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

   1 /*
   2  * Copyright (c) 2017 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
  27 namespace Toolkit
  28 {
  29
  30 namespace Text
  31 {
  32
  33 namespace
  34 {
  35 /**
  36  * Implementation of the XHTML Entity matching
  37  */
  38 struct XHTMLEntityLookup
  39 {
  40     const char* const entityName;  // XHTML Named Entity string
  41     const char* const entityCode;  // Corresponding UTF-8
  42 };
  43
  44 /* table of html name entities supported in DALi
  45  *
  46  * these are stored as pair with Named entity as Key and
  47  * its utf 8 as value
  48  */
  49 const XHTMLEntityLookup XHTMLEntityLookupTable[] =
  50   {
  51   { "&quot;\0"    ,"\x22\0" },
  52   { "&amp;\0"     ,"\x26\0" },
  53   { "&apos;\0"    ,"\x27\0" },
  54   { "&lt;\0"      ,"\x3c\0" },
  55   { "&gt;\0"      ,"\x3e\0" },
  56   { "&nbsp;\0"    ,"\xc2\xa0\0" },
  57   { "&iexcl;\0"   ,"\xc2\xa1\0" },
  58   { "&cent;\0"    ,"\xc2\xa2\0" },
  59   { "&pound;\0"   ,"\xc2\xa3\0" },
  60   { "&curren;\0"  ,"\xc2\xa4\0" },
  61   { "&yen;\0"     ,"\xc2\xa5\0" },
  62   { "&brvbar;\0"  ,"\xc2\xa6\0" },
  63   { "&sect;\0"    ,"\xc2\xa7\0" },
  64   { "&uml;\0"     ,"\xc2\xa8\0" },
  65   { "&copy;\0"    ,"\xc2\xa9\0" },
  66   { "&ordf;\0"    ,"\xc2\xaa\0" },
  67   { "&laquo;\0"   ,"\xc2\xab\0" },
  68   { "&not;\0"     ,"\xc2\xac\0" },
  69   { "&shy;\0"     ,"\xc2\xad\0" },
  70   { "&reg;\0"     ,"\xc2\xae\0" },
  71   { "&macr;\0"    ,"\xc2\xaf\0" },
  72   { "&deg;\0"     ,"\xc2\xb0\0" },
  73   { "&plusmn;\0"  ,"\xc2\xb1\0" },
  74   { "&sup2;\0"    ,"\xc2\xb2\0" },
  75   { "&sup3;\0"    ,"\xc2\xb3\0" },
  76   { "&acute;\0"   ,"\xc2\xb4\0" },
  77   { "&micro;\0"   ,"\xc2\xb5\0" },
  78   { "&para;\0"    ,"\xc2\xb6\0" },
  79   { "&middot;\0"  ,"\xc2\xb7\0" },
  80   { "&cedil;\0"   ,"\xc2\xb8\0" },
  81   { "&sup1;\0"    ,"\xc2\xb9\0" },
  82   { "&ordm;\0"    ,"\xc2\xba\0" },
  83   { "&raquo;\0"   ,"\xc2\xbb\0" },
  84   { "&frac14;\0"  ,"\xc2\xbc\0" },
  85   { "&frac12;\0"  ,"\xc2\xbd\0" },
  86   { "&frac34;\0"  ,"\xc2\xbe\0" },
  87   { "&iquest;\0"  ,"\xc2\xbf\0" },
  88   { "&Agrave;\0"  ,"\xc3\x80\0" },
  89   { "&Aacute;\0"  ,"\xc3\x81\0" },
  90   { "&Acirc;\0"   ,"\xc3\x82\0" },
  91   { "&Atilde;\0"  ,"\xc3\x83\0" },
  92   { "&Auml;\0"    ,"\xc3\x84\0" },
  93   { "&Aring;\0"   ,"\xc3\x85\0" },
