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