icuSources/common/uniset_closure.cpp

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