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 #ifdef __GNUG__
  12 #pragma implementation
  13 #endif
  14
  15 #include "wx/wxprec.h"
  16
  17 #include "wx/defs.h"
  18 #if wxUSE_HTML
  19
  20 #ifdef __BORDLANDC__
  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 <stdio.h> // for vsscanf
  30 #include <stdarg.h>
  31
  32
  33 //-----------------------------------------------------------------------------
  34 // wxHtmlTagsCache
  35 //-----------------------------------------------------------------------------
  36
  37 struct wxHtmlCacheItem
  38 {
  39     // this is "pos" value passed to wxHtmlTag's constructor.
  40     // it is position of '<' character of the tag
  41     int Key;
  42
  43     // end positions for the tag:
  44     // end1 is '<' of ending tag,
  45     // end2 is '>' or both are
  46     // -1 if there is no ending tag for this one...
  47     // or -2 if this is ending tag  </...>
  48     int End1, End2;
  49
  50     // name of this tag
  51     wxChar *Name;
  52 };
  53
  54
  55 IMPLEMENT_CLASS(wxHtmlTagsCache,wxObject)
  56
  57 #define CACHE_INCREMENT  64
  58
  59 wxHtmlTagsCache::wxHtmlTagsCache(const wxString& source)
  60 {
  61     const wxChar *src = source.c_str();
  62     int i, tg, pos, stpos;
  63     int lng = source.Length();
  64     wxChar dummy[256];
  65
  66     m_Cache = NULL;
  67     m_CacheSize = 0;
  68     m_CachePos = 0;
  69
  70     pos = 0;
  71     while (pos < lng)
  72     {
  73         if (src[pos] == wxT('<'))   // tag found:
  74         {
  75             if (m_CacheSize % CACHE_INCREMENT == 0)
  76                 m_Cache = (wxHtmlCacheItem*) realloc(m_Cache, (m_CacheSize + CACHE_INCREMENT) * sizeof(wxHtmlCacheItem));
  77             tg = m_CacheSize++;
  78             m_Cache[tg].Key = stpos = pos++;
  79             dummy[0] = 0; i = 0;
  80             while (pos < lng &&
  81                    src[pos] != wxT('>') &&
  82                    src[pos] != wxT(' ') && src[pos] != wxT('\r') &&
  83                    src[pos] != wxT('\n') && src[pos] != wxT('\t'))
  84             {
  85                 dummy[i] = src[pos++];
  86                 if ((dummy[i] >= wxT('a')) && (dummy[i] <= wxT('z'))) dummy[i] -= (wxT('a') - wxT('A'));
  87                 i++;
  88             }
  89             dummy[i] = 0;
  90             m_Cache[tg].Name = new wxChar[i+1];
  91             memcpy(m_Cache[tg].Name, dummy, (i+1)*sizeof(wxChar));
  92
  93             while (pos < lng && src[pos] != wxT('>')) pos++;
  94
  95             if (src[stpos+1] == wxT('/')) // ending tag:
  96             {
  97                 m_Cache[tg].End1 = m_Cache[tg].End2 = -2;
  98                 // find matching begin tag:
  99                 for (i = tg; i >= 0; i--)
 100                     if ((m_Cache[i].End1 == -1) && (wxStrcmp(m_Cache[i].Name, dummy+1) == 0))
 101                     {
 102                         m_Cache[i].End1 = stpos;
 103                         m_Cache[i].End2 = pos + 1;
 104                         break;
 105                     }
 106             }
 107             else
 108             {
 109                 m_Cache[tg].End1 = m_Cache[tg].End2 = -1;
 110             }
 111         }
 112
 113         pos++;
 114     }
 115
 116     // ok, we're done, now we'll free .Name members of cache - we don't need it anymore:
 117     for (i = 0; i < m_CacheSize; i++)
 118     {
 119         delete[] m_Cache[i].Name;
 120         m_Cache[i].Name = NULL;
