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 U_NAMESPACE_BEGIN
  35
  36 // TODO memory debugging provided inside uniset.cpp
  37 // could be made available here but probably obsolete with use of modern
  38 // memory leak checker tools
  39 #define _dbgct(me)
  40
  41 //----------------------------------------------------------------
  42 // Constructors &c
  43 //----------------------------------------------------------------
  44
  45 UnicodeSet::UnicodeSet(const UnicodeString& pattern,
  46                        uint32_t options,
  47                        const SymbolTable* symbols,
  48                        UErrorCode& status) {
  49     applyPattern(pattern, options, symbols, status);
  50     _dbgct(this);
  51 }
  52
  53 UnicodeSet::UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
  54                        uint32_t options,
  55                        const SymbolTable* symbols,
  56                        UErrorCode& status) {
  57     applyPattern(pattern, pos, options, symbols, status);
  58     _dbgct(this);
  59 }
  60
  61 //----------------------------------------------------------------
  62 // Public API
  63 //----------------------------------------------------------------
  64
  65 UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
  66                                      uint32_t options,
  67                                      const SymbolTable* symbols,
  68                                      UErrorCode& status) {
  69     ParsePosition pos(0);
  70     applyPattern(pattern, pos, options, symbols, status);
  71     if (U_FAILURE(status)) return *this;
  72
  73     int32_t i = pos.getIndex();
  74
  75     if (options & USET_IGNORE_SPACE) {
  76         // Skip over trailing whitespace
  77         ICU_Utility::skipWhitespace(pattern, i, TRUE);
  78     }
  79
  80     if (i != pattern.length()) {
  81         status = U_ILLEGAL_ARGUMENT_ERROR;
  82     }
  83     return *this;
  84 }
  85
  86 UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
  87                               ParsePosition& pos,
  88                               uint32_t options,
  89                               const SymbolTable* symbols,
  90                               UErrorCode& status) {
  91     if (U_FAILURE(status)) {
  92         return *this;
  93     }
  94     if (isFrozen()) {
  95         status = U_NO_WRITE_PERMISSION;
  96         return *this;
  97     }
  98     // Need to build the pattern in a temporary string because
  99     // _applyPattern calls add() etc., which set pat to empty.
 100     UnicodeString rebuiltPat;
 101     RuleCharacterIterator chars(pattern, symbols, pos);
 102     applyPattern(chars, symbols, rebuiltPat, options, &UnicodeSet::closeOver, 0, status);
 103     if (U_FAILURE(status)) return *this;
 104     if (chars.inVariable()) {
 105         // syntaxError(chars, "Extra chars in variable value");
 106         status = U_MALFORMED_SET;
 107         return *this;
 108     }
 109     setPattern(rebuiltPat);
 110     return *this;
 111 }
 112
 113 // USetAdder implementation
 114 // Does not use uset.h to reduce code dependencies
 115 static void U_CALLCONV
 116 _set_add(USet *set, UChar32 c) {
 117     ((UnicodeSet *)set)->add(c);
 118 }
 119
 120 static void U_CALLCONV
 121 _set_addRange(USet *set, UChar32 start, UChar32 end) {
 122     ((UnicodeSet *)set)->add(start, end);
 123 }
 124
 125 static void U_CALLCONV
 126 _set_addString(USet *set, const UChar *str, int32_t length) {
 127     ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length));
 128 }
 129
 130 //----------------------------------------------------------------
 131 // Case folding API
 132 //----------------------------------------------------------------
 133
 134 // add the result of a full case mapping to the set
 135 // use str as a temporary string to avoid constructing one
 136 static inline void
 137 addCaseMapping(UnicodeSet &set, int32_t result, const UChar *full, UnicodeString &str) {
 138     if(result >= 0) {
 139         if(result > UCASE_MAX_STRING_LENGTH) {
 140             // add a single-code point case mapping
 141             set.add(result);
 142         } else {
 143             // add a string case mapping from full with length result
 144             str.setTo((UBool)FALSE, full, result);
 145             set.add(str);
 146         }
 147     }
 148     // result < 0: the code point mapped to itself, no need to add it
 149     // see ucase.h
 150 }
