icuSources/i18n/casetrn.cpp

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2001-2004, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  casetrn.cpp
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2004sep03
  14 *   created by: Markus W. Scherer
  15 *
  16 *   Implementation class for lower-/upper-/title-casing transliterators.
  17 */
  18
  19 #include "unicode/utypes.h"
  20
  21 #if !UCONFIG_NO_TRANSLITERATION
  22
  23 #include "unicode/uchar.h"
  24 #include "unicode/ustring.h"
  25 #include "tolowtrn.h"
  26 #include "ucase.h"
  27 #include "cpputils.h"
  28
  29 /* case context iterator using a Replaceable */
  30 U_CFUNC UChar32 U_CALLCONV
  31 utrans_rep_caseContextIterator(void *context, int8_t dir)
  32 {
  33     UCaseContext *csc=(UCaseContext *)context;
  34     Replaceable *rep=(Replaceable *)csc->p;
  35     UChar32 c;
  36
  37     if(dir<0) {
  38         /* reset for backward iteration */
  39         csc->index=csc->cpStart;
  40         csc->dir=dir;
  41     } else if(dir>0) {
  42         /* reset for forward iteration */
  43         csc->index=csc->cpLimit;
  44         csc->dir=dir;
  45     } else {
  46         /* continue current iteration direction */
  47         dir=csc->dir;
  48     }
  49
  50     // automatically adjust start and limit if the Replaceable disagrees
  51     // with the original values
  52     if(dir<0) {
  53         if(csc->start<csc->index) {
  54             c=rep->char32At(csc->index-1);
  55             if(c<0) {
  56                 csc->start=csc->index;
  57             } else {
  58                 csc->index-=U16_LENGTH(c);
  59                 return c;
  60             }
  61         }
  62     } else {
  63         // detect, and store in csc->b1, if we hit the limit
  64         if(csc->index<csc->limit) {
  65             c=rep->char32At(csc->index);
  66             if(c<0) {
  67                 csc->limit=csc->index;
  68                 csc->b1=TRUE;
  69             } else {
  70                 csc->index+=U16_LENGTH(c);
  71                 return c;
  72             }
  73         } else {
  74             csc->b1=TRUE;
  75         }
  76     }
  77     return U_SENTINEL;
  78 }
  79
  80 U_NAMESPACE_BEGIN
  81
  82 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CaseMapTransliterator)
  83
  84 /**
  85  * Constructs a transliterator.
  86  */
  87 CaseMapTransliterator::CaseMapTransliterator(const Locale &loc, const UnicodeString &id, UCaseMapFull *map) :
  88     Transliterator(id, 0),
  89     fLoc(loc), fLocName(NULL),
  90     fCsp(NULL),
  91     fMap(map)
  92 {
  93     UErrorCode errorCode = U_ZERO_ERROR;
  94     fCsp = ucase_getSingleton(&errorCode); // expect to get NULL if failure
  95     fLocName=fLoc.getName();
  96
  97     // TODO test incremental mode with context-sensitive text (e.g. greek sigma)
  98     // TODO need to call setMaximumContextLength()?!
  99 }
 100
 101 /**
 102  * Destructor.
 103  */
 104 CaseMapTransliterator::~CaseMapTransliterator() {
 105 }
 106
 107 /**
 108  * Copy constructor.
 109  */
 110 CaseMapTransliterator::CaseMapTransliterator(const CaseMapTransliterator& o) :
 111     Transliterator(o),
 112     fLoc(o.fLoc), fLocName(NULL), fCsp(o.fCsp), fMap(o.fMap)
 113 {
 114     fLocName=fLoc.getName();
 115 }
 116
 117 /**
 118  * Assignment operator.
 119  */
 120 CaseMapTransliterator& CaseMapTransliterator::operator=(const CaseMapTransliterator& o) {
 121     Transliterator::operator=(o);
 122     fLoc = o.fLoc;
 123     fLocName = fLoc.getName();
 124     fCsp = o.fCsp;
 125     fMap = o.fMap;
 126     return *this;
 127 }
 128
 129 /**
 130  * Transliterator API.
 131  */
 132 Transliterator* CaseMapTransliterator::clone(void) const {
 133     return new CaseMapTransliterator(*this);
 134 }
 135
 136 /**
 137  * Implements {@link Transliterator#handleTransliterate}.
 138  */
 139 void CaseMapTransliterator::handleTransliterate(Replaceable& text,
 140                                  UTransPosition& offsets,
 141                                  UBool isIncremental) const
 142 {
 143     if (offsets.start >= offsets.limit) {
 144         return;
 145     }
 146
 147     UCaseContext csc;
 148     uprv_memset(&csc, 0, sizeof(csc));
 149     csc.p = &text;
 150     csc.start = offsets.contextStart;
 151     csc.limit = offsets.contextLimit;
 152
 153     UnicodeString tmp;
 154     const UChar *s;
 155     UChar32 c;
 156     int32_t textPos, delta, result, locCache=0;
 157
 158     for(textPos=offsets.start; textPos<offsets.limit;) {
 159         csc.cpStart=textPos;
 160         c=text.char32At(textPos);
 161         csc.cpLimit=textPos+=U16_LENGTH(c);
 162
 163         result=fMap(fCsp, c, utrans_rep_caseContextIterator, &csc, &s, fLocName, &locCache);
 164
 165         if(csc.b1 && isIncremental) {
 166             // fMap() tried to look beyond the context limit
 167             // wait for more input
 168             break;
 169         }
 170
 171         if(result>=0) {
 172             // replace the current code point with its full case mapping result
 173             // see UCASE_MAX_STRING_LENGTH
 174             if(result<=UCASE_MAX_STRING_LENGTH) {
 175                 // string s[result]
 176                 tmp.setTo(FALSE, s, result);
 177                 delta=result-U16_LENGTH(c);
 178             } else {
 179                 // single code point
 180                 tmp.setTo(result);
 181                 delta=tmp.length()-U16_LENGTH(c);
 182             }
 183             text.handleReplaceBetween(csc.cpStart, textPos, tmp);
 184             if(delta!=0) {
 185                 textPos+=delta;
 186                 csc.limit=offsets.contextLimit+=delta;
 187                 offsets.limit+=delta;
 188             }
 189         }
 190     }
 191     offsets.start=textPos;
 192 }
 193
 194 U_NAMESPACE_END
 195
 196 #endif /* #if !UCONFIG_NO_TRANSLITERATION */