*** empty log message ***

ase/awk/awk.c

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.c,v 1.5 2005-12-29 12:04:51 bacon Exp $
+ * $Id: awk.c,v 1.6 2006-01-09 12:51:46 bacon Exp $
  */
 
 #include <xp/awk/awk.h>
@@ -20,6 +20,7 @@ xp_awk_t* xp_awk_open (xp_awk_t* awk)
 		return XP_NULL;
 	}
 
+	awk->tree = XP_NULL;
 	awk->errnum = XP_AWK_ENOERR;
 
 	awk->src_func = XP_NULL;
@@ -38,6 +39,10 @@ xp_awk_t* xp_awk_open (xp_awk_t* awk)
 
 int xp_awk_close (xp_awk_t* awk)
 {
+	/* TODO: collapse the tree */
+
+	if (awk->tree != XP_NULL) __collapse_tree (awk->tree);
+
 	if (xp_awk_detsrc(awk) == -1) return -1;
 	xp_str_close (&awk->token.name);
 	if (awk->__dynamic) xp_free (awk);

ase/awk/awk.h

@@ -1,5 +1,5 @@
 /* 
- * $Id: awk.h,v 1.6 2005-12-29 12:04:51 bacon Exp $
+ * $Id: awk.h,v 1.7 2006-01-09 12:51:47 bacon Exp $
  */
 
 #ifndef _XP_AWK_AWK_H_
@@ -8,6 +8,7 @@
 #include <xp/types.h>
 #include <xp/macros.h>
 #include <xp/bas/str.h>
+#include <xp/awk/tree.h>
 
 enum
 {
@@ -16,8 +17,11 @@ enum
 	XP_AWK_ESRCOP,
 	XP_AWK_ESRCCL,
 	XP_AWK_ESRCDT, /* error in reading source */
+
 	XP_AWK_ELXCHR, /* lexer came accross an wrong character */
-	XP_AWK_ELXUNG  /* lexer failed to unget a character */
+	XP_AWK_ELXUNG, /* lexer failed to unget a character */
+
+	XP_AWK_ELBRACE /* left brace expected */
 };
 
 /*
@@ -40,6 +44,9 @@ enum
 
 struct xp_awk_t
 {
+	/* parse tree */
+	xp_awk_node_t* tree;
+
 	/* io functions */
 	xp_awk_io_t src_func;
 	xp_awk_io_t inp_func;

ase/awk/parse.c

@@ -1,8 +1,10 @@
 /*
- * $Id: parse.c,v 1.9 2005-12-29 12:04:51 bacon Exp $
+ * $Id: parse.c,v 1.10 2006-01-09 12:51:47 bacon Exp $
  */
 
 #include <xp/awk/awk.h>
+#include <xp/awk/tree.h>
+#include <xp/awk/tree.h>
 #include <xp/bas/memory.h>
 #include <xp/bas/ctype.h>
 #include <xp/bas/string.h>
@@ -26,8 +28,8 @@ enum
 	TOKEN_RPAREN,
 	TOKEN_LBRACE,
 	TOKEN_RBRACE,
-	TOKEN_LBRAKET,
+	TOKEN_LBRACKET,
-	TOKEN_RBRAKET,
+	TOKEN_RBRACKET,
 
 	TOKEN_STRING,
 	TOKEN_REGEX,
@@ -37,15 +39,29 @@ enum
 	TOKEN_END,
 	TOKEN_FUNCTION,
 	TOKEN_IF,
-	TOKEN_DO,
 	TOKEN_WHILE,
 	TOKEN_FOR,
+	TOKEN_DO,
+	TOKEN_BREAK,
 	TOKEN_CONTINUE,
-	TOKEN_BREAK
+	TOKEN_RETURN,
+	TOKEN_EXIT,
+	TOKEN_DELETE,
+	TOKEN_NEXT,
+	TOKEN_NEXTFILE
 };
 
