src/html/winpars.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        src/html/winpars.cpp
   3 // Purpose:     wxHtmlParser class (generic parser)
   4 // Author:      Vaclav Slavik
   5 // RCS-ID:      $Id$
   6 // Copyright:   (c) 1999 Vaclav Slavik
   7 // Licence:     wxWindows licence
   8 /////////////////////////////////////////////////////////////////////////////
   9
  10 #include "wx/wxprec.h"
  11
  12 #ifdef __BORLANDC__
  13     #pragma hdrstop
  14 #endif
  15
  16 #if wxUSE_HTML && wxUSE_STREAMS
  17
  18 #ifndef WXPRECOMP
  19     #include "wx/intl.h"
  20     #include "wx/dc.h"
  21     #include "wx/log.h"
  22     #include "wx/settings.h"
  23 #endif
  24
  25 #include "wx/html/htmldefs.h"
  26 #include "wx/html/winpars.h"
  27 #include "wx/html/htmlwin.h"
  28 #include "wx/fontmap.h"
  29 #include "wx/uri.h"
  30
  31
  32 //-----------------------------------------------------------------------------
  33 // wxHtmlWinParser
  34 //-----------------------------------------------------------------------------
  35
  36 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinParser, wxHtmlParser)
  37
  38 wxList wxHtmlWinParser::m_Modules;
  39
  40 wxHtmlWinParser::wxHtmlWinParser(wxHtmlWindowInterface *wndIface)
  41 {
  42     m_tmpStrBuf = NULL;
  43     m_tmpStrBufSize = 0;
  44     m_windowInterface = wndIface;
  45     m_Container = NULL;
  46     m_DC = NULL;
  47     m_CharHeight = m_CharWidth = 0;
  48     m_UseLink = false;
  49 #if !wxUSE_UNICODE
  50     m_EncConv = NULL;
  51     m_InputEnc = wxFONTENCODING_ISO8859_1;
  52     m_OutputEnc = wxFONTENCODING_DEFAULT;
  53 #endif
  54     m_lastWordCell = NULL;
  55
  56     {
  57         int i, j, k, l, m;
  58         for (i = 0; i < 2; i++)
  59             for (j = 0; j < 2; j++)
  60                 for (k = 0; k < 2; k++)
  61                     for (l = 0; l < 2; l++)
  62                         for (m = 0; m < 7; m++)
  63                         {
  64                             m_FontsTable[i][j][k][l][m] = NULL;
  65                             m_FontsFacesTable[i][j][k][l][m] = wxEmptyString;
  66 #if !wxUSE_UNICODE
  67                             m_FontsEncTable[i][j][k][l][m] = wxFONTENCODING_DEFAULT;
  68 #endif
  69                         }
  70
  71         SetFonts(wxEmptyString, wxEmptyString, NULL);
  72     }
  73
  74     // fill in wxHtmlParser's tables:
  75     wxList::compatibility_iterator node = m_Modules.GetFirst();
  76     while (node)
  77     {
  78         wxHtmlTagsModule *mod = (wxHtmlTagsModule*) node->GetData();
  79         mod->FillHandlersTable(this);
  80         node = node->GetNext();
  81     }
  82 }
  83
  84 wxHtmlWinParser::~wxHtmlWinParser()
  85 {
  86     int i, j, k, l, m;
  87
  88     for (i = 0; i < 2; i++)
  89         for (j = 0; j < 2; j++)
  90             for (k = 0; k < 2; k++)
  91                 for (l = 0; l < 2; l++)
  92                     for (m = 0; m < 7; m++)
  93                     {
  94                         if (m_FontsTable[i][j][k][l][m] != NULL)
  95                             delete m_FontsTable[i][j][k][l][m];
  96                     }
  97 #if !wxUSE_UNICODE
  98     delete m_EncConv;
  99 #endif
 100     delete[] m_tmpStrBuf;
 101 }
 102
 103 void wxHtmlWinParser::AddModule(wxHtmlTagsModule *module)
 104 {
 105     m_Modules.Append(module);
 106 }
 107
 108 void wxHtmlWinParser::RemoveModule(wxHtmlTagsModule *module)
 109 {
 110     m_Modules.DeleteObject(module);
 111 }
 112
 113 // build all HTML font sizes (1..7) from the given base size
 114 static void wxBuildFontSizes(int *sizes, int size)
 115 {
 116     // using a fixed factor (1.2, from CSS2) is a bad idea as explained at
 117     // http://www.w3.org/TR/CSS21/fonts.html#font-size-props but this is by far
 118     // simplest thing to do so still do it like this for now
 119     sizes[0] = int(size * 0.75); // exception to 1.2 rule, otherwise too small
 120     sizes[1] = int(size * 0.83);
 121     sizes[2] = size;