 121     }
 122 }
 123
 124 void wxHtmlTagsCache::QueryTag(int at, int* end1, int* end2)
 125 {
 126     if (m_Cache == NULL) return;
 127     if (m_Cache[m_CachePos].Key != at)
 128     {
 129         int delta = (at < m_Cache[m_CachePos].Key) ? -1 : 1;
 130         do
 131         {
 132             m_CachePos += delta;
 133         }
 134         while (m_Cache[m_CachePos].Key != at);
 135     }
 136     *end1 = m_Cache[m_CachePos].End1;
 137     *end2 = m_Cache[m_CachePos].End2;
 138 }
 139
 140
 141
 142
 143 //-----------------------------------------------------------------------------
 144 // wxHtmlTag
 145 //-----------------------------------------------------------------------------
 146
 147 IMPLEMENT_CLASS(wxHtmlTag,wxObject)
 148
 149 wxHtmlTag::wxHtmlTag(wxHtmlTag *parent,
 150                      const wxString& source, int pos, int end_pos,
 151                      wxHtmlTagsCache *cache,
 152                      wxHtmlEntitiesParser *entParser) : wxObject()
 153 {
 154     /* Setup DOM relations */
 155
 156     m_Next = NULL;
 157     m_FirstChild = m_LastChild = NULL;
 158     m_Parent = parent;
 159     if (parent)
 160     {
 161         m_Prev = m_Parent->m_LastChild;
 162         if (m_Prev == NULL)
 163             m_Parent->m_FirstChild = this;
 164         else
 165             m_Prev->m_Next = this;
 166         m_Parent->m_LastChild = this;
 167     }
 168     else
 169         m_Prev = NULL;
 170
 171     /* Find parameters and their values: */
 172
 173     int i;
 174     wxChar c;
 175
 176     // fill-in name, params and begin pos:
 177     i = pos+1;
 178
 179     // find tag's name and convert it to uppercase:
 180     while ((i < end_pos) &&
 181            ((c = source[i++]) != wxT(' ') && c != wxT('\r') &&
 182              c != wxT('\n') && c != wxT('\t') &&
 183              c != wxT('>')))
 184     {
 185         if ((c >= wxT('a')) && (c <= wxT('z')))
 186             c -= (wxT('a') - wxT('A'));
 187         m_Name << c;
 188     }
 189
 190     // if the tag has parameters, read them and "normalize" them,
 191     // i.e. convert to uppercase, replace whitespaces by spaces and
 192     // remove whitespaces around '=':
 193     if (source[i-1] != wxT('>'))
 194     {
 195         #define IS_WHITE(c) (c == wxT(' ') || c == wxT('\r') || \
 196                              c == wxT('\n') || c == wxT('\t'))
 197         wxString pname, pvalue;
 198         wxChar quote;
 199         enum
 200         {
 201             ST_BEFORE_NAME = 1,
 202             ST_NAME,
 203             ST_BEFORE_EQ,
 204             ST_BEFORE_VALUE,
 205             ST_VALUE
 206         } state;
 207
 208         quote = 0;
 209         state = ST_BEFORE_NAME;
 210         while (i < end_pos)
 211         {
 212             c = source[i++];
 213
 214             if (c == wxT('>') && !(state == ST_VALUE && quote != 0))
 215             {
 216                 if (state == ST_BEFORE_EQ || state == ST_NAME)
 217                 {
 218                     m_ParamNames.Add(pname);
 219                     m_ParamValues.Add(wxEmptyString);
 220                 }
 221                 else if (state == ST_VALUE && quote == 0)
 222                 {
 223                     m_ParamNames.Add(pname);
 224                     if (entParser)
 225                         m_ParamValues.Add(entParser->Parse(pvalue));
 226                     else
 227                         m_ParamValues.Add(pvalue);
 228                 }