 static int __parse (xp_awk_t* awk);
-static int __parse_program (xp_awk_t* awk);
+
+static xp_awk_node_t* __parse_program (xp_awk_t* awk);
+static xp_awk_node_t* __parse_block (xp_awk_t* awk);
+static xp_awk_node_t* __parse_if (xp_awk_t* awk);
+static xp_awk_node_t* __parse_while (xp_awk_t* awk);
+static xp_awk_node_t* __parse_for (xp_awk_t* awk);
+static xp_awk_node_t* __parse_do (xp_awk_t* awk);
+static xp_awk_node_t* __parse_break (xp_awk_t* awk);
+static xp_awk_node_t* __parse_continue (xp_awk_t* awk);
+
 static int __get_token (xp_awk_t* awk);
 static int __get_char (xp_awk_t* awk);
 static int __unget_char (xp_awk_t* awk, xp_cint_t c);
@@ -65,11 +81,16 @@ static struct __kwent __kwtab[] =
 	{ XP_TEXT("END"),      TOKEN_END },
 	{ XP_TEXT("function"), TOKEN_FUNCTION },
 	{ XP_TEXT("if"),       TOKEN_IF },
-	{ XP_TEXT("do"),       TOKEN_DO },
 	{ XP_TEXT("while"),    TOKEN_WHILE },
 	{ XP_TEXT("for"),      TOKEN_FOR },
-	{ XP_TEXT("continue"), TOKEN_CONTINUE },
+	{ XP_TEXT("do"),       TOKEN_DO },
 	{ XP_TEXT("break"),    TOKEN_BREAK },
+	{ XP_TEXT("continue"), TOKEN_CONTINUE },
+	{ XP_TEXT("return"),   TOKEN_RETURN },
+	{ XP_TEXT("exit"),     TOKEN_EXIT },
+	{ XP_TEXT("delete"),   TOKEN_DELETE },
+	{ XP_TEXT("next"),     TOKEN_NEXT },
+	{ XP_TEXT("nextfile"), TOKEN_NEXTFILE },
 	{ XP_NULL,             0 },
 };
 
@@ -98,20 +119,36 @@ static struct __kwent __kwtab[] =
 #define GET_TOKEN(awk) \
 	do { if (__get_token(awk) == -1) return -1; } while(0)
 
+#define MATCH(awk,token_type) ((awk)->token.type == (token_type))
+#define CONSUME(awk) \
+	do { if (__get_token(awk) == -1) return XP_NULL; } while(0)
+
+#define PANIC(awk,code) do { (awk)->errnum = (code);  return XP_NULL; } while (0);
+
 int xp_awk_parse (xp_awk_t* awk)
 {
+	xp_awk_node_t* node;
+
 	GET_CHAR (awk);
 	GET_TOKEN (awk);
-	return __parse_program (awk);
+
+	node = __parse_program(awk);
+	if (node == XP_NULL) return -1;
+
+	awk->tree = node;
+	return 0;
 }
 
-static int __parse_program (xp_awk_t* awk)
+static xp_awk_node_t* __parse_program (xp_awk_t* awk)
 {
 	/*
 	pattern { action }
 	function name (parameter-list) { statement }
 	*/
 
+	xp_awk_node_t* node;
+
+	/*
 	while (1) {
 		if (awk->token.type == TOKEN_FUNCTION) {
 			if (__parse_function_declaration(awk) == -1) return -1;
@@ -120,16 +157,18 @@ static int __parse_program (xp_awk_t* awk)
 			if (__parse_pattern_action(awk) == -1) return -1;
 		}
 	}
+	*/
 
-	return 0;
+	return node;
 }
 
-static int __parse_function_declaration (xp_awk_t* awk)
+
+static xp_awk_node_t* __parse_function_declaration (xp_awk_t* awk)
 {
-	return -1;
+	return XP_NULL;
 }
 