 122     sizes[3] = int(size * 1.2);
 123     sizes[4] = int(size * 1.44);
 124     sizes[5] = int(size * 1.73);
 125     sizes[6] = int(size * 2);
 126 }
 127
 128 static int wxGetDefaultHTMLFontSize()
 129 {
 130     // base the default font size on the size of the default system font but
 131     // also ensure that we have a font of reasonable size, otherwise small HTML
 132     // fonts are unreadable
 133     int size = wxNORMAL_FONT->GetPointSize();
 134     if ( size < 10 )
 135         size = 10;
 136     return size;
 137 }
 138
 139 void wxHtmlWinParser::SetFonts(const wxString& normal_face,
 140                                const wxString& fixed_face,
 141                                const int *sizes)
 142 {
 143     static int default_sizes[7] = { 0 };
 144     if ( !sizes )
 145     {
 146         if ( !default_sizes[0] )
 147             wxBuildFontSizes(default_sizes, wxGetDefaultHTMLFontSize());
 148
 149         sizes = default_sizes;
 150     }
 151
 152     int i, j, k, l, m;
 153
 154     for (i = 0; i < 7; i++)
 155         m_FontsSizes[i] = sizes[i];
 156
 157     m_FontFaceFixed = fixed_face;
 158     m_FontFaceNormal = normal_face;
 159
 160 #if !wxUSE_UNICODE
 161     SetInputEncoding(m_InputEnc);
 162 #endif
 163
 164     for (i = 0; i < 2; i++)
 165         for (j = 0; j < 2; j++)
 166             for (k = 0; k < 2; k++)
 167                 for (l = 0; l < 2; l++)
 168                     for (m = 0; m < 7; m++) {
 169                         if (m_FontsTable[i][j][k][l][m] != NULL)
 170                         {
 171                             delete m_FontsTable[i][j][k][l][m];
 172                             m_FontsTable[i][j][k][l][m] = NULL;
 173                         }
 174                     }
 175 }
 176
 177 void wxHtmlWinParser::SetStandardFonts(int size,
 178                                        const wxString& normal_face,
 179                                        const wxString& fixed_face)
 180 {
 181     if (size == -1)
 182         size = wxGetDefaultHTMLFontSize();
 183
 184     int f_sizes[7];
 185     wxBuildFontSizes(f_sizes, size);
 186
 187     wxString normal = normal_face;
 188     if ( normal.empty() )
 189         normal = wxNORMAL_FONT->GetFaceName();
 190
 191     SetFonts(normal, fixed_face, f_sizes);
 192 }
 193
 194 void wxHtmlWinParser::InitParser(const wxString& source)
 195 {
 196     wxHtmlParser::InitParser(source);
 197     wxASSERT_MSG(m_DC != NULL, wxT("no DC assigned to wxHtmlWinParser!!"));
 198
 199     m_FontBold = m_FontItalic = m_FontUnderlined = m_FontFixed = FALSE;
 200     m_FontSize = 3; //default one
 201     CreateCurrentFont();           // we're selecting default font into
 202     m_DC->GetTextExtent( wxT("H"), &m_CharWidth, &m_CharHeight);
 203                 /* NOTE : we're not using GetCharWidth/Height() because
 204                    of differences under X and win
 205                  */
 206
 207     m_UseLink = false;
 208     m_Link = wxHtmlLinkInfo( wxEmptyString );
 209     m_LinkColor.Set(0, 0, 0xFF);
 210     m_ActualColor.Set(0, 0, 0);
 211     m_Align = wxHTML_ALIGN_LEFT;
 212     m_ScriptMode = wxHTML_SCRIPT_NORMAL;
 213     m_ScriptBaseline = 0;
 214     m_tmpLastWasSpace = false;
 215     m_lastWordCell = NULL;
 216
 217     // open the toplevel container that contains everything else and that
 218     // is never closed (this makes parser's life easier):
 219     OpenContainer();
 220
 221     // then open the first container into which page's content will go:
 222     OpenContainer();
 223
 224 #if !wxUSE_UNICODE
 225     wxString charset = ExtractCharsetInformation(source);
 226     if (!charset.empty())
 227     {
 228         wxFontEncoding enc = wxFontMapper::Get()->CharsetToEncoding(charset);
 229         if (enc != wxFONTENCODING_SYSTEM)
 230           SetInputEncoding(enc);
 231     }
 232 #endif
 233
 234     m_Container->InsertCell(new wxHtmlColourCell(m_ActualColor));
 235     wxColour windowColour = wxSystemSettings::GetColour(wxSYS_COLOUR_WINDOW) ;
