src/html/htmltag.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        htmltag.cpp
   3 // Purpose:     wxHtmlTag class (represents single tag)
   4 // Author:      Vaclav Slavik
   5 // RCS-ID:      $Id$
   6 // Copyright:   (c) 1999 Vaclav Slavik
   7 // Licence:     wxWindows licence
   8 /////////////////////////////////////////////////////////////////////////////
   9
  10
  11 #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
  12 #pragma implementation "htmltag.h"
  13 #endif
  14
  15 #include "wx/wxprec.h"
  16
  17 #include "wx/defs.h"
  18 #if wxUSE_HTML
  19
  20 #ifdef __BORLANDC__
  21 #pragma hdrstop
  22 #endif
  23
  24 #ifndef WXPRECOMP
  25 #endif
  26
  27 #include "wx/html/htmltag.h"
  28 #include "wx/html/htmlpars.h"
  29 #include "wx/colour.h"
  30 #include <stdio.h> // for vsscanf
  31 #include <stdarg.h>
  32
  33
  34 //-----------------------------------------------------------------------------
  35 // wxHtmlTagsCache
  36 //-----------------------------------------------------------------------------
  37
  38 struct wxHtmlCacheItem
  39 {
  40     // this is "pos" value passed to wxHtmlTag's constructor.
  41     // it is position of '<' character of the tag
  42     int Key;
  43
  44     // end positions for the tag:
  45     // end1 is '<' of ending tag,
  46     // end2 is '>' or both are
  47     // -1 if there is no ending tag for this one...
  48     // or -2 if this is ending tag  </...>
  49     int End1, End2;
  50
  51     // name of this tag
  52     wxChar *Name;
  53 };
  54
  55
  56 IMPLEMENT_CLASS(wxHtmlTagsCache,wxObject)
  57
  58 #define CACHE_INCREMENT  64
  59
  60 bool wxIsCDATAElement(const wxChar *tag)
  61 {
  62     return (wxStrcmp(tag, _T("SCRIPT")) == 0) ||
  63            (wxStrcmp(tag, _T("STYLE")) == 0);
  64 }
  65
  66 wxHtmlTagsCache::wxHtmlTagsCache(const wxString& source)
  67 {
  68     const wxChar *src = source.c_str();
  69     int lng = source.Length();
  70     wxChar tagBuffer[256];
  71
  72     m_Cache = NULL;
  73     m_CacheSize = 0;
  74     m_CachePos = 0;
  75
  76     int pos = 0;
  77     while (pos < lng)
  78     {
  79         if (src[pos] == wxT('<'))   // tag found:
  80         {
  81             if (m_CacheSize % CACHE_INCREMENT == 0)
  82                 m_Cache = (wxHtmlCacheItem*) realloc(m_Cache, (m_CacheSize + CACHE_INCREMENT) * sizeof(wxHtmlCacheItem));
  83             int tg = m_CacheSize++;
  84             int stpos = pos++;
  85             m_Cache[tg].Key = stpos;
  86
  87             int i;
  88             for ( i = 0;
  89                   pos < lng && i < (int)WXSIZEOF(tagBuffer) - 1 &&
  90                   src[pos] != wxT('>') && !wxIsspace(src[pos]);
  91                   i++, pos++ )
  92             {
  93                 tagBuffer[i] = (wxChar)wxToupper(src[pos]);
  94             }
  95             tagBuffer[i] = _T('\0');
  96
  97             m_Cache[tg].Name = new wxChar[i+1];
  98             memcpy(m_Cache[tg].Name, tagBuffer, (i+1)*sizeof(wxChar));
  99
 100             while (pos < lng && src[pos] != wxT('>')) pos++;
 101
 102             if (src[stpos+1] == wxT('/')) // ending tag:
 103             {
 104                 m_Cache[tg].End1 = m_Cache[tg].End2 = -2;
 105                 // find matching begin tag:
 106                 for (i = tg; i >= 0; i--)
 107                     if ((m_Cache[i].End1 == -1) && (wxStrcmp(m_Cache[i].Name, tagBuffer+1) == 0))
 108                     {
 109                         m_Cache[i].End1 = stpos;
 110                         m_Cache[i].End2 = pos + 1;
 111                         break;
 112                     }
 113             }
 114             else
 115             {
 116                 m_Cache[tg].End1 = m_Cache[tg].End2 = -1;
 117
 118                 if (wxIsCDATAElement(tagBuffer))
 119                 {
 120                     // store the orig pos in case we are missing the closing
 121                     // tag (see below)
 122                     wxInt32 old_pos = pos;
 123                     bool foundCloseTag = false;
 124
 125                     // find next matching tag
 126                     int tag_len = wxStrlen(tagBuffer);
 127                     while (pos < lng)
 128                     {
 129                         // find the ending tag
 130                         while (pos + 1 < lng &&
 131                                (src[pos] != '<' || src[pos+1] != '/'))
 132                             ++pos;
 133                         if (src[pos] == '<')
 134                             ++pos;
 135
 136                         // see if it matches