  94   { "&AElig;\0"   ,"\xc3\x86\0" },
  95   { "&Ccedil;\0"  ,"\xc3\x87\0" },
  96   { "&Egrave;\0"  ,"\xc3\x88\0" },
  97   { "&Eacute;\0"  ,"\xc3\x89\0" },
  98   { "&Ecirc;\0"   ,"\xc3\x8a\0" },
  99   { "&Euml;\0"    ,"\xc3\x8b\0" },
 100   { "&Igrave;\0"  ,"\xc3\x8c\0" },
 101   { "&Iacute;\0"  ,"\xc3\x8d\0" },
 102   { "&Icirc;\0"   ,"\xc3\x8e\0" },
 103   { "&Iuml;\0"    ,"\xc3\x8f\0" },
 104   { "&ETH;\0"     ,"\xc3\x90\0" },
 105   { "&Ntilde;\0"  ,"\xc3\x91\0" },
 106   { "&Ograve;\0"  ,"\xc3\x92\0" },
 107   { "&Oacute;\0"  ,"\xc3\x93\0" },
 108   { "&Ocirc;\0"   ,"\xc3\x94\0" },
 109   { "&Otilde;\0"  ,"\xc3\x95\0" },
 110   { "&Ouml;\0"    ,"\xc3\x96\0" },
 111   { "&times;\0"   ,"\xc3\x97\0" },
 112   { "&Oslash;\0"  ,"\xc3\x98\0" },
 113   { "&Ugrave;\0"  ,"\xc3\x99\0" },
 114   { "&Uacute;\0"  ,"\xc3\x9a\0" },
 115   { "&Ucirc;\0"   ,"\xc3\x9b\0" },
 116   { "&Uuml;\0"    ,"\xc3\x9c\0" },
 117   { "&Yacute;\0"  ,"\xc3\x9d\0" },
 118   { "&THORN;\0"   ,"\xc3\x9e\0" },
 119   { "&szlig;\0"   ,"\xc3\x9f\0" },
 120   { "&agrave;\0"  ,"\xc3\xa0\0" },
 121   { "&aacute;\0"  ,"\xc3\xa1\0" },
 122   { "&acirc;\0"   ,"\xc3\xa2\0" },
 123   { "&atilde;\0"  ,"\xc3\xa3\0" },
 124   { "&auml;\0"    ,"\xc3\xa4\0" },
 125   { "&aring;\0"   ,"\xc3\xa5\0" },
 126   { "&aelig;\0"   ,"\xc3\xa6\0" },
 127   { "&ccedil;\0"  ,"\xc3\xa7\0" },
 128   { "&egrave;\0"  ,"\xc3\xa8\0" },
 129   { "&eacute;\0"  ,"\xc3\xa9\0" },
 130   { "&ecirc;\0"   ,"\xc3\xaa\0" },
 131   { "&euml;\0"    ,"\xc3\xab\0" },
 132   { "&igrave;\0"  ,"\xc3\xac\0" },
 133   { "&iacute;\0"  ,"\xc3\xad\0" },
 134   { "&icirc;\0"   ,"\xc3\xae\0" },
 135   { "&iuml;\0"    ,"\xc3\xaf\0" },
 136   { "&eth;\0"     ,"\xc3\xb0\0" },
 137   { "&ntilde;\0"  ,"\xc3\xb1\0" },
 138   { "&ograve;\0"  ,"\xc3\xb2\0" },
 139   { "&oacute;\0"  ,"\xc3\xb3\0" },
 140   { "&ocirc;\0"   ,"\xc3\xb4\0" },
 141   { "&otilde;\0"  ,"\xc3\xb5\0" },
 142   { "&ouml;\0"    ,"\xc3\xb6\0" },
 143   { "&divide;\0"  ,"\xc3\xb7\0" },
 144   { "&oslash;\0"  ,"\xc3\xb8\0" },
 145   { "&ugrave;\0"  ,"\xc3\xb9\0" },
 146   { "&uacute;\0"  ,"\xc3\xba\0" },
 147   { "&ucirc;\0"   ,"\xc3\xbb\0" },
 148   { "&uuml;\0"    ,"\xc3\xbc\0" },
 149   { "&yacute;\0"  ,"\xc3\xbd\0" },
 150   { "&thorn;\0"   ,"\xc3\xbe\0" },
 151   { "&yuml;\0"    ,"\xc3\xbf\0" },
 152   { "&OElig;\0"   ,"\xc5\x92\0" },
 153   { "&oelig;\0"   ,"\xc5\x93\0" },
