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 #include <wctype.h>
  22 #include <unicode/uchar.h>
  23 #include <FBaseSysLog.h>
  24 #include "FBase_CharacterImpl.h"
  25
  26 namespace Tizen { namespace Base
  27 {
  28
  29 _CharacterImpl::~_CharacterImpl(void)
  30 {
  31 }
  32
  33 wchar_t
  34 _CharacterImpl::ToLowerCase(wchar_t ch)
  35 {
  36         return u_tolower(ch);
  37 }
  38
  39 wchar_t
  40 _CharacterImpl::ToUpperCase(wchar_t ch)
  41 {
  42         return u_toupper(ch);
  43 }
  44
  45 UnicodeCategory
  46 _CharacterImpl::GetUnicodeCategory(wchar_t ch)
  47 {
  48         if (_CharacterImpl::IsSurrogate(ch))
  49         {
  50                 return UNICODE_SURROGATE;
  51         }
  52         else if (_CharacterImpl::IsModifier(ch))
  53         {
  54                 return UNICODE_MODIFIER;
  55         }
  56         else if (_CharacterImpl::IsArrow(ch))
  57         {
  58                 return UNICODE_ARROW;
  59         }
  60         else if (_CharacterImpl::IsWhitespace(ch))
  61         {
  62                 return UNICODE_SPACE;
  63         }
  64         else if (_CharacterImpl::IsPunctuation(ch))
  65         {
  66                 return UNICODE_PUNCTUATION;
  67         }
  68         else if (_CharacterImpl::IsControl(ch))
  69         {
  70                 return UNICODE_CONTROL;
  71         }
  72         else if (_CharacterImpl::IsMath(ch))
  73         {
  74                 return UNICODE_MATH;
  75         }
  76         else if (_CharacterImpl::IsDigit(ch))
  77         {
  78                 return UNICODE_DIGIT;
  79         }
  80         else if (_CharacterImpl::IsHangul(ch))
  81         {
  82                 return UNICODE_HANGUL;
  83         }
  84         else if (_CharacterImpl::IsCJK(ch))
  85         {
  86                 return UNICODE_HANJA;
  87         }
  88         else if (_CharacterImpl::IsCombining(ch))
  89         {
  90                 return UNICODE_COMBINING;
  91         }
  92         else if (_CharacterImpl::IsLanguage(ch))
  93         {
  94                 return UNICODE_LANGUAGE;
  95         }
  96         else
  97         {
  98                 return UNICODE_OTHER;
  99         }
 100 }
 101
 102 bool
 103 _CharacterImpl::IsDigit(wchar_t ch)
 104 {
 105         return(u_isdigit(ch) == TRUE);
 106 }
 107
 108 bool
 109 _CharacterImpl::IsLetter(wchar_t ch)
 110 {
 111         return(u_isalpha(ch) == TRUE);
 112 }
 113
 114 int
 115 _CharacterImpl::ToDigit(wchar_t ch, int radix)
 116 {
 117         return u_digit(ch, radix);
 118 }
 119
 120 wchar_t
 121 _CharacterImpl::ForDigit(int digit, int radix)
 122 {
 123         return u_forDigit(digit, radix);
 124 }
 125
 126 double
 127 _CharacterImpl::GetNumericValue(wchar_t ch)
 128 {
 129         return u_getNumericValue(ch);
 130 }
 131
 132 bool
 133 _CharacterImpl::IsDefined(wchar_t ch)
 134 {
 135         return(u_isdefined(ch) == TRUE);
 136 }
 137
 138 bool
 139 _CharacterImpl::IsWhitespace(wchar_t ch)
 140 {
 141         return(u_isWhitespace(ch) == TRUE);
 142 }
 143
 144 bool
 145 _CharacterImpl::IsTitleCase(wchar_t ch)
 146 {
 147         return(u_istitle(ch) == TRUE);
 148 }
 149
 150 wchar_t
 151 _CharacterImpl::ToTitleCase(wchar_t ch)
