X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-http;a=blobdiff_plain;f=php_http_utf8.h;h=f9008032dff0eea98abc8c59cb1134b35ab3b7c3;hp=c7bcb49428c53a79234593390c3705d9ca791af5;hb=refs%2Fheads%2Fv2.4.x;hpb=caddd8ffdc0b23544671c64df711c04d3da1b4b4

diff --git a/php_http_utf8.h b/php_http_utf8.h
index c7bcb49..f900803 100644
--- a/php_http_utf8.h
+++ b/php_http_utf8.h
@@ -555,7 +555,7 @@ static inline size_t utf8towc(unsigned *wc, const unsigned char *uc, size_t len)
 {
 	unsigned char ub = utf8_mblen[*uc];
 
-	if (!ub || ub > len || ub > 3) {
+	if (!ub || ub > len || ub > 4) {
 		return 0;
 	}
 
@@ -595,9 +595,9 @@ static inline size_t utf8towc(unsigned *wc, const unsigned char *uc, size_t len)
 
 static inline zend_bool isualpha(unsigned ch)
 {
-	unsigned i, j;
+	unsigned i = 0, j;
 
-	for (i = 0; i < sizeof(utf8_ranges)/sizeof(utf8_range_t); ++i) {
+	PHP_HTTP_DUFF(sizeof(utf8_ranges)/sizeof(utf8_range_t),
 		if (utf8_ranges[i].start == ch) {
 			return 1;
 		} else if (utf8_ranges[i].start <= ch && utf8_ranges[i].end >= ch) {
@@ -611,7 +611,8 @@ static inline zend_bool isualpha(unsigned ch)
 			}
 			return 0;
 		}
-	}
+		++i;
+	);
 	return 0;
 }
 
@@ -624,6 +625,39 @@ static inline zend_bool isualnum(unsigned ch)
 	return isualpha(ch);
 }
 
+static inline size_t wctoutf16(unsigned short u16[2], unsigned wc)
+{
+	if (wc > 0x10ffff || (wc >= 0xd800 && wc <= 0xdfff)) {
+		return 0;
+	}
+
+	if (wc <= 0xffff) {
+		u16[0] = (unsigned short) wc;
+		return 1;
+	}
+
+	wc -= 0x10000;
+	u16[0] = (unsigned short) ((wc >> 10) + 0xd800);
+	u16[1] = (unsigned short) ((wc & 0x3ff) + 0xdc00);
+	return 2;
+}
+
+static inline size_t utf16towc(unsigned *wc, unsigned short *u16_str, size_t u16_len)
+{
+	if (u16_len < 1) {
+		return 0;
+	}
+	if (u16_str[0] - 0xd800 >= 0x800) {
+		*wc = u16_str[0];
+		return 1;
+	}
+	if (u16_len < 2 || (u16_str[0] & 0xfffffc00) != 0xd800 || (u16_str[1] & 0xfffffc00) != 0xdc00) {
+		return 0;
+	}
+	*wc = (u16_str[0] << 10) + u16_str[1] - 0x35fdc00;
+	return 2;
+}
+
 #endif	/* PHP_HTTP_UTF8_H */
 
 /*