enhanced index and match

qse/doc/page/awk.doc

@@ -13,10 +13,10 @@ into an application written in C and/or C++. A hosting application can
 - change language features supported by setting options.
 
 The interpreter implements the language described in the book 
-The AWK Programming Language (http://cm.bell-labs.com/cm/cs/awkbook/) with
+The AWK Programming Language(http://cm.bell-labs.com/cm/cs/awkbook/) with
 some extensions.
 
-@section awk_ext EXTENSION
+@section awk_ext EXTENSIONS
 Some language extensions are implemented and they can be enabled by setting the
 corresponding options.
 
@@ -83,13 +83,30 @@ BEGIN {
 }
 @endcode
 
-@subsectin awk_ext_fnc EXTENDED FUNCTIONS
+@subsection awk_ext_return RETURN
-index() and match() can accept the third parameter indicating the position
+The return statement is valid in BEGIN blocks, END blocks, and pattern-action 
-where the search should begin. The negative position enables backward search.
+blocks as well as in functions. The execution of a calling block is aborted
+once the return statement is executed.
+
+@subsection awk_ext_comment COMMENT
+You can use the C-style comment as well as the pound comment.
+
+@subsection awk_ext_fnc EXTENDED FUNCTIONS
+index() and match() can accept the third parameter indicating the position 
+where the search begins. A negative value indicates a position from the back.
 
 @code
 BEGIN {
+	xstr = "abcdefabcdefabcdef";
+	xsub = "abc";
+	xlen = length(xsub);
 
+	i = 1;
+	while ((i = index(xstr, xsub, i)) > 0)
+	{
+		print i, substr(xstr, i, xlen);
+		i += xlen;
+	}
 }
 @endcode
 

qse/include/qse/cmn/str.h

@@ -1,5 +1,5 @@
 /*
- * $Id: str.h 287 2009-09-15 10:01:02Z hyunghwan.chung $
+ * $Id: str.h 289 2009-09-16 06:35:29Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -250,7 +250,7 @@ int qse_strcasecmp (const qse_char_t* s1, const qse_char_t* s2);
  * represented by its beginning pointer and length. 
  *
  * For two strings to be equal, they need to have the same length and all
- * characters in the first string should be equal to their counterpart in the
+ * characters in the first string must be equal to their counterpart in the
  * second string.
  *
  * The following code snippet compares "foo" and "FoO" case-insenstively.
@@ -258,31 +258,84 @@ int qse_strcasecmp (const qse_char_t* s1, const qse_char_t* s2);
  * qse_strxncasecmp (QSE_T("foo"), 3, QSE_T("FoO"), 3);
  * @endcode
  *
- * @return
+ * @return 0 if two strings are equal, 
- * The qse_strxncasecmp() returns 0 if two strings are equal, a positive
+ *         a positive number if the first string is larger, 
- * number if the first string is larger, -1 if the second string is larger.
+ *         -1 if the second string is larger.
  *
  */
 int qse_strxncasecmp (
-	const qse_char_t* s1   /* the pointer to the first string */,
+	const qse_char_t* s1,   /**< pointer to the first string */
-	qse_size_t        len1 /* the length of the first string */, 
+	qse_size_t        len1, /**< length of the first string */ 
-	const qse_char_t* s2   /* the pointer to the second string */,
+	const qse_char_t* s2,   /**< pointer to the second string */
-	qse_size_t        len2 /* the length of the second string */
+	qse_size_t        len2  /**< length of the second string */
 );
 
-qse_char_t* qse_strdup (const qse_char_t* str, qse_mmgr_t* mmgr);
+qse_char_t* qse_strdup (
-qse_char_t* qse_strxdup (
+	const qse_char_t* str,
-	const qse_char_t* str, qse_size_t len, qse_mmgr_t* mmgr);
+	qse_mmgr_t*       mmgr
-qse_char_t* qse_strxdup2 (
+);
-	const qse_char_t* str1, qse_size_t len1,
+
-	const qse_char_t* str2, qse_size_t len2, qse_mmgr_t* mmgr);
+qse_char_t* qse_strxdup (
+	const qse_char_t* str,
+	qse_size_t        len, 
+	qse_mmgr_t*       mmgr
+);
+
+qse_char_t* qse_strxdup2 (
+	const qse_char_t* str1,
+	qse_size_t        len1,
+	const qse_char_t* str2,
+	qse_size_t        len2,
+	qse_mmgr_t*       mmgr
+);
+
+/**
+ * The qse_strstr() function searchs a string @a str for the first occurrence 
+ * of a substring @a sub
+ * @return pointer to the first occurrence in @a str if @a sub is found, 
+ *         QSE_NULL if not.
+ */
+qse_char_t* qse_strstr (
+	const qse_char_t* str, 
+	const qse_char_t* sub
+);
 