 137                         int match_pos = 0;
 138                         while (pos < lng && match_pos < tag_len && src[pos] != '>' && src[pos] != '<') {
 139                             // cast to wxChar needed to suppress warning in
 140                             // Unicode build
 141                             if ((wxChar)wxToupper(src[pos]) == tagBuffer[match_pos]) {
 142                                 ++match_pos;
 143                             }
 144                             else if (src[pos] == wxT(' ') || src[pos] == wxT('\n') ||
 145                                 src[pos] == wxT('\r') || src[pos] == wxT('\t')) {
 146                                 // need to skip over these
 147                             }
 148                             else {
 149                                 match_pos = 0;
 150                             }
 151                             ++pos;
 152                         }
 153
 154                         // found a match
 155                         if (match_pos == tag_len)
 156                         {
 157                             pos = pos - tag_len - 3;
 158                             foundCloseTag = true;
 159                             break;
 160                         }
 161                         else // keep looking for the closing tag
 162                         {
 163                             ++pos;
 164                         }
 165                     }
 166                     if (!foundCloseTag)
 167                     {
 168                         // we didn't find closing tag; this means the markup
 169                         // is incorrect and the best thing we can do is to
 170                         // ignore the unclosed tag and continue parsing as if
 171                         // it didn't exist:
 172                         pos = old_pos;
 173                     }
 174                 }
 175             }
 176         }
 177
 178         pos++;
 179     }
 180
 181     // ok, we're done, now we'll free .Name members of cache - we don't need it anymore:
 182     for (int i = 0; i < m_CacheSize; i++)
 183     {
 184         delete[] m_Cache[i].Name;
 185         m_Cache[i].Name = NULL;
 186     }
 187 }
 188
 189 void wxHtmlTagsCache::QueryTag(int at, int* end1, int* end2)
 190 {
 191     if (m_Cache == NULL) return;
 192     if (m_Cache[m_CachePos].Key != at)
 193     {
 194         int delta = (at < m_Cache[m_CachePos].Key) ? -1 : 1;
 195         do
 196         {
 197             m_CachePos += delta;
 198         }
 199         while (m_Cache[m_CachePos].Key != at);
 200     }
 201     *end1 = m_Cache[m_CachePos].End1;
 202     *end2 = m_Cache[m_CachePos].End2;
 203 }
 204
 205
 206
 207
 208 //-----------------------------------------------------------------------------
 209 // wxHtmlTag
 210 //-----------------------------------------------------------------------------
 211
 212 IMPLEMENT_CLASS(wxHtmlTag,wxObject)
 213
 214 wxHtmlTag::wxHtmlTag(wxHtmlTag *parent,
 215                      const wxString& source, int pos, int end_pos,
 216                      wxHtmlTagsCache *cache,
 217                      wxHtmlEntitiesParser *entParser) : wxObject()
 218 {
 219     /* Setup DOM relations */
 220
 221     m_Next = NULL;
 222     m_FirstChild = m_LastChild = NULL;
 223     m_Parent = parent;
 224     if (parent)
 225     {
 226         m_Prev = m_Parent->m_LastChild;
 227         if (m_Prev == NULL)
 228             m_Parent->m_FirstChild = this;
 229         else
 230             m_Prev->m_Next = this;
 231         m_Parent->m_LastChild = this;
 232     }
 233     else
 234         m_Prev = NULL;
 235
 236     /* Find parameters and their values: */
 237
 238     int i;
 239     wxChar c;
 240
 241     // fill-in name, params and begin pos:
 242     i = pos+1;
 243
 244     // find tag's name and convert it to uppercase:
 245     while ((i < end_pos) &&
 246            ((c = source[i++]) != wxT(' ') && c != wxT('\r') &&
 247              c != wxT('\n') && c != wxT('\t') &&
 248              c != wxT('>')))
 249     {
 250         if ((c >= wxT('a')) && (c <= wxT('z')))
 251             c -= (wxT('a') - wxT('A'));
 252         m_Name << c;
 253     }
 254
 255     // if the tag has parameters, read them and "normalize" them,
 256     // i.e. convert to uppercase, replace whitespaces by spaces and
 257     // remove whitespaces around '=':
 258     if (source[i-1] != wxT('>'))
 259     {
 260         #define IS_WHITE(c) (c == wxT(' ') || c == wxT('\r') || \
 261                              c == wxT('\n') || c == wxT('\t'))
