*** empty log message ***

ase/awk/extio.c

@@ -1,5 +1,5 @@
 /*
- * $Id: extio.c,v 1.37 2006-08-29 15:01:44 bacon Exp $
+ * $Id: extio.c,v 1.38 2006-08-30 07:15:14 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -254,7 +254,8 @@ int xp_awk_readextio (
 		if (rs_ptr == XP_NULL)
 		{
 			/* separate by a new line */
-			if (c == XP_T('\n')) break; 
+			/* TODO: handle different line terminator like \r\n */
+			if (c == XP_T('\n')) break;
 		}
 		else if (rs_len == 0)
 		{
@@ -320,13 +321,50 @@ int xp_awk_readextio (
 			break;
 		}
 
-		/* TODO: handle difference line terminator like \r\n */
+		/* TODO: handle different line terminator like \r\n */
 		if (c == XP_T('\n')) line_len = 0;
 		else line_len = line_len + 1;
 	}
 
 	if (rs_ptr != XP_NULL && rs->type != XP_AWK_VAL_STR) xp_free (rs_ptr);
 	xp_awk_refdownval (run, rs);
+
+	/* increment NR */
+	if (ret != -1)
+	{
+		xp_awk_val_t* nr;
+		xp_long_t lv;
+		xp_real_t rv;
+
+		nr = xp_awk_getglobal (run, XP_AWK_GLOBAL_NR);
+		xp_awk_refupval (nr);
+
+		n = xp_awk_valtonum (nr, &lv, &rv);
+		xp_awk_refdownval (run, nr);
+
+		if (n == -1)
+		{
+			run->errnum = XP_AWK_EVALTYPE;
+			ret = -1;
+		}
+		else
+		{
+			if (n == 1) lv = (xp_long_t)rv;
+
+			nr = xp_awk_makeintval (run, lv + 1);
+			if (nr == XP_NULL) 
+			{
+				run->errnum = XP_AWK_ENOMEM;
+				ret = -1;
+			}
+			else 
+			{
+				if (xp_awk_setglobal (
+					run, XP_AWK_GLOBAL_NR, nr) == -1) ret = -1;
+			}
+		}
+	}
+
 	return ret;
 }
 

ase/awk/parse.c

@@ -1,5 +1,5 @@
 /*
- * $Id: parse.c,v 1.169 2006-08-29 15:01:44 bacon Exp $
+ * $Id: parse.c,v 1.170 2006-08-30 07:15:14 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -3700,7 +3700,7 @@ static int __get_charstr (xp_awk_t* awk)
 {
 	if (awk->src.lex.curc != XP_T('\"')) 
 	{
-		/* the starting quote has been comsumed before this function
+		/* the starting quote has been consumed before this function
 		 * has been called */
 		ADD_TOKEN_CHAR (awk, awk->src.lex.curc);
 	}
@@ -3709,13 +3709,18 @@ static int __get_charstr (xp_awk_t* awk)
 
 static int __get_rexstr (xp_awk_t* awk)
 {
-	if (awk->src.lex.curc != XP_T('/')) 
+	if (awk->src.lex.curc == XP_T('/')) 
 	{
-		/* the starting slash has been comsumed before this function
+		/* this part of the function is different from __get_charstr
-		 * has been called */
+		 * because of the way this function is called */
-		ADD_TOKEN_CHAR (awk, awk->src.lex.curc);
+		GET_CHAR (awk);
+		return 0;
+	}
+	else 
+	{
+		ADD_TOKEN_CHAR (awk, awk->src.lex.curc);
+		return __get_string (awk, XP_T('/'), XP_T('\\'), xp_true);
 	}
-	return __get_string (awk, XP_T('/'), XP_T('\\'), xp_true);
 }
 
 static int __get_string (

ase/awk/rex.c

@@ -1,5 +1,5 @@
 /*
- * $Id: rex.c,v 1.23 2006-08-16 15:21:17 bacon Exp $
+ * $Id: rex.c,v 1.24 2006-08-30 07:15:14 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -348,9 +348,28 @@ int xp_awk_safematchrex (void* code,
 
 void xp_awk_freerex (void* code)
 {
+	xp_assert (code != XP_NULL);
 	xp_free (code);
 }
 
