icuSources/common/ucasemap_imp.h

   1 // © 2017 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3
   4 // ucasemap_imp.h
   5 // created: 2017feb08 Markus W. Scherer
   6
   7 #ifndef __UCASEMAP_IMP_H__
   8 #define __UCASEMAP_IMP_H__
   9
  10 #include "unicode/utypes.h"
  11 #include "unicode/ucasemap.h"
  12 #include "ucase.h"
  13
  14 #ifndef U_COMPARE_IGNORE_CASE
  15 /* see also unorm.h */
  16 /**
  17  * Option bit for unorm_compare:
  18  * Perform case-insensitive comparison.
  19  */
  20 #define U_COMPARE_IGNORE_CASE       0x10000
  21 #endif
  22
  23 /**
  24  * Internal API, used by u_strcasecmp() etc.
  25  * Compare strings case-insensitively,
  26  * in code point order or code unit order.
  27  */
  28 U_CFUNC int32_t
  29 u_strcmpFold(const UChar *s1, int32_t length1,
  30              const UChar *s2, int32_t length2,
  31              uint32_t options,
  32              UErrorCode *pErrorCode);
  33
  34 /**
  35  * Interanl API, used for detecting length of
  36  * shared prefix case-insensitively.
  37  * @param s1            input string 1
  38  * @param length1       length of string 1, or -1 (NULL terminated)
  39  * @param s2            input string 2
  40  * @param length2       length of string 2, or -1 (NULL terminated)
  41  * @param options       compare options
  42  * @param matchLen1     (output) length of partial prefix match in s1
  43  * @param matchLen2     (output) length of partial prefix match in s2
  44  * @param pErrorCode    receives error status
  45  */
  46 U_CAPI void
  47 u_caseInsensitivePrefixMatch(const UChar *s1, int32_t length1,
  48                              const UChar *s2, int32_t length2,
  49                              uint32_t options,
  50                              int32_t *matchLen1, int32_t *matchLen2,
  51                              UErrorCode *pErrorCode);
  52
  53 /**
  54  * Are the Unicode properties loaded?
  55  * This must be used before internal functions are called that do
  56  * not perform this check.
  57  * Generate a debug assertion failure if data is not loaded.
  58  */
  59 U_CFUNC UBool
  60 uprv_haveProperties(UErrorCode *pErrorCode);
  61
  62 #ifdef __cplusplus
  63
  64 #include "unicode/unistr.h"  // for UStringCaseMapper
  65
  66 /*
  67  * Internal string casing functions implementing
  68  * ustring.h/ustrcase.cpp and UnicodeString case mapping functions.
  69  */
  70
  71 struct UCaseMap : public icu::UMemory {
  72     /** Implements most of ucasemap_open(). */
  73     UCaseMap(const char *localeID, uint32_t opts, UErrorCode *pErrorCode);
  74     ~UCaseMap();
  75
  76 #if !UCONFIG_NO_BREAK_ITERATION
  77     icu::BreakIterator *iter;  /* We adopt the iterator, so we own it. */
  78 #endif
  79     char locale[32];
  80     int32_t caseLocale;
  81     uint32_t options;
  82 };
  83
  84 #if UCONFIG_NO_BREAK_ITERATION
  85 #   define UCASEMAP_BREAK_ITERATOR_PARAM
  86 #   define UCASEMAP_BREAK_ITERATOR_UNUSED
  87 #   define UCASEMAP_BREAK_ITERATOR
  88 #   define UCASEMAP_BREAK_ITERATOR_NULL
  89 #else
  90 #   define UCASEMAP_BREAK_ITERATOR_PARAM icu::BreakIterator *iter,
  91 #   define UCASEMAP_BREAK_ITERATOR_UNUSED icu::BreakIterator *,
  92 #   define UCASEMAP_BREAK_ITERATOR iter,
  93 #   define UCASEMAP_BREAK_ITERATOR_NULL NULL,
  94 #endif
  95
  96 U_CFUNC int32_t
  97 ustrcase_getCaseLocale(const char *locale);
  98
  99 // TODO: swap src / dest if approved for new public api
 100 /** Implements UStringCaseMapper. */
 101 U_CFUNC int32_t U_CALLCONV
 102 ustrcase_internalToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 103                          UChar *dest, int32_t destCapacity,
 104                          const UChar *src, int32_t srcLength,
 105                          icu::Edits *edits,
 106                          UErrorCode &errorCode);
 107
 108 /** Implements UStringCaseMapper. */
 109 U_CFUNC int32_t U_CALLCONV
 110 ustrcase_internalToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 111                          UChar *dest, int32_t destCapacity,
 112                          const UChar *src, int32_t srcLength,
 113                          icu::Edits *edits,
 114                          UErrorCode &errorCode);
 115
 116 #if !UCONFIG_NO_BREAK_ITERATION
 117
 118 /** Implements UStringCaseMapper. */
 119 U_CFUNC int32_t U_CALLCONV
 120 ustrcase_internalToTitle(int32_t caseLocale, uint32_t options,
 121                          icu::BreakIterator *iter,
 122                          UChar *dest, int32_t destCapacity,
 123                          const UChar *src, int32_t srcLength,
 124                          icu::Edits *edits,
 125                          UErrorCode &errorCode);
 126
 127 #endif
 128
 129 /** Implements UStringCaseMapper. */
 130 U_CFUNC int32_t U_CALLCONV
 131 ustrcase_internalFold(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 132                       UChar *dest, int32_t destCapacity,
 133                       const UChar *src, int32_t srcLength,
 134                       icu::Edits *edits,
 135                       UErrorCode &errorCode);
 136
 137 /**
 138  * Common string case mapping implementation for ucasemap_toXyz() and UnicodeString::toXyz().