 262         wxString pname, pvalue;
 263         wxChar quote;
 264         enum
 265         {
 266             ST_BEFORE_NAME = 1,
 267             ST_NAME,
 268             ST_BEFORE_EQ,
 269             ST_BEFORE_VALUE,
 270             ST_VALUE
 271         } state;
 272
 273         quote = 0;
 274         state = ST_BEFORE_NAME;
 275         while (i < end_pos)
 276         {
 277             c = source[i++];
 278
 279             if (c == wxT('>') && !(state == ST_VALUE && quote != 0))
 280             {
 281                 if (state == ST_BEFORE_EQ || state == ST_NAME)
 282                 {
 283                     m_ParamNames.Add(pname);
 284                     m_ParamValues.Add(wxEmptyString);
 285                 }
 286                 else if (state == ST_VALUE && quote == 0)
 287                 {
 288                     m_ParamNames.Add(pname);
 289                     if (entParser)
 290                         m_ParamValues.Add(entParser->Parse(pvalue));
 291                     else
 292                         m_ParamValues.Add(pvalue);
 293                 }
 294                 break;
 295             }
 296             switch (state)
 297             {
 298                 case ST_BEFORE_NAME:
 299                     if (!IS_WHITE(c))
 300                     {
 301                         pname = c;
 302                         state = ST_NAME;
 303                     }
 304                     break;
 305                 case ST_NAME:
 306                     if (IS_WHITE(c))
 307                         state = ST_BEFORE_EQ;
 308                     else if (c == wxT('='))
 309                         state = ST_BEFORE_VALUE;
 310                     else
 311                         pname << c;
 312                     break;
 313                 case ST_BEFORE_EQ:
 314                     if (c == wxT('='))
 315                         state = ST_BEFORE_VALUE;
 316                     else if (!IS_WHITE(c))
 317                     {
 318                         m_ParamNames.Add(pname);
 319                         m_ParamValues.Add(wxEmptyString);
 320                         pname = c;
 321                         state = ST_NAME;
 322                     }
 323                     break;
 324                 case ST_BEFORE_VALUE:
 325                     if (!IS_WHITE(c))
 326                     {
 327                         if (c == wxT('"') || c == wxT('\''))
 328                             quote = c, pvalue = wxEmptyString;
 329                         else
 330                             quote = 0, pvalue = c;
 331                         state = ST_VALUE;
 332                     }
 333                     break;
 334                 case ST_VALUE:
 335                     if ((quote != 0 && c == quote) ||
 336                         (quote == 0 && IS_WHITE(c)))
 337                     {
 338                         m_ParamNames.Add(pname);
 339                         if (quote == 0)
 340                         {
 341                             // VS: backward compatibility, no real reason,
 342                             //     but wxHTML code relies on this... :(
 343                             pvalue.MakeUpper();
 344                         }
 345                         if (entParser)
 346                             m_ParamValues.Add(entParser->Parse(pvalue));
 347                         else
 348                             m_ParamValues.Add(pvalue);
 349                         state = ST_BEFORE_NAME;
 350                     }
 351                     else
 352                         pvalue << c;
 353                     break;
 354             }
 355         }
 356
 357         #undef IS_WHITE
 358    }
 359    m_Begin = i;
 360
 361    cache->QueryTag(pos, &m_End1, &m_End2);
 362    if (m_End1 > end_pos) m_End1 = end_pos;
 363    if (m_End2 > end_pos) m_End2 = end_pos;
 364 }
 365
 366 wxHtmlTag::~wxHtmlTag()
 367 {
 368     wxHtmlTag *t1, *t2;
 369     t1 = m_FirstChild;
 370     while (t1)
 371     {
 372         t2 = t1->GetNextSibling();
 373         delete t1;
 374         t1 = t2;
 375     }
 376 }
 377
 378 bool wxHtmlTag::HasParam(const wxString& par) const
 379 {
 380     return (m_ParamNames.Index(par, false) != wxNOT_FOUND);
 381 }
 382
 383 wxString wxHtmlTag::GetParam(const wxString& par, bool with_commas) const
 384 {
 385     int index = m_ParamNames.Index(par, false);
 386     if (index == wxNOT_FOUND)
 387         return wxEmptyString;
 388     if (with_commas)
 389     {
 390         // VS: backward compatibility, seems to be never used by wxHTML...
