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 WX_PRECOMP
  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
 203     // we're not using GetCharWidth/Height() because of
 204     // differences under X and win
 205     wxCoord w,h;
 206     m_DC->GetTextExtent( wxT("H"), &w, &h);
 207     m_CharWidth = w;
 208     m_CharHeight = h;
 209
 210     m_UseLink = false;
 211     m_Link = wxHtmlLinkInfo( wxEmptyString );
 212     m_LinkColor.Set(0, 0, 0xFF);
 213     m_ActualColor.Set(0, 0, 0);
 214     m_Align = wxHTML_ALIGN_LEFT;
 215     m_ScriptMode = wxHTML_SCRIPT_NORMAL;
 216     m_ScriptBaseline = 0;
 217     m_tmpLastWasSpace = false;
 218     m_lastWordCell = NULL;
 219
 220     // open the toplevel container that contains everything else and that
 221     // is never closed (this makes parser's life easier):
 222     OpenContainer();
 223
 224     // then open the first container into which page's content will go:
 225     OpenContainer();
 226
 227 #if !wxUSE_UNICODE
 228     wxString charset = ExtractCharsetInformation(source);
 229     if (!charset.empty())
 230     {
 231         wxFontEncoding enc = wxFontMapper::Get()->CharsetToEncoding(charset);
 232         if (enc != wxFONTENCODING_SYSTEM)
 233           SetInputEncoding(enc);
 234     }
 235 #endif
 236
 237     m_Container->InsertCell(new wxHtmlColourCell(m_ActualColor));
 238     wxColour windowColour = wxSystemSettings::GetColour(wxSYS_COLOUR_WINDOW) ;
 239
 240     m_Container->InsertCell
 241                  (
 242                    new wxHtmlColourCell
 243                        (
 244                          m_windowInterface
 245                             ? m_windowInterface->GetHTMLBackgroundColour()
 246                             : windowColour,
 247                          wxHTML_CLR_BACKGROUND
 248                        )
 249                   );
 250
 251     m_Container->InsertCell(new wxHtmlFontCell(CreateCurrentFont()));
 252 }
 253
 254 void wxHtmlWinParser::DoneParser()
 255 {
 256     m_Container = NULL;
 257 #if !wxUSE_UNICODE
 258     SetInputEncoding(wxFONTENCODING_ISO8859_1); // for next call
 259 #endif
 260     wxHtmlParser::DoneParser();
 261 }
 262
 263 #if WXWIN_COMPATIBILITY_2_6
 264 wxHtmlWindow *wxHtmlWinParser::GetWindow()
 265 {
 266     if (!m_windowInterface)
 267         return NULL;
 268     return wxDynamicCast(m_windowInterface->GetHTMLWindow(), wxHtmlWindow);
 269 }
 270 #endif
 271
 272 wxObject* wxHtmlWinParser::GetProduct()
 273 {
 274     wxHtmlContainerCell *top;
 275
 276     CloseContainer();
 277     OpenContainer();
 278
 279     top = m_Container;
 280     while (top->GetParent()) top = top->GetParent();
 281     top->RemoveExtraSpacing(true, true);
 282
 283     return top;
 284 }
 285
 286 wxFSFile *wxHtmlWinParser::OpenURL(wxHtmlURLType type,
 287                                    const wxString& url) const
 288 {
 289     if ( !m_windowInterface )
 290         return wxHtmlParser::OpenURL(type, url);
 291
 292     wxString myurl(url);
 293     wxHtmlOpeningStatus status;
 294     for (;;)
 295     {
 296         wxString myfullurl(myurl);
 297
 298         // consider url as absolute path first
 299         wxURI current(myurl);
 300         myfullurl = current.BuildUnescapedURI();
 301
 302         // if not absolute then ...
 303         if( current.IsReference() )
 304         {
 305             wxString basepath = GetFS()->GetPath();
 306             wxURI base(basepath);
 307
 308             // ... try to apply base path if valid ...
 309             if( !base.IsReference() )
 310             {
 311                 wxURI path(myfullurl);
 312                 path.Resolve( base );
 313                 myfullurl = path.BuildUnescapedURI();
 314             }
 315             else
 316             {
 317                 // ... or force such addition if not included already
 318                 if( !current.GetPath().Contains(base.GetPath()) )
 319                 {
 320                     basepath += myurl;
 321                     wxURI connected( basepath );
 322                     myfullurl = connected.BuildUnescapedURI();
 323                 }
 324             }
 325         }
 326
 327         wxString redirect;
 328         status = m_windowInterface->OnHTMLOpeningURL(type, myfullurl, &redirect);
 329         if ( status != wxHTML_REDIRECT )
 330             break;
 331
 332         myurl = redirect;
 333     }
 334
 335     if ( status == wxHTML_BLOCK )
 336         return NULL;
 337
 338     int flags = wxFS_READ;
 339     if (type == wxHTML_URL_IMAGE)
 340         flags |= wxFS_SEEKABLE;
 341
 342     return GetFS()->OpenFile(myurl, flags);
 343 }
 344
 345 void wxHtmlWinParser::AddText(const wxString& txt)
 346 {
 347     register wxChar d;
 348     int templen = 0;
 349     wxChar nbsp = GetEntitiesParser()->GetCharForCode(160 /* nbsp */);
 350
 351     size_t lng = txt.length();
 352     if (lng+1 > m_tmpStrBufSize)
