Wed, 31 Dec 2014 06:09:35 +0100
Cloned upstream origin tor-browser at tor-browser-31.3.0esr-4.5-1-build1
revision ID fc1c9ff7c1b2defdbc039f12214767608f46423f for hacking purpose.
michael@0 | 1 | /* |
michael@0 | 2 | ******************************************************************************* |
michael@0 | 3 | * Copyright (C) 2001-2011, International Business Machines |
michael@0 | 4 | * Corporation and others. All Rights Reserved. |
michael@0 | 5 | ******************************************************************************* |
michael@0 | 6 | * |
michael@0 | 7 | * File ucoleitr.cpp |
michael@0 | 8 | * |
michael@0 | 9 | * Modification History: |
michael@0 | 10 | * |
michael@0 | 11 | * Date Name Description |
michael@0 | 12 | * 02/15/2001 synwee Modified all methods to process its own function |
michael@0 | 13 | * instead of calling the equivalent c++ api (coleitr.h) |
michael@0 | 14 | *******************************************************************************/ |
michael@0 | 15 | |
michael@0 | 16 | #ifndef UCOLEITR_H |
michael@0 | 17 | #define UCOLEITR_H |
michael@0 | 18 | |
michael@0 | 19 | #include "unicode/utypes.h" |
michael@0 | 20 | |
michael@0 | 21 | #if !UCONFIG_NO_COLLATION |
michael@0 | 22 | |
michael@0 | 23 | /** |
michael@0 | 24 | * This indicates an error has occured during processing or if no more CEs is |
michael@0 | 25 | * to be returned. |
michael@0 | 26 | * @stable ICU 2.0 |
michael@0 | 27 | */ |
michael@0 | 28 | #define UCOL_NULLORDER ((int32_t)0xFFFFFFFF) |
michael@0 | 29 | |
michael@0 | 30 | #ifndef U_HIDE_INTERNAL_API |
michael@0 | 31 | /** |
michael@0 | 32 | * This indicates an error has occured during processing or there are no more CEs |
michael@0 | 33 | * to be returned. |
michael@0 | 34 | * |
michael@0 | 35 | * @internal |
michael@0 | 36 | */ |
michael@0 | 37 | #define UCOL_PROCESSED_NULLORDER ((int64_t)U_INT64_MAX) |
michael@0 | 38 | #endif /* U_HIDE_INTERNAL_API */ |
michael@0 | 39 | |
michael@0 | 40 | #include "unicode/ucol.h" |
michael@0 | 41 | |
michael@0 | 42 | /** |
michael@0 | 43 | * The UCollationElements struct. |
michael@0 | 44 | * For usage in C programs. |
michael@0 | 45 | * @stable ICU 2.0 |
michael@0 | 46 | */ |
michael@0 | 47 | typedef struct UCollationElements UCollationElements; |
michael@0 | 48 | |
michael@0 | 49 | /** |
michael@0 | 50 | * \file |
michael@0 | 51 | * \brief C API: UCollationElements |
michael@0 | 52 | * |
michael@0 | 53 | * The UCollationElements API is used as an iterator to walk through each |
michael@0 | 54 | * character of an international string. Use the iterator to return the |
michael@0 | 55 | * ordering priority of the positioned character. The ordering priority of a |
michael@0 | 56 | * character, which we refer to as a key, defines how a character is collated |
michael@0 | 57 | * in the given collation object. |
michael@0 | 58 | * For example, consider the following in Spanish: |
michael@0 | 59 | * <pre> |
michael@0 | 60 | * . "ca" -> the first key is key('c') and second key is key('a'). |
michael@0 | 61 | * . "cha" -> the first key is key('ch') and second key is key('a'). |
michael@0 | 62 | * </pre> |
michael@0 | 63 | * And in German, |
michael@0 | 64 | * <pre> |
michael@0 | 65 | * . "<ae ligature>b"-> the first key is key('a'), the second key is key('e'), and |
michael@0 | 66 | * . the third key is key('b'). |
michael@0 | 67 | * </pre> |
michael@0 | 68 | * <p>Example of the iterator usage: (without error checking) |
michael@0 | 69 | * <pre> |
michael@0 | 70 | * . void CollationElementIterator_Example() |
michael@0 | 71 | * . { |
michael@0 | 72 | * . UChar *s; |
michael@0 | 73 | * . t_int32 order, primaryOrder; |
michael@0 | 74 | * . UCollationElements *c; |