 236
 237     m_Container->InsertCell
 238                  (
 239                    new wxHtmlColourCell
 240                        (
 241                          m_windowInterface
 242                             ? m_windowInterface->GetHTMLBackgroundColour()
 243                             : windowColour,
 244                          wxHTML_CLR_BACKGROUND
 245                        )
 246                   );
 247
 248     m_Container->InsertCell(new wxHtmlFontCell(CreateCurrentFont()));
 249 }
 250
 251 void wxHtmlWinParser::DoneParser()
 252 {
 253     m_Container = NULL;
 254 #if !wxUSE_UNICODE
 255     SetInputEncoding(wxFONTENCODING_ISO8859_1); // for next call
 256 #endif
 257     wxHtmlParser::DoneParser();
 258 }
 259
 260 #if WXWIN_COMPATIBILITY_2_6
 261 wxHtmlWindow *wxHtmlWinParser::GetWindow()
 262 {
 263     if (!m_windowInterface)
 264         return NULL;
 265     return wxDynamicCast(m_windowInterface->GetHTMLWindow(), wxHtmlWindow);
 266 }
 267 #endif
 268
 269 wxObject* wxHtmlWinParser::GetProduct()
 270 {
 271     wxHtmlContainerCell *top;
 272
 273     CloseContainer();
 274     OpenContainer();
 275
 276     top = m_Container;
 277     while (top->GetParent()) top = top->GetParent();
 278     top->RemoveExtraSpacing(true, true);
 279
 280     return top;
 281 }
 282
 283 wxFSFile *wxHtmlWinParser::OpenURL(wxHtmlURLType type,
 284                                    const wxString& url) const
 285 {
 286     if ( !m_windowInterface )
 287         return wxHtmlParser::OpenURL(type, url);
 288
 289     wxString myurl(url);
 290     wxHtmlOpeningStatus status;
 291     for (;;)
 292     {
 293         wxString myfullurl(myurl);
 294
 295         // consider url as absolute path first
 296         wxURI current(myurl);
 297         myfullurl = current.BuildUnescapedURI();
 298
 299         // if not absolute then ...
 300         if( current.IsReference() )
 301         {
 302             wxString basepath = GetFS()->GetPath();
 303             wxURI base(basepath);
 304
 305             // ... try to apply base path if valid ...
 306             if( !base.IsReference() )
 307             {
 308                 wxURI path(myfullurl);
 309                 path.Resolve( base );
 310                 myfullurl = path.BuildUnescapedURI();
 311             }
 312             else
 313             {
 314                 // ... or force such addition if not included already
 315                 if( !current.GetPath().Contains(base.GetPath()) )
 316                 {
 317                     basepath += myurl;
 318                     wxURI connected( basepath );
 319                     myfullurl = connected.BuildUnescapedURI();
 320                 }
 321             }
 322         }
 323
 324         wxString redirect;
 325         status = m_windowInterface->OnHTMLOpeningURL(type, myfullurl, &redirect);
 326         if ( status != wxHTML_REDIRECT )
 327             break;
 328
 329         myurl = redirect;
 330     }
 331
 332     if ( status == wxHTML_BLOCK )
 333         return NULL;
 334
 335     return GetFS()->OpenFile(myurl);
 336 }
 337
 338 void wxHtmlWinParser::AddText(const wxChar* txt)
 339 {
 340     size_t i = 0,
 341            x,
 342            lng = wxStrlen(txt);
 343     register wxChar d;
 344     int templen = 0;
 345     wxChar nbsp = GetEntitiesParser()->GetCharForCode(160 /* nbsp */);
 346
 347     if (lng+1 > m_tmpStrBufSize)
 348     {
 349         delete[] m_tmpStrBuf;
 350         m_tmpStrBuf = new wxChar[lng+1];
 351         m_tmpStrBufSize = lng+1;
 352     }
 353     wxChar *temp = m_tmpStrBuf;
 354
 355     if (m_tmpLastWasSpace)
 356     {
 357         while ((i < lng) &&
 358                ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) || (txt[i] == wxT(' ')) ||
 359                 (txt[i] == wxT('\t')))) i++;
 360     }
 361
 362     while (i < lng)
 363     {
 364         x = 0;
 365         d = temp[templen++] = txt[i];
 366         if ((d == wxT('\n')) || (d == wxT('\r')) || (d == wxT(' ')) || (d == wxT('\t')))
