libbb: robustify isXXXX(). +39 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

coreutils/sort.c

@@ -128,7 +128,7 @@ static char *get_key(char *str, struct sort_key *key, int flags)
 	/* Handle -i */
 	if (flags & FLAG_i) {
 		for (start = end = 0; str[end]; end++)
-			if (isprint(str[end]))
+			if (isprint_asciionly(str[end]))
 				str[start++] = str[end];
 		str[start] = '\0';
 	}

coreutils/tr.c

@@ -153,7 +153,7 @@ static unsigned expand(const char *arg, char **buffer_p)
 				}
 				if (j == CLASS_punct || j == CLASS_cntrl) {
 					for (i = '\0'; i < ASCII; i++) {
-						if ((j == CLASS_punct && isprint(i) && !isalnum(i) && !isspace(i))
+						if ((j == CLASS_punct && isprint_asciionly(i) && !isalnum(i) && !isspace(i))
 						 || (j == CLASS_cntrl && iscntrl(i))
 						) {
 							buffer[pos++] = i;

editors/diff.c

@@ -749,7 +749,7 @@ static int asciifile(FILE *f)
 	rewind(f);
 	cnt = fread(g_read_buf, 1, COMMON_BUFSIZE, f);
 	for (i = 0; i < cnt; i++) {
-		if (!isprint(g_read_buf[i])
+		if (!isprint_asciionly(g_read_buf[i])
 		 && !isspace(g_read_buf[i])
 		) {
 			return 0;

include/libbb.h

@@ -1606,10 +1606,11 @@ extern const char bb_default_login_shell[];
 
 /* We save ~500 bytes on isdigit alone.
  * BTW, x86 likes (unsigned char) cast more than (unsigned). */
-#define isdigit(a) ((unsigned char)((a) - '0') <= 9)
+
+/* These work the same for ASCII and Unicode,
+ * assuming no one asks "is this a *Unicode* letter?" using isalpha(letter) */
 #define isascii(a) ((unsigned char)(a) <= 0x7f)
-#define isgraph(a) ((unsigned char)(a) > ' ')
-#define isprint(a) ((unsigned char)(a) >= ' ')
+#define isdigit(a) ((unsigned char)((a) - '0') <= 9)
 #define isupper(a) ((unsigned char)((a) - 'A') <= ('Z' - 'A'))
 #define islower(a) ((unsigned char)((a) - 'a') <= ('z' - 'a'))
 #define isalpha(a) ((unsigned char)(((a)|0x20) - 'a') <= ('z' - 'a'))
@@ -1619,9 +1620,9 @@ extern const char bb_default_login_shell[];
  * "\t\n\v\f\r" happen to have ASCII codes 9,10,11,12,13.
  */
 #define isspace(a) ({ unsigned char bb__isspace = (a) - 9; bb__isspace == (' ' - 9) || bb__isspace <= (13 - 9); })
-
-// Bigger code:
-//#define isalnum(a) ({ unsigned char bb__isalnum = (a) - '0'; bb__isalnum <= 9 || ((bb__isalnum - ('A' - '0')) & 0xdf) <= 25; })
+// Unsafe wrt NUL: #define ispunct(a) (strchr("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", (a)) != NULL)
+#define ispunct(a) (strchrnul("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", (a))[0])
+// Bigger code: #define isalnum(a) ({ unsigned char bb__isalnum = (a) - '0'; bb__isalnum <= 9 || ((bb__isalnum - ('A' - '0')) & 0xdf) <= 25; })
 #define isalnum(a) bb_ascii_isalnum(a)
 static ALWAYS_INLINE int bb_ascii_isalnum(unsigned char a)
 {
@@ -1640,11 +1641,6 @@ static ALWAYS_INLINE int bb_ascii_isxdigit(unsigned char a)
 	b = (a|0x20) - 'a';
 	return b <= 'f' - 'a';
 }
-
-// Unsafe wrt NUL!
-//#define ispunct(a) (strchr("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", (a)) != NULL)
-#define ispunct(a) (strchrnul("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", (a))[0])
-
 #define toupper(a) bb_ascii_toupper(a)
 static ALWAYS_INLINE unsigned char bb_ascii_toupper(unsigned char a)
 {
@@ -1662,6 +1658,14 @@ static ALWAYS_INLINE unsigned char bb_ascii_tolower(unsigned char a)
 	return a;
 }
 
+/* In ASCII and Unicode, these are likely to be very different.
+ * Let's prevent ambiguous usage from the start */
+#define isgraph(a) isgraph_is_ambiguous_dont_use(a)
+#define isprint(a) isprint_is_ambiguous_dont_use(a)
+/* NB: must not treat EOF as isgraph or isprint */
+#define isgraph_asciionly(a) ((unsigned)((a) - 0x21) <= 0x7e - 0x21)
+#define isprint_asciionly(a) ((unsigned)((a) - 0x20) <= 0x7e - 0x20)
+
 
 POP_SAVED_FUNCTION_VISIBILITY
 

libbb/dump.c

@@ -492,7 +492,7 @@ static void conv_c(PR *pr, unsigned char *p)
 		str += 4;
 	} while (*str);
 
-	if (isprint(*p)) {
+	if (isprint_asciionly(*p)) {
 		*pr->cchar = 'c';
 		printf(pr->fmt, *p);
 	} else {
@@ -519,7 +519,7 @@ static void conv_u(PR *pr, unsigned char *p)
 	} else if (*p == 0x7f) {
 		*pr->cchar = 's';
 		printf(pr->fmt, "del");
-	} else if (isprint(*p)) {
+	} else if (*p < 0x7f) { /* isprint() */
 		*pr->cchar = 'c';
 		printf(pr->fmt, *p);
 	} else {
@@ -609,7 +609,7 @@ static void display(priv_dumper_t* dumper)
 							break;
 						}
 						case F_P:
-							printf(pr->fmt, isprint(*bp) ? *bp : '.');
+							printf(pr->fmt, isprint_asciionly(*bp) ? *bp : '.');
 							break;
 						case F_STR:
 							printf(pr->fmt, (char *) bp);

loginutils/getty.c

@@ -487,7 +487,7 @@ static char *get_logname(char *logname, unsigned size_logname,
 			case CTL('D'):
 				exit(EXIT_SUCCESS);
 			default:
-				if (!isprint(ascval)) {
+				if (ascval < ' ') {
 					/* ignore garbage characters */
 				} else if ((int)(bp - logname) >= size_logname - 1) {
 					bb_error_msg_and_die("%s: input overrun", op->tty);

miscutils/strings.c

@@ -49,7 +49,7 @@ int strings_main(int argc UNUSED_PARAM, char **argv)
 		count = 0;
 		do {
 			c = fgetc(file);
-			if (isprint(c) || c == '\t') {
+			if (isprint_asciionly(c) || c == '\t') {
 				if (count > n) {
 					bb_putchar(c);
 				} else {

networking/inetd.c

@@ -1475,8 +1475,7 @@ static void init_ring(void)
 	int i;
 
 	end_ring = ring;
-	for (i = 0; i <= 128; ++i)
-		if (isprint(i))
+	for (i = ' '; i < 127; i++)
 		*end_ring++ = i;
 }
 /* Character generator. MMU arches only. */