michael@0 | 75 | * . UCollatorOld *coll; |
michael@0 | 76 | * . UErrorCode success = U_ZERO_ERROR; |
michael@0 | 77 | * . s=(UChar*)malloc(sizeof(UChar) * (strlen("This is a test")+1) ); |
michael@0 | 78 | * . u_uastrcpy(s, "This is a test"); |
michael@0 | 79 | * . coll = ucol_open(NULL, &success); |
michael@0 | 80 | * . c = ucol_openElements(coll, str, u_strlen(str), &status); |
michael@0 | 81 | * . order = ucol_next(c, &success); |
michael@0 | 82 | * . ucol_reset(c); |
michael@0 | 83 | * . order = ucol_prev(c, &success); |
michael@0 | 84 | * . free(s); |
michael@0 | 85 | * . ucol_close(coll); |
michael@0 | 86 | * . ucol_closeElements(c); |
michael@0 | 87 | * . } |
michael@0 | 88 | * </pre> |
michael@0 | 89 | * <p> |
michael@0 | 90 | * ucol_next() returns the collation order of the next. |
michael@0 | 91 | * ucol_prev() returns the collation order of the previous character. |
michael@0 | 92 | * The Collation Element Iterator moves only in one direction between calls to |
michael@0 | 93 | * ucol_reset. That is, ucol_next() and ucol_prev can not be inter-used. |
michael@0 | 94 | * Whenever ucol_prev is to be called after ucol_next() or vice versa, |
michael@0 | 95 | * ucol_reset has to be called first to reset the status, shifting pointers to |
michael@0 | 96 | * either the end or the start of the string. Hence at the next call of |
michael@0 | 97 | * ucol_prev or ucol_next, the first or last collation order will be returned. |
michael@0 | 98 | * If a change of direction is done without a ucol_reset, the result is |
michael@0 | 99 | * undefined. |
michael@0 | 100 | * The result of a forward iterate (ucol_next) and reversed result of the |
michael@0 | 101 | * backward iterate (ucol_prev) on the same string are equivalent, if |
michael@0 | 102 | * collation orders with the value UCOL_IGNORABLE are ignored. |
michael@0 | 103 | * Character based on the comparison level of the collator. A collation order |
michael@0 | 104 | * consists of primary order, secondary order and tertiary order. The data |
michael@0 | 105 | * type of the collation order is <strong>t_int32</strong>. |
michael@0 | 106 | * |
michael@0 | 107 | * @see UCollator |
michael@0 | 108 | */ |
michael@0 | 109 | |
michael@0 | 110 | /** |
michael@0 | 111 | * Open the collation elements for a string. |
michael@0 | 112 | * |
michael@0 | 113 | * @param coll The collator containing the desired collation rules. |
michael@0 | 114 | * @param text The text to iterate over. |
michael@0 | 115 | * @param textLength The number of characters in text, or -1 if null-terminated |
michael@0 | 116 | * @param status A pointer to an UErrorCode to receive any errors. |
michael@0 | 117 | * @return a struct containing collation element information |
michael@0 | 118 | * @stable ICU 2.0 |
michael@0 | 119 | */ |
michael@0 | 120 | U_STABLE UCollationElements* U_EXPORT2 |
michael@0 | 121 | ucol_openElements(const UCollator *coll, |
michael@0 | 122 | const UChar *text, |
michael@0 | 123 | int32_t textLength, |
michael@0 | 124 | UErrorCode *status); |
michael@0 | 125 | |
michael@0 | 126 | |
michael@0 | 127 | /** |
michael@0 | 128 | * get a hash code for a key... Not very useful! |
michael@0 | 129 | * @param key the given key. |
michael@0 | 130 | * @param length the size of the key array. |
michael@0 | 131 | * @return the hash code. |
michael@0 | 132 | * @stable ICU 2.0 |
michael@0 | 133 | */ |
michael@0 | 134 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 135 | ucol_keyHashCode(const uint8_t* key, int32_t length); |
michael@0 | 136 | |
michael@0 | 137 | /** |
michael@0 | 138 | * Close a UCollationElements. |
michael@0 | 139 | * Once closed, a UCollationElements may no longer be used. |
