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 tg, stpos;
  70     int lng = source.Length();
  71     wxChar tagBuffer[256];
  72
  73     m_Cache = NULL;
  74     m_CacheSize = 0;
  75     m_CachePos = 0;
  76
  77     int pos = 0;
  78     while (pos < lng)
  79     {
  80         if (src[pos] == wxT('<'))   // tag found:
  81         {
  82             if (m_CacheSize % CACHE_INCREMENT == 0)
  83                 m_Cache = (wxHtmlCacheItem*) realloc(m_Cache, (m_CacheSize + CACHE_INCREMENT) * sizeof(wxHtmlCacheItem));
  84             tg = m_CacheSize++;
  85             m_Cache[tg].Key = stpos = pos++;
  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                             pos = pos - tag_len - 3;
 152                             stpos = pos;
 153                             break;
 154                         }
 155                         else {
 156                             ++pos;
 157                         }
 158                     }
 159                 }
 160             }
 161         }
 162
 163         pos++;
 164     }
 165
 166     // ok, we're done, now we'll free .Name members of cache - we don't need it anymore:
 167     for (int i = 0; i < m_CacheSize; i++)
 168     {
 169         delete[] m_Cache[i].Name;
 170         m_Cache[i].Name = NULL;
 171     }
 172 }
 173
 174 void wxHtmlTagsCache::QueryTag(int at, int* end1, int* end2)
 175 {
 176     if (m_Cache == NULL) return;
 177     if (m_Cache[m_CachePos].Key != at)
 178     {
 179         int delta = (at < m_Cache[m_CachePos].Key) ? -1 : 1;
 180         do
 181         {
 182             m_CachePos += delta;
 183         }
 184         while (m_Cache[m_CachePos].Key != at);
 185     }
 186     *end1 = m_Cache[m_CachePos].End1;
 187     *end2 = m_Cache[m_CachePos].End2;
 188 }
 189
 190
 191
 192
 193 //-----------------------------------------------------------------------------
 194 // wxHtmlTag
 195 //-----------------------------------------------------------------------------
 196
 197 IMPLEMENT_CLASS(wxHtmlTag,wxObject)
 198
 199 wxHtmlTag::wxHtmlTag(wxHtmlTag *parent,
 200                      const wxString& source, int pos, int end_pos,
 201                      wxHtmlTagsCache *cache,
 202                      wxHtmlEntitiesParser *entParser) : wxObject()
 203 {
 204     /* Setup DOM relations */
 205
 206     m_Next = NULL;
 207     m_FirstChild = m_LastChild = NULL;
 208     m_Parent = parent;
 209     if (parent)
 210     {
 211         m_Prev = m_Parent->m_LastChild;
 212         if (m_Prev == NULL)
 213             m_Parent->m_FirstChild = this;
 214         else
 215             m_Prev->m_Next = this;
 216         m_Parent->m_LastChild = this;
 217     }
 218     else
 219         m_Prev = NULL;
 220
 221     /* Find parameters and their values: */
 222
 223     int i;
 224     wxChar c;
 225
 226     // fill-in name, params and begin pos:
 227     i = pos+1;
 228
 229     // find tag's name and convert it to uppercase:
 230     while ((i < end_pos) &&
 231            ((c = source[i++]) != wxT(' ') && c != wxT('\r') &&
 232              c != wxT('\n') && c != wxT('\t') &&
 233              c != wxT('>')))
 234     {
 235         if ((c >= wxT('a')) && (c <= wxT('z')))
 236             c -= (wxT('a') - wxT('A'));
 237         m_Name << c;
 238     }
 239
 240     // if the tag has parameters, read them and "normalize" them,
 241     // i.e. convert to uppercase, replace whitespaces by spaces and
 242     // remove whitespaces around '=':
 243     if (source[i-1] != wxT('>'))
 244     {
 245         #define IS_WHITE(c) (c == wxT(' ') || c == wxT('\r') || \
 246                              c == wxT('\n') || c == wxT('\t'))
 247         wxString pname, pvalue;
 248         wxChar quote;
 249         enum
 250         {
 251             ST_BEFORE_NAME = 1,
 252             ST_NAME,
 253             ST_BEFORE_EQ,
 254             ST_BEFORE_VALUE,
 255             ST_VALUE
 256         } state;
 257
 258         quote = 0;
 259         state = ST_BEFORE_NAME;
 260         while (i < end_pos)
 261         {
 262             c = source[i++];
 263
