src/base/FBase_CharacterImpl.cpp

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