src/base/FBase_CharacterImpl.cpp

   1 //
   2 // Copyright (c) 2012 Samsung Electronics Co., Ltd.
   3 //
   4 // Licensed under the Apache License, Version 2.0 (the License);
   5 // you may not use this file except in compliance with the License.
   6 // You may obtain a copy of the License at
   7 //
   8 //     http://www.apache.org/licenses/LICENSE-2.0
   9 //
  10 // Unless required by applicable law or agreed to in writing, software
  11 // distributed under the License is distributed on an "AS IS" BASIS,
  12 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13 // See the License for the specific language governing permissions and
  14 // limitations under the License.
  15 //
  16
  17 /**
  18  * @file                FBase_CharacterImpl.cpp
  19  * @brief               This is the implementation file for _CharImpl class.
  20  */
  21
  22 #include <wctype.h>
  23 #include <unicode/uchar.h>
  24 #include <FBaseSysLog.h>
  25 #include "FBase_CharacterImpl.h"
  26
  27
  28 namespace Tizen { namespace Base
  29 {
  30
  31 _CharacterImpl::~_CharacterImpl(void)
  32 {
  33 }
  34
  35 wchar_t
  36 _CharacterImpl::ToLowerCase(wchar_t ch)
  37 {
  38         return u_tolower(ch);
  39 }
  40
  41 wchar_t
  42 _CharacterImpl::ToUpperCase(wchar_t ch)
  43 {
  44         return u_toupper(ch);
  45 }
  46
  47 UnicodeCategory
  48 _CharacterImpl::GetUnicodeCategory(wchar_t ch)
  49 {
  50         if (_CharacterImpl::IsSurrogate(ch))
  51         {
  52                 return UNICODE_SURROGATE;
  53         }
  54         else if (_CharacterImpl::IsModifier(ch))
  55         {
  56                 return UNICODE_MODIFIER;
  57         }
  58         else if (_CharacterImpl::IsArrow(ch))
  59         {
  60                 return UNICODE_ARROW;
  61         }
  62         else if (_CharacterImpl::IsWhitespace(ch))
  63         {
  64                 return UNICODE_SPACE;
  65         }
  66         else if (_CharacterImpl::IsPunctuation(ch))
  67         {
  68                 return UNICODE_PUNCTUATION;
  69         }
  70         else if (_CharacterImpl::IsControl(ch))
  71         {
  72                 return UNICODE_CONTROL;
  73         }
  74         else if (_CharacterImpl::IsMath(ch))
  75         {
  76                 return UNICODE_MATH;
  77         }
  78         else if (_CharacterImpl::IsDigit(ch))
  79         {
  80                 return UNICODE_DIGIT;
  81         }
  82         else if (_CharacterImpl::IsHangul(ch))
  83         {
  84                 return UNICODE_HANGUL;
  85         }
  86         else if (_CharacterImpl::IsCJK(ch))
  87         {
  88                 return UNICODE_HANJA;
  89         }
  90         else if (_CharacterImpl::IsCombining(ch))
  91         {
  92                 return UNICODE_COMBINING;
  93         }
  94         else if (_CharacterImpl::IsLanguage(ch))
  95         {
  96                 return UNICODE_LANGUAGE;
  97         }
  98         else
  99         {
 100                 return UNICODE_OTHER;
 101         }
 102 }
 103
 104 bool
 105 _CharacterImpl::IsDigit(wchar_t ch)
 106 {
 107         return(u_isdigit(ch) == TRUE);
 108 }
 109
 110 bool
 111 _CharacterImpl::IsLetter(wchar_t ch)
 112 {
 113         return(u_isalpha(ch) == TRUE);
 114 }
 115
 116 int
 117 _CharacterImpl::ToDigit(wchar_t ch, int radix)
 118 {
 119         return u_digit(ch, radix);
 120 }
 121
 122 wchar_t
 123 _CharacterImpl::ForDigit(int digit, int radix)
 124 {
 125         return u_forDigit(digit, radix);