 229                 break;
 230             }
 231             switch (state)
 232             {
 233                 case ST_BEFORE_NAME:
 234                     if (!IS_WHITE(c))
 235                     {
 236                         pname = c;
 237                         state = ST_NAME;
 238                     }
 239                     break;
 240                 case ST_NAME:
 241                     if (IS_WHITE(c))
 242                         state = ST_BEFORE_EQ;
 243                     else if (c == wxT('='))
 244                         state = ST_BEFORE_VALUE;
 245                     else
 246                         pname << c;
 247                     break;
 248                 case ST_BEFORE_EQ:
 249                     if (c == wxT('='))
 250                         state = ST_BEFORE_VALUE;
 251                     else if (!IS_WHITE(c))
 252                     {
 253                         m_ParamNames.Add(pname);
 254                         m_ParamValues.Add(wxEmptyString);
 255                         pname = c;
 256                         state = ST_NAME;
 257                     }
 258                     break;
 259                 case ST_BEFORE_VALUE:
 260                     if (!IS_WHITE(c))
 261                     {
 262                         if (c == wxT('"') || c == wxT('\''))
 263                             quote = c, pvalue = wxEmptyString;
 264                         else
 265                             quote = 0, pvalue = c;
 266                         state = ST_VALUE;
 267                     }
 268                     break;
 269                 case ST_VALUE:
 270                     if ((quote != 0 && c == quote) ||
 271                         (quote == 0 && IS_WHITE(c)))
 272                     {
 273                         m_ParamNames.Add(pname);
 274                         if (quote == 0)
 275                         {
 276                             // VS: backward compatibility, no real reason,
 277                             //     but wxHTML code relies on this... :(
 278                             pvalue.MakeUpper();
 279                         }
 280                         if (entParser)
 281                             m_ParamValues.Add(entParser->Parse(pvalue));
 282                         else
 283                             m_ParamValues.Add(pvalue);
 284                         state = ST_BEFORE_NAME;
 285                     }
 286                     else
 287                         pvalue << c;
 288                     break;
 289             }
 290         }
 291
 292         #undef IS_WHITE
 293    }
 294    m_Begin = i;
 295
 296    cache->QueryTag(pos, &m_End1, &m_End2);
 297    if (m_End1 > end_pos) m_End1 = end_pos;
 298    if (m_End2 > end_pos) m_End2 = end_pos;
 299 }
 300
 301 wxHtmlTag::~wxHtmlTag()
 302 {
 303     wxHtmlTag *t1, *t2;
 304     t1 = m_FirstChild;
 305     while (t1)
 306     {
 307         t2 = t1->GetNextSibling();
 308         delete t1;
 309         t1 = t2;
 310     }
 311 }
 312
 313 bool wxHtmlTag::HasParam(const wxString& par) const
 314 {
 315     return (m_ParamNames.Index(par, FALSE) != wxNOT_FOUND);
 316 }
 317
 318 wxString wxHtmlTag::GetParam(const wxString& par, bool with_commas) const
 319 {
 320     int index = m_ParamNames.Index(par, FALSE);
 321     if (index == wxNOT_FOUND)
 322         return wxEmptyString;
 323     if (with_commas)
 324     {
 325         // VS: backward compatibility, seems to be never used by wxHTML...