+xp_bool_t xp_awk_isemptyrex (void* code)
+{
+	const xp_byte_t* p = code;
+	xp_size_t nb, el;
+
+	xp_assert (p != XP_NULL);
+
+	nb = *(xp_size_t*)p; p += xp_sizeof(nb);
+	el = *(xp_size_t*)p; p += xp_sizeof(el);
+
+	/* an empty regular expression look like:
+	 *  | expression                     | 
+	 *  | header         | branch        |
+	 *  |                | branch header |
+	 *  | NB(1) | EL(16) | NA(1) | BL(8) | */
+	return (nb == 1 && el == xp_sizeof(xp_size_t)*4)? xp_true: xp_false;
+}
+
 static int __build_pattern (__builder_t* builder)
 {
 	int n;
@@ -1574,7 +1593,7 @@ static const xp_byte_t* __print_pattern (const xp_byte_t* p)
 
 	nb = *(xp_size_t*)p; p += xp_sizeof(nb);
 	el = *(xp_size_t*)p; p += xp_sizeof(el);
-//xp_printf (XP_T("NA = %u, EL = %u\n"), (unsigned int)nb, (unsigned int)el);
+xp_printf (XP_T("NB = %u, EL = %u\n"), (unsigned int)nb, (unsigned int)el);
 
 	for (i = 0; i < nb; i++)
 	{
@@ -1591,7 +1610,7 @@ static const xp_byte_t* __print_branch (const xp_byte_t* p)
 
 	na = *(xp_size_t*)p; p += xp_sizeof(na);
 	bl = *(xp_size_t*)p; p += xp_sizeof(bl);
-//xp_printf (XP_T("NA = %u, BL = %u\n"), (unsigned int) na, (unsigned int)bl);
+xp_printf (XP_T("NA = %u, BL = %u\n"), (unsigned int) na, (unsigned int)bl);
 
 	for (i = 0; i < na; i++)
 	{

ase/awk/rex.h

@@ -1,5 +1,5 @@
 /*
- * $Id: rex.h,v 1.13 2006-08-16 11:35:53 bacon Exp $
+ * $Id: rex.h,v 1.14 2006-08-30 07:15:14 bacon Exp $
  **/
 
 #ifndef _XP_AWK_REX_H_
@@ -62,6 +62,8 @@ int xp_awk_safematchrex (void* code,
 
 void xp_awk_freerex (void* code);
 
+xp_bool_t xp_awk_isemptyrex (void* code);
+
 #ifndef XP_AWK_NTDDK
 void xp_awk_printrex (void* code);
 #endif

ase/awk/run.c

@@ -1,5 +1,5 @@
 /*
- * $Id: run.c,v 1.182 2006-08-29 15:01:45 bacon Exp $
+ * $Id: run.c,v 1.183 2006-08-30 07:15:14 bacon Exp $
  */
 
 #include <xp/awk/awk_i.h>
@@ -622,12 +622,21 @@ static int __run_main (xp_awk_run_t* run)
 			/* restore the stack_top with the saved value
 			 * instead of calling __raw_pop as many times as
 			 * the successful __raw_push. it is ok because
-			 * the values pushed so fare are all xp_awk_val_nil */
+			 * the values pushed so far are all xp_awk_val_nil */
 			run->stack_top = saved_stack_top;
 			PANIC_I (run, XP_AWK_ENOMEM);
 		}
 	}	
 
+	if (xp_awk_setglobal (run, XP_AWK_GLOBAL_NR, xp_awk_val_zero) == -1)
+	{
+		/* it can simply restore the top of the stack this way
+		 * because the values pused onto the stack so far are
+		 * all xp_awk_val_nils */
+		run->stack_top = saved_stack_top;
+		return -1;
+	}
+
 	run->exit_level = EXIT_NONE;
 
 	if (run->awk->option & XP_AWK_RUNMAIN)
@@ -1825,22 +1834,34 @@ static xp_awk_val_t* __eval_expression (xp_awk_run_t* run, xp_awk_nde_t* nde)
 
 	if (v->type == XP_AWK_VAL_REX)
 	{
-		xp_assert (run->inrec.d0->type == XP_AWK_VAL_STR);
+		if (run->inrec.d0->type == XP_AWK_VAL_NIL)
-
-		xp_awk_refupval (v);
-		n = xp_awk_matchrex (
-			((xp_awk_val_rex_t*)v)->code,
-			((xp_awk_val_str_t*)run->inrec.d0)->buf,
-			((xp_awk_val_str_t*)run->inrec.d0)->len,
-			XP_NULL, XP_NULL, &errnum);
-
-		if (n == -1) 
 		{
-			xp_awk_refdownval (run, v);
+			/* the record has never been read. 
-			PANIC (run, errnum);
+			 * probably, this functions has been triggered
+			 * by the statements in the BEGIN block */
+			n = xp_awk_isemptyrex(
+				((xp_awk_val_rex_t*)v)->code)? 1: 0;
 		}
+		else
+		{
+			xp_assert (run->inrec.d0->type == XP_AWK_VAL_STR);
 