 126 }
 127
 128 double
 129 _CharacterImpl::GetNumericValue(wchar_t ch)
 130 {
 131         return u_getNumericValue(ch);
 132 }
 133
 134 bool
 135 _CharacterImpl::IsDefined(wchar_t ch)
 136 {
 137         return(u_isdefined(ch) == TRUE);
 138 }
 139
 140 bool
 141 _CharacterImpl::IsWhitespace(wchar_t ch)
 142 {
 143         return(u_isWhitespace(ch) == TRUE);
 144 }
 145
 146 bool
 147 _CharacterImpl::IsTitleCase(wchar_t ch)
 148 {
 149         return(u_istitle(ch) == TRUE);
 150 }
 151
 152 wchar_t
 153 _CharacterImpl::ToTitleCase(wchar_t ch)
 154 {
 155         return(u_totitle(ch));
 156 }
 157
 158 bool
 159 _CharacterImpl::IsISOControl(wchar_t ch)
 160 {
 161         return(u_isISOControl(ch) == TRUE);
 162 }
 163
 164 bool
 165 _CharacterImpl::IsControl(wchar_t ch)
 166 {
 167         return(u_iscntrl(ch) == TRUE);
 168 }
 169
 170 bool
 171 _CharacterImpl::IsPunctuation(wchar_t ch)
 172 {
 173         return(u_ispunct(ch) == TRUE);
 174 }
 175
 176 bool
 177 _CharacterImpl::IsSurrogate(wchar_t ch)
 178 {
 179         if (0xD800 <= ch && ch <= 0xDFFF)
 180         {
 181                 return true;
 182         }
 183         else
 184         {
 185                 return false;
 186         }
 187 }
 188
 189 bool
 190 _CharacterImpl::IsArrow(wchar_t ch)
 191 {
 192         if ((0x2190 <= ch && ch <= 0x21FF)          // Arrows
 193                 || (0x27F0 <= ch && ch <= 0x27FF)       // Supplemental Arrows A
 194                 || (0x2900 <= ch && ch <= 0x297F)       // Supplemental Arrows B
 195                 || (0x2B00 <= ch && ch <= 0x2B11)       // white and black arrows
 196                 || (0x2B30 <= ch && ch <= 0x2B4C))      // math arrows
 197         {
 198                 return true;
 199         }
 200         else
 201         {
 202                 return false;
 203         }
 204 }
 205
 206 bool
 207 _CharacterImpl::IsMath(wchar_t ch)
 208 {
 209         if (u_charType(ch) == U_MATH_SYMBOL)
 210         {
 211                 return true;
 212         }
 213         else
 214         {
 215                 return false;
 216         }
 217 }
 218
 219 bool
 220 _CharacterImpl::IsModifier(wchar_t ch)
 221 {
 222         if (u_charType(ch) == U_MODIFIER_LETTER)
 223         {
 224                 return true;
 225         }
 226         else if (u_charType(ch) == U_MODIFIER_SYMBOL)
 227         {
 228                 return true;
 229         }
 230         else
 231         {
 232                 return false;
 233         }
 234 }
 235
 236 bool
 237 _CharacterImpl::IsHangul(wchar_t ch)
 238 {
 239         if ((0xAC00 <= ch && ch <= 0xD7AF)          // Hangul Syllable
 240                 || (0x1100 <= ch && ch <= 0x11FF)       // Hangul Jamo
 241                 || (0x3130 <= ch && ch <= 0x318F))      // Hangul Compatibility Jamo
 242         {
 243                 return true;
 244         }
 245         else
 246         {
 247                 return false;
 248         }
 249 }
 250
 251 bool
 252 _CharacterImpl::IsCJK(wchar_t ch)
 253 {
 254         if ((0x4E00 <= ch && ch <= 0x9FFF)      // CJK Unified Ideographs
 255                 || (0xF900 <= ch && ch <= 0xFAFF)   // CJK Compatibility Ideographs
 256                 || (0x3400 <= ch && ch <= 0x4DBF))  // CJK Unified Ideographs, Extension A
 257         {
 258                 return true;
 259         }
 260         else
 261         {
 262                 return false;
 263         }
 264 }
 265
 266 bool
 267 _CharacterImpl::IsCombining(wchar_t ch)