 367         {
 368             i++, x++;
 369             while ((i < lng) && ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) ||
 370                                  (txt[i] == wxT(' ')) || (txt[i] == wxT('\t')))) i++, x++;
 371         }
 372         else i++;
 373
 374         if (x)
 375         {
 376             temp[templen-1] = wxT(' ');
 377             DoAddText(temp, templen, nbsp);
 378             m_tmpLastWasSpace = true;
 379         }
 380     }
 381
 382     if (templen && (templen > 1 || temp[0] != wxT(' ')))
 383     {
 384         DoAddText(temp, templen, nbsp);
 385         m_tmpLastWasSpace = false;
 386     }
 387 }
 388
 389 void wxHtmlWinParser::DoAddText(wxChar *temp, int& templen, wxChar nbsp)
 390 {
 391     temp[templen] = 0;
 392     templen = 0;
 393 #if !wxUSE_UNICODE
 394     if (m_EncConv)
 395         m_EncConv->Convert(temp);
 396 #endif
 397     size_t len = wxStrlen(temp);
 398     for (size_t j = 0; j < len; j++)
 399     {
 400         if (temp[j] == nbsp)
 401             temp[j] = wxT(' ');
 402     }
 403
 404     wxHtmlCell *c = new wxHtmlWordCell(temp, *(GetDC()));
 405
 406     ApplyStateToCell(c);
 407
 408     m_Container->InsertCell(c);
 409     ((wxHtmlWordCell*)c)->SetPreviousWord(m_lastWordCell);
 410     m_lastWordCell = (wxHtmlWordCell*)c;
 411 }
 412
 413
 414
 415 wxHtmlContainerCell* wxHtmlWinParser::OpenContainer()
 416 {
 417     m_Container = new wxHtmlContainerCell(m_Container);
 418     m_Container->SetAlignHor(m_Align);
 419     m_tmpLastWasSpace = true;
 420         /* to avoid space being first character in paragraph */
 421     return m_Container;
 422 }
 423
 424
 425
 426 wxHtmlContainerCell* wxHtmlWinParser::SetContainer(wxHtmlContainerCell *c)
 427 {
 428     m_tmpLastWasSpace = true;
 429         /* to avoid space being first character in paragraph */
 430     return m_Container = c;
 431 }
 432
 433
 434
 435 wxHtmlContainerCell* wxHtmlWinParser::CloseContainer()
 436 {
 437     m_Container = m_Container->GetParent();
 438     return m_Container;
 439 }
 440
 441
 442 void wxHtmlWinParser::SetFontSize(int s)
 443 {
 444     if (s < 1) s = 1;
 445     else if (s > 7) s = 7;
 446     m_FontSize = s;
 447 }
 448
 449
 450
 451 wxFont* wxHtmlWinParser::CreateCurrentFont()
 452 {
 453     int fb = GetFontBold(),
 454         fi = GetFontItalic(),
 455         fu = GetFontUnderlined(),
 456         ff = GetFontFixed(),
 457         fs = GetFontSize() - 1 /*remap from <1;7> to <0;6>*/ ;
 458
 459     wxString face = ff ? m_FontFaceFixed : m_FontFaceNormal;
 460     wxString *faceptr = &(m_FontsFacesTable[fb][fi][fu][ff][fs]);
 461     wxFont **fontptr = &(m_FontsTable[fb][fi][fu][ff][fs]);
 462 #if !wxUSE_UNICODE
 463     wxFontEncoding *encptr = &(m_FontsEncTable[fb][fi][fu][ff][fs]);
 464 #endif
 465
 466     if (*fontptr != NULL && (*faceptr != face
 467 #if !wxUSE_UNICODE
 468                              || *encptr != m_OutputEnc
 469 #endif
 470                             ))
 471     {
 472         delete *fontptr;
 473         *fontptr = NULL;
 474     }
 475
 476     if (*fontptr == NULL)
 477     {
 478         *faceptr = face;
 479         *fontptr = new wxFont(
 480                        (int) (m_FontsSizes[fs] * m_PixelScale),
 481                        ff ? wxMODERN : wxSWISS,
 482                        fi ? wxITALIC : wxNORMAL,
 483                        fb ? wxBOLD : wxNORMAL,
 484                        fu ? true : false, face
 485 #if wxUSE_UNICODE
 486                        );
 487 #else
 488                        , m_OutputEnc);
 489         *encptr = m_OutputEnc;
 490 #endif
 491     }
 492     m_DC->SetFont(**fontptr);
 493     return (*fontptr);
 494 }
 495
 496
 497
 498 void wxHtmlWinParser::SetLink(const wxHtmlLinkInfo& link)
 499 {
 500     m_Link = link;
 501     m_UseLink = (link.GetHref() != wxEmptyString);
 502 }
 503
 504 void wxHtmlWinParser::SetFontFace(const wxString& face)
 505 {
