dali-toolkit/internal/text/xhtml-entities.cpp

   1 /*
   2  * Copyright (c) 2021 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 // EXTERNAL INCLUDES
  19 #include <cstring> // for strlen()
  20
  21 // FILE HEADER
  22 #include "xhtml-entities.h"
  23
  24 namespace Dali
  25 {
  26 namespace Toolkit
  27 {
  28 namespace Text
  29 {
  30 namespace
  31 {
  32 /**
  33  * Implementation of the XHTML Entity matching
  34  */
  35 struct XHTMLEntityLookup
  36 {
  37   const char* const entityName; // XHTML Named Entity string
  38   const char* const entityCode; // Corresponding UTF-8
  39 };
  40
  41 /* table of html name entities supported in DALi
  42  *
  43  * these are stored as pair with Named entity as Key and
  44  * its utf 8 as value
  45  */
  46 // clang-format off
  47 const XHTMLEntityLookup XHTMLEntityLookupTable[] =
  48 {
  49   {"&quot;\0"    ,"\x22\0"         },
  50   {"&amp;\0"     ,"\x26\0"         },
  51   {"&apos;\0"    ,"\x27\0"         },
  52   {"&lt;\0"      ,"\x3c\0"         },
  53   {"&gt;\0"      ,"\x3e\0"         },
  54   {"&nbsp;\0"    ,"\xc2\xa0\0"     },
  55   {"&iexcl;\0"   ,"\xc2\xa1\0"     },
  56   {"&cent;\0"    ,"\xc2\xa2\0"     },
  57   {"&pound;\0"   ,"\xc2\xa3\0"     },
  58   {"&curren;\0"  ,"\xc2\xa4\0"     },
  59   {"&yen;\0"     ,"\xc2\xa5\0"     },
  60   {"&brvbar;\0"  ,"\xc2\xa6\0"     },
  61   {"&sect;\0"    ,"\xc2\xa7\0"     },
  62   {"&uml;\0"     ,"\xc2\xa8\0"     },
  63   {"&copy;\0"    ,"\xc2\xa9\0"     },
  64   {"&ordf;\0"    ,"\xc2\xaa\0"     },
  65   {"&laquo;\0"   ,"\xc2\xab\0"     },
  66   {"&not;\0"     ,"\xc2\xac\0"     },
  67   {"&shy;\0"     ,"\xc2\xad\0"     },
  68   {"&reg;\0"     ,"\xc2\xae\0"     },
  69   {"&macr;\0"    ,"\xc2\xaf\0"     },
  70   {"&deg;\0"     ,"\xc2\xb0\0"     },
  71   {"&plusmn;\0"  ,"\xc2\xb1\0"     },
  72   {"&sup2;\0"    ,"\xc2\xb2\0"     },
  73   {"&sup3;\0"    ,"\xc2\xb3\0"     },
  74   {"&acute;\0"   ,"\xc2\xb4\0"     },
  75   {"&micro;\0"   ,"\xc2\xb5\0"     },
  76   {"&para;\0"    ,"\xc2\xb6\0"     },
  77   {"&middot;\0"  ,"\xc2\xb7\0"     },
  78   {"&cedil;\0"   ,"\xc2\xb8\0"     },
  79   {"&sup1;\0"    ,"\xc2\xb9\0"     },
  80   {"&ordm;\0"    ,"\xc2\xba\0"     },
  81   {"&raquo;\0"   ,"\xc2\xbb\0"     },
  82   {"&frac14;\0"  ,"\xc2\xbc\0"     },
  83   {"&frac12;\0"  ,"\xc2\xbd\0"     },
  84   {"&frac34;\0"  ,"\xc2\xbe\0"     },
  85   {"&iquest;\0"  ,"\xc2\xbf\0"     },
  86   {"&Agrave;\0"  ,"\xc3\x80\0"     },
  87   {"&Aacute;\0"  ,"\xc3\x81\0"     },