 391         wxString s;
 392         s << wxT('"') << m_ParamValues[index] << wxT('"');
 393         return s;
 394     }
 395     else
 396         return m_ParamValues[index];
 397 }
 398
 399 int wxHtmlTag::ScanParam(const wxString& par,
 400                          const wxChar *format,
 401                          void *param) const
 402 {
 403     wxString parval = GetParam(par);
 404     return wxSscanf(parval, format, param);
 405 }
 406
 407 bool wxHtmlTag::GetParamAsColour(const wxString& par, wxColour *clr) const
 408 {
 409     wxString str = GetParam(par);
 410
 411     if (str.empty()) return false;
 412     if (str.GetChar(0) == wxT('#'))
 413     {
 414         unsigned long tmp;
 415         if (ScanParam(par, wxT("#%lX"), &tmp) != 1)
 416             return false;
 417         *clr = wxColour((unsigned char)((tmp & 0xFF0000) >> 16),
 418                         (unsigned char)((tmp & 0x00FF00) >> 8),
 419                         (unsigned char)(tmp & 0x0000FF));
 420         return true;
 421     }
 422     else
 423     {
 424         // Handle colours defined in HTML 4.0:
 425         #define HTML_COLOUR(name,r,g,b)                 \
 426             if (str.IsSameAs(wxT(name), false))         \
 427                 { *clr = wxColour(r,g,b); return true; }
 428         HTML_COLOUR("black",   0x00,0x00,0x00)
 429         HTML_COLOUR("silver",  0xC0,0xC0,0xC0)
 430         HTML_COLOUR("gray",    0x80,0x80,0x80)
 431         HTML_COLOUR("white",   0xFF,0xFF,0xFF)
 432         HTML_COLOUR("maroon",  0x80,0x00,0x00)
 433         HTML_COLOUR("red",     0xFF,0x00,0x00)
 434         HTML_COLOUR("purple",  0x80,0x00,0x80)
 435         HTML_COLOUR("fuchsia", 0xFF,0x00,0xFF)
 436         HTML_COLOUR("green",   0x00,0x80,0x00)
 437         HTML_COLOUR("lime",    0x00,0xFF,0x00)
 438         HTML_COLOUR("olive",   0x80,0x80,0x00)
 439         HTML_COLOUR("yellow",  0xFF,0xFF,0x00)
 440         HTML_COLOUR("navy",    0x00,0x00,0x80)
 441         HTML_COLOUR("blue",    0x00,0x00,0xFF)
 442         HTML_COLOUR("teal",    0x00,0x80,0x80)
 443         HTML_COLOUR("aqua",    0x00,0xFF,0xFF)
 444         #undef HTML_COLOUR
 445     }
 446
 447     return false;
 448 }
 449
 450 bool wxHtmlTag::GetParamAsInt(const wxString& par, int *clr) const
 451 {
 452     if (!HasParam(par)) return false;
 453     long i;
 454     bool succ = GetParam(par).ToLong(&i);
 455     *clr = (int)i;
 456     return succ;
 457 }
 458
 459 wxString wxHtmlTag::GetAllParams() const
 460 {
 461     // VS: this function is for backward compatibility only,
 462     //     never used by wxHTML
 463     wxString s;
 464     size_t cnt = m_ParamNames.GetCount();
 465     for (size_t i = 0; i < cnt; i++)
 466     {
 467         s << m_ParamNames[i];
 468         s << wxT('=');
 469         if (m_ParamValues[i].Find(wxT('"')) != wxNOT_FOUND)
 470             s << wxT('\'') << m_ParamValues[i] << wxT('\'');
 471         else
 472             s << wxT('"') << m_ParamValues[i] << wxT('"');
 473     }
 474     return s;
 475 }
 476
 477 wxHtmlTag *wxHtmlTag::GetFirstSibling() const
 478 {
 479     if (m_Parent)
 480         return m_Parent->m_FirstChild;
 481     else
 482     {
 483         wxHtmlTag *cur = (wxHtmlTag*)this;
 484         while (cur->m_Prev)
 485             cur = cur->m_Prev;
 486         return cur;
 487     }
 488 }
 489
 490 wxHtmlTag *wxHtmlTag::GetLastSibling() const
 491 {
 492     if (m_Parent)
 493         return m_Parent->m_LastChild;
 494     else
 495     {
 496         wxHtmlTag *cur = (wxHtmlTag*)this;
 497         while (cur->m_Next)
 498             cur = cur->m_Next;
 499         return cur;
 500     }
 501 }
 502
 503 wxHtmlTag *wxHtmlTag::GetNextTag() const
 504 {
 505     if (m_FirstChild) return m_FirstChild;
 506     if (m_Next) return m_Next;
 507     wxHtmlTag *cur = m_Parent;
 508     if (!cur) return NULL;
 509     while (cur->m_Parent && !cur->m_Next)
 510         cur = cur->m_Parent;
 511     return cur->m_Next;
 512 }
 513
 514 #if WXWIN_COMPATIBILITY_2_2
 515
 516 bool wxHtmlTag::IsEnding() const
 517 {
 518     return false;
 519 }
 520
 521 #endif // WXWIN_COMPATIBILITY_2_2
 522
 523 #endif