 506     if (GetFontFixed()) m_FontFaceFixed = face;
 507     else m_FontFaceNormal = face;
 508
 509 #if !wxUSE_UNICODE
 510     if (m_InputEnc != wxFONTENCODING_DEFAULT)
 511         SetInputEncoding(m_InputEnc);
 512 #endif
 513 }
 514
 515 void wxHtmlWinParser::ApplyStateToCell(wxHtmlCell *cell)
 516 {
 517     // set the link:
 518     if (m_UseLink)
 519         cell->SetLink(GetLink());
 520
 521     // apply current script mode settings:
 522     cell->SetScriptMode(GetScriptMode(), GetScriptBaseline());
 523 }
 524
 525
 526 #if !wxUSE_UNICODE
 527 void wxHtmlWinParser::SetInputEncoding(wxFontEncoding enc)
 528 {
 529     m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 530     if (m_EncConv)
 531     {
 532         delete m_EncConv;
 533         m_EncConv = NULL;
 534     }
 535
 536     if (enc == wxFONTENCODING_DEFAULT) return;
 537
 538     wxFontEncoding altfix, altnorm;
 539     bool availfix, availnorm;
 540
 541     // exact match?
 542     availnorm = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceNormal);
 543     availfix = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceFixed);
 544     if (availnorm && availfix)
 545         m_OutputEnc = enc;
 546
 547     // alternatives?
 548     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false) &&
 549              wxFontMapper::Get()->GetAltForEncoding(enc, &altfix, m_FontFaceFixed, false) &&
 550              altnorm == altfix)
 551         m_OutputEnc = altnorm;
 552
 553     // at least normal face?
 554     else if (availnorm)
 555         m_OutputEnc = enc;
 556     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false))
 557         m_OutputEnc = altnorm;
 558
 559     else
 560     {
 561 #ifndef __WXMAC__
 562         // okay, let's convert to ISO_8859-1, available always
 563         m_OutputEnc = wxFONTENCODING_DEFAULT;
 564 #else
 565         m_OutputEnc = wxLocale::GetSystemEncoding() ;
 566 #endif
 567     }
 568
 569     m_InputEnc = enc;
 570     if (m_OutputEnc == wxFONTENCODING_DEFAULT)
 571         GetEntitiesParser()->SetEncoding(wxFONTENCODING_SYSTEM);
 572     else
 573         GetEntitiesParser()->SetEncoding(m_OutputEnc);
 574
 575     if (m_InputEnc == m_OutputEnc) return;
 576
 577     m_EncConv = new wxEncodingConverter();
 578     if (!m_EncConv->Init(m_InputEnc,
 579                            (m_OutputEnc == wxFONTENCODING_DEFAULT) ?
 580                                       wxFONTENCODING_ISO8859_1 : m_OutputEnc,
 581                            wxCONVERT_SUBSTITUTE))
 582     { // total failure :-(
 583         wxLogError(_("Failed to display HTML document in %s encoding"),
 584                    wxFontMapper::GetEncodingName(enc).c_str());
 585         m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 586         delete m_EncConv;
 587         m_EncConv = NULL;
 588     }
 589 }
 590 #endif
 591
 592
 593
 594
 595 //-----------------------------------------------------------------------------
 596 // wxHtmlWinTagHandler
 597 //-----------------------------------------------------------------------------
 598
 599 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinTagHandler, wxHtmlTagHandler)
 600
 601 //-----------------------------------------------------------------------------
 602 // wxHtmlTagsModule
 603 //-----------------------------------------------------------------------------
 604
 605 // NB: This is *NOT* winpars.cpp's initialization and shutdown code!!
 606 //     This module is an ancestor for tag handlers modules defined
 607 //     in m_*.cpp files with TAGS_MODULE_BEGIN...TAGS_MODULE_END construct.
 608 //
 609 //     Do not add any winpars.cpp shutdown or initialization code to it,
 610 //     create a new module instead!
 611
 612 IMPLEMENT_DYNAMIC_CLASS(wxHtmlTagsModule, wxModule)
 613
 614 bool wxHtmlTagsModule::OnInit()
 615 {
 616     wxHtmlWinParser::AddModule(this);
 617     return true;
 618 }
 619
 620 void wxHtmlTagsModule::OnExit()
 621 {
 622     wxHtmlWinParser::RemoveModule(this);
 623 }
 624
 625 #endif