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] = 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                     // find next matching tag
 121                     int tag_len = wxStrlen(tagBuffer);
 122                     while (pos < lng)
 123                     {
 124                         // find the ending tag
 125                         while (pos + 1 < lng &&
 126                                (src[pos] != '<' || src[pos+1] != '/'))
 127                             ++pos;
 128                         if (src[pos] == '<')
 129                             ++pos;
 130
 131                         // see if it matches
 132                         int match_pos = 0;
 133                         while (pos < lng && match_pos < tag_len && src[pos] != '>' && src[pos] != '<') {
 134                             // cast to wxChar needed to suppress warning in
 135                             // Unicode build
 136                             if ((wxChar)wxToupper(src[pos]) == tagBuffer[match_pos]) {
 137                                 ++match_pos;
 138                             }
 139                             else if (src[pos] == wxT(' ') || src[pos] == wxT('\n') ||
 140                                 src[pos] == wxT('\r') || src[pos] == wxT('\t')) {
 141                                 // need to skip over these
 142                             }
 143                             else {
 144                                 match_pos = 0;
 145                             }
 146                             ++pos;
 147                         }
 148
 149                         // found a match
 150                         if (match_pos == tag_len) {
 151                             // These 2 lines commented:
 152                             //pos = pos - tag_len - 3; pos is not used outside of while loop
 153                             //stpos = pos; Value is local to while loop, no need to set it
 154                             break;
 155                         }
 156                         else {
 157                             ++pos;
 158                         }
 159                     }
 160                 }
 161             }
 162         }
 163
 164         pos++;
 165     }
 166
 167     // ok, we're done, now we'll free .Name members of cache - we don't need it anymore:
 168     for (int i = 0; i < m_CacheSize; i++)
 169     {
 170         delete[] m_Cache[i].Name;
 171         m_Cache[i].Name = NULL;
 172     }
 173 }
 174
 175 void wxHtmlTagsCache::QueryTag(int at, int* end1, int* end2)
 176 {
 177     if (m_Cache == NULL) return;
 178     if (m_Cache[m_CachePos].Key != at)
 179     {
 180         int delta = (at < m_Cache[m_CachePos].Key) ? -1 : 1;
 181         do
 182         {
 183             m_CachePos += delta;
 184         }
 185         while (m_Cache[m_CachePos].Key != at);
 186     }
 187     *end1 = m_Cache[m_CachePos].End1;
 188     *end2 = m_Cache[m_CachePos].End2;
 189 }
 190
 191
 192
 193
 194 //-----------------------------------------------------------------------------
 195 // wxHtmlTag
 196 //-----------------------------------------------------------------------------
 197
 198 IMPLEMENT_CLASS(wxHtmlTag,wxObject)
 199
 200 wxHtmlTag::wxHtmlTag(wxHtmlTag *parent,
 201                      const wxString& source, int pos, int end_pos,
 202                      wxHtmlTagsCache *cache,
 203                      wxHtmlEntitiesParser *entParser) : wxObject()
 204 {
 205     /* Setup DOM relations */
 206
 207     m_Next = NULL;
 208     m_FirstChild = m_LastChild = NULL;
 209     m_Parent = parent;
 210     if (parent)
 211     {
 212         m_Prev = m_Parent->m_LastChild;
 213         if (m_Prev == NULL)
 214             m_Parent->m_FirstChild = this;
 215         else
 216             m_Prev->m_Next = this;
 217         m_Parent->m_LastChild = this;
 218     }
 219     else
 220         m_Prev = NULL;
 221
 222     /* Find parameters and their values: */
 223
 224     int i;
 225     wxChar c;
 226
 227     // fill-in name, params and begin pos:
 228     i = pos+1;
 229
 230     // find tag's name and convert it to uppercase:
 231     while ((i < end_pos) &&
 232            ((c = source[i++]) != wxT(' ') && c != wxT('\r') &&
 233              c != wxT('\n') && c != wxT('\t') &&
 234              c != wxT('>')))
 235     {
 236         if ((c >= wxT('a')) && (c <= wxT('z')))
 237             c -= (wxT('a') - wxT('A'));
 238         m_Name << c;
 239     }
 240
 241     // if the tag has parameters, read them and "normalize" them,
 242     // i.e. convert to uppercase, replace whitespaces by spaces and
 243     // remove whitespaces around '=':
 244     if (source[i-1] != wxT('>'))
 245     {
