icuSources/samples/translit/unaccent.cpp

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1999-2003, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6
   7 #include "unaccent.h"
   8
   9 const char UnaccentTransliterator::fgClassID = 0;
  10
  11 /**
  12  * Constructor
  13  */
  14 UnaccentTransliterator::UnaccentTransliterator() :
  15     normalizer("", UNORM_NFD),
  16     Transliterator("Unaccent", 0) {
  17 }
  18
  19 /**
  20  * Destructor
  21  */
  22 UnaccentTransliterator::~UnaccentTransliterator() {
  23 }
  24
  25 /**
  26  * Remove accents from a character using Normalizer.
  27  */
  28 UChar UnaccentTransliterator::unaccent(UChar c) const {
  29     UnicodeString str(c);
  30     UErrorCode status = U_ZERO_ERROR;
  31     UnaccentTransliterator* t = (UnaccentTransliterator*)this;
  32
  33     t->normalizer.setText(str, status);
  34     if (U_FAILURE(status)) {
  35         return c;
  36     }
  37     return (UChar) t->normalizer.next();
  38 }
  39
  40 /**
  41  * Implement Transliterator API
  42  */
  43 void UnaccentTransliterator::handleTransliterate(Replaceable& text,
  44                                                  UTransPosition& index,
  45                                                  UBool incremental) const {
  46     UnicodeString str("a");
  47     while (index.start < index.limit) {
  48         UChar c = text.charAt(index.start);
  49         UChar d = unaccent(c);
  50         if (c != d) {
  51             str.setCharAt(0, d);
  52             text.handleReplaceBetween(index.start, index.start+1, str);
  53         }
  54         index.start++;
  55     }
  56 }