 353     {
 354         delete[] m_tmpStrBuf;
 355         m_tmpStrBuf = new wxChar[lng+1];
 356         m_tmpStrBufSize = lng+1;
 357     }
 358     wxChar *temp = m_tmpStrBuf;
 359
 360     wxString::const_iterator i = txt.begin();
 361     wxString::const_iterator end = txt.end();
 362
 363     if (m_tmpLastWasSpace)
 364     {
 365         while ( (i < end) &&
 366                 (*i == wxT('\n') || *i == wxT('\r') || *i == wxT(' ') ||
 367                  *i == wxT('\t')) )
 368         {
 369             ++i;
 370         }
 371     }
 372
 373     while (i < end)
 374     {
 375         size_t x = 0;
 376         d = temp[templen++] = *i;
 377         if ((d == wxT('\n')) || (d == wxT('\r')) || (d == wxT(' ')) || (d == wxT('\t')))
 378         {
 379             ++i, ++x;
 380             while ( (i < end) &&
 381                     (*i == wxT('\n') || *i == wxT('\r') ||
 382                      *i == wxT(' ') || *i == wxT('\t')) )
 383             {
 384                 ++i;
 385                 ++x;
 386             }
 387         }
 388         else
 389             ++i;
 390
 391         if (x)
 392         {
 393             temp[templen-1] = wxT(' ');
 394             DoAddText(temp, templen, nbsp);
 395             m_tmpLastWasSpace = true;
 396         }
 397     }
 398
 399     if (templen && (templen > 1 || temp[0] != wxT(' ')))
 400     {
 401         DoAddText(temp, templen, nbsp);
 402         m_tmpLastWasSpace = false;
 403     }
 404 }
 405
 406 void wxHtmlWinParser::DoAddText(wxChar *temp, int& templen, wxChar nbsp)
 407 {
 408     temp[templen] = 0;
 409     templen = 0;
 410 #if !wxUSE_UNICODE
 411     if (m_EncConv)
 412         m_EncConv->Convert(temp);
 413 #endif
 414     size_t len = wxStrlen(temp);
 415     for (size_t j = 0; j < len; j++)
 416     {
 417         if (temp[j] == nbsp)
 418             temp[j] = wxT(' ');
 419     }
 420
 421     wxHtmlCell *c = new wxHtmlWordCell(temp, *(GetDC()));
 422
 423     ApplyStateToCell(c);
 424
 425     m_Container->InsertCell(c);
 426     ((wxHtmlWordCell*)c)->SetPreviousWord(m_lastWordCell);
 427     m_lastWordCell = (wxHtmlWordCell*)c;
 428 }
 429
 430
 431
 432 wxHtmlContainerCell* wxHtmlWinParser::OpenContainer()
 433 {
 434     m_Container = new wxHtmlContainerCell(m_Container);
 435     m_Container->SetAlignHor(m_Align);
 436     m_tmpLastWasSpace = true;
 437         /* to avoid space being first character in paragraph */
 438     return m_Container;
 439 }
 440
 441
 442
 443 wxHtmlContainerCell* wxHtmlWinParser::SetContainer(wxHtmlContainerCell *c)
 444 {
 445     m_tmpLastWasSpace = true;
 446         /* to avoid space being first character in paragraph */
 447     return m_Container = c;
 448 }
 449
 450
 451
 452 wxHtmlContainerCell* wxHtmlWinParser::CloseContainer()
 453 {
 454     m_Container = m_Container->GetParent();
 455     return m_Container;
 456 }
 457
 458
 459 void wxHtmlWinParser::SetFontSize(int s)
 460 {
 461     if (s < 1) s = 1;
 462     else if (s > 7) s = 7;
 463     m_FontSize = s;
 464 }
 465
 466
 467
 468 wxFont* wxHtmlWinParser::CreateCurrentFont()
 469 {
 470     int fb = GetFontBold(),
 471         fi = GetFontItalic(),
 472         fu = GetFontUnderlined(),
 473         ff = GetFontFixed(),
 474         fs = GetFontSize() - 1 /*remap from <1;7> to <0;6>*/ ;
 475
 476     wxString face = ff ? m_FontFaceFixed : m_FontFaceNormal;
 477     wxString *faceptr = &(m_FontsFacesTable[fb][fi][fu][ff][fs]);
 478     wxFont **fontptr = &(m_FontsTable[fb][fi][fu][ff][fs]);
 479 #if !wxUSE_UNICODE
 480     wxFontEncoding *encptr = &(m_FontsEncTable[fb][fi][fu][ff][fs]);
 481 #endif
 482
 483     if (*fontptr != NULL && (*faceptr != face
 484 #if !wxUSE_UNICODE
 485                              || *encptr != m_OutputEnc
 486 #endif
 487                             ))
 488     {
 489         delete *fontptr;
 490         *fontptr = NULL;
 491     }
 492
 493     if (*fontptr == NULL)
 494     {
 495         *faceptr = face;
 496         *fontptr = new wxFont(
 497                        (int) (m_FontsSizes[fs] * m_PixelScale),
 498                        ff ? wxMODERN : wxSWISS,
 499                        fi ? wxITALIC : wxNORMAL,
 500                        fb ? wxBOLD : wxNORMAL,
 501                        fu ? true : false, face
 502 #if wxUSE_UNICODE
 503                        );
 504 #else
 505                        , m_OutputEnc);
 506         *encptr = m_OutputEnc;
 507 #endif
 508     }
 509     m_DC->SetFont(**fontptr);
 510     return (*fontptr);
 511 }
 512
 513
 514
 515 void wxHtmlWinParser::SetLink(const wxHtmlLinkInfo& link)
 516 {
 517     m_Link = link;
 518     m_UseLink = (link.GetHref() != wxEmptyString);
 519 }
 520
 521 void wxHtmlWinParser::SetFontFace(const wxString& face)
 522 {
 523     if (GetFontFixed()) m_FontFaceFixed = face;
 524     else m_FontFaceNormal = face;
 525
 526 #if !wxUSE_UNICODE
 527     if (m_InputEnc != wxