*** empty log message ***

ase/awk/awk.h

@@ -1,5 +1,5 @@
 /* 
- * $Id: awk.h,v 1.80 2006-07-25 16:41:40 bacon Exp $
+ * $Id: awk.h,v 1.81 2006-07-26 16:43:35 bacon Exp $
  */
 
 #ifndef _XP_AWK_AWK_H_
@@ -127,7 +127,8 @@ enum
 	XP_AWK_ETOOFEWARGS,    /* too few arguments */
 	XP_AWK_ETOOMANYARGS,   /* too many arguments */
 	XP_AWK_EGETLINE,       /* getline expected */
-	XP_AWK_EREXCMPL,       /* cannot compile the regular expressio n*/
+	XP_AWK_EREXBUILD,      /* cannot build regexp */
+	XP_AWK_EREXMATCH,      /* an error occurred in matching regexp */
 
 	/* run time error */
 	XP_AWK_EDIVBYZERO,       /* divide by zero */
@@ -166,6 +167,8 @@ int xp_awk_close (xp_awk_t* awk);
 
 int xp_awk_geterrnum (xp_awk_t* awk);
 const xp_char_t* xp_awk_geterrstr (xp_awk_t* awk);
+int xp_awk_getsuberrnum (xp_awk_t* awk);
+const xp_char_t* xp_awk_getsuberrstr (xp_awk_t* awk);
 
 void xp_awk_clear (xp_awk_t* awk);
 void xp_awk_setparseopt (xp_awk_t* awk, int opt);

ase/awk/awk_i.h

@@ -1,5 +1,5 @@
 /*
- * $Id: awk_i.h,v 1.33 2006-07-26 05:19:45 bacon Exp $
+ * $Id: awk_i.h,v 1.34 2006-07-26 16:43:35 bacon Exp $
  */
 
 #ifndef _XP_AWK_AWKI_H_
@@ -138,7 +138,8 @@ struct xp_awk_t
 	} bfn;
 
 	/* housekeeping */
-	int errnum;
+	short errnum;
+	short suberrnum;
 };
 
 struct xp_awk_chain_t
@@ -190,7 +191,8 @@ struct xp_awk_run_t
 	xp_awk_extio_t* extio;
 
 	int opt;
-	int errnum;
+	short errnum;
+	short suberrnum;
 
 	/*xp_awk_tree_t* tree;
 	xp_size_t nglobals;*/

ase/awk/err.c

@@ -1,5 +1,5 @@
 /*
- * $Id: err.c,v 1.27 2006-07-25 16:41:40 bacon Exp $
+ * $Id: err.c,v 1.28 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -9,6 +9,17 @@ int xp_awk_geterrnum (xp_awk_t* awk)
 	return awk->errnum;
 }
 
+int xp_awk_getsuberrnum (xp_awk_t* awk)
+{
+	if (awk->errnum == XP_AWK_EREXBUILD ||
+	    awk->errnum == XP_AWK_EREXMATCH)
+	{
+		return awk->suberrnum;
+	}
+
+	return XP_AWK_ENOERR;
+}
+
 const xp_char_t* xp_awk_geterrstr (xp_awk_t* awk)
 {
 	static const xp_char_t* __errstr[] =
@@ -59,7 +70,8 @@ const xp_char_t* xp_awk_geterrstr (xp_awk_t* awk)
 		XP_T("too few arguments"),
 		XP_T("too many arguments"),
 		XP_T("getline expected"),
-		XP_T("cannot compile the regular expression"),
+		XP_T("cannot build the regular expression"),
+		XP_T("an error occurred in the regular expression match"),
 
 		XP_T("divide by zero"),
 		XP_T("invalid operand"),
@@ -83,3 +95,14 @@ const xp_char_t* xp_awk_geterrstr (xp_awk_t* awk)
 
 	return XP_T("unknown error");
 }
+
+const xp_char_t* xp_awk_getsuberrstr (xp_awk_t* awk)
+{
+	if (awk->errnum == XP_AWK_EREXBUILD ||
+	    awk->errnum == XP_AWK_EREXMATCH)
+	{
+		return xp_awk_getrexerrstr (awk->suberrnum);
+	}
+
+	return XP_T("no error");
+}

ase/awk/parse.c

@@ -1,5 +1,5 @@
 /*
- * $Id: parse.c,v 1.141 2006-07-26 14:59:59 bacon Exp $
+ * $Id: parse.c,v 1.142 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -294,6 +294,9 @@ static struct __bvent __bvtab[] =
 #define PANIC(awk,code) \
 	do { (awk)->errnum = (code); return XP_NULL; } while (0)
 