 246         #define IS_WHITE(c) (c == wxT(' ') || c == wxT('\r') || \
 247                              c == wxT('\n') || c == wxT('\t'))
 248         wxString pname, pvalue;
 249         wxChar quote;
 250         enum
 251         {
 252             ST_BEFORE_NAME = 1,
 253             ST_NAME,
 254             ST_BEFORE_EQ,
 255             ST_BEFORE_VALUE,
 256             ST_VALUE
 257         } state;
 258
 259         quote = 0;
 260         state = ST_BEFORE_NAME;
 261         while (i < end_pos)
 262         {
 263             c = source[i++];
 264
 265             if (c == wxT('>') && !(state == ST_VALUE && quote != 0))
 266             {
 267                 if (state == ST_BEFORE_EQ || state == ST_NAME)
 268                 {
 269                     m_ParamNames.Add(pname);
 270                     m_ParamValues.Add(wxEmptyString);
 271                 }
 272                 else if (state == ST_VALUE && quote == 0)
 273                 {
 274                     m_ParamNames.Add(pname);
 275                     if (entParser)
 276                         m_ParamValues.Add(entParser->Parse(pvalue));
 277                     else
 278                         m_ParamValues.Add(pvalue);
 279                 }
 280                 break;
 281             }
 282             switch (state)
 283             {
 284                 case ST_BEFORE_NAME:
 285                     if (!IS_WHITE(c))
 286                     {
 287                         pname = c;
 288                         state = ST_NAME;
 289                     }
 290                     break;
 291                 case ST_NAME:
 292                     if (IS_WHITE(c))
 293                         state = ST_BEFORE_EQ;
 294                     else if (c == wxT('='))
 295                         state = ST_BEFORE_VALUE;
 296                     else
 297                         pname << c;
 298                     break;
 299                 case ST_BEFORE_EQ:
 300                     if (c == wxT('='))
 301                         state = ST_BEFORE_VALUE;
 302                     else if (!IS_WHITE(c))
 303                     {
 304                         m_ParamNames.Add(pname);
 305                         m_ParamValues.Add(wxEmptyString);
 306                         pname = c;
 307                         state = ST_NAME;
 308                     }
 309                     break;
 310                 case ST_BEFORE_VALUE:
 311                     if (!IS_WHITE(c))
 312                     {
 313                         if (c == wxT('"') || c == wxT('\''))
 314                             quote = c, pvalue = wxEmptyString;
 315                         else
 316                             quote = 0, pvalue = c;
 317                         state = ST_VALUE;
 318                     }
 319                     break;
 320                 case ST_VALUE:
 321                     if ((quote != 0 && c == quote) ||
 322                         (quote == 0 && IS_WHITE(c)))
 323                     {
 324                         m_ParamNames.Add(pname);
 325                         if (quote == 0)
 326                         {
 327                             // VS: backward compatibility, no real reason,
 328                             //     but wxHTML code relies on this... :(
 329                             pvalue.MakeUpper();
 330                         }
 331                         if (entParser)
 332                             m_ParamValues.Add(entParser->Parse(pvalue));
 333                         else
 334                             m_ParamValues.Add(pvalue);
 335                         state = ST_BEFORE_NAME;
 336                     }
 337                     else
 338                         pvalue << c;
 339                     break;
 340             }
 341         }
 342
 343         #undef IS_WHITE
 344    }
 345    m_Begin = i;
 346
 347    cache->QueryTag(pos, &m_End1, &m_End2);
 348    if (m_End1 > end_pos) m_End1 = end_pos;
 349    if (m_End2 > end_pos) m_End2 = end_pos;
 350 }
 351
 352 wxHtmlTag::~wxHtmlTag()
 353 {
 354     wxHtmlTag *t1, *t2;
 355     t1 = m_FirstChild;
 356     while (t1)
 357     {
 358         t2 = t1->GetNextSibling();
 359         delete t1;
 360         t1 = t2;
 361     }
 362 }
 363
 364 bool wxHtmlTag::HasParam(const wxString& par) const
 365 {
 366     return (m_ParamNames.Index(par, false) != wxNOT_FOUND);
 367 }
 368
 369 wxString wxHtmlTag::GetParam(const wxString& par, bool with_commas) const
 370 {
 371     int index = m_ParamNames.Index(par, false);
 372     if (index == wxNOT_FOUND)
 373         return wxEmptyString;
 374     if (with_commas)
 375     {
 376         // VS: backward compatibility, seems to be never used by wxHTML...
