diff options
Diffstat (limited to 'deps/icu-small/source/common/ucase.h')
| -rw-r--r-- | deps/icu-small/source/common/ucase.h | 35 |
1 files changed, 35 insertions, 0 deletions
diff --git a/deps/icu-small/source/common/ucase.h b/deps/icu-small/source/common/ucase.h index 9d6365eadf..a7a8c9f00d 100644 --- a/deps/icu-small/source/common/ucase.h +++ b/deps/icu-small/source/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 @@ -148,6 +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 @@ -308,6 +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 { @@ -320,10 +351,14 @@ 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_HAS_EXCEPTION(props) ((props)&UCASE_EXCEPTION) + #define UCASE_DOT_MASK 0x60 enum { UCASE_NO_DOT=0, /* normal characters with cc=0 */ |