michael@0 | 140 | * @param elems The UCollationElements to close. |
michael@0 | 141 | * @stable ICU 2.0 |
michael@0 | 142 | */ |
michael@0 | 143 | U_STABLE void U_EXPORT2 |
michael@0 | 144 | ucol_closeElements(UCollationElements *elems); |
michael@0 | 145 | |
michael@0 | 146 | /** |
michael@0 | 147 | * Reset the collation elements to their initial state. |
michael@0 | 148 | * This will move the 'cursor' to the beginning of the text. |
michael@0 | 149 | * Property settings for collation will be reset to the current status. |
michael@0 | 150 | * @param elems The UCollationElements to reset. |
michael@0 | 151 | * @see ucol_next |
michael@0 | 152 | * @see ucol_previous |
michael@0 | 153 | * @stable ICU 2.0 |
michael@0 | 154 | */ |
michael@0 | 155 | U_STABLE void U_EXPORT2 |
michael@0 | 156 | ucol_reset(UCollationElements *elems); |
michael@0 | 157 | |
michael@0 | 158 | #ifndef U_HIDE_INTERNAL_API |
michael@0 | 159 | /** |
michael@0 | 160 | * Set the collation elements to use implicit ordering for Han |
michael@0 | 161 | * even if they've been tailored. This will also force Hangul |
michael@0 | 162 | * syllables to be ordered by decomposing them to their component |
michael@0 | 163 | * Jamo. |
michael@0 | 164 | * |
michael@0 | 165 | * @param elems The UCollationElements containing the text. |
michael@0 | 166 | * @param status A pointer to a UErrorCode to reveive any errors. |
michael@0 | 167 | * |
michael@0 | 168 | * @internal |
michael@0 | 169 | */ |
michael@0 | 170 | U_INTERNAL void U_EXPORT2 |
michael@0 | 171 | ucol_forceHanImplicit(UCollationElements *elems, UErrorCode *status); |
michael@0 | 172 | #endif /* U_HIDE_INTERNAL_API */ |
michael@0 | 173 | |
michael@0 | 174 | /** |
michael@0 | 175 | * Get the ordering priority of the next collation element in the text. |
michael@0 | 176 | * A single character may contain more than one collation element. |
michael@0 | 177 | * @param elems The UCollationElements containing the text. |
michael@0 | 178 | * @param status A pointer to an UErrorCode to receive any errors. |
michael@0 | 179 | * @return The next collation elements ordering, otherwise returns NULLORDER |
michael@0 | 180 | * if an error has occured or if the end of string has been reached |
michael@0 | 181 | * @stable ICU 2.0 |
michael@0 | 182 | */ |
michael@0 | 183 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 184 | ucol_next(UCollationElements *elems, UErrorCode *status); |
michael@0 | 185 | |
michael@0 | 186 | /** |
michael@0 | 187 | * Get the ordering priority of the previous collation element in the text. |
michael@0 | 188 | * A single character may contain more than one collation element. |
michael@0 | 189 | * Note that internally a stack is used to store buffered collation elements. |
michael@0 | 190 | * It is very rare that the stack will overflow, however if such a case is |
michael@0 | 191 | * encountered, the problem can be solved by increasing the size |
michael@0 | 192 | * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h. |
michael@0 | 193 | * @param elems The UCollationElements containing the text. |
michael@0 | 194 | * @param status A pointer to an UErrorCode to receive any errors. Noteably |
michael@0 | 195 | * a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack |
michael@0 | 196 | * buffer has been exhausted. |
michael@0 | 197 | * @return The previous collation elements ordering, otherwise returns |
michael@0 | 198 | * NULLORDER if an error has occured or if the start of string has |
michael@0 | 199 | * been reached. |
michael@0 | 200 | * @stable ICU 2.0 |
michael@0 | 201 | */ |
michael@0 | 202 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 203 | ucol_previous(UCollationElements *elems, UErrorCode *status); |