  88   {"&Acirc;\0"   ,"\xc3\x82\0"     },
  89   {"&Atilde;\0"  ,"\xc3\x83\0"     },
  90   {"&Auml;\0"    ,"\xc3\x84\0"     },
  91   {"&Aring;\0"   ,"\xc3\x85\0"     },
  92   {"&AElig;\0"   ,"\xc3\x86\0"     },
  93   {"&Ccedil;\0"  ,"\xc3\x87\0"     },
  94   {"&Egrave;\0"  ,"\xc3\x88\0"     },
  95   {"&Eacute;\0"  ,"\xc3\x89\0"     },
  96   {"&Ecirc;\0"   ,"\xc3\x8a\0"     },
  97   {"&Euml;\0"    ,"\xc3\x8b\0"     },
  98   {"&Igrave;\0"  ,"\xc3\x8c\0"     },
  99   {"&Iacute;\0"  ,"\xc3\x8d\0"     },
 100   {"&Icirc;\0"   ,"\xc3\x8e\0"     },
 101   {"&Iuml;\0"    ,"\xc3\x8f\0"     },
 102   {"&ETH;\0"     ,"\xc3\x90\0"     },
 103   {"&Ntilde;\0"  ,"\xc3\x91\0"     },
 104   {"&Ograve;\0"  ,"\xc3\x92\0"     },
 105   {"&Oacute;\0"  ,"\xc3\x93\0"     },
 106   {"&Ocirc;\0"   ,"\xc3\x94\0"     },
 107   {"&Otilde;\0"  ,"\xc3\x95\0"     },
 108   {"&Ouml;\0"    ,"\xc3\x96\0"     },
 109   {"&times;\0"   ,"\xc3\x97\0"     },
 110   {"&Oslash;\0"  ,"\xc3\x98\0"     },
 111   {"&Ugrave;\0"  ,"\xc3\x99\0"     },
 112   {"&Uacute;\0"  ,"\xc3\x9a\0"     },
 113   {"&Ucirc;\0"   ,"\xc3\x9b\0"     },
 114   {"&Uuml;\0"    ,"\xc3\x9c\0"     },
 115   {"&Yacute;\0"  ,"\xc3\x9d\0"     },
 116   {"&THORN;\0"   ,"\xc3\x9e\0"     },
 117   {"&szlig;\0"   ,"\xc3\x9f\0"     },
 118   {"&agrave;\0"  ,"\xc3\xa0\0"     },
 119   {"&aacute;\0"  ,"\xc3\xa1\0"     },
 120   {"&acirc;\0"   ,"\xc3\xa2\0"     },
 121   {"&atilde;\0"  ,"\xc3\xa3\0"     },
 122   {"&auml;\0"    ,"\xc3\xa4\0"     },
 123   {"&aring;\0"   ,"\xc3\xa5\0"     },
 124   {"&aelig;\0"   ,"\xc3\xa6\0"     },
 125   {"&ccedil;\0"  ,"\xc3\xa7\0"     },
 126   {"&egrave;\0"  ,"\xc3\xa8\0"     },
 127   {"&eacute;\0"  ,"\xc3\xa9\0"     },
 128   {"&ecirc;\0"   ,"\xc3\xaa\0"     },
 129   {"&euml;\0"    ,"\xc3\xab\0"     },
 130   {"&igrave;\0"  ,"\xc3\xac\0"     },
 131   {"&iacute;\0"  ,"\xc3\xad\0"     },
 132   {"&icirc;\0"   ,"\xc3\xae\0"     },
 133   {"&iuml;\0"    ,"\xc3\xaf\0"     },
 134   {"&eth;\0"     ,"\xc3\xb0\0"     },
 135   {"&ntilde;\0"  ,"\xc3\xb1\0"     },
 136   {"&ograve;\0"  ,"\xc3\xb2\0"     },
 137   {"&oacute;\0"  ,"\xc3\xb3\0"     },
 138   {"&ocirc;\0"   ,"\xc3\xb4\0"     },
 139   {"&otilde;\0"  ,"\xc3\xb5\0"     },
 140   {"&ouml;\0"    ,"\xc3\xb6\0"     },
 141   {"&divide;\0"  ,"\xc3\xb7\0"     },
 142   {"&oslash;\0"  ,"\xc3\xb8\0"     },
 143   {"&ugrave;\0"  ,"\xc3\xb9\0"     },
 144   {"&uacute;\0"  ,"\xc3\xba\0"     },
 145   {"&ucirc;\0"   ,"\xc3\xbb\0"     },
