icuSources/common/uniset_closure.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2011, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  uniset_closure.cpp
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2011may30
  16 *   created by: Markus W. Scherer
  17 *
  18 *   UnicodeSet::closeOver() and related methods moved here from uniset_props.cpp
  19 *   to simplify dependencies.
  20 *   In particular, this depends on the BreakIterator, but the BreakIterator
  21 *   code also builds UnicodeSets from patterns and needs uniset_props.
  22 */
  23
  24 #include "unicode/brkiter.h"
  25 #include "unicode/locid.h"
  26 #include "unicode/parsepos.h"
  27 #include "unicode/uniset.h"
  28 #include "cmemory.h"
  29 #include "ruleiter.h"
  30 #include "ucase.h"
  31 #include "util.h"
  32 #include "uvector.h"
  33
  34 // initial storage. Must be >= 0
  35 // *** same as in uniset.cpp ! ***
  36 #define START_EXTRA 16
  37
  38 U_NAMESPACE_BEGIN
  39
  40 // TODO memory debugging provided inside uniset.cpp
  41 // could be made available here but probably obsolete with use of modern
  42 // memory leak checker tools
  43 #define _dbgct(me)
  44
  45 //----------------------------------------------------------------
  46 // Constructors &c
  47 //----------------------------------------------------------------
  48
  49 UnicodeSet::UnicodeSet(const UnicodeString& pattern,
  50                        uint32_t options,
  51                        const SymbolTable* symbols,
  52                        UErrorCode& status) :
  53     len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0),
  54     bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL),
  55     fFlags(0)
  56 {
  57     if(U_SUCCESS(status)){
  58         list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity);
  59         /* test for NULL */
  60         if(list == NULL) {
  61             status = U_MEMORY_ALLOCATION_ERROR;
  62         }else{
  63             allocateStrings(status);
  64             applyPattern(pattern, options, symbols, status);
  65         }
  66     }
  67     _dbgct(this);
  68 }
  69
  70 UnicodeSet::UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
  71                        uint32_t options,
  72                        const SymbolTable* symbols,
  73                        UErrorCode& status) :
  74     len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0),
  75     bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL),
  76     fFlags(0)
  77 {
  78     if(U_SUCCESS(status)){
  79         list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity);
  80         /* test for NULL */
  81         if(list == NULL) {
  82             status = U_MEMORY_ALLOCATION_ERROR;
  83         }else{
  84             allocateStrings(status);
  85             applyPattern(pattern, pos, options, symbols, status);
  86         }
  87     }
  88     _dbgct(this);
  89 }
  90
  91 //----------------------------------------------------------------
  92 // Public API
  93 //----------------------------------------------------------------
  94
  95 UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
  96                                      uint32_t options,
  97                                      const SymbolTable* symbols,
  98                                      UErrorCode& status) {
  99     ParsePosition pos(0);
 100     applyPattern(pattern, pos, options, symbols, status);
 101     if (U_FAILURE(status)) return *this;
 102
 103     int32_t i = pos.getIndex();
 104
 105     if (options & USET_IGNORE_SPACE) {
 106         // Skip over trailing whitespace
 107         ICU_Utility::skipWhitespace(pattern, i, TRUE);
 108     }
 109
 110     if (i != pattern.length()) {
 111         status = U_ILLEGAL_ARGUMENT_ERROR;
 112     }
 113     return *this;
 114 }
 115
 116 UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
 117                               ParsePosition& pos,
 118                               uint32_t options,
 119                               const SymbolTable* symbols,
 120                               UErrorCode& status) {
 121     if (U_FAILURE(status)) {
 122         return *this;
 123     }
 124     if (isFrozen()) {
 125         status = U_NO_WRITE_PERMISSION;
 126         return *this;
 127     }
 128     // Need to build the pattern in a temporary string because
 129     // _applyPattern calls add() etc., which set pat to empty.