michael@0 | 204 | |
michael@0 | 205 | #ifndef U_HIDE_INTERNAL_API |
michael@0 | 206 | /** |
michael@0 | 207 | * Get the processed ordering priority of the next collation element in the text. |
michael@0 | 208 | * A single character may contain more than one collation element. |
michael@0 | 209 | * |
michael@0 | 210 | * @param elems The UCollationElements containing the text. |
michael@0 | 211 | * @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE. |
michael@0 | 212 | * @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE. |
michael@0 | 213 | * @param status A pointer to an UErrorCode to receive any errors. |
michael@0 | 214 | * @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER |
michael@0 | 215 | * if an error has occured or if the end of string has been reached |
michael@0 | 216 | * |
michael@0 | 217 | * @internal |
michael@0 | 218 | */ |
michael@0 | 219 | U_INTERNAL int64_t U_EXPORT2 |
michael@0 | 220 | ucol_nextProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status); |
michael@0 | 221 | |
michael@0 | 222 | /** |
michael@0 | 223 | * Get the processed ordering priority of the previous collation element in the text. |
michael@0 | 224 | * A single character may contain more than one collation element. |
michael@0 | 225 | * Note that internally a stack is used to store buffered collation elements. |
michael@0 | 226 | * It is very rare that the stack will overflow, however if such a case is |
michael@0 | 227 | * encountered, the problem can be solved by increasing the size |
michael@0 | 228 | * UCOL_EXPAND_CE_BUFFER_SIZE in ucol_imp.h. |
michael@0 | 229 | * |
michael@0 | 230 | * @param elems The UCollationElements containing the text. |
michael@0 | 231 | * @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE |
michael@0 | 232 | * @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE |
michael@0 | 233 | * @param status A pointer to an UErrorCode to receive any errors. Noteably |
michael@0 | 234 | * a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack |
michael@0 | 235 | * buffer has been exhausted. |
michael@0 | 236 | * @return The previous collation elements ordering, otherwise returns |
michael@0 | 237 | * UCOL_PROCESSED_NULLORDER if an error has occured or if the start of |
michael@0 | 238 | * string has been reached. |
michael@0 | 239 | * |
michael@0 | 240 | * @internal |
michael@0 | 241 | */ |
michael@0 | 242 | U_INTERNAL int64_t U_EXPORT2 |
michael@0 | 243 | ucol_previousProcessed(UCollationElements *elems, int32_t *ixLow, int32_t *ixHigh, UErrorCode *status); |
michael@0 | 244 | #endif /* U_HIDE_INTERNAL_API */ |
michael@0 | 245 | |
michael@0 | 246 | /** |
michael@0 | 247 | * Get the maximum length of any expansion sequences that end with the |
michael@0 | 248 | * specified comparison order. |
michael@0 | 249 | * This is useful for .... ? |
michael@0 | 250 | * @param elems The UCollationElements containing the text. |
michael@0 | 251 | * @param order A collation order returned by previous or next. |
michael@0 | 252 | * @return maximum size of the expansion sequences ending with the collation |
michael@0 | 253 | * element or 1 if collation element does not occur at the end of any |
michael@0 | 254 | * expansion sequence |
michael@0 | 255 | * @stable ICU 2.0 |
michael@0 | 256 | */ |
michael@0 | 257 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 258 | ucol_getMaxExpansion(const UCollationElements *elems, int32_t order); |
michael@0 | 259 | |
michael@0 | 260 | /** |
michael@0 | 261 | * Set the text containing the collation elements. |
michael@0 | 262 | * Property settings for collation will remain the same. |