+#define PANIC2(awk,code,subcode) \
+	do { (awk)->errnum = (code); (awk)->suberrnum = (subcode); return XP_NULL; } while (0)
+
 /* TODO: remove stdio.h */
 #ifndef XP_AWK_STAND_ALONE
 #include <xp/bas/stdio.h>
@@ -1902,6 +1905,7 @@ static xp_awk_nde_t* __parse_primary (xp_awk_t* awk)
 	else if (MATCH(awk,TOKEN_DIV))
 	{
 		xp_awk_nde_rex_t* nde;
+		int errnum;
 
 		/* the regular expression is tokenized here because 
 		 * of the context-sensitivity of the slash symbol */
@@ -1928,17 +1932,19 @@ static xp_awk_nde_t* __parse_primary (xp_awk_t* awk)
 
 		nde->code = xp_awk_buildrex (
 			XP_STR_BUF(&awk->token.name), 
-			XP_STR_LEN(&awk->token.name));
+			XP_STR_LEN(&awk->token.name), 
+			&errnum);
 		if (nde->code == XP_NULL)
 		{
-			/* TODO: get the proper errnum */
 			xp_free (nde->buf);
 			xp_free (nde);
-			PANIC (awk, XP_AWK_ENOMEM);
+			PANIC2 (awk, XP_AWK_EREXBUILD, errnum);
 		}
 
 		if (__get_token(awk) == -1) 
 		{
+			xp_free (nde->buf);
+			xp_free (nde->code);
 			xp_free (nde);
 			return XP_NULL;			
 		}

ase/awk/rex.c

@@ -1,5 +1,5 @@
 /*
- * $Id: rex.c,v 1.16 2006-07-26 15:00:00 bacon Exp $
+ * $Id: rex.c,v 1.17 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -225,7 +225,6 @@ const xp_char_t* xp_awk_getrexerrstr (int errnum)
 	{
 		XP_T("no error"),
 		XP_T("out of memory"),
-		XP_T("no pattern built"),
 		XP_T("a right parenthesis is expected"),
 		XP_T("a right bracket is expected"),
 		XP_T("a right brace is expected"),
@@ -233,8 +232,8 @@ const xp_char_t* xp_awk_getrexerrstr (int errnum)
 		XP_T("invalid character range"),
 		XP_T("invalid character class"),
 		XP_T("invalid boundary range"),
-		XP_T("unexpected end of the pattern"),
+		XP_T("unexpected end of the regular expression"),
-		XP_T("garbage after the pattern")
+		XP_T("garbage after the regular expression")
 	};
 
 	if (errnum >= 0 && errnum < xp_countof(__errstr)) 
@@ -245,14 +244,18 @@ const xp_char_t* xp_awk_getrexerrstr (int errnum)
 	return XP_T("unknown error");
 }
 
-void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len)
+void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len, int* errnum)
 {
 	__builder_t builder;
 
 	builder.code.capa = 512;
 	builder.code.size = 0;
 	builder.code.buf = (xp_byte_t*) xp_malloc (builder.code.capa);
-	if (builder.code.buf == XP_NULL) return XP_NULL;
+	if (builder.code.buf == XP_NULL) 
+	{
+		*errnum = XP_AWK_REX_ENOMEM;
+		return XP_NULL;
+	}
 
 	builder.ptn.ptr = ptn;
 	builder.ptn.end = builder.ptn.ptr + len;
@@ -264,19 +267,21 @@ void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len)
 	//NEXT_CHAR (&builder, LEVEL_TOP);
 	if (__next_char (&builder, LEVEL_TOP) == -1) 
 	{
+		if (errnum != XP_NULL) *errnum = builder.errnum;
 		xp_free (builder.code.buf);
 		return XP_NULL;
 	}
 
 	if (__build_pattern (&builder) == -1) 
 	{
+		if (errnum != XP_NULL) *errnum = builder.errnum;
 		xp_free (builder.code.buf);
 		return XP_NULL;
 	}
 
 	if (builder.ptn.curc.type != CT_EOF)
 	{
-		/* garbage after the pattern */
+		if (errnum != XP_NULL) *errnum = XP_AWK_REX_EGARBAGE;
 		xp_free (builder.code.buf);
 		return XP_NULL;
 	}
@@ -286,7 +291,7 @@ void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len)
 