 152 {
 153         return(u_totitle(ch));
 154 }
 155
 156 bool
 157 _CharacterImpl::IsISOControl(wchar_t ch)
 158 {
 159         return(u_isISOControl(ch) == TRUE);
 160 }
 161
 162 bool
 163 _CharacterImpl::IsControl(wchar_t ch)
 164 {
 165         return(u_iscntrl(ch) == TRUE);
 166 }
 167
 168 bool
 169 _CharacterImpl::IsPunctuation(wchar_t ch)
 170 {
 171         return(u_ispunct(ch) == TRUE);
 172 }
 173
 174 bool
 175 _CharacterImpl::IsSurrogate(wchar_t ch)
 176 {
 177         if (0xD800 <= ch && ch <= 0xDFFF)
 178         {
 179                 return true;
 180         }
 181         else
 182         {
 183                 return false;
 184         }
 185 }
 186
 187 bool
 188 _CharacterImpl::IsArrow(wchar_t ch)
 189 {
 190         if ((0x2190 <= ch && ch <= 0x21FF)          // Arrows
 191                 || (0x27F0 <= ch && ch <= 0x27FF)       // Supplemental Arrows A
 192                 || (0x2900 <= ch && ch <= 0x297F)       // Supplemental Arrows B
 193                 || (0x2B00 <= ch && ch <= 0x2B11)       // white and black arrows
 194                 || (0x2B30 <= ch && ch <= 0x2B4C))      // math arrows
 195         {
 196                 return true;
 197         }
 198         else
 199         {
 200                 return false;
 201         }
 202 }
 203
 204 bool
 205 _CharacterImpl::IsMath(wchar_t ch)
 206 {
 207         if (u_charType(ch) == U_MATH_SYMBOL)
 208         {
 209                 return true;
 210         }
 211         else
 212         {
 213                 return false;
 214         }
 215 }
 216
 217 bool
 218 _CharacterImpl::IsModifier(wchar_t ch)
 219 {
 220         if (u_charType(ch) == U_MODIFIER_LETTER)
 221         {
 222                 return true;
 223         }
 224         else if (u_charType(ch) == U_MODIFIER_SYMBOL)
 225         {
 226                 return true;
 227         }
 228         else
 229         {
 230                 return false;
 231         }
 232 }
 233
 234 bool
 235 _CharacterImpl::IsHangul(wchar_t ch)
 236 {
 237         if ((0xAC00 <= ch && ch <= 0xD7AF)          // Hangul Syllable
 238                 || (0x1100 <= ch && ch <= 0x11FF)       // Hangul Jamo
 239                 || (0x3130 <= ch && ch <= 0x318F))      // Hangul Compatibility Jamo
 240         {
 241                 return true;
 242         }
 243         else
 244         {
 245                 return false;
 246         }
 247 }
 248
 249 bool
 250 _CharacterImpl::IsCJK(wchar_t ch)
 251 {
 252         if ((0x4E00 <= ch && ch <= 0x9FFF)      // CJK Unified Ideographs
 253                 || (0xF900 <= ch && ch <= 0xFAFF)   // CJK Compatibility Ideographs
 254                 || (0x3400 <= ch && ch <= 0x4DBF))  // CJK Unified Ideographs, Extension A
 255         {
 256                 return true;
 257         }
 258         else
 259         {
 260                 return false;
 261         }
 262 }
 263
 264 bool
 265 _CharacterImpl::IsCombining(wchar_t ch)
 266 {
 267         if ((0x0300 <= ch && ch <= 0x036F) || (0x20D0 <= ch && ch <= 0x20FF)
 268                 || (0x1DC0 <= ch && ch <= 0x1DFF) || (0xFE20 <= ch && ch <= 0xFE2F))
 269         {
 270                 return true;
 271         }
 272         else
