
+#define	PROGRAM 			(1UL << 0)
+#define COMPLETE_COMMANDS	(1UL << 1)
+#define COMPLETE_COMMAND	(1UL << 2)
+#define LIST				(1UL << 3)
+#define AND_OR				(1UL << 4)
+#define PIPELINE			(1UL << 5)
+#define PIPE_SEQUENCE		(1UL << 6)
+#define COMMAND				(1UL << 7)
+#define COMPOUND_COMMAND	(1UL << 8)
+#define SUBSHELL			(1UL << 9)
+#define COMPOUND_LIST		(1UL << 10)
+#define TERM				(1UL << 11)
+#define FOR_CLAUSE			(1UL << 12)
+#define NAME				(1UL << 13)
+#define IN					(1UL << 14)
+#define WORDLIST			(1UL << 15)
+#define CASE_CLAUSE			(1UL << 16)
+#define CASE_LIST_NS		(1UL << 17)
+#define CASE_LIST			(1UL << 18)
+#define CASE_ITEM_NS		(1UL << 19)
+#define PATTERN				(1UL << 20)
+#define IF_CLAUSE			(1UL << 21)
+#define ELSE_PART			(1UL << 22)
+#define WHILE_CLAUSE		(1UL << 23)
+#define UNTIL_CLAUSE		(1UL << 24)
+#define FUNCTION_DEFINITION	(1UL << 25)
+#define FUNCTION_BODY		(1UL << 26)
+#define FNAME				(1UL << 27)
+#define BRACE_GROUP			(1UL << 28)
+#define DO_GROUP			(1UL << 29)
+#define SIMPLE_COMMAND		(1UL << 30)
+#define CMD_NAME			(1UL << 31)
+#define CMD_WORD			(1UL << 32)
+#define CMD_PREFIX			(1UL << 33)
+#define CMD_SUFFIX			(1UL << 34)
+#define REDIRECT_LIST		(1UL << 35)
+#define IO_REDIRECT			(1UL << 36)
+#define IO_FILE				(1UL << 37)
+#define FILENAME			(1UL << 38)
+#define IO_HERE				(1UL << 39)
+#define HERE_END			(1UL << 40)
+#define NEWLINE_LIST		(1UL << 41)
+#define LINEBREAK			(1UL << 42)
+#define SEPARATOR_OP		(1UL << 43)
+#define SEPARATOR			(1UL << 44)
+#define SEQUENTIAL_SEP		(1UL << 45)
+
+#define SYM_DLESS			(1UL << 46)
+#define SYM_DGREAT			(1UL << 47)
+#define SYM_GREATAND		(1UL << 48)
+#define SYM_GREAT			(1UL << 49)
+#define SYM_LESSAND			(1UL << 50)
+#define SYM_LESS			(1UL << 51)
+
+
+int		ft_parse(t_btree **ast, t_list **token);
+int		produce_prim_sym(t_sym *new_sym, t_list **lst);
+
+int		ft_read_stack(t_sym stack[], int size);
+char	*read_state(t_sym current);
+
+enum	e_parstate
+{
+	UNDEFINED,
+	ERROR,
+	SUCCESS,
+};
+
+typedef enum e_parstate t_parstate;
+
+/*
+ * Build AST
+ * 
+*/
 
 typedef struct s_parser		t_parser;
 typedef struct	s_ld		t_ld;
@@ -64,7 +143,6 @@ extern t_parser		g_parser[];
 
 int		parse(t_btree **ast, t_list **token);
 
-int		ft_parse(t_btree **ast, t_list **token);
 int		get_instruction(t_list **lst);
 int		get_sub_instruction(t_btree **ast, t_list **start, t_list **lst);
 

42sh/sample/stack.sh

@@ -0,0 +1 @@
+ls > file1

42sh/src/lexer/ft_lexer.c

@@ -6,7 +6,7 @@
 /*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2017/02/02 15:30:59 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/06 20:27:07 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:36:59 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 

42sh/src/lexer/ft_tokenize.c

@@ -6,7 +6,7 @@
 /*   By: jhalford <marvin@42.fr>                    +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/11/10 13:37:11 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/03 20:04:33 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:37:13 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 

42sh/src/lexer/lexer_less.c

@@ -6,7 +6,7 @@
 /*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/12/03 12:06:53 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/03 14:50:00 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:00:13 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 

42sh/src/lexer/lexer_word.c

@@ -6,7 +6,7 @@
 /*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/12/03 12:07:11 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/03 19:52:48 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:37:21 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 