 int xp_awk_matchrex (void* code,
 	const xp_char_t* str, xp_size_t len, 
-	const xp_char_t** match_ptr, xp_size_t* match_len)
+	const xp_char_t** match_ptr, xp_size_t* match_len, int* errnum)
 {
 	__matcher_t matcher;
 	__match_t mat;
@@ -303,8 +308,11 @@ int xp_awk_matchrex (void* code,
 
 	while (mat.match_ptr < matcher.match.str.end)
 	{
-		if (__match_pattern (
+		if (__match_pattern (&matcher, code, &mat) == XP_NULL) 
-			&matcher, code, &mat) == XP_NULL) return -1;
+		{
+			if (errnum != XP_NULL) *errnum = matcher.errnum;
+			return -1;
+		}
 
 		if (mat.matched)
 		{

ase/awk/rex.h

@@ -1,15 +1,10 @@
 /*
- * $Id: rex.h,v 1.9 2006-07-26 05:19:45 bacon Exp $
+ * $Id: rex.h,v 1.10 2006-07-26 16:43:35 bacon Exp $
  **/
 
 #ifndef _XP_AWK_REX_H_
 #define _XP_AWK_REX_H_
 
-/*
-#ifndef _XP_AWK_AWK_H_
-#error Never include this file directly. Include <xp/awk/awk.h> instead
-#endif
-*/
 #include <xp/types.h>
 #include <xp/macros.h>
 
@@ -27,9 +22,9 @@
  *   | nb | el | na | bl | cmd | arg | cmd | arg | na | bl | cmd | arg | na | bl | cmd |
  *
  *   nb: the number of branches
- *   el: the length of a expression excluding the length of nb and el
+ *   el: the length of a expression including the length of nb and el
  *   na: the number of atoms
- *   bl: the length of a branch excluding the length of bl
+ *   bl: the length of a branch including the length of na and bl
  *   cmd: The command and repetition info encoded together. 
  *      Some commands require an argument to follow them but some other don't.
  *      It is encoded as follows:
@@ -46,7 +41,6 @@ enum
 {
 	XP_AWK_REX_ENOERR,    /* no error */
 	XP_AWK_REX_ENOMEM,    /* ran out of memory */
-	XP_AWK_REX_ENOPTN,    /* no pattern compiled */
 	XP_AWK_REX_ERPAREN,   /* a right parenthesis is expected */
 	XP_AWK_REX_ERBRACKET, /* a right bracket is expected */
 	XP_AWK_REX_ERBRACE,   /* a right brace is expected */
@@ -58,22 +52,22 @@ enum
 	XP_AWK_REX_EGARBAGE   /* garbage after the pattern */
 };
 
-#define XP_AWK_REXNA(code) (*(xp_size_t*)(code))
+#define XP_AWK_REX_NA(code) (*(xp_size_t*)(code))
 
-#define XP_AWK_REXLEN(code) \
+#define XP_AWK_REX_LEN(code) \
 	(*(xp_size_t*)((xp_byte_t*)(code)+xp_sizeof(xp_size_t)))
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-const xp_char_t* xp_awk_rex_geterrstr (int errnum);
+const xp_char_t* xp_awk_getrexerrstr (int errnum);
 
-void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len);
+void* xp_awk_buildrex (const xp_char_t* ptn, xp_size_t len, int* errnum);
 
 int xp_awk_matchrex (void* code,
 	const xp_char_t* str, xp_size_t len, 
-	const xp_char_t** match_ptr, xp_size_t* match_len);
+	const xp_char_t** match_ptr, xp_size_t* match_len, int* errnum);
 
 void xp_awk_printrex (void* code);
 

ase/awk/run.c

@@ -1,5 +1,5 @@
 /*
- * $Id: run.c,v 1.139 2006-07-26 15:00:00 bacon Exp $
+ * $Id: run.c,v 1.140 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -39,8 +39,13 @@ enum
 