 154   { "&Scaron;\0"  ,"\xc5\xa0\0" },
 155   { "&scaron;\0"  ,"\xc5\xa1\0" },
 156   { "&Yuml;\0"    ,"\xc5\xb8\0" },
 157   { "&fnof;\0"    ,"\xc6\x92\0" },
 158   { "&circ;\0"    ,"\xcb\x86\0" },
 159   { "&tilde;\0"   ,"\xcb\x9c\0" },
 160   { "&Alpha;\0"   ,"\xce\x91\0" },
 161   { "&Beta;\0"    ,"\xce\x92\0" },
 162   { "&Gamma;\0"   ,"\xce\x93\0" },
 163   { "&Delta;\0"   ,"\xce\x94\0" },
 164   { "&Epsilon;\0" ,"\xce\x95\0" },
 165   { "&Zeta;\0"    ,"\xce\x96\0" },
 166   { "&Eta;\0"     ,"\xce\x97\0" },
 167   { "&Theta;\0"   ,"\xce\x98\0" },
 168   { "&Iota;\0"    ,"\xce\x99\0" },
 169   { "&Kappa;\0"   ,"\xce\x9a\0" },
 170   { "&Lambda;\0"  ,"\xce\x9b\0" },
 171   { "&Mu;\0"      ,"\xce\x9c\0" },
 172   { "&Nu;\0"      ,"\xce\x9d\0" },
 173   { "&Xi;\0"      ,"\xce\x9e\0" },
 174   { "&Omicron;\0" ,"\xce\x9f\0" },
 175   { "&Pi;\0"      ,"\xce\xa0\0" },
 176   { "&Rho;\0"     ,"\xce\xa1\0" },
 177   { "&Sigma;\0"   ,"\xce\xa3\0" },
 178   { "&Tau;\0"     ,"\xce\xa4\0" },
 179   { "&Upsilon;\0" ,"\xce\xa5\0" },
 180   { "&Phi;\0"     ,"\xce\xa6\0" },
 181   { "&Chi;\0"     ,"\xce\xa7\0" },
 182   { "&Psi;\0"     ,"\xce\xa8\0" },
 183   { "&Omega;\0"   ,"\xce\xa9\0" },
 184   { "&alpha;\0"   ,"\xce\xb1\0" },
 185   { "&beta;\0"    ,"\xce\xb2\0" },
 186   { "&gamma;\0"   ,"\xce\xb3\0" },
 187   { "&delta;\0"   ,"\xce\xb4\0" },
 188   { "&epsilon;\0" ,"\xce\xb5\0" },
 189   { "&zeta;\0"    ,"\xce\xb6\0" },
 190   { "&eta;\0"     ,"\xce\xb7\0" },
 191   { "&theta;\0"   ,"\xce\xb8\0" },
 192   { "&iota;\0"    ,"\xce\xb9\0" },
 193   { "&kappa;\0"   ,"\xce\xba\0" },
 194   { "&lambda;\0"  ,"\xce\xbb\0" },
 195   { "&mu;\0"      ,"\xce\xbc\0" },
 196   { "&nu;\0"      ,"\xce\xbd\0" },
 197   { "&xi;\0"      ,"\xce\xbe\0" },
 198   { "&omicron;\0" ,"\xce\xbf\0" },
 199   { "&pi;\0"      ,"\xcf\x80\0" },
 200   { "&rho;\0"     ,"\xcf\x81\0" },
 201   { "&sigmaf;\0"  ,"\xcf\x82\0" },
 202   { "&sigma;\0"   ,"\xcf\x83\0" },
 203   { "&tau;\0"     ,"\xcf\x84\0" },
 204   { "&upsilon;\0" ,"\xcf\x85\0" },
 205   { "&phi;\0"     ,"\xcf\x86\0" },
 206   { "&chi;\0"     ,"\xcf\x87\0" },
 207   { "&psi;\0"     ,"\xcf\x88\0" },
 208   { "&omega;\0"   ,"\xcf\x89\0" },
 209   { "&thetasym;\0","\xcf\x91\0" },
 210   { "&upsih;\0"   ,"\xcf\x92\0" },
 211   { "&piv;\0"     ,"\xcf\x96\0" },
 212   { "&ensp;\0"    ,"\xe2\x80\x82\0" },
 213   { "&emsp;\0"    ,"\xe2\x80\x83\0" },