 146   {"&uuml;\0"    ,"\xc3\xbc\0"     },
 147   {"&yacute;\0"  ,"\xc3\xbd\0"     },
 148   {"&thorn;\0"   ,"\xc3\xbe\0"     },
 149   {"&yuml;\0"    ,"\xc3\xbf\0"     },
 150   {"&OElig;\0"   ,"\xc5\x92\0"     },
 151   {"&oelig;\0"   ,"\xc5\x93\0"     },
 152   {"&Scaron;\0"  ,"\xc5\xa0\0"     },
 153   {"&scaron;\0"  ,"\xc5\xa1\0"     },
 154   {"&Yuml;\0"    ,"\xc5\xb8\0"     },
 155   {"&fnof;\0"    ,"\xc6\x92\0"     },
 156   {"&circ;\0"    ,"\xcb\x86\0"     },
 157   {"&tilde;\0"   ,"\xcb\x9c\0"     },
 158   {"&Alpha;\0"   ,"\xce\x91\0"     },
 159   {"&Beta;\0"    ,"\xce\x92\0"     },
 160   {"&Gamma;\0"   ,"\xce\x93\0"     },
 161   {"&Delta;\0"   ,"\xce\x94\0"     },
 162   {"&Epsilon;\0" ,"\xce\x95\0"     },
 163   {"&Zeta;\0"    ,"\xce\x96\0"     },
 164   {"&Eta;\0"     ,"\xce\x97\0"     },
 165   {"&Theta;\0"   ,"\xce\x98\0"     },
 166   {"&Iota;\0"    ,"\xce\x99\0"     },
 167   {"&Kappa;\0"   ,"\xce\x9a\0"     },
 168   {"&Lambda;\0"  ,"\xce\x9b\0"     },
 169   {"&Mu;\0"      ,"\xce\x9c\0"     },
 170   {"&Nu;\0"      ,"\xce\x9d\0"     },
 171   {"&Xi;\0"      ,"\xce\x9e\0"     },
 172   {"&Omicron;\0" ,"\xce\x9f\0"     },
 173   {"&Pi;\0"      ,"\xce\xa0\0"     },
 174   {"&Rho;\0"     ,"\xce\xa1\0"     },
 175   {"&Sigma;\0"   ,"\xce\xa3\0"     },
 176   {"&Tau;\0"     ,"\xce\xa4\0"     },
 177   {"&Upsilon;\0" ,"\xce\xa5\0"     },
 178   {"&Phi;\0"     ,"\xce\xa6\0"     },
 179   {"&Chi;\0"     ,"\xce\xa7\0"     },
 180   {"&Psi;\0"     ,"\xce\xa8\0"     },
 181   {"&Omega;\0"   ,"\xce\xa9\0"     },
 182   {"&alpha;\0"   ,"\xce\xb1\0"     },
 183   {"&beta;\0"    ,"\xce\xb2\0"     },
 184   {"&gamma;\0"   ,"\xce\xb3\0"     },
 185   {"&delta;\0"   ,"\xce\xb4\0"     },
 186   {"&epsilon;\0" ,"\xce\xb5\0"     },
 187   {"&zeta;\0"    ,"\xce\xb6\0"     },
 188   {"&eta;\0"     ,"\xce\xb7\0"     },
 189   {"&theta;\0"   ,"\xce\xb8\0"     },
 190   {"&iota;\0"    ,"\xce\xb9\0"     },
 191   {"&kappa;\0"   ,"\xce\xba\0"     },
 192   {"&lambda;\0"  ,"\xce\xbb\0"     },
 193   {"&mu;\0"      ,"\xce\xbc\0"     },
 194   {"&nu;\0"      ,"\xce\xbd\0"     },
 195   {"&xi;\0"      ,"\xce\xbe\0"     },
 196   {"&omicron;\0" ,"\xce\xbf\0"     },
 197   {"&pi;\0"      ,"\xcf\x80\0"     },
 198   {"&rho;\0"     ,"\xcf\x81\0"     },
 199   {"&sigmaf;\0"  ,"\xcf\x82\0"     },
 200   {"&sigma;\0"   ,"\xcf\x83\0"     },
 201   {"&tau;\0"     ,"\xcf\x84\0"     },
 202   {"&upsilon;\0" ,"\xcf\x85\0"     },
 203   {"&phi;\0"     ,"\xcf\x86\0"     },
 204   {"&chi;\0"     ,"\xcf\x87\0"     },