-		xp_awk_refdownval (run, v);
+			xp_awk_refupval (v);
+			n = xp_awk_matchrex (
+				((xp_awk_val_rex_t*)v)->code,
+				((xp_awk_val_str_t*)run->inrec.d0)->buf,
+				((xp_awk_val_str_t*)run->inrec.d0)->len,
+				XP_NULL, XP_NULL, &errnum);
+	
+			if (n == -1) 
+			{
+				xp_awk_refdownval (run, v);
+				PANIC (run, errnum);
+			}
+
+			xp_awk_refdownval (run, v);
+
+		}
 
 		v = xp_awk_makeintval (run, (n != 0));
 		if (v == XP_NULL) PANIC (run, XP_AWK_ENOMEM);

ase/awk/sa.h

@@ -1,5 +1,5 @@
 /*
- * $Id: sa.h,v 1.31 2006-08-16 11:35:54 bacon Exp $
+ * $Id: sa.h,v 1.32 2006-08-30 07:15:14 bacon Exp $
  */
 
 #ifndef _XP_AWK_SA_H_
@@ -96,6 +96,7 @@
 #define XP_STR_SIZE(x) ((x)->size + 1)
 #define XP_STR_CAPA(x) ((x)->capa)
 #define XP_STR_BUF(x)  ((x)->buf)
+#define XP_STR_CHAR(x,idx) ((x)->buf[idx])
 
 typedef struct xp_str_t xp_str_t;
 

ase/test/awk/awk.c

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.c,v 1.78 2006-08-28 14:30:08 bacon Exp $
+ * $Id: awk.c,v 1.79 2006-08-30 07:15:14 bacon Exp $
  */
 
 #include <xp/awk/awk.h>
@@ -567,7 +567,8 @@ static int __main (int argc, xp_char_t* argv[])
 		return -1;
 	}
 
-	opt = XP_AWK_EXPLICIT | XP_AWK_UNIQUE | XP_AWK_DBLSLASHES |
+	opt = XP_AWK_EXPLICIT | XP_AWK_UNIQUE | XP_AWK_HASHSIGN |
+		/*XP_AWK_DBLSLASHES |*/
 		XP_AWK_SHADING | XP_AWK_IMPLICIT | XP_AWK_SHIFT | 
 		XP_AWK_EXTIO | XP_AWK_BLOCKLESS | XP_AWK_STRINDEXONE;
 

ase/test/awk/rex2.c

@@ -19,7 +19,8 @@ int xp_main (int argc, const xp_char_t* argv[])
 	//ptn = XP_T("^he.llo(jo(in|kk)s|com)+h*e{1,40}abc|[^abc][de-f]|^he.llo(jo(in|kk)s|com)+h*e{1,40}abc|[^abc][de-f]");
 	//ptn = XP_T("^he.llo(jo(in|kk)s|com)+h*e{1,40}abc|[^abc][de-f]");
 	//ptn = XP_T("^he.llo(jo(in|kk)s|com)|[^x[:space:][:alpha:]j][^abc][de-f]|^he.llo(jo(in|kk)s|com)|[^x[:space:][:alpha:]j][^abc][de-f]");
-	ptn = XP_T("^.{0,2}.z[^[:space:]]+(abc|zzz){1,2}khg");
+	//ptn = XP_T("^.{0,2}.z[^[:space:]]+(abc|zzz){1,2}khg");
+	ptn = XP_T("");
 
 	rex = xp_awk_buildrex (ptn, xp_strlen(ptn), &errnum);
 	if (rex == XP_NULL)
@@ -28,6 +29,7 @@ int xp_main (int argc, const xp_char_t* argv[])
 		return -1;
 	}
 
+	xp_printf (XP_T("isemptyrex => %d\n"), xp_awk_isemptyrex (rex));
 	xp_printf (XP_T("NA: %u\n"), (unsigned int)XP_AWK_REX_NA(rex));
 	xp_printf (XP_T("LEN: %u\n"), (unsigned int)XP_AWK_REX_LEN(rex));
 	xp_awk_printrex (rex);

ase/test/awk/t32.awk

@@ -1,2 +1,2 @@
-BEGIN { /*RS = "Asia";*/ /*RS=746;*/ RS=/USA/; }
+BEGIN { /*RS = "Asia";*/ /*RS=746;*/ /*RS="";*/ RS=/USA/; }
-{ print "RECORD: ", $0; }
+{ print NR, " ", $0; }

ase/test/awk/t33.awk

@@ -5,7 +5,7 @@ BEGIN {
 	delete NF;
 	print "NF[1]=", NF[1];
 
-	NF[1] = 20;  // this line should not be allowed
+	NF[1] = 20;  # this line should not be allowed
 	print "AWK IMPLEMENTATION ERROR: hey... NF[1] = 20 has succeeded in the BEGIN block. your interpreter must be wrong";
 	print "NF[1]=", NF[1];
 }