 214   { "&thinsp;\0"  ,"\xe2\x80\x89\0" },
 215   { "&zwnj;\0"    ,"\xe2\x80\x8c\0" },
 216   { "&zwj;\0"     ,"\xe2\x80\x8d\0" },
 217   { "&lrm;\0"     ,"\xe2\x80\x8e\0" },
 218   { "&rlm;\0"     ,"\xe2\x80\x8f\0" },
 219   { "&ndash;\0"   ,"\xe2\x80\x93\0" },
 220   { "&mdash;\0"   ,"\xe2\x80\x94\0" },
 221   { "&lsquo;\0"   ,"\xe2\x80\x98\0" },
 222   { "&rsquo;\0"   ,"\xe2\x80\x99\0" },
 223   { "&sbquo;\0"   ,"\xe2\x80\x9a\0" },
 224   { "&ldquo;\0"   ,"\xe2\x80\x9c\0" },
 225   { "&rdquo;\0"   ,"\xe2\x80\x9d\0" },
 226   { "&bdquo;\0"   ,"\xe2\x80\x9e\0" },
 227   { "&dagger;\0"  ,"\xe2\x80\xa0\0" },
 228   { "&Dagger;\0"  ,"\xe2\x80\xa1\0" },
 229   { "&bull;\0"    ,"\xe2\x80\xa2\0" },
 230   { "&hellip;\0"  ,"\xe2\x80\xa6\0" },
 231   { "&permil;\0"  ,"\xe2\x80\xb0\0" },
 232   { "&prime;\0"   ,"\xe2\x80\xb2\0" },
 233   { "&Prime;\0"   ,"\xe2\x80\xb3\0" },
 234   { "&lsaquo;\0"  ,"\xe2\x80\xb9\0" },
 235   { "&rsaquo;\0"  ,"\xe2\x80\xba\0" },
 236   { "&oline;\0"   ,"\xe2\x80\xbe\0" },
 237   { "&frasl;\0"   ,"\xe2\x81\x84\0" },
 238   { "&euro;\0"    ,"\xe2\x82\xac\0" },
 239   { "&image;\0"   ,"\xe2\x84\x91\0" },
 240   { "&weierp;\0"  ,"\xe2\x84\x98\0" },
 241   { "&real;\0"    ,"\xe2\x84\x9c\0" },
 242   { "&trade;\0"   ,"\xe2\x84\xa2\0" },
 243   { "&alefsym;\0" ,"\xe2\x84\xb5\0" },
 244   { "&larr;\0"    ,"\xe2\x86\x90\0" },
 245   { "&uarr;\0"    ,"\xe2\x86\x91\0" },
 246   { "&rarr;\0"    ,"\xe2\x86\x92\0" },
 247   { "&darr;\0"    ,"\xe2\x86\x93\0" },
 248   { "&harr;\0"    ,"\xe2\x86\x94\0" },
 249   { "&crarr;\0"   ,"\xe2\x86\xb5\0" },
 250   { "&lArr;\0"    ,"\xe2\x87\x90\0" },
 251   { "&uArr;\0"    ,"\xe2\x87\x91\0" },
 252   { "&rArr;\0"    ,"\xe2\x87\x92\0" },
 253   { "&dArr;\0"    ,"\xe2\x87\x93\0" },
 254   { "&hArr;\0"    ,"\xe2\x87\x94\0" },
 255   { "&forall;\0"  ,"\xe2\x88\x80\0" },
 256   { "&part;\0"    ,"\xe2\x88\x82\0" },
 257   { "&exist;\0"   ,"\xe2\x88\x83\0" },
 258   { "&empty;\0"   ,"\xe2\x88\x85\0" },
 259   { "&nabla;\0"   ,"\xe2\x88\x87\0" },
 260   { "&isin;\0"    ,"\xe2\x88\x88\0" },
 261   { "&notin;\0"   ,"\xe2\x88\x89\0" },
 262   { "&ni;\0"      ,"\xe2\x88\x8b\0" },
 263   { "&prod;\0"    ,"\xe2\x88\x8f\0" },
 264   { "&sum;\0"     ,"\xe2\x88\x91\0" },
 265   { "&minus;\0"   ,"\xe2\x88\x92\0" },
 266   { "&lowast;\0"  ,"\xe2\x88\x97\0" },
 267   { "&radic;\0"   ,"\xe2\x88\x9a\0" },
 268   { "&prop;\0"    ,"\xe2\x88\x9d\0" },
