ICU-64260.0.1.tar.gz

[apple/icu.git] / icuSources / common / ustr_imp.h

diff --git a/icuSources/common/ustr_imp.h b/icuSources/common/ustr_imp.h
index 9815915ff5294589f6845dd8e0f3134c317b6cf1..34ca2f7428283685565b7be1fc85980ed804904f 100644 (file)
--- a/icuSources/common/ustr_imp.h
+++ b/icuSources/common/ustr_imp.h
@@ -18,6 +18,7 @@
 #define __USTR_IMP_H__
 
 #include "unicode/utypes.h"
+#include "unicode/utf8.h"
 
 /**
  * Internal option for unorm_cmpEquivFold() for strncmp style.
@@ -81,4 +82,75 @@ u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorC
 U_CAPI int32_t U_EXPORT2
 u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
 
+/**
+ * Counts the bytes of any whole valid sequence for a UTF-8 lead byte.
+ * Returns 1 for ASCII 0..0x7f.
+ * Returns 0 for 0x80..0xc1 as well as for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @return 0..4
+ */
+#define U8_COUNT_BYTES(leadByte) \
+    (U8_IS_SINGLE(leadByte) ? 1 : U8_COUNT_BYTES_NON_ASCII(leadByte))
+
+/**
+ * Counts the bytes of any whole valid sequence for a UTF-8 lead byte.
+ * Returns 0 for 0x00..0xc1 as well as for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @return 0 or 2..4
+ */
+#define U8_COUNT_BYTES_NON_ASCII(leadByte) \
+    (U8_IS_LEAD(leadByte) ? ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+2 : 0)
+
+#ifdef __cplusplus
+
+U_NAMESPACE_BEGIN
+
+class UTF8 {
+public:
+    UTF8() = delete;  // all static
+
+    /**
+     * Is t a valid UTF-8 trail byte?
+     *
+     * @param prev Must be the preceding lead byte if i==1 and length>=3;
+     *             otherwise ignored.
+     * @param t The i-th byte following the lead byte.
+     * @param i The index (1..3) of byte t in the byte sequence. 0<i<length
+     * @param length The length (2..4) of the byte sequence according to the lead byte.
+     * @return TRUE if t is a valid trail byte in this context.
+     */
+    static inline UBool isValidTrail(int32_t prev, uint8_t t, int32_t i, int32_t length) {
+        // The first trail byte after a 3- or 4-byte lead byte
+        // needs to be validated together with its lead byte.
+        if (length <= 2 || i > 1) {
+            return U8_IS_TRAIL(t);
+        } else if (length == 3) {
+            return U8_IS_VALID_LEAD3_AND_T1(prev, t);
+        } else {  // length == 4
+            return U8_IS_VALID_LEAD4_AND_T1(prev, t);
+        }
+    }
+};
+
+U_NAMESPACE_END
+
+#endif  // __cplusplus
+
+/**
+ * Check whether a code point has the emoji property.
+ * (implemented in uchar.cpp as reduced version of u_hasBinaryProperty)
+ *
+ * @param c Code point to test.
+ * @return TRUE or FALSE according to whether c has the emoji property.
+ *
+ * @internal Apple only
+ */
+U_CAPI UBool U_EXPORT2
+u_isEmoji(UChar32 c);
+
+
 #endif