 #define PANIC(run,code) \
 	do { (run)->errnum = (code); return XP_NULL; } while (0)
+#define PANIC2(awk,code,subcode) \
+	do { (awk)->errnum = (code); (awk)->suberrnum = (subcode); return XP_NULL; } while (0)
+
 #define PANIC_I(run,code) \
 	do { (run)->errnum = (code); return -1; } while (0)
+#define PANIC2_I(awk,code,subcode) \
+	do { (awk)->errnum = (code); (awk)->suberrnum = (subcode); return -1; } while (0)
 
 static int __open_run (
 	xp_awk_run_t* run, xp_awk_t* awk, xp_awk_io_t txtio, void* txtio_arg);
@@ -634,33 +639,25 @@ static int __run_pattern_block_chain (xp_awk_run_t* run, xp_awk_chain_t* chain)
 
 static int __handle_pattern (xp_awk_run_t* run, xp_awk_val_t* val)
 {
-	int n;
+	int n, errnum;
 
 	if (val->type == XP_AWK_VAL_REX)
 	{
 		xp_assert (run->inrec.d0->type == XP_AWK_VAL_STR);
 
-/* TODO: do it properly  match value...*/
-		//xp_awk_rex_setpattern (v->buf, v->len);
-
 		n = xp_awk_matchrex (
 			((xp_awk_val_rex_t*)val)->code,
 			((xp_awk_val_str_t*)run->inrec.d0)->buf,
 			((xp_awk_val_str_t*)run->inrec.d0)->len,
-			XP_NULL, XP_NULL);
+			XP_NULL, XP_NULL, &errnum);
 
-		if (n == -1)
+		if (n == -1) PANIC2_I (run, XP_AWK_EREXMATCH, errnum);
-		{
-// TODO: convert rex_matcher->errnum to run->errnum.
-			PANIC_I (run, XP_AWK_ENOMEM);
-		};
 	}
 	else n = xp_awk_valtobool(val)? 1: 0;
 
 	return n;
 }
 
-
 static int __run_block (xp_awk_run_t* run, xp_awk_nde_blk_t* nde)
 {
 	xp_awk_nde_t* p;

ase/awk/val.c

@@ -1,5 +1,5 @@
 /*
- * $Id: val.c,v 1.47 2006-07-26 15:00:00 bacon Exp $
+ * $Id: val.c,v 1.48 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -153,7 +153,7 @@ xp_awk_val_t* xp_awk_makerexval (
 		return XP_NULL;
 	}
 
-	val->code = xp_malloc (XP_AWK_REXLEN(code));
+	val->code = xp_malloc (XP_AWK_REX_LEN(code));
 	if (val->code == XP_NULL)
 	{
 		xp_free (val->buf);
@@ -161,7 +161,7 @@ xp_awk_val_t* xp_awk_makerexval (
 		return XP_NULL;
 	}
 
-	xp_memcpy (val->code, code, XP_AWK_REXLEN(code));
+	xp_memcpy (val->code, code, XP_AWK_REX_LEN(code));
 	return (xp_awk_val_t*)val;
 }
 

ase/test/awk/awk.c

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.c,v 1.52 2006-07-06 13:57:31 bacon Exp $
+ * $Id: awk.c,v 1.53 2006-07-26 16:43:35 bacon Exp $
  */
 
 #include <xp/awk/awk.h>
@@ -504,15 +504,24 @@ static int __main (int argc, xp_char_t* argv[])
 	{
 #if defined(__STAND_ALONE) && !defined(_WIN32) && defined(XP_CHAR_IS_WCHAR)
 		xp_printf (
-			XP_T("error: cannot parse program - line %u [%d] %ls\n"), 
+			XP_T("ERROR: cannot parse program - line %u [%d] %ls"), 
 			(unsigned int)xp_awk_getsrcline(awk), 
 			xp_awk_geterrnum(awk), xp_awk_geterrstr(awk));
+		if (xp_awk_getsuberrnum(awk) != XP_AWK_ENOERR)
+		{
+			xp_printf (XP_T(" - %ls\n"), xp_awk_getsuberrstr(awk));
+		}
 #else
 		xp_printf (
-			XP_T("error: cannot parse program - line %u [%d] %s\n"), 
+			XP_T("ERROR: cannot parse program - line %u [%d] %s"), 
 			(unsigned int)xp_awk_getsrcline(awk), 
 			xp_awk_geterrnum(awk), xp_awk_geterrstr(awk));
+		if (xp_awk_getsuberrnum(awk) != XP_AWK_ENOERR)
+		{
+			xp_printf (XP_T(" - %s\n"), xp_awk_getsuberrstr(awk));
+		}
 #endif
+		xp_printf (XP_T("\n"));
 		xp_awk_close (awk);
 		return -1;
 	}

ase/test/awk/t17.awk

@@ -1 +1 @@
-/hello/ { print $0; }
+/hello[[:space:]]/ { print $0; }