michael@0 | 263 | * In order to reset the iterator to the current collation property settings, |
michael@0 | 264 | * the API reset() has to be called. |
michael@0 | 265 | * @param elems The UCollationElements to set. |
michael@0 | 266 | * @param text The source text containing the collation elements. |
michael@0 | 267 | * @param textLength The length of text, or -1 if null-terminated. |
michael@0 | 268 | * @param status A pointer to an UErrorCode to receive any errors. |
michael@0 | 269 | * @see ucol_getText |
michael@0 | 270 | * @stable ICU 2.0 |
michael@0 | 271 | */ |
michael@0 | 272 | U_STABLE void U_EXPORT2 |
michael@0 | 273 | ucol_setText( UCollationElements *elems, |
michael@0 | 274 | const UChar *text, |
michael@0 | 275 | int32_t textLength, |
michael@0 | 276 | UErrorCode *status); |
michael@0 | 277 | |
michael@0 | 278 | /** |
michael@0 | 279 | * Get the offset of the current source character. |
michael@0 | 280 | * This is an offset into the text of the character containing the current |
michael@0 | 281 | * collation elements. |
michael@0 | 282 | * @param elems The UCollationElements to query. |
michael@0 | 283 | * @return The offset of the current source character. |
michael@0 | 284 | * @see ucol_setOffset |
michael@0 | 285 | * @stable ICU 2.0 |
michael@0 | 286 | */ |
michael@0 | 287 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 288 | ucol_getOffset(const UCollationElements *elems); |
michael@0 | 289 | |
michael@0 | 290 | /** |
michael@0 | 291 | * Set the offset of the current source character. |
michael@0 | 292 | * This is an offset into the text of the character to be processed. |
michael@0 | 293 | * Property settings for collation will remain the same. |
michael@0 | 294 | * In order to reset the iterator to the current collation property settings, |
michael@0 | 295 | * the API reset() has to be called. |
michael@0 | 296 | * @param elems The UCollationElements to set. |
michael@0 | 297 | * @param offset The desired character offset. |
michael@0 | 298 | * @param status A pointer to an UErrorCode to receive any errors. |
michael@0 | 299 | * @see ucol_getOffset |
michael@0 | 300 | * @stable ICU 2.0 |
michael@0 | 301 | */ |
michael@0 | 302 | U_STABLE void U_EXPORT2 |
michael@0 | 303 | ucol_setOffset(UCollationElements *elems, |
michael@0 | 304 | int32_t offset, |
michael@0 | 305 | UErrorCode *status); |
michael@0 | 306 | |
michael@0 | 307 | /** |
michael@0 | 308 | * Get the primary order of a collation order. |
michael@0 | 309 | * @param order the collation order |
michael@0 | 310 | * @return the primary order of a collation order. |
michael@0 | 311 | * @stable ICU 2.6 |
michael@0 | 312 | */ |
michael@0 | 313 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 314 | ucol_primaryOrder (int32_t order); |
michael@0 | 315 | |
michael@0 | 316 | /** |
michael@0 | 317 | * Get the secondary order of a collation order. |
michael@0 | 318 | * @param order the collation order |
michael@0 | 319 | * @return the secondary order of a collation order. |
michael@0 | 320 | * @stable ICU 2.6 |
michael@0 | 321 | */ |
michael@0 | 322 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 323 | ucol_secondaryOrder (int32_t order); |
michael@0 | 324 | |
michael@0 | 325 | /** |
michael@0 | 326 | * Get the tertiary order of a collation order. |
michael@0 | 327 | * @param order the collation order |
michael@0 | 328 | * @return the tertiary order of a collation order. |
michael@0 | 329 | * @stable ICU 2.6 |
michael@0 | 330 | */ |
michael@0 | 331 | U_STABLE int32_t U_EXPORT2 |
michael@0 | 332 | ucol_tertiaryOrder (int32_t order); |
michael@0 | 333 | |
michael@0 | 334 | #endif /* #if !UCONFIG_NO_COLLATION */ |
michael@0 | 335 | |
michael@0 | 336 | #endif |