 326         wxString s;
 327         s << wxT('"') << m_ParamValues[index] << wxT('"');
 328         return s;
 329     }
 330     else
 331         return m_ParamValues[index];
 332 }
 333
 334 int wxHtmlTag::ScanParam(const wxString& par, wxChar *format, void *param) const
 335 {
 336     wxString parval = GetParam(par);
 337     return wxSscanf(parval, format, param);
 338 }
 339
 340 bool wxHtmlTag::GetParamAsColour(const wxString& par, wxColour *clr) const
 341 {
 342     wxString str = GetParam(par);
 343
 344     if (str.IsEmpty()) return FALSE;
 345     if (str.GetChar(0) == wxT('#'))
 346     {
 347         unsigned long tmp;
 348         if (ScanParam(par, wxT("#%lX"), &tmp) != 1)
 349             return FALSE;
 350         *clr = wxColour((unsigned char)((tmp & 0xFF0000) >> 16),
 351                                             (unsigned char)((tmp & 0x00FF00) >> 8),
 352                                             (unsigned char)(tmp & 0x0000FF));
 353         return TRUE;
 354     }
 355     else
 356     {
 357         // Handle colours defined in HTML 4.0:
 358         #define HTML_COLOUR(name,r,g,b)                 \
 359             if (str.IsSameAs(wxT(name), FALSE))         \
 360                 { *clr = wxColour(r,g,b); return TRUE; }
 361         HTML_COLOUR("black",   0x00,0x00,0x00)
 362         HTML_COLOUR("silver",  0xC0,0xC0,0xC0)
 363         HTML_COLOUR("gray",    0x80,0x80,0x80)
 364         HTML_COLOUR("white",   0xFF,0xFF,0xFF)
 365         HTML_COLOUR("maroon",  0x80,0x00,0x00)
 366         HTML_COLOUR("red",     0xFF,0x00,0x00)
 367         HTML_COLOUR("purple",  0x80,0x00,0x80)
 368         HTML_COLOUR("fuchsia", 0xFF,0x00,0xFF)
 369         HTML_COLOUR("green",   0x00,0x80,0x00)
 370         HTML_COLOUR("lime",    0x00,0xFF,0x00)
 371         HTML_COLOUR("olive",   0x80,0x80,0x00)
 372         HTML_COLOUR("yellow",  0xFF,0xFF,0x00)
 373         HTML_COLOUR("navy",    0x00,0x00,0x80)
 374         HTML_COLOUR("blue",    0x00,0x00,0xFF)
 375         HTML_COLOUR("teal",    0x00,0x80,0x80)
 376         HTML_COLOUR("aqua",    0x00,0xFF,0xFF)
 377         #undef HTML_COLOUR
 378         return FALSE;
 379     }
 380 }
 381
 382 bool wxHtmlTag::GetParamAsInt(const wxString& par, int *clr) const
 383 {
 384     if (!HasParam(par)) return FALSE;
 385     long i;
 386     bool succ = GetParam(par).ToLong(&i);
 387     *clr = (int)i;
 388     return succ;
 389 }
 390
 391 wxString wxHtmlTag::GetAllParams() const
 392 {
 393     // VS: this function is for backward compatiblity only,
 394     //     never used by wxHTML
 395     wxString s;
 396     size_t cnt = m_ParamNames.GetCount();
 397     for (size_t i = 0; i < cnt; i++)
 398     {
 399         s << m_ParamNames[i];
 400         s << wxT('=');
 401         if (m_ParamValues[i].Find(wxT('"')) != wxNOT_FOUND)
 402             s << wxT('\'') << m_ParamValues[i] << wxT('\'');
 403         else
 404             s << wxT('"') << m_ParamValues[i] << wxT('"');
 405     }
 406     return s;
 407 }
 408
 409 wxHtmlTag *wxHtmlTag::GetFirstSibling() const
 410 {
 411     if (m_Parent)
 412         return m_Parent->m_FirstChild;
 413     else
 414     {
 415         wxHtmlTag *cur = (wxHtmlTag*)this;
 416         while (cur->m_Prev)
 417             cur = cur->m_Prev;
 418         return cur;
 419     }
 420 }
 421
 422 wxHtmlTag *wxHtmlTag::GetLastSibling() const
 423 {
 424     if (m_Parent)
 425         return m_Parent->m_LastChild;
 426     else
 427     {
 428         wxHtmlTag *cur = (wxHtmlTag*)this;
 429         while (cur->m_Next)
 430             cur = cur->m_Next;
 431         return cur;
 432     }
 433 }
 434
 435 wxHtmlTag *wxHtmlTag::GetNextTag() const
 436 {
 437     if (m_FirstChild) return m_FirstChild;
 438     if (m_Next) return m_Next;
 439     wxHtmlTag *cur = m_Parent;
 440     if (!cur) return NULL;
 441     while (cur->m_Parent && !cur->m_Next)
 442         cur = cur->m_Parent;
 443     return cur->m_Next;
 444 }
 445
 446 #endif