 377         wxString s;
 378         s << wxT('"') << m_ParamValues[index] << wxT('"');
 379         return s;
 380     }
 381     else
 382         return m_ParamValues[index];
 383 }
 384
 385 int wxHtmlTag::ScanParam(const wxString& par,
 386                          const wxChar *format,
 387                          void *param) const
 388 {
 389     wxString parval = GetParam(par);
 390     return wxSscanf(parval, format, param);
 391 }
 392
 393 bool wxHtmlTag::GetParamAsColour(const wxString& par, wxColour *clr) const
 394 {
 395     wxString str = GetParam(par);
 396
 397     if (str.IsEmpty()) return false;
 398     if (str.GetChar(0) == wxT('#'))
 399     {
 400         unsigned long tmp;
 401         if (ScanParam(par, wxT("#%lX"), &tmp) != 1)
 402             return false;
 403         *clr = wxColour((unsigned char)((tmp & 0xFF0000) >> 16),
 404                         (unsigned char)((tmp & 0x00FF00) >> 8),
 405                         (unsigned char)(tmp & 0x0000FF));
 406         return true;
 407     }
 408     else
 409     {
 410         // Handle colours defined in HTML 4.0:
 411         #define HTML_COLOUR(name,r,g,b)                 \
 412             if (str.IsSameAs(wxT(name), false))         \
 413                 { *clr = wxColour(r,g,b); return true; }
 414         HTML_COLOUR("black",   0x00,0x00,0x00)
 415         HTML_COLOUR("silver",  0xC0,0xC0,0xC0)
 416         HTML_COLOUR("gray",    0x80,0x80,0x80)
 417         HTML_COLOUR("white",   0xFF,0xFF,0xFF)
 418         HTML_COLOUR("maroon",  0x80,0x00,0x00)
 419         HTML_COLOUR("red",     0xFF,0x00,0x00)
 420         HTML_COLOUR("purple",  0x80,0x00,0x80)
 421         HTML_COLOUR("fuchsia", 0xFF,0x00,0xFF)
 422         HTML_COLOUR("green",   0x00,0x80,0x00)
 423         HTML_COLOUR("lime",    0x00,0xFF,0x00)
 424         HTML_COLOUR("olive",   0x80,0x80,0x00)
 425         HTML_COLOUR("yellow",  0xFF,0xFF,0x00)
 426         HTML_COLOUR("navy",    0x00,0x00,0x80)
 427         HTML_COLOUR("blue",    0x00,0x00,0xFF)
 428         HTML_COLOUR("teal",    0x00,0x80,0x80)
 429         HTML_COLOUR("aqua",    0x00,0xFF,0xFF)
 430         #undef HTML_COLOUR
 431     }
 432
 433     return false;
 434 }
 435
 436 bool wxHtmlTag::GetParamAsInt(const wxString& par, int *clr) const
 437 {
 438     if (!HasParam(par)) return false;
 439     long i;
 440     bool succ = GetParam(par).ToLong(&i);
 441     *clr = (int)i;
 442     return succ;
 443 }
 444
 445 wxString wxHtmlTag::GetAllParams() const
 446 {
 447     // VS: this function is for backward compatiblity only,
 448     //     never used by wxHTML
 449     wxString s;
 450     size_t cnt = m_ParamNames.GetCount();
 451     for (size_t i = 0; i < cnt; i++)
 452     {
 453         s << m_ParamNames[i];
 454         s << wxT('=');
 455         if (m_ParamValues[i].Find(wxT('"')) != wxNOT_FOUND)
 456             s << wxT('\'') << m_ParamValues[i] << wxT('\'');
 457         else
 458             s << wxT('"') << m_ParamValues[i] << wxT('"');
 459     }
 460     return s;
 461 }
 462
 463 wxHtmlTag *wxHtmlTag::GetFirstSibling() const
 464 {
 465     if (m_Parent)
 466         return m_Parent->m_FirstChild;
 467     else
 468     {
 469         wxHtmlTag *cur = (wxHtmlTag*)this;
 470         while (cur->m_Prev)
 471             cur = cur->m_Prev;
 472         return cur;
 473     }
 474 }
 475
 476 wxHtmlTag *wxHtmlTag::GetLastSibling() const
 477 {
 478     if (m_Parent)
 479         return m_Parent->m_LastChild;
 480     else
 481     {
 482         wxHtmlTag *cur = (wxHtmlTag*)this;
 483         while (cur->m_Next)
 484             cur = cur->m_Next;
 485         return cur;
 486     }
 487 }
 488
 489 wxHtmlTag *wxHtmlTag::GetNextTag() const
 490 {
 491     if (m_FirstChild) return m_FirstChild;
 492     if (m_Next) return m_Next;
 493     wxHtmlTag *cur = m_Parent;
 494     if (!cur) return NULL;
 495     while (cur->m_Parent && !cur->m_Next)
 496         cur = cur->m_Parent;
 497     return cur->m_Next;
 498 }
 499
 500 #endif