 264             if (c == wxT('>') && !(state == ST_VALUE && quote != 0))
 265             {
 266                 if (state == ST_BEFORE_EQ || state == ST_NAME)
 267                 {
 268                     m_ParamNames.Add(pname);
 269                     m_ParamValues.Add(wxEmptyString);
 270                 }
 271                 else if (state == ST_VALUE && quote == 0)
 272                 {
 273                     m_ParamNames.Add(pname);
 274                     if (entParser)
 275                         m_ParamValues.Add(entParser->Parse(pvalue));
 276                     else
 277                         m_ParamValues.Add(pvalue);
 278                 }
 279                 break;
 280             }
 281             switch (state)
 282             {
 283                 case ST_BEFORE_NAME:
 284                     if (!IS_WHITE(c))
 285                     {
 286                         pname = c;
 287                         state = ST_NAME;
 288                     }
 289                     break;
 290                 case ST_NAME:
 291                     if (IS_WHITE(c))
 292                         state = ST_BEFORE_EQ;
 293                     else if (c == wxT('='))
 294                         state = ST_BEFORE_VALUE;
 295                     else
 296                         pname << c;
 297                     break;
 298                 case ST_BEFORE_EQ:
 299                     if (c == wxT('='))
 300                         state = ST_BEFORE_VALUE;
 301                     else if (!IS_WHITE(c))
 302                     {
 303                         m_ParamNames.Add(pname);
 304                         m_ParamValues.Add(wxEmptyString);
 305                         pname = c;
 306                         state = ST_NAME;
 307                     }
 308                     break;
 309                 case ST_BEFORE_VALUE:
 310                     if (!IS_WHITE(c))
 311                     {
 312                         if (c == wxT('"') || c == wxT('\''))
 313                             quote = c, pvalue = wxEmptyString;
 314                         else
 315                             quote = 0, pvalue = c;
 316                         state = ST_VALUE;
 317                     }
 318                     break;
 319                 case ST_VALUE:
 320                     if ((quote != 0 && c == quote) ||
 321                         (quote == 0 && IS_WHITE(c)))
 322                     {
 323                         m_ParamNames.Add(pname);
 324                         if (quote == 0)
 325                         {
 326                             // VS: backward compatibility, no real reason,
 327                             //     but wxHTML code relies on this... :(
 328                             pvalue.MakeUpper();
 329                         }
 330                         if (entParser)
 331                             m_ParamValues.Add(entParser->Parse(pvalue));
 332                         else
 333                             m_ParamValues.Add(pvalue);
 334                         state = ST_BEFORE_NAME;
 335                     }
 336                     else
 337                         pvalue << c;
 338                     break;
 339             }
 340         }
 341
 342         #undef IS_WHITE
 343    }
 344    m_Begin = i;
 345
 346    cache->QueryTag(pos, &m_End1, &m_End2);
 347    if (m_End1 > end_pos) m_End1 = end_pos;
 348    if (m_End2 > end_pos) m_End2 = end_pos;
 349 }
 350
 351 wxHtmlTag::~wxHtmlTag()
 352 {
 353     wxHtmlTag *t1, *t2;
 354     t1 = m_FirstChild;
 355     while (t1)
 356     {
 357         t2 = t1->GetNextSibling();
 358         delete t1;
 359         t1 = t2;
 360     }
 361 }
 362
 363 bool wxHtmlTag::HasParam(const wxString& par) const
 364 {
 365     return (m_ParamNames.Index(par, FALSE) != wxNOT_FOUND);
 366 }
 367
 368 wxString wxHtmlTag::GetParam(const wxString& par, bool with_commas) const
 369 {
 370     int index = m_ParamNames.Index(par, FALSE);
 371     if (index == wxNOT_FOUND)
 372         return wxEmptyString;
 373     if (with_commas)
 374     {
 375         // VS: backward compatibility, seems to be never used by wxHTML...