 139  * Implements argument checking.
 140  */
 141 U_CFUNC int32_t
 142 ustrcase_map(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 143              UChar *dest, int32_t destCapacity,
 144              const UChar *src, int32_t srcLength,
 145              UStringCaseMapper *stringCaseMapper,
 146              icu::Edits *edits,
 147              UErrorCode &errorCode);
 148
 149 /**
 150  * Common string case mapping implementation for old-fashioned u_strToXyz() functions
 151  * that allow the source string to overlap the destination buffer.
 152  * Implements argument checking and internally works with an intermediate buffer if necessary.
 153  */
 154 U_CFUNC int32_t
 155 ustrcase_mapWithOverlap(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 156                         UChar *dest, int32_t destCapacity,
 157                         const UChar *src, int32_t srcLength,
 158                         UStringCaseMapper *stringCaseMapper,
 159                         UErrorCode &errorCode);
 160
 161 /**
 162  * UTF-8 string case mapping function type, used by ucasemap_mapUTF8().
 163  * UTF-8 version of UStringCaseMapper.
 164  * All error checking must be done.
 165  * The UCaseMap must be fully initialized, with locale and/or iter set as needed.
 166  * src and dest must not overlap.
 167  */
 168 typedef int32_t U_CALLCONV
 169 UTF8CaseMapper(int32_t caseLocale, uint32_t options,
 170 #if !UCONFIG_NO_BREAK_ITERATION
 171                icu::BreakIterator *iter,
 172 #endif
 173                uint8_t *dest, int32_t destCapacity,
 174                const uint8_t *src, int32_t srcLength,
 175                icu::Edits *edits,
 176                UErrorCode &errorCode);
 177
 178 #if !UCONFIG_NO_BREAK_ITERATION
 179
 180 /** Implements UTF8CaseMapper. */
 181 U_CFUNC int32_t U_CALLCONV
 182 ucasemap_internalUTF8ToTitle(int32_t caseLocale, uint32_t options,
 183         icu::BreakIterator *iter,
 184         uint8_t *dest, int32_t destCapacity,
 185         const uint8_t *src, int32_t srcLength,
 186         icu::Edits *edits,
 187         UErrorCode &errorCode);
 188
 189 #endif
 190
 191 /**
 192  * Implements argument checking and buffer handling
 193  * for UTF-8 string case mapping as a common function.
 194  */
 195 U_CFUNC int32_t
 196 ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
 197                  uint8_t *dest, int32_t destCapacity,
 198                  const uint8_t *src, int32_t srcLength,
 199                  UTF8CaseMapper *stringCaseMapper,
 200                  icu::Edits *edits,
 201                  UErrorCode &errorCode);
 202
 203 U_NAMESPACE_BEGIN
 204 namespace GreekUpper {
 205
 206 // Data bits.
 207 static const uint32_t UPPER_MASK = 0x3ff;
 208 static const uint32_t HAS_VOWEL = 0x1000;
 209 static const uint32_t HAS_YPOGEGRAMMENI = 0x2000;
 210 static const uint32_t HAS_ACCENT = 0x4000;
 211 static const uint32_t HAS_DIALYTIKA = 0x8000;
 212 // Further bits during data building and processing, not stored in the data map.
 213 static const uint32_t HAS_COMBINING_DIALYTIKA = 0x10000;
 214 static const uint32_t HAS_OTHER_GREEK_DIACRITIC = 0x20000;
 215
 216 static const uint32_t HAS_VOWEL_AND_ACCENT = HAS_VOWEL | HAS_ACCENT;
 217 static const uint32_t HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA =
 218         HAS_VOWEL_AND_ACCENT | HAS_DIALYTIKA;
 219 static const uint32_t HAS_EITHER_DIALYTIKA = HAS_DIALYTIKA | HAS_COMBINING_DIALYTIKA;
 220
 221 // State bits.
 222 static const uint32_t AFTER_CASED = 1;
 223 static const uint32_t AFTER_VOWEL_WITH_ACCENT = 2;
 224
 225 uint32_t getLetterData(UChar32 c);
 226
 227 /**
 228  * Returns a non-zero value for each of the Greek combining diacritics
 229  * listed in The Unicode Standard, version 8, chapter 7.2 Greek,
 230  * plus some perispomeni look-alikes.
 231  */
 232 uint32_t getDiacriticData(UChar32 c);
 233
 234 }  // namespace GreekUpper
 235 U_NAMESPACE_END
 236
 237 #endif  // __cplusplus
 238
 239 #endif  // __UCASEMAP_IMP_H__