Make single byte caching more elegant.

5 files changed, 21 insertions, 10 deletions

diff --git a/ChangeLog b/ChangeLog
index 18067fae..666a435f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,13 @@
+Wed Jan 19 20:31:17 2011  Arnold D. Robbins  <arnold@skeeve.com>
+
+	* awk.h (is_valid_character): Add `& 0XFF' and remove casts to
+	unsigned chars in other files. Remove definition of this macro
+	in not MBS_SUPPORT case, since it wasn't being used.
+	(btowc_cache): New macro to index into the array and use the
+	same trick. Relies on ANSI C preprocessor semantics.
+	Fix all uses.
+	* builtin.c, node.c, io.c: Fix uses of these macros.
+
 Wed Jan 19 20:19:29 2011  Arnold D. Robbins  <arnold@skeeve.com>
 
 	* node.c (wstr2str): New function.
@@ -7,6 +17,7 @@ Wed Jan 19 20:19:29 2011  Arnold D. Robbins  <arnold@skeeve.com>
 	simplify wide character case conversions.
 	(do_tolower, do_toupper): Use wide_tolower, wide_toupper in multibyte
 	case.
+	(do_substr): Simplify code a little bit.
 
 Mon Jan 17 22:48:48 2011  Arnold D. Robbins  <arnold@skeeve.com>
 
diff --git a/awk.h b/awk.h
index 664b4f7c..13674448 100644
--- a/awk.h
+++ b/awk.h
@@ -1329,11 +1329,11 @@ extern const wchar_t *wcasestrstr(const wchar_t *haystack, size_t hs_len,
 		const wchar_t *needle, size_t needle_len);
 extern void free_wstr(NODE *n);
 extern wint_t btowc_cache[];
+#define btowc_cache(x) btowc_cache[(x)&0xFF]
 extern void init_btowc_cache();
-#define is_valid_character(b)	(btowc_cache[b] != WEOF)
+#define is_valid_character(b)	(btowc_cache[(b)&0xFF] != WEOF)
 #else
 #define free_wstr(NODE)	/* empty */
-#define is_valid_character(c)	(TRUE)
 #endif
 /* re.c */
 extern Regexp *make_regexp(const char *s, size_t len, int ignorecase, int dfa, int canfatal);
diff --git a/builtin.c b/builtin.c
index a9ece11a..8bb10486 100644
--- a/builtin.c
+++ b/builtin.c
@@ -231,26 +231,26 @@ strncasecmpmbs(const unsigned char *s1, const unsigned char *s2, size_t n)
 	for (i1 = i2 = 0 ; i1 < n && i2 < n ;i1 += mbclen1, i2 += mbclen2) {
 		if (is_valid_character(s1[i1])) {
 			mbclen1 = 1;
-			wc1 = btowc_cache[s1[i1]];
+			wc1 = btowc_cache(s1[i1]);
 		} else {
 			mbclen1 = mbrtowc(& wc1, (const char *)s1 + i1,
 					  n - i1, & mbs1);
 			if (mbclen1 == (size_t) -1 || mbclen1 == (size_t) -2 || mbclen1 == 0) {
 				/* We treat it as a singlebyte character. */
 				mbclen1 = 1;
-				wc1 = btowc_cache[s1[i1]];
+				wc1 = btowc_cache(s1[i1]);
 			}
 		}
 		if (is_valid_character(s2[i2])) {
 			mbclen2 = 1;
-			wc2 = btowc_cache[s2[i2]];
+			wc2 = btowc_cache(s2[i2]);
 		} else {
 			mbclen2 = mbrtowc(& wc2, (const char *)s2 + i2,
 					  n - i2, & mbs2);
 			if (mbclen2 == (size_t) -1 || mbclen2 == (size_t) -2 || mbclen2 == 0) {
 				/* We treat it as a singlebyte character. */
 				mbclen2 = 1;
-				wc2 = btowc_cache[s2[i2]];
+				wc2 = btowc_cache(s2[i2]);
 			}
 		}
 		if ((gap = towlower(wc1) - towlower(wc2)) != 0)
@@ -313,8 +313,8 @@ do_index(int nargs)
 	const char *p1, *p2;
 	size_t l1, l2;
 	long ret;
-	int do_single_byte = FALSE;
 #ifdef MBS_SUPPORT
+	int do_single_byte = FALSE;
 	mbstate_t mbs1, mbs2;
 
 	if (gawk_mb_cur_max > 1) {
diff --git a/io.c b/io.c
index af15aa02..7cf48911 100644
--- a/io.c
+++ b/io.c
@@ -2689,7 +2689,7 @@ rs1scan(IOBUF *iop, struct recmatch *recm, SCANSTATE *state)
 		do {
 			if (*bp == rs)
 				found = 1;
-			if (is_valid_character((unsigned char)*bp))
+			if (is_valid_character(*bp))
 				mbclen = 1;
 			else
 				mbclen = mbrlen(bp, len, &mbs);
diff --git a/node.c b/node.c
index cf16f794..a8805176 100644
--- a/node.c
+++ b/node.c
@@ -708,9 +708,9 @@ str2wstr(NODE *n, size_t **ptr)
 		 * big speed up. Thanks to Ulrich Drepper for the tip.
 		 * 11/2010: Thanks to Paolo Bonzini for some even faster code.
 		 */
-		if (is_valid_character((unsigned char)*sp)) {
+		if (is_valid_character(*sp)) {
 			count = 1;
-			wc = btowc_cache[(unsigned char)*sp];
+			wc = btowc_cache(*sp);
 		} else
 			count = mbrtowc(& wc, sp, src_count, & mbs);
 		switch (count) {