 205   {"&psi;\0"     ,"\xcf\x88\0"     },
 206   {"&omega;\0"   ,"\xcf\x89\0"     },
 207   {"&thetasym;\0","\xcf\x91\0"     },
 208   {"&upsih;\0"   ,"\xcf\x92\0"     },
 209   {"&piv;\0"     ,"\xcf\x96\0"     },
 210   {"&ensp;\0"    ,"\xe2\x80\x82\0" },
 211   {"&emsp;\0"    ,"\xe2\x80\x83\0" },
 212   {"&thinsp;\0"  ,"\xe2\x80\x89\0" },
 213   {"&zwnj;\0"    ,"\xe2\x80\x8c\0" },
 214   {"&zwj;\0"     ,"\xe2\x80\x8d\0" },
 215   {"&lrm;\0"     ,"\xe2\x80\x8e\0" },
 216   {"&rlm;\0"     ,"\xe2\x80\x8f\0" },
 217   {"&ndash;\0"   ,"\xe2\x80\x93\0" },
 218   {"&mdash;\0"   ,"\xe2\x80\x94\0" },
 219   {"&lsquo;\0"   ,"\xe2\x80\x98\0" },
 220   {"&rsquo;\0"   ,"\xe2\x80\x99\0" },
 221   {"&sbquo;\0"   ,"\xe2\x80\x9a\0" },
 222   {"&ldquo;\0"   ,"\xe2\x80\x9c\0" },
 223   {"&rdquo;\0"   ,"\xe2\x80\x9d\0" },
 224   {"&bdquo;\0"   ,"\xe2\x80\x9e\0" },
 225   {"&dagger;\0"  ,"\xe2\x80\xa0\0" },
 226   {"&Dagger;\0"  ,"\xe2\x80\xa1\0" },
 227   {"&bull;\0"    ,"\xe2\x80\xa2\0" },
 228   {"&hellip;\0"  ,"\xe2\x80\xa6\0" },
 229   {"&permil;\0"  ,"\xe2\x80\xb0\0" },
 230   {"&prime;\0"   ,"\xe2\x80\xb2\0" },
 231   {"&Prime;\0"   ,"\xe2\x80\xb3\0" },
 232   {"&lsaquo;\0"  ,"\xe2\x80\xb9\0" },
 233   {"&rsaquo;\0"  ,"\xe2\x80\xba\0" },
 234   {"&oline;\0"   ,"\xe2\x80\xbe\0" },
 235   {"&frasl;\0"   ,"\xe2\x81\x84\0" },
 236   {"&euro;\0"    ,"\xe2\x82\xac\0" },
 237   {"&image;\0"   ,"\xe2\x84\x91\0" },
 238   {"&weierp;\0"  ,"\xe2\x84\x98\0" },
 239   {"&real;\0"    ,"\xe2\x84\x9c\0" },
 240   {"&trade;\0"   ,"\xe2\x84\xa2\0" },
 241   {"&alefsym;\0" ,"\xe2\x84\xb5\0" },
 242   {"&larr;\0"    ,"\xe2\x86\x90\0" },
 243   {"&uarr;\0"    ,"\xe2\x86\x91\0" },
 244   {"&rarr;\0"    ,"\xe2\x86\x92\0" },
 245   {"&darr;\0"    ,"\xe2\x86\x93\0" },
 246   {"&harr;\0"    ,"\xe2\x86\x94\0" },
 247   {"&crarr;\0"   ,"\xe2\x86\xb5\0" },
 248   {"&lArr;\0"    ,"\xe2\x87\x90\0" },
 249   {"&uArr;\0"    ,"\xe2\x87\x91\0" },
 250   {"&rArr;\0"    ,"\xe2\x87\x92\0" },
 251   {"&dArr;\0"    ,"\xe2\x87\x93\0" },
 252   {"&hArr;\0"    ,"\xe2\x87\x94\0" },
 253   {"&forall;\0"  ,"\xe2\x88\x80\0" },
 254   {"&part;\0"    ,"\xe2\x88\x82\0" },
 255   {"&exist;\0"   ,"\xe2\x88\x83\0" },
 256   {"&empty;\0"   ,"\xe2\x88\x85\0" },
 257   {"&nabla;\0"   ,"\xe2\x88\x87\0" },
 258   {"&isin;\0"    ,"\xe2\x88\x88\0" },
 259   {"&notin;\0"   ,"\xe2\x88\x89\0" },
 260   {"&ni;\0"      ,"\xe2\x88\x8b\0" },
 261   {"&prod;\0"    ,"\xe2\x88\x8f\0" },
 262   {"&sum;\0"     ,"\xe2\x88\x91\0" },
 263   {"&minus;\0"   ,"\xe2\x88\x92\0" },