-static int __parse_pattern_action (xp_awk_t* awk)
+static xp_awk_node_t* __parse_pattern_action (xp_awk_t* awk)
 {
 	/* 
 	BEGIN
@@ -143,12 +182,147 @@ static int __parse_pattern_action (xp_awk_t* awk)
 	pattern, pattern
 	*/
 
-	if (awk->token.type == TOKEN_BEGIN) {
+	xp_awk_node_t* node;
+
+	if (MATCH(awk,TOKEN_BEGIN)) {
+		CONSUME (awk);
 	}
-	else if (awk->token.type == TOKEN_END) {
+	else if (MATCH(awk,TOKEN_END)) {
+		CONSUME (awk);
+	}
+	/* patterns ...
+	 * etc */
+
+	if (!MATCH(awk,TOKEN_LBRACE))  {
+		PANIC (awk, XP_AWK_ELBRACE);
 	}
 
-	return -1;
+	CONSUME (awk);
+	node = __parse_block(awk);
+	if (node == XP_NULL) return XP_NULL;
+
+	return node;
+}
+
+
+/* TODO: what is the best name for the parsing routine for the outermost block? */
+static xp_awk_node_t* __parse_block (xp_awk_t* awk) 
+{
+	xp_awk_node_block_t* node;
+
+	node = (xp_awk_node_block_t*) xp_malloc (xp_sizeof(xp_awk_node_block_t));
+	if (node == XP_NULL) {
+		/* TODO: do some clean-up */
+		PANIC (awk, XP_AWK_ENOMEM);
+	}
+
+	node->type = XP_AWK_NODE_BLOCK;
+	node->sbls = XP_NULL;
+	node->body = XP_NULL;
+
+	while (1) {
+		if (MATCH(awk,TOKEN_RBRACE)) {
+			/* TODO: should finalize a block */
+			CONSUME (awk);
+			break;
+		}
+
+		if (MATCH(awk,TOKEN_LBRACE)) {
+			/* nested block */
+			CONSUME (awk);
+			if (__parse_block(awk) == XP_NULL) return XP_NULL;
+		}
+		else if (MATCH(awk,TOKEN_IF)) {
+			CONSUME (awk);
+			if (__parse_if(awk) == XP_NULL) return XP_NULL;
+		}
+		else if (MATCH(awk,TOKEN_WHILE)) {
+			CONSUME (awk);
+			if (__parse_while(awk) == XP_NULL) return XP_NULL;
+		}
+		else if (MATCH(awk,TOKEN_FOR)) {
+			CONSUME (awk);
+			if (__parse_for(awk) == XP_NULL) return XP_NULL;
+		}
+		else if (MATCH(awk,TOKEN_DO)) {
+			CONSUME (awk);
+			if (__parse_do(awk) == XP_NULL) return XP_NULL;
+		}
+		else if (MATCH(awk,TOKEN_BREAK)) {
+			CONSUME (awk);
+			node->body = __parse_break(awk);
+		}
+		else if (MATCH(awk,TOKEN_CONTINUE)) {
+			CONSUME (awk);
+			node->body = __parse_continue(awk);
+		}
+		else if (MATCH(awk,TOKEN_RETURN)) {
+			CONSUME (awk);
+			/* TOOD: */
+		}
+		else if (MATCH(awk, TOKEN_EXIT)) {
+			CONSUME (awk);
+			/* TOOD: */
+		}
+		else if (MATCH(awk, TOKEN_DELETE)) {
+			CONSUME (awk);
+			/* TOOD: */
+		}
+		else if (MATCH(awk, TOKEN_NEXT)) {
+			CONSUME (awk);
+			/* TOOD: */
+		}
+		else if (MATCH(awk, TOKEN_NEXTFILE)) {
+			CONSUME (awk);
+			/* TOOD: */
+		}
+	}
+
+	return 0;
+}
+
+static xp_awk_node_t* __parse_if (xp_awk_t* awk)
+{
+	return XP_NULL;
+}
+
+static xp_awk_node_t* __parse_while (xp_awk_t* awk)
+{
+	return XP_NULL;
+}
+
+static xp_awk_node_t* __parse_for (xp_awk_t* awk)
+{
+	return XP_NULL;
+}
+
+static xp_awk_node_t* __parse_do (xp_awk_t* awk)
+{
+	return XP_NULL;
+}
+
+static xp_awk_node_t* __parse_break (xp_awk_t* awk)
+{
+	xp_awk_node_t* node;
+
+	node = (xp_awk_node_t*) xp_malloc (xp_sizeof(xp_awk_node_t));
+	if (node == XP_NULL) PANIC (awk, XP_AWK_ENOMEM);
+	node->type = XP_AWK_NODE_BREAK;
+	
+	/* TODO: do i have to consume a semicolon here???? */
+	return node;
+}
+
+static xp_awk_node_t* __parse_continue (xp_awk_t* awk)
+{
+	xp_awk_node_t* node;
+
+	node = (xp_awk_node_t*) xp_malloc (xp_sizeof(xp_awk_node_t));
+	if (node == XP_NULL) PANIC (awk, XP_AWK_ENOMEM);
+	node->type = XP_AWK_NODE_CONTINUE;
+	
+	/* TODO: do i have to consume a semicolon here???? */
+	return node;
 }
 
