ICU-64243.0.1.tar.gz

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

diff --git a/icuSources/common/ucase.h b/icuSources/common/ucase.h
index e15bae6604daef02f06df71ec5594666ce56411d..b0a453b87e8afa9936f4ef6780548ee391fb5bd0 100644 (file)
--- a/icuSources/common/ucase.h
+++ b/icuSources/common/ucase.h
@@ -26,6 +26,7 @@
 #include "putilimp.h"
 #include "uset_imp.h"
 #include "udataswp.h"
+#include "utrie2.h"
 
 #ifdef __cplusplus
 U_NAMESPACE_BEGIN
@@ -61,7 +62,7 @@ enum {
 /**
  * Bit mask for getting just the options from a string compare options word
  * that are relevant for case-insensitive string comparison.
- * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
+ * See stringoptions.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
  * @internal
  */
 #define _STRCASECMP_OPTIONS_MASK 0xffff
@@ -69,10 +70,16 @@ enum {
 /**
  * Bit mask for getting just the options from a string compare options word
  * that are relevant for case folding (of a single string or code point).
- * See uchar.h.
+ *
+ * Currently only bit 0 for U_FOLD_CASE_EXCLUDE_SPECIAL_I.
+ * It is conceivable that at some point we might use one more bit for using uppercase sharp s.
+ * It is conceivable that at some point we might want the option to use only simple case foldings
+ * when operating on strings.
+ *
+ * See stringoptions.h.
  * @internal
  */
-#define _FOLD_CASE_OPTIONS_MASK 0xff
+#define _FOLD_CASE_OPTIONS_MASK 7
 
 /* single-code point functions */
 
@@ -142,6 +149,33 @@ private:
     int32_t rowCpIndex;
 };
 
+/**
+ * Fast case mapping data for ASCII/Latin.
+ * Linear arrays of delta bytes: 0=no mapping; EXC=exception.
+ * Deltas must not cross the ASCII boundary, or else they cannot be easily used
+ * in simple UTF-8 code.
+ */
+namespace LatinCase {
+
+/** Case mapping/folding data for code points up to U+017F. */
+constexpr UChar LIMIT = 0x180;
+/** U+017F case-folds and uppercases crossing the ASCII boundary. */
+constexpr UChar LONG_S = 0x17f;
+/** Exception: Complex mapping, or too-large delta. */
+constexpr int8_t EXC = -0x80;
+
+/** Deltas for lowercasing for most locales, and default case folding. */
+extern const int8_t TO_LOWER_NORMAL[LIMIT];
+/** Deltas for lowercasing for tr/az/lt, and Turkic case folding. */
+extern const int8_t TO_LOWER_TR_LT[LIMIT];
+
+/** Deltas for uppercasing for most locales. */
+extern const int8_t TO_UPPER_NORMAL[LIMIT];
+/** Deltas for uppercasing for tr/az. */
+extern const int8_t TO_UPPER_TR[LIMIT];
+
+}  // namespace LatinCase
+
 U_NAMESPACE_END
 #endif
 
@@ -302,6 +336,9 @@ enum {
 
 /* definitions for 16-bit case properties word ------------------------------ */
 
+U_CFUNC const UTrie2 * U_EXPORT2
+ucase_getTrie();
+
 /* 2-bit constants for types of cased characters */
 #define UCASE_TYPE_MASK     3
 enum {
@@ -314,9 +351,13 @@ enum {
 #define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
 #define UCASE_GET_TYPE_AND_IGNORABLE(props) ((props)&7)
 
+#define UCASE_IS_UPPER_OR_TITLE(props) ((props)&2)
+
 #define UCASE_IGNORABLE         4
-#define UCASE_SENSITIVE         8
-#define UCASE_EXCEPTION         0x10
+#define UCASE_EXCEPTION         8
+#define UCASE_SENSITIVE         0x10
+
+#define UCASE_HAS_EXCEPTION(props) ((props)&UCASE_EXCEPTION)
 
 #define UCASE_DOT_MASK      0x60
 enum {
@@ -338,9 +379,9 @@ enum {
 #   define UCASE_GET_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UCASE_DELTA_SHIFT)|0xfe00) : ((uint16_t)(props)>>UCASE_DELTA_SHIFT))
 #endif
 
-/* exception: bits 15..5 are an unsigned 11-bit index into the exceptions array */
-#define UCASE_EXC_SHIFT     5
-#define UCASE_EXC_MASK      0xffe0
+/* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */
+#define UCASE_EXC_SHIFT     4
+#define UCASE_EXC_MASK      0xfff0
 #define UCASE_MAX_EXCEPTIONS ((UCASE_EXC_MASK>>UCASE_EXC_SHIFT)+1)
 
 /* definitions for 16-bit main exceptions word ------------------------------ */
@@ -351,7 +392,7 @@ enum {
     UCASE_EXC_FOLD,
     UCASE_EXC_UPPER,
     UCASE_EXC_TITLE,
-    UCASE_EXC_4,            /* reserved */
+    UCASE_EXC_DELTA,
     UCASE_EXC_5,            /* reserved */
     UCASE_EXC_CLOSURE,
     UCASE_EXC_FULL_MAPPINGS,
@@ -361,7 +402,11 @@ enum {
 /* each slot is 2 uint16_t instead of 1 */
 #define UCASE_EXC_DOUBLE_SLOTS      0x100
 
-/* reserved: exception bits 11..9 */
+enum {
+    UCASE_EXC_NO_SIMPLE_CASE_FOLDING=0x200,
+    UCASE_EXC_DELTA_IS_NEGATIVE=0x400,
+    UCASE_EXC_SENSITIVE=0x800
+};
 
 /* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
 #define UCASE_EXC_DOT_SHIFT     7