 130     UnicodeString rebuiltPat;
 131     RuleCharacterIterator chars(pattern, symbols, pos);
 132     applyPattern(chars, symbols, rebuiltPat, options, &UnicodeSet::closeOver, 0, status);
 133     if (U_FAILURE(status)) return *this;
 134     if (chars.inVariable()) {
 135         // syntaxError(chars, "Extra chars in variable value");
 136         status = U_MALFORMED_SET;
 137         return *this;
 138     }
 139     setPattern(rebuiltPat);
 140     return *this;
 141 }
 142
 143 // USetAdder implementation
 144 // Does not use uset.h to reduce code dependencies
 145 static void U_CALLCONV
 146 _set_add(USet *set, UChar32 c) {
 147     ((UnicodeSet *)set)->add(c);
 148 }
 149
 150 static void U_CALLCONV
 151 _set_addRange(USet *set, UChar32 start, UChar32 end) {
 152     ((UnicodeSet *)set)->add(start, end);
 153 }
 154
 155 static void U_CALLCONV
 156 _set_addString(USet *set, const UChar *str, int32_t length) {
 157     ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length));
 158 }
 159
 160 //----------------------------------------------------------------
 161 // Case folding API
 162 //----------------------------------------------------------------
 163
 164 // add the result of a full case mapping to the set
 165 // use str as a temporary string to avoid constructing one
 166 static inline void
 167 addCaseMapping(UnicodeSet &set, int32_t result, const UChar *full, UnicodeString &str) {
 168     if(result >= 0) {
 169         if(result > UCASE_MAX_STRING_LENGTH) {
 170             // add a single-code point case mapping
 171             set.add(result);
 172         } else {
 173             // add a string case mapping from full with length result
 174             str.setTo((UBool)FALSE, full, result);
 175             set.add(str);
 176         }
 177     }
 178     // result < 0: the code point mapped to itself, no need to add it
 179     // see ucase.h
 180 }
 181
 182 UnicodeSet& UnicodeSet::closeOver(int32_t attribute) {
 183     if (isFrozen() || isBogus()) {
 184         return *this;
 185     }
 186     if (attribute & (USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS)) {
 187         {
 188             UnicodeSet foldSet(*this);
 189             UnicodeString str;
 190             USetAdder sa = {
 191                 foldSet.toUSet(),
 192                 _set_add,
 193                 _set_addRange,
 194                 _set_addString,
 195                 NULL, // don't need remove()
 196                 NULL // don't need removeRange()
 197             };
 198
 199             // start with input set to guarantee inclusion
 200             // USET_CASE: remove strings because the strings will actually be reduced (folded);
 201             //            therefore, start with no strings and add only those needed
 202             if (attribute & USET_CASE_INSENSITIVE) {
 203                 foldSet.strings->removeAllElements();
 204             }
 205
 206             int32_t n = getRangeCount();
 207             UChar32 result;
 208             const UChar *full;
 209
 210             for (int32_t i=0; i<n; ++i) {
 211                 UChar32 start = getRangeStart(i);
 212                 UChar32 end   = getRangeEnd(i);
 213
 214                 if (attribute & USET_CASE_INSENSITIVE) {
 215                     // full case closure
 216                     for (UChar32 cp=start; cp<=end; ++cp) {
 217                         ucase_addCaseClosure(cp, &sa);
 218                     }
 219                 } else {
 220                     // add case mappings
 221                     // (does not add long s for regular s, or Kelvin for k, for example)
 222                     for (UChar32 cp=start; cp<=end; ++cp) {
 223                         result = ucase_toFullLower(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 224                         addCaseMapping(foldSet, result, full, str);
 225
 226                         result = ucase_toFullTitle(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 227                         addCaseMapping(foldSet, result, full, str);
 228
 229                         result = ucase_toFullUpper(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 230                         addCaseMapping(foldSet, result, full, str);
 231
 232                         result = ucase_toFullFolding(cp, &full, 0);
 233                         addCaseMapping(foldSet, result, full, str);
 234                     }
 235                 }
 236             }
 237             if (strings != NULL && strings->size() > 0) {
 238                 if (attribute & USET_CASE_INSENSITIVE) {
 239                     for (int32_t j=0; j<strings->size(); ++j) {
 240                         str = *(const UnicodeString *) strings->elementAt(j);
 241                         str.foldCase();
 242                         if(!ucase_addStringCaseClosure(str.getBuffer(), str.length(), &sa)) {
 243                             foldSet.add(str); // does not map to code points: add the folded string itself
 244                         }
 245                     }
 246                 } else {
 247                     Locale root("");
 248 #if !UCONFIG_NO_BREAK_ITERATION
 249                     UErrorCode status = U_ZERO_ERROR;
 250                     BreakIterator *bi = BreakIterator::createWordInstance(root, status);
 251                     if (U_SUCCESS(status)) {
 252 #endif
 253                         const UnicodeString *pStr;
 254
 255                         for (int32_t j=0; j<strings->size(); ++j) {
 256                             pStr = (const UnicodeString *) strings->elementAt(j);
 257                             (str = *pStr).toLower(root);
 258                             foldSet.add(str);
 259 #if !UCONFIG_NO_BREAK_ITERATION
 260                             (str = *pStr).toTitle(bi, root);
 261                             foldSet.add(str);
 262 #endif
 263                             (str = *pStr).toUpper(root);
 264                             foldSet.add(str);
 265                             (str = *pStr).foldCase();
 266                             foldSet.add(str);
 267                         }
 268 #if !UCONFIG_NO_BREAK_ITERATION
 269                     }
 270                     delete bi;
 271 #endif
 272                 }
 273             }
 274             *this = foldSet;
 275         }
 276     }
 277     return *this;
 278 }
 279
 280 U_NAMESPACE_END