42sh/src/main/shell_script.c

@@ -6,7 +6,7 @@
 /*   By: ariard <ariard@student.42.fr>              +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2017/01/22 23:06:34 by ariard            #+#    #+#             */
-/*   Updated: 2017/02/06 22:05:34 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:26:55 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
@@ -29,16 +29,12 @@ int		shell_script()
 			return (1);
 		DG("after post_tokenize");
 		token_print(token);
-
+		if (ft_parse(&ast, &token))
-		while (token)
+			return (1);
-		{
+//		btree_print(STDBUG, ast, &ft_putast);
-			if (parse(&ast, &token))
+//		if (ft_exec(&ast))
-				return (1);
+//			return (1);
-			btree_print(STDBUG, ast, &ft_putast);
+//		ast = NULL;
-			if (ft_exec(&ast))
-				return (1);
-			ast = NULL;	
-		}
 		script->size = 0;
 		get_script_content(script);	
 	}

42sh/src/parser/ft_parse.c

@@ -3,56 +3,44 @@
 /*                                                        :::      ::::::::   */
 /*   ft_parse.c                                         :+:      :+:    :+:   */
 /*                                                    +:+ +:+         +:+     */
-/*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
+/*   By: ariard <ariard@student.42.fr>              +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
-/*   Created: 2016/11/30 17:14:58 by jhalford          #+#    #+#             */
+/*   Created: 2017/02/09 14:30:22 by ariard            #+#    #+#             */
-/*   Updated: 2017/02/06 20:37:26 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 16:02:49 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 
 #include "parser.h"
 
-t_parser	g_parser[] =
+int			ft_parse(t_btree **ast, t_list **token)
 {
-	{INSTRUCTION, &get_sub_instruction},
+	t_sym			*new_sym;
-	{TK_AND_IF | TK_OR_IF, &parse_separator},
+	t_parstate		state;
-	{TK_AMP, &parse_separator},
-	{TK_PIPE, &parse_separator},
-	{TK_LESS, &parse_less},
-	{TK_GREAT, &parse_great},
-	{TK_DLESS, &parse_dless},
-	{TK_DGREAT, &parse_dgreat},
-	{TK_LESSAND, &parse_lessand},
-	{TK_GREATAND, &parse_greatand},
-	{TK_SUBSHELL, &parse_subshell},
-	{TK_WORD, &parse_word},
-	{0, 0},
-};
 
-int		ft_parse(t_btree **ast, t_list **start)
+	(void)ast;
-{
+	state = UNDEFINED;
-	t_list		*lst;
+	new_sym = ft_memalloc(sizeof(t_sym)); 
-	t_astnode	item;
+	while (*token)
-	int			i;
-
-	i = 0;
-	if (!*start)
-		return (0);
-	if (!*ast)
 	{
-		*ast = btree_create_node(&item, sizeof(item));
+		produce_prim_sym(new_sym, token);
-		((t_astnode *)(*ast)->item)->data.token = NULL;
+		DG("new sym : %s", read_state(*new_sym));
-		((t_astnode *)(*ast)->item)->type = 0;
+/*		if (eval_sym(head_stack, new_sym))
-	}
+			state = ERROR;
-	while (g_parser[i].type)
+		else
-	{
-		if ((lst = ft_lst_find(*start, &g_parser[i].type, &token_cmp_type)))
 		{
-			if (g_parser[i].f)
+			aggregate_sym(head_stack, new_sym, struct_sym);
-				(*g_parser[i].f)(ast, start, &lst);
+			if (struct_sym)
-			return (0);
+				pop(struct_sym.sym);
+			else if (!same_sym(new_sym, head_stack))
+				push(new_sym);
 		}
-		i++;
+		build_tree(token, ast);
-	}
+		if (head_stack == PROGRAM)
+			state = PROGRAM;
+		if (state == ERROR)
+			return (error_syntax(token));
+		if (state == PROGRAM)
+			return (0);
+*/	}
 	return (0);
 }