 273         {
 274                 return false;
 275         }
 276 }
 277 bool
 278 _CharacterImpl::IsLanguage(wchar_t ch)
 279 {
 280         if ((0x0041 <= ch && ch <= 0x005A) || (0x0061 <= ch && ch <= 0x007A) || (0x0100 <= ch && ch <= 0x024F)
 281                 || (0x0250 <= ch && ch <= 0x02AF)   // IPA Extensions
 282                 || (0x0370 <= ch && ch <= 0x03FF)   // Greek
 283                 || (0x0400 <= ch && ch <= 0x04FF)   // Cyrillic
 284                 || (0x0530 <= ch && ch <= 0x058F)   // Armenian
 285                 || (0x0590 <= ch && ch <= 0x05FF)   // Hebrew
 286                 || (0x0600 <= ch && ch <= 0x06FF)   // Arabic
 287                 || (0x0700 <= ch && ch <= 0x074F)   // Syriac
 288                 || (0x0780 <= ch && ch <= 0x07BF)   // Thaana
 289                 || (0x0900 <= ch && ch <= 0x0D7F)   // Devanagari ~ Malayalam
 290                 || (0x0D80 <= ch && ch <= 0x0DFF)   // Sinhala
 291                 || (0x0E00 <= ch && ch <= 0x0FFF)   // Thai, Lao, Tibetan
 292                 || (0x1000 <= ch && ch <= 0x109F)   // Myanmar
 293                 || (0x10A0 <= ch && ch <= 0x10FF)   // Georgian
 294                 || (0x1100 <= ch && ch <= 0x11FF)   // Hangul Jamo
 295                 || (0x1200 <= ch && ch <= 0x137F)   // Ethiopic
 296                 || (0x13A0 <= ch && ch <= 0x13FF)   // Cherokee
 297                 || (0x1400 <= ch && ch <= 0x167F)   // Unified Canadian Aboriginal Syllabics
 298                 || (0x1680 <= ch && ch <= 0x169F)   // Ogham
 299                 || (0x16A0 <= ch && ch <= 0x16FF)   // Runic
 300                 || (0x1780 <= ch && ch <= 0x17FF)   // Khmer
 301                 || (0x1800 <= ch && ch <= 0x18AF)   // Mongolian
 302                 || (0x1E00 <= ch && ch <= 0x1EFF)   // Latin Extended Additional
 303                 || (0x1F00 <= ch && ch <= 0x1FFF)   // Greek Extended
 304                 || (0x3040 <= ch && ch <= 0x30FF)   // Hiragana, Katakana
 305                 || (0x3100 <= ch && ch <= 0x312F)   // Bopomofo
 306                 || (0x3130 <= ch && ch <= 0x318F)   // Hangul Compatibility Jamo
 307                 || (0x3190 <= ch && ch <= 0x319F)   // Kanbun
 308                 || (0x31A0 <= ch && ch <= 0x31BF)   // Bopomofo Extended
 309                 || (0x3400 <= ch && ch <= 0x4DB5)   // CJK Unified Ideographs Extension A
 310                 || (0x4E00 <= ch && ch <= 0x9FFF)   // CJK Unified Ideographs
 311                 || (0xA000 <= ch && ch <= 0xA4CF)   // Yi Syllables, Radicals
 312                 || (0xAC00 <= ch && ch <= 0xD7A3)   // Hangul Syllables
 313                 || (0xF900 <= ch && ch <= 0xFAFF)   // CJK Compatibility Ideographs
 314                 || (0xFB00 <= ch && ch <= 0xFB4F)   // Alphabetic Presentation Forms
 315                 || (0xFB50 <= ch && ch <= 0xFDFF)   // Arabic Presentation Forms-A
 316                 || (0xFE70 <= ch && ch <= 0xFEFE)   // Arabic Presentation Forms-B
 317                 || (0xFF00 <= ch && ch <= 0xFFEF)
 318                 )
 319         {
 320                 return true;
 321         }
 322         else
 323         {
 324                 return false;
 325         }
 326 }
 327
 328 }} // Tizen::Base