 264   {"&lowast;\0"  ,"\xe2\x88\x97\0" },
 265   {"&radic;\0"   ,"\xe2\x88\x9a\0" },
 266   {"&prop;\0"    ,"\xe2\x88\x9d\0" },
 267   {"&infin;\0"   ,"\xe2\x88\x9e\0" },
 268   {"&ang;\0"     ,"\xe2\x88\xa0\0" },
 269   {"&and;\0"     ,"\xe2\x88\xa7\0" },
 270   {"&or;\0"      ,"\xe2\x88\xa8\0" },
 271   {"&cap;\0"     ,"\xe2\x88\xa9\0" },
 272   {"&cup;\0"     ,"\xe2\x88\xaa\0" },
 273   {"&int;\0"     ,"\xe2\x88\xab\0" },
 274   {"&there4;\0"  ,"\xe2\x88\xb4\0" },
 275   {"&sim;\0"     ,"\xe2\x88\xbc\0" },
 276   {"&cong;\0"    ,"\xe2\x89\x85\0" },
 277   {"&asymp;\0"   ,"\xe2\x89\x88\0" },
 278   {"&ne;\0"      ,"\xe2\x89\xa0\0" },
 279   {"&equiv;\0"   ,"\xe2\x89\xa1\0" },
 280   {"&le;\0"      ,"\xe2\x89\xa4\0" },
 281   {"&ge;\0"      ,"\xe2\x89\xa5\0" },
 282   {"&sub;\0"     ,"\xe2\x8a\x82\0" },
 283   {"&sup;\0"     ,"\xe2\x8a\x83\0" },
 284   {"&nsub;\0"    ,"\xe2\x8a\x84\0" },
 285   {"&sube;\0"    ,"\xe2\x8a\x86\0" },
 286   {"&supe;\0"    ,"\xe2\x8a\x87\0" },
 287   {"&oplus;\0"   ,"\xe2\x8a\x95\0" },
 288   {"&otimes;\0"  ,"\xe2\x8a\x97\0" },
 289   {"&perp;\0"    ,"\xe2\x8a\xa5\0" },
 290   {"&sdot;\0"    ,"\xe2\x8b\x85\0" },
 291   {"&lceil;\0"   ,"\xe2\x8c\x88\0" },
 292   {"&rceil;\0"   ,"\xe2\x8c\x89\0" },
 293   {"&lfloor;\0"  ,"\xe2\x8c\x8a\0" },
 294   {"&rfloor;\0"  ,"\xe2\x8c\x8b\0" },
 295   {"&loz;\0"     ,"\xe2\x97\x8a\0" },
 296   {"&spades;\0"  ,"\xe2\x99\xa0\0" },
 297   {"&clubs;\0"   ,"\xe2\x99\xa3\0" },
 298   {"&hearts;\0"  ,"\xe2\x99\xa5\0" },
 299   {"&diams;\0"   ,"\xe2\x99\xa6\0" },
 300   {"&lang;\0"    ,"\xe2\x9f\xa8\0" },
 301   {"&rang;\0"    ,"\xe2\x9f\xa9\0" }
 302 };
 303 // clang-format on
 304
 305 const std::size_t XHTMLENTITY_LOOKUP_COUNT = (sizeof(XHTMLEntityLookupTable)) / (sizeof(XHTMLEntityLookup));
 306
 307 } // unnamed namespace
 308
 309 const char* const NamedEntityToUtf8(const char* const markupText, unsigned int len)
 310 {
 311   // finding if given XHTML named entity is supported or not
 312   for(size_t i = 0; i < XHTMLENTITY_LOOKUP_COUNT; ++i)
 313   {
 314     unsigned int entityLen = strlen(XHTMLEntityLookupTable[i].entityName);
 315     if(len == entityLen)
 316     {
 317       if(strncmp(markupText, XHTMLEntityLookupTable[i].entityName, len) == 0) // if named Entity found in table
 318       {
 319         return XHTMLEntityLookupTable[i].entityCode;
 320       }
 321     }
 322   }
 323   return NULL;
 324 }
 325
 326 } // namespace  Text
 327
 328 } // namespace  Toolkit
 329
 330 } // namespace  Dali