42sh/src/parser/old_parse.c

@@ -0,0 +1,58 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   ft_parse.c                                         :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: jhalford <jack@crans.org>                  +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2016/11/30 17:14:58 by jhalford          #+#    #+#             */
+/*   Updated: 2017/02/09 15:24:15 by ariard           ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+#include "parser.h"
+
+t_parser	g_parser[] =
+{
+	{INSTRUCTION, &get_sub_instruction},
+	{TK_AND_IF | TK_OR_IF, &parse_separator},
+	{TK_AMP, &parse_separator},
+	{TK_PIPE, &parse_separator},
+	{TK_LESS, &parse_less},
+	{TK_GREAT, &parse_great},
+	{TK_DLESS, &parse_dless},
+	{TK_DGREAT, &parse_dgreat},
+	{TK_LESSAND, &parse_lessand},
+	{TK_GREATAND, &parse_greatand},
+	{TK_SUBSHELL, &parse_subshell},
+	{TK_WORD, &parse_word},
+	{0, 0},
+};
+
+int		ft_parse(t_btree **ast, t_list **start)
+{
+	t_list		*lst;
+	t_astnode	item;
+	int			i;
+
+	i = 0;
+	if (!*start)
+		return (0);
+	if (!*ast)
+	{
+		*ast = btree_create_node(&item, sizeof(item));
+		((t_astnode *)(*ast)->item)->data.token = NULL;
+		((t_astnode *)(*ast)->item)->type = 0;
+	}
+	while (g_parser[i].type)
+	{
+		if ((lst = ft_lst_find(*start, &g_parser[i].type, &token_cmp_type)))
+		{
+			if (g_parser[i].f)
+				(*g_parser[i].f)(ast, start, &lst);
+			return (0);
+		}
+		i++;
+	}
+	return (0);
+}

42sh/src/parser/parse_word.c

@@ -6,7 +6,7 @@
 /*   By: jhalford <marvin@42.fr>                    +#+  +:+       +#+        */
 /*                                                +#+#+#+#+#+   +#+           */
 /*   Created: 2016/11/14 12:49:45 by jhalford          #+#    #+#             */
-/*   Updated: 2017/02/03 19:37:29 by ariard           ###   ########.fr       */
+/*   Updated: 2017/02/09 15:24:49 by ariard           ###   ########.fr       */
 /*                                                                            */
 /* ************************************************************************** */
 

42sh/src/parser/produce_prim_sym.c

@@ -0,0 +1,38 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   produce_prim_sym.c                                 :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: ariard <ariard@student.42.fr>              +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2017/02/09 14:55:10 by ariard            #+#    #+#             */
+/*   Updated: 2017/02/09 16:05:27 by ariard           ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+#include "parser.h"
+
+int		produce_prim_sym(t_sym *new_sym, t_list **lst)
+{
+	t_token		*token;
+
+	token = (*lst)->content;
+	if (token->type == TK_N_WORD)
+		*new_sym = CMD_NAME;
+	else if (token->type == TK_NEWLINE)
+		*new_sym = NEWLINE_LIST;
+	else if (token->type == TK_DLESS)
+		*new_sym = SYM_DLESS;
+	else if (token->type == TK_DGREAT)
+		*new_sym = SYM_DGREAT;
+	else if (token->type == TK_GREATAND)
+		*new_sym = SYM_GREATAND;
+	else if (token->type == TK_GREAT)
+	   *new_sym = SYM_GREAT;
+	else if (token->type == TK_LESSAND)
+		*new_sym = SYM_LESSAND;
+	else if (token->type == TK_LESS)
+		*new_sym = SYM_LESS;
+	ft_lst_delif(lst, (*lst)->content, &ft_addrcmp, &token_free);
+	return (0);
+}

42sh/src/parser/read_stack.c

@@ -0,0 +1,41 @@
+/* ************************************************************************** */
+/*                                                                            */
+/*                                                        :::      ::::::::   */
+/*   read_stack.c                                       :+:      :+:    :+:   */
+/*                                                    +:+ +:+         +:+     */
+/*   By: ariard <ariard@student.42.fr>              +#+  +:+       +#+        */
+/*                                                +#+#+#+#+#+   +#+           */
+/*   Created: 2017/02/09 15:32:10 by ariard            #+#    #+#             */
+/*   Updated: 2017/02/09 16:02:30 by ariard           ###   ########.fr       */
+/*                                                                            */
+/* ************************************************************************** */
+
+#include "parser.h"
+
+char			*read_state(t_sym current)
+{
+	if (current == CMD_NAME)
+		return ("CMD_NAME");
+	if (current == NEWLINE_LIST)
+		return ("NEWLINE_LIST");
+	if (current == SYM_DLESS)
+		return ("DLESS");
+	if (current == SYM_DGREAT)
+		return ("DGREAT");
+	if (current == SYM_GREATAND)
+		return ("GREATAND");
+	if (current == SYM_GREAT)
+		return ("GREAT");
+	if (current == SYM_LESSAND)
+		return ("LESSAND");
+	if (current == SYM_LESS)
+		return ("LESS");
+	return (NULL);
+}
+
+int			ft_read_stack(t_sym	stack[], int size)
+{
+	while (stack[size])
+		ft_putstr(read_state(stack[size--]));
+	return (0);
+}