 376         wxString s;
 377         s << wxT('"') << m_ParamValues[index] << wxT('"');
 378         return s;
 379     }
 380     else
 381         return m_ParamValues[index];
 382 }
 383
 384 int wxHtmlTag::ScanParam(const wxString& par,
 385                          const wxChar *format,
 386                          void *param) const
 387 {
 388     wxString parval = GetParam(par);
 389     return wxSscanf(parval, format, param);
 390 }
 391
 392 bool wxHtmlTag::GetParamAsColour(const wxString& par, wxColour *clr) const
 393 {
 394     wxString str = GetParam(par);
 395
 396     if (str.IsEmpty()) return FALSE;
 397     if (str.GetChar(0) == wxT('#'))
 398     {
 399         unsigned long tmp;
 400         if (ScanParam(par, wxT("#%lX"), &tmp) != 1)
 401             return FALSE;
 402         *clr = wxColour((unsigned char)((tmp & 0xFF0000) >> 16),
 403                                             (unsigned char)((tmp & 0x00FF00) >> 8),
 404                                             (unsigned char)(tmp & 0x0000FF));
 405         return TRUE;
 406     }
 407     else
 408     {
 409         // Handle colours defined in HTML 4.0:
 410         #define HTML_COLOUR(name,r,g,b)                 \
 411             if (str.IsSameAs(wxT(name), FALSE))         \
 412                 { *clr = wxColour(r,g,b); return TRUE; }
 413         HTML_COLOUR("black",   0x00,0x00,0x00)
 414         HTML_COLOUR("silver",  0xC0,0xC0,0xC0)
 415         HTML_COLOUR("gray",    0x80,0x80,0x80)
 416         HTML_COLOUR("white",   0xFF,0xFF,0xFF)
 417         HTML_COLOUR("maroon",  0x80,0x00,0x00)
 418         HTML_COLOUR("red",     0xFF,0x00,0x00)
 419         HTML_COLOUR("purple",  0x80,0x00,0x80)
 420         HTML_COLOUR("fuchsia", 0xFF,0x00,0xFF)
 421         HTML_COLOUR("green",   0x00,0x80,0x00)
 422         HTML_COLOUR("lime",    0x00,0xFF,0x00)
 423         HTML_COLOUR("olive",   0x80,0x80,0x00)
 424         HTML_COLOUR("yellow",  0xFF,0xFF,0x00)
 425         HTML_COLOUR("navy",    0x00,0x00,0x80)
 426         HTML_COLOUR("blue",    0x00,0x00,0xFF)
 427         HTML_COLOUR("teal",    0x00,0x80,0x80)
 428         HTML_COLOUR("aqua",    0x00,0xFF,0xFF)
 429         #undef HTML_COLOUR
 430     }
 431
 432     return FALSE;
 433 }
 434
 435 bool wxHtmlTag::GetParamAsInt(const wxString& par, int *clr) const
 436 {
 437     if (!HasParam(par)) return FALSE;
 438     long i;
 439     bool succ = GetParam(par).ToLong(&i);
 440     *clr = (int)i;
 441     return succ;
 442 }
 443
 444 wxString wxHtmlTag::GetAllParams() const
 445 {
 446     // VS: this function is for backward compatiblity only,
 447     //     never used by wxHTML
 448     wxString s;
 449     size_t cnt = m_ParamNames.GetCount();
 450     for (size_t i = 0; i < cnt; i++)
 451     {
 452         s << m_ParamNames[i];
 453         s << wxT('=');
 454         if (m_ParamValues[i].Find(wxT('"')) != wxNOT_FOUND)
 455             s << wxT('\'') << m_ParamValues[i] << wxT('\'');
 456         else
 457             s << wxT('"') << m_ParamValues[i] << wxT('"');
 458     }
 459     return s;
 460 }
 461
 462 wxHtmlTag *wxHtmlTag::GetFirstSibling() const
 463 {
 464     if (m_Parent)
 465         return m_Parent->m_FirstChild;
 466     else
 467     {
 468         wxHtmlTag *cur = (wxHtmlTag*)this;
 469         while (cur->m_Prev)
 470             cur = cur->m_Prev;
 471         return cur;
 472     }
 473 }
 474
 475 wxHtmlTag *wxHtmlTag::GetLastSibling() const
 476 {
 477     if (m_Parent)
 478         return m_Parent->m_LastChild;
 479     else
 480     {
 481         wxHtmlTag *cur = (wxHtmlTag*)this;
 482         while (cur->m_Next)
 483             cur = cur->m_Next;
 484         return cur;
 485     }
 486 }
 487
 488 wxHtmlTag *wxHtmlTag::GetNextTag() const
 489 {
 490     if (m_FirstChild) return m_FirstChild;
 491     if (m_Next) return m_Next;
 492     wxHtmlTag *cur = m_Parent;
 493     if (!cur) return NULL;
 494     while (cur->m_Parent && !cur->m_Next)
 495         cur = cur->m_Parent;
 496     return cur->m_Next;
 497 }
 498
 499 #endif