 268 {
 269         if ((0x0300 <= ch && ch <= 0x036F) || (0x20D0 <= ch && ch <= 0x20FF)
 270                 || (0x1DC0 <= ch && ch <= 0x1DFF) || (0xFE20 <= ch && ch <= 0xFE2F))
 271         {
 272                 return true;
 273         }
 274         else
 275         {
 276                 return false;
 277         }
 278 }
 279 bool
 280 _CharacterImpl::IsLanguage(wchar_t ch)
 281 {
 282         if ((0x0041 <= ch && ch <= 0x005A) || (0x0061 <= ch && ch <= 0x007A) || (0x0100 <= ch && ch <= 0x024F)
 283                 || (0x0250 <= ch && ch <= 0x02AF)   // IPA Extensions
 284                 || (0x0370 <= ch && ch <= 0x03FF)   // Greek
 285                 || (0x0400 <= ch && ch <= 0x04FF)   // Cyrillic
 286                 || (0x0530 <= ch && ch <= 0x058F)   // Armenian
 287                 || (0x0590 <= ch && ch <= 0x05FF)   // Hebrew
 288                 || (0x0600 <= ch && ch <= 0x06FF)   // Arabic
 289                 || (0x0700 <= ch && ch <= 0x074F)   // Syriac
 290                 || (0x0780 <= ch && ch <= 0x07BF)   // Thaana
 291                 || (0x0900 <= ch && ch <= 0x0D7F)   // Devanagari ~ Malayalam
 292                 || (0x0D80 <= ch && ch <= 0x0DFF)   // Sinhala
 293                 || (0x0E00 <= ch && ch <= 0x0FFF)   // Thai, Lao, Tibetan
 294                 || (0x1000 <= ch && ch <= 0x109F)   // Myanmar
 295                 || (0x10A0 <= ch && ch <= 0x10FF)   // Georgian
 296                 || (0x1100 <= ch && ch <= 0x11FF)   // Hangul Jamo
 297                 || (0x1200 <= ch && ch <= 0x137F)   // Ethiopic
 298                 || (0x13A0 <= ch && ch <= 0x13FF)   // Cherokee
 299                 || (0x1400 <= ch && ch <= 0x167F)   // Unified Canadian Aboriginal Syllabics
 300                 || (0x1680 <= ch && ch <= 0x169F)   // Ogham
 301                 || (0x16A0 <= ch && ch <= 0x16FF)   // Runic
 302                 || (0x1780 <= ch && ch <= 0x17FF)   // Khmer
 303                 || (0x1800 <= ch && ch <= 0x18AF)   // Mongolian
 304                 || (0x1E00 <= ch && ch <= 0x1EFF)   // Latin Extended Additional
 305                 || (0x1F00 <= ch && ch <= 0x1FFF)   // Greek Extended
 306                 || (0x3040 <= ch && ch <= 0x30FF)   // Hiragana, Katakana
 307                 || (0x3100 <= ch && ch <= 0x312F)   // Bopomofo
 308                 || (0x3130 <= ch && ch <= 0x318F)   // Hangul Compatibility Jamo
 309                 || (0x3190 <= ch && ch <= 0x319F)   // Kanbun
 310                 || (0x31A0 <= ch && ch <= 0x31BF)   // Bopomofo Extended
 311                 || (0x3400 <= ch && ch <= 0x4DB5)   // CJK Unified Ideographs Extension A
 312                 || (0x4E00 <= ch && ch <= 0x9FFF)   // CJK Unified Ideographs
 313                 || (0xA000 <= ch && ch <= 0xA4CF)   // Yi Syllables, Radicals
 314                 || (0xAC00 <= ch && ch <= 0xD7A3)   // Hangul Syllables
 315                 || (0xF900 <= ch && ch <= 0xFAFF)   // CJK Compatibility Ideographs
 316                 || (0xFB00 <= ch && ch <= 0xFB4F)   // Alphabetic Presentation Forms
 317                 || (0xFB50 <= ch && ch <= 0xFDFF)   // Arabic Presentation Forms-A
 318                 || (0xFE70 <= ch && ch <= 0xFEFE)   // Arabic Presentation Forms-B
 319                 || (0xFF00 <= ch && ch <= 0xFFEF)
 320                 )
 321         {
 322                 return true;
 323         }
 324         else
 325         {
 326                 return false;
 327         }
 328 }
 329
 330 }} // Tizen::Base