 269   { "&infin;\0"   ,"\xe2\x88\x9e\0" },
 270   { "&ang;\0"     ,"\xe2\x88\xa0\0" },
 271   { "&and;\0"     ,"\xe2\x88\xa7\0" },
 272   { "&or;\0"      ,"\xe2\x88\xa8\0" },
 273   { "&cap;\0"     ,"\xe2\x88\xa9\0" },
 274   { "&cup;\0"     ,"\xe2\x88\xaa\0" },
 275   { "&int;\0"     ,"\xe2\x88\xab\0" },
 276   { "&there4;\0"  ,"\xe2\x88\xb4\0" },
 277   { "&sim;\0"     ,"\xe2\x88\xbc\0" },
 278   { "&cong;\0"    ,"\xe2\x89\x85\0" },
 279   { "&asymp;\0"   ,"\xe2\x89\x88\0" },
 280   { "&ne;\0"      ,"\xe2\x89\xa0\0" },
 281   { "&equiv;\0"   ,"\xe2\x89\xa1\0" },
 282   { "&le;\0"      ,"\xe2\x89\xa4\0" },
 283   { "&ge;\0"      ,"\xe2\x89\xa5\0" },
 284   { "&sub;\0"     ,"\xe2\x8a\x82\0" },
 285   { "&sup;\0"     ,"\xe2\x8a\x83\0" },
 286   { "&nsub;\0"    ,"\xe2\x8a\x84\0" },
 287   { "&sube;\0"    ,"\xe2\x8a\x86\0" },
 288   { "&supe;\0"    ,"\xe2\x8a\x87\0" },
 289   { "&oplus;\0"   ,"\xe2\x8a\x95\0" },
 290   { "&otimes;\0"  ,"\xe2\x8a\x97\0" },
 291   { "&perp;\0"    ,"\xe2\x8a\xa5\0" },
 292   { "&sdot;\0"    ,"\xe2\x8b\x85\0" },
 293   { "&lceil;\0"   ,"\xe2\x8c\x88\0" },
 294   { "&rceil;\0"   ,"\xe2\x8c\x89\0" },
 295   { "&lfloor;\0"  ,"\xe2\x8c\x8a\0" },
 296   { "&rfloor;\0"  ,"\xe2\x8c\x8b\0" },
 297   { "&loz;\0"     ,"\xe2\x97\x8a\0" },
 298   { "&spades;\0"  ,"\xe2\x99\xa0\0" },
 299   { "&clubs;\0"   ,"\xe2\x99\xa3\0" },
 300   { "&hearts;\0"  ,"\xe2\x99\xa5\0" },
 301   { "&diams;\0"   ,"\xe2\x99\xa6\0" },
 302   { "&lang;\0"    ,"\xe2\x9f\xa8\0" },
 303   { "&rang;\0"    ,"\xe2\x9f\xa9\0" }
 304 };
 305
 306 const std::size_t XHTMLENTITY_LOOKUP_COUNT = (sizeof( XHTMLEntityLookupTable))/ (sizeof(XHTMLEntityLookup));
 307
 308 } // unnamed namespace
 309
 310 const char* const  NamedEntityToUtf8( const char* const markupText, unsigned int len )
 311 {
 312   // finding if given XHTML named entity is supported or not
 313   for( size_t i = 0; i < XHTMLENTITY_LOOKUP_COUNT ; ++i )
 314   {
 315     unsigned int entityLen = strlen(XHTMLEntityLookupTable[i].entityName);
 316     if( len == entityLen )
 317     {
 318       if( strncmp( markupText, XHTMLEntityLookupTable[i].entityName, len )  == 0 )  // if named Entity found in table
 319         {
 320           return XHTMLEntityLookupTable[i].entityCode;
 321         }
 322     }
 323   }
 324   return NULL;
 325 }
 326
 327 } // namespace  Text
 328
 329 } // namespace  Toolkit
 330
 331 } // namespace  Dali