vcs.maemo.org Git - mdictionary/blob - trunk/src/includes/AccentsNormalizer.h

   1 /*******************************************************************************
   2
   3     This file is part of mDictionary.
   4
   5     mDictionary is free software: you can redistribute it and/or modify
   6     it under the terms of the GNU General Public License as published by
   7     the Free Software Foundation, either version 3 of the License, or
   8     (at your option) any later version.
   9
  10     mDictionary is distributed in the hope that it will be useful,
  11     but WITHOUT ANY WARRANTY; without even the implied warranty of
  12     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13     GNU General Public License for more details.
  14
  15     You should have received a copy of the GNU General Public License
  16     along with mDictionary.  If not, see <http://www.gnu.org/licenses/>.
  17
  18     Copyright 2010 Comarch S.A.
  19
  20 *******************************************************************************/
  21
  22 /*! /file AccentsNormalizer.h
  23 \brief Common funcionality for all classes that need to remove accents from strings
  24
  25 \author Bartosz Szatkowski <bulislaw@linux.com>
  26 */
  27
  28 #ifndef ACCENTSNORMALIZER_H
  29 #define ACCENTSNORMALIZER_H
  30
  31 #include <QMap>
  32 #include <QString>
  33 #include <QRegExp>
  34
  35
  36 class AccentsNormalizer {
  37   protected:
  38     QMap<QChar, QRegExp> letters;
  39     QRegExp noLetter;
  40
  41     QString removeAccents(QString string) {
  42        string = string.toLower();
  43        QString normalized = string.normalized(QString::NormalizationForm_KD);
  44        foreach(QChar let, letters.keys())
  45            normalized.replace(letters[let], QString(let));
  46
  47        normalized.replace(noLetter, "");
  48        //qDebug() << "NORMALIZED: " << normalized;
  49        return normalized;
  50     }
  51     void initAccents() {
  52         letters['a'] = QRegExp(QString::fromUtf8("[ÀàÁáÂÂâÃãÄäÅåæÆĀāĂăĄą]"), Qt::CaseInsensitive);
  53         letters['c'] = QRegExp(QString::fromUtf8("[ÇçÈçŒĆćĈĉĊċČč]"), Qt::CaseInsensitive);
  54         letters['d'] = QRegExp(QString::fromUtf8("[ÐĐđĎď]"), Qt::CaseInsensitive);
  55         letters['e'] = QRegExp(QString::fromUtf8("[ÈéèÉÊêËëĒēĔĕĖėĘęĚěē]"), Qt::CaseInsensitive);
  56         letters['f'] = QRegExp(QString::fromUtf8("[ſ]"), Qt::CaseInsensitive);
  57         letters['g'] = QRegExp(QString::fromUtf8("[ĠġĢģĜĝĞğ]"), Qt::CaseInsensitive);
  58         letters['h'] = QRegExp(QString::fromUtf8("[ħĤĥĦ]"), Qt::CaseInsensitive);
  59         letters['i'] = QRegExp(QString::fromUtf8("[ÌìÍíÎîÏïİĳĨĩĪīĬĭĮį]"), Qt::CaseInsensitive);
  60         letters['j'] = QRegExp(QString::fromUtf8("[Ĵĵ]"), Qt::CaseInsensitive);
  61         letters['k'] = QRegExp(QString::fromUtf8("[Ķķĸ]"), Qt::CaseInsensitive);
  62         letters['l'] = QRegExp(QString::fromUtf8("[ŀŁłĹĺĻļĽľĿ]"), Qt::CaseInsensitive);
  63         letters['n'] = QRegExp(QString::fromUtf8("[ÑñŃńŅņŇňŉŊŋ]"), Qt::CaseInsensitive);
  64         letters['o'] = QRegExp(QString::fromUtf8("[ÒòÓóÔÕõÖöØøŐőœŌōŎŏ]"), Qt::CaseInsensitive);
  65         letters['r'] = QRegExp(QString::fromUtf8("[ŕŖŖŗŘř]"), Qt::CaseInsensitive);
  66         letters['s'] = QRegExp(QString::fromUtf8("[ßŠšŚśŜŝŞş]"), Qt::CaseInsensitive);
  67         letters['t'] = QRegExp(QString::fromUtf8("[ŢţŤťŦŧ]"), Qt::CaseInsensitive);
  68         letters['u'] = QRegExp(QString::fromUtf8("[ÙùÚúÛûÜüŰűŲųŨũŪūŬŮů]"), Qt::CaseInsensitive);
  69         letters['w'] = QRegExp(QString::fromUtf8("[Ŵŵ]"), Qt::CaseInsensitive);
  70         letters['y'] = QRegExp(QString::fromUtf8("[ÿÝýŶŷŸ]"), Qt::CaseInsensitive);
  71         letters['z'] = QRegExp(QString::fromUtf8("[ŹźŻżŽž]"), Qt::CaseInsensitive);
  72         noLetter = QRegExp("[^a-z ]", Qt::CaseInsensitive);
  73     }
  74 };
  75
  76 #endif // ACCENTSNORMALIZER_H