 static int __get_token (xp_awk_t* awk)
@@ -266,11 +440,11 @@ static int __get_token (xp_awk_t* awk)
 		ADD_TOKEN_CHAR (awk, c);
 	}
 	else if (c == XP_CHAR('[')) {
-		SET_TOKEN_TYPE (awk, TOKEN_LBRAKET);
+		SET_TOKEN_TYPE (awk, TOKEN_LBRACKET);
 		ADD_TOKEN_CHAR (awk, c);
 	}
 	else if (c == XP_CHAR(']')) {
-		SET_TOKEN_TYPE (awk, TOKEN_RBRAKET);
+		SET_TOKEN_TYPE (awk, TOKEN_RBRACKET);
 		ADD_TOKEN_CHAR (awk, c);
 	}
 	else {

ase/awk/tree.h

@@ -1,7 +1,14 @@
+/*
+ * $Id: tree.h,v 1.2 2006-01-09 12:51:47 bacon Exp $
+ */
 
+#ifndef _XP_AWK_TREE_H_
+#define _XP_AWK_TREE_H_
+
+/*
 enum TokenType
 {
-	IF, END, ID, NUM, READ, WrITE, UNTIL, ....
+	IF, END, ID, NUM, READ, WRITE, UNTIL, ....
 }
 
 
@@ -30,3 +37,53 @@ struct treenode
 
 	exptype type; <- for type checking...
 };
+
+struct node_t
+{
+	int type;
+
+};
+*/
+
+enum
+{
+	XP_AWK_NODE_BLOCK,
+	XP_AWK_NODE_BREAK,
+	XP_AWK_NODE_CONTINUE
+};
+
+typedef struct xp_awk_node_t xp_awk_node_t;
+typedef struct xp_awk_node_block_t xp_awk_node_block_t;
+typedef struct xp_awk_node_if_t xp_awk_node_if_t;
+
+#define XP_AWK_NODE_HDR \
+	int type; \
+	xp_awk_node_t* sbls
+
+struct xp_awk_node_t
+{
+	XP_AWK_NODE_HDR;
+};
+
+/*
+struct xp_awk_node_plain_t
+{
+	XP_AWK_NODE_HDR;
+};
+*/
+
+struct xp_awk_node_block_t
+{
+	XP_AWK_NODE_HDR;
+	xp_awk_node_t* body;
+};
+
+struct xp_awk_node_if_t
+{
+	XP_AWK_NODE_HDR;
+	xp_awk_node_t* cond;
+	xp_awk_node_t* if_part;
+	xp_awk_node_t* else_part;
+};
+
+#endif