icuSources/samples/translit/answers/unaccent.cpp

   1 /***********************************************************************
   2  * © 2016 and later: Unicode, Inc. and others.
   3  * License & terms of use: http://www.unicode.org/copyright.html#License
   4  ***********************************************************************
   5  * COPYRIGHT:
   6  * Copyright (c) 1999-2002, International Business Machines Corporation and
   7  * others. All Rights Reserved.
   8  ***********************************************************************/
   9
  10 #include "unaccent.h"
  11
  12 /**
  13  * Constructor
  14  */
  15 UnaccentTransliterator::UnaccentTransliterator() :
  16     normalizer("", Normalizer::DECOMP),
  17     Transliterator("Unaccent", 0) {
  18 }
  19
  20 /**
  21  * Destructor
  22  */
  23 UnaccentTransliterator::~UnaccentTransliterator() {
  24 }
  25
  26 /**
  27  * Remove accents from a character using Normalizer.
  28  */
  29 UChar UnaccentTransliterator::unaccent(UChar c) const {
  30     UnicodeString str(c);
  31     UErrorCode status = U_ZERO_ERROR;
  32     UnaccentTransliterator* t = (UnaccentTransliterator*)this;
  33
  34     t->normalizer.setText(str, status);
  35     if (U_FAILURE(status)) {
  36         return c;
  37     }
  38     return (UChar) t->normalizer.next();
  39 }
  40
  41 /**
  42  * Implement Transliterator API
  43  */
  44 void UnaccentTransliterator::handleTransliterate(Replaceable& text,
  45                                                  UTransPosition& index,
  46                                                  UBool incremental) const {
  47     UnicodeString str("a");
  48     while (index.start < index.limit) {
  49         UChar c = text.charAt(index.start);
  50         UChar d = unaccent(c);
  51         if (c != d) {
  52             str.setCharAt(0, d);
  53             text.handleReplaceBetween(index.start, index.start+1, str);
  54         }
  55         index.start++;
  56     }
  57 }