 151
 152 UnicodeSet& UnicodeSet::closeOver(int32_t attribute) {
 153     if (isFrozen() || isBogus()) {
 154         return *this;
 155     }
 156     if (attribute & (USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS)) {
 157         {
 158             UnicodeSet foldSet(*this);
 159             UnicodeString str;
 160             USetAdder sa = {
 161                 foldSet.toUSet(),
 162                 _set_add,
 163                 _set_addRange,
 164                 _set_addString,
 165                 NULL, // don't need remove()
 166                 NULL // don't need removeRange()
 167             };
 168
 169             // start with input set to guarantee inclusion
 170             // USET_CASE: remove strings because the strings will actually be reduced (folded);
 171             //            therefore, start with no strings and add only those needed
 172             if ((attribute & USET_CASE_INSENSITIVE) && foldSet.hasStrings()) {
 173                 foldSet.strings->removeAllElements();
 174             }
 175
 176             int32_t n = getRangeCount();
 177             UChar32 result;
 178             const UChar *full;
 179
 180             for (int32_t i=0; i<n; ++i) {
 181                 UChar32 start = getRangeStart(i);
 182                 UChar32 end   = getRangeEnd(i);
 183
 184                 if (attribute & USET_CASE_INSENSITIVE) {
 185                     // full case closure
 186                     for (UChar32 cp=start; cp<=end; ++cp) {
 187                         ucase_addCaseClosure(cp, &sa);
 188                     }
 189                 } else {
 190                     // add case mappings
 191                     // (does not add long s for regular s, or Kelvin for k, for example)
 192                     for (UChar32 cp=start; cp<=end; ++cp) {
 193                         result = ucase_toFullLower(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 194                         addCaseMapping(foldSet, result, full, str);
 195
 196                         result = ucase_toFullTitle(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 197                         addCaseMapping(foldSet, result, full, str);
 198
 199                         result = ucase_toFullUpper(cp, NULL, NULL, &full, UCASE_LOC_ROOT);
 200                         addCaseMapping(foldSet, result, full, str);
 201
 202                         result = ucase_toFullFolding(cp, &full, 0);
 203                         addCaseMapping(foldSet, result, full, str);
 204                     }
 205                 }
 206             }
 207             if (hasStrings()) {
 208                 if (attribute & USET_CASE_INSENSITIVE) {
 209                     for (int32_t j=0; j<strings->size(); ++j) {
 210                         str = *(const UnicodeString *) strings->elementAt(j);
 211                         str.foldCase();
 212                         if(!ucase_addStringCaseClosure(str.getBuffer(), str.length(), &sa)) {
 213                             foldSet.add(str); // does not map to code points: add the folded string itself
 214                         }
 215                     }
 216                 } else {
 217                     Locale root("");
 218 #if !UCONFIG_NO_BREAK_ITERATION
 219                     UErrorCode status = U_ZERO_ERROR;
 220                     BreakIterator *bi = BreakIterator::createWordInstance(root, status);
 221                     if (U_SUCCESS(status)) {
 222 #endif
 223                         const UnicodeString *pStr;
 224
 225                         for (int32_t j=0; j<strings->size(); ++j) {
 226                             pStr = (const UnicodeString *) strings->elementAt(j);
 227                             (str = *pStr).toLower(root);
 228                             foldSet.add(str);
 229 #if !UCONFIG_NO_BREAK_ITERATION
 230                             (str = *pStr).toTitle(bi, root);
 231                             foldSet.add(str);
 232 #endif
 233                             (str = *pStr).toUpper(root);
 234                             foldSet.add(str);
 235                             (str = *pStr).foldCase();
 236                             foldSet.add(str);
 237                         }
 238 #if !UCONFIG_NO_BREAK_ITERATION
 239                     }
 240                     delete bi;
 241 #endif
 242                 }
 243             }
 244             *this = foldSet;
 245         }
 246     }
 247     return *this;
 248 }
 249
 250 U_NAMESPACE_END