-qse_char_t* qse_strstr (const qse_char_t* str, const qse_char_t* sub);
 qse_char_t* qse_strxstr (
-	const qse_char_t* str, qse_size_t size, const qse_char_t* sub);
+	const qse_char_t* str,
+	qse_size_t        size,
+	const qse_char_t* sub
+);
+
 qse_char_t* qse_strxnstr (
-	const qse_char_t* str, qse_size_t strsz, 
+	const qse_char_t* str,
-	const qse_char_t* sub, qse_size_t subsz);
+	qse_size_t        strsz, 
+	const qse_char_t* sub,
+	qse_size_t        subsz
+);
+
+/**
+ * The qse_strstr() function searchs a string @a str for the last occurrence 
+ * of a substring @a sub
+ * @return pointer to the last occurrence in @a str if @a sub is found, 
+ *         QSE_NULL if not.
+ */
+qse_char_t* qse_strrstr (
+	const qse_char_t* str,
+	const qse_char_t* sub
+);
+
+qse_char_t* qse_strxrstr (
+	const qse_char_t* str,
+	qse_size_t        size,
+	const qse_char_t* sub
+);
+
+qse_char_t* qse_strxnrstr (
+	const qse_char_t* str,
+	qse_size_t        strsz, 
+	const qse_char_t* sub,
+	qse_size_t        subsz
+);
 
 qse_char_t* qse_strchr (const qse_char_t* str, qse_cint_t c);
 qse_char_t* qse_strxchr (const qse_char_t* str, qse_size_t len, qse_cint_t c);

qse/include/qse/macros.h

@@ -1,5 +1,5 @@
 /*
- * $Id: macros.h 287 2009-09-15 10:01:02Z hyunghwan.chung $
+ * $Id: macros.h 289 2009-09-16 06:35:29Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -45,8 +45,19 @@
 #	define QSE_NULL ((void*)0)
 #endif
 
+/**
+ * The QSE_MCHAR_EOF macro defines an EOF character.
+ */
 #define QSE_MCHAR_EOF ((qse_mcint_t)-1)
+
+/**
+ * The QSE_WCHAR_EOF macro defines an EOF character.
+ */
 #define QSE_WCHAR_EOF ((qse_wcint_t)-1)
+
+/**
+ * The QSE_CHAR_EOF macro defines an EOF character.
+ */
 #define QSE_CHAR_EOF  ((qse_cint_t)-1)
 
 /**

qse/lib/awk/fnc.c

@@ -1,5 +1,5 @@
 /*
- * $Id: fnc.c 288 2009-09-15 14:03:15Z hyunghwan.chung $
+ * $Id: fnc.c 289 2009-09-16 06:35:29Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -462,7 +462,6 @@ static int fnc_index (qse_awk_rtx_t* rtx, const qse_cstr_t* fnm)
 		n = qse_awk_rtx_valtonum (rtx, a2, &start, &rv);
 		if (n <= -1) return -1;
 		if (n >= 1) start = (qse_long_t)rv;
-		if (start < 1) start = 1;
 	}
 
 	if (a0->type == QSE_AWK_VAL_STR)
@@ -492,9 +491,14 @@ static int fnc_index (qse_awk_rtx_t* rtx, const qse_cstr_t* fnm)
 		}
 	}
 
-	ptr = (start > len0)? 
+/* TODO: ignorecase... */
+	if (start == 0) start = 1;
+	else if (start < 0) start = len0 + start + 1;
+
+	ptr = (start > len0 || start <= 0)? 
 		QSE_NULL:
 		qse_strxnstr (&str0[start-1], len0-start+1, str1, len1);
+
 	idx = (ptr == QSE_NULL)? 0: ((qse_long_t)(ptr-str0) + 1);
 
 	if (a0->type != QSE_AWK_VAL_STR) QSE_AWK_FREE (rtx->awk, str0);
@@ -1305,10 +1309,8 @@ static int fnc_match (qse_awk_rtx_t* rtx, const qse_cstr_t* fnm)
 		n = qse_awk_rtx_valtonum (rtx, a2, &start, &rv);
 		if (n <= -1) return -1;
 		if (n >= 1) start = (qse_long_t)rv;
-		if (start < 1) start = 1;
 	}
 
-
 	if (a0->type == QSE_AWK_VAL_STR)
 	{
 		str0 = ((qse_awk_val_str_t*)a0)->ptr;
@@ -1356,7 +1358,10 @@ static int fnc_match (qse_awk_rtx_t* rtx, const qse_cstr_t* fnm)
 		if (a1->type != QSE_AWK_VAL_STR) QSE_AWK_FREE (rtx->awk, str1);
 	}
 
-	if (start > len0) n = 0;
+	if (start == 0) start = 1;
+	else if (start < 0) start = len0 + start + 1;
+
+	if (start > len0 || start <= 0) n = 0;
 	else
 	{
 		n = QSE_AWK_MATCHREX (

qse/lib/cmn/map.c

@@ -1,5 +1,5 @@
 /*
- * $Id: map.c 287 2009-09-15 10:01:02Z hyunghwan.chung $
+ * $Id: map.c 289 2009-09-16 06:35:29Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -57,17 +57,20 @@ static int reorganize (map_t* map);
 
 static size_t hash_key (map_t* map, const void* kptr, size_t klen)
 {
-	size_t n = 0;
+	/*size_t h = 2166136261;*/
+	/*size_t h = 0;*/
+	size_t h = 5381;
 	const byte_t* p = (const byte_t*)kptr;
 	const byte_t* bound = p + klen;
 
 	while (p < bound)
 	{
-		n = n * 31 + *p++;
+		/*h = (h * 16777619) ^ *p++;*/
-		p++;
+		/*h = h * 31 + *p++;*/
+		h = ((h << 5) + h) + *p++;
 	}	
 
-	return n;
+	return h ; 
 }
 
 static int comp_key (map_t* map, 

qse/lib/cmn/str_bas.c

@@ -1,5 +1,5 @@
 /*
- * $Id: str_bas.c 287 2009-09-15 10:01:02Z hyunghwan.chung $
+ * $Id: str_bas.c 289 2009-09-16 06:35:29Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -543,6 +543,47 @@ qse_char_t* qse_strxnstr (
 	return QSE_NULL;
 }
 
+qse_char_t* qse_strrstr (const qse_char_t* str, const qse_char_t* sub)
+{
+	return qse_strxnrstr (str, qse_strlen(str), sub, qse_strlen(sub));
+}
+
+qse_char_t* qse_strxrstr (
+	const qse_char_t* str, qse_size_t size, const qse_char_t* sub)
+{
+	return qse_strxnrstr (str, size, sub, qse_strlen(sub));
+}
+
+qse_char_t* qse_strxnrstr (
+	const qse_char_t* str, qse_size_t strsz, 
+	const qse_char_t* sub, qse_size_t subsz)
+{
+	const qse_char_t* p = str + strsz;
+	const qse_char_t* subp = sub + subsz;
+
+	if (subsz == 0) return (qse_char_t*)p;
+	if (strsz < subsz) return QSE_NULL;
+
+	p = p - subsz;
+
+	while (p >= str) 
+	{
+		const qse_char_t* x = p;
+		const qse_char_t* y = sub;
+
+		while (1) 
+		{
+			if (y >= subp) return (qse_char_t*)p;
+			if (*x != *y) break;
+			x++; y++;
+		}	
+
+		p--;
+	}
+
+	return QSE_NULL;
+}
+
 qse_char_t* qse_strchr (const qse_char_t* str, qse_cint_t c)
 {
 	while (*str != QSE_T('\0')) 

qse/regress/awk/lang-038.awk

@@ -0,0 +1,25 @@
+#
+# test the third parameter(starting position) of index and match
+#
+
+BEGIN {
+	xstr = "abcdefabcdefabcdef";
+	xsub = "abc";
+	xlen = length(xsub);
+
+	i = 1;
+	while ((i = index(xstr, xsub, i)) > 0)
+	{
+		print i, substr(xstr, i, xlen);
+		i += xlen;
+	}
+
+	print "----------------";
+
+	i = 1;
+	while (match(xstr, xsub, i) > 0)
+	{
+		print RSTART, substr(xstr, RSTART, RLENGTH);
+		i = RSTART + RLENGTH;
+	}
+}

qse/regress/awk/regress.out

@@ -1990,6 +1990,35 @@ pq...r AAA2
 
 kbs ddd
 dif cccc
+--------------------------------------------------------------------------------
+ ../../cmd/awk/.libs/qseawk --newline=on -o- -f lang-038.awk  </dev/stdin 2>&1 
+--------------------------------------------------------------------------------
+BEGIN {
+	xstr = "abcdefabcdefabcdef";
+	xsub = "abc";
+	xlen = length (xsub);
+	i = 1;
+	while ((i = index (xstr,xsub,i) > 0))
+	{
+		print i,substr (xstr,i,xlen);
+		i += xlen;
+	}
+	print "----------------";
+	i = 1;
+	while ((match (xstr,xsub,i) > 0))
+	{
+		print RSTART,substr (xstr,RSTART,RLENGTH);
+		i = (RSTART + RLENGTH);
+	}
+}
+
+1 abc
+7 abc
+13 abc
+----------------
+1 abc
+7 abc
+13 abc
 --------------------------------------------------------------------------------
  ../../cmd/awk/.libs/qseawk  -f quicksort.awk quicksort.dat </dev/stdin 2>&1 
 --------------------------------------------------------------------------------

qse/regress/awk/regress.sh

@@ -148,6 +148,7 @@ PROGS="
 	lang-035.awk/lang-035.dat2//--newline=on -o- -vdatafile=lang-035.dat1 -vgroupname=lang-035
 	lang-036.awk/lang-036.dat//--newline=on -o-
 	lang-037.awk/lang-037.dat//--newline=on -o-
+	lang-038.awk///--newline=on -o-
 
 	quicksort.awk/quicksort.dat//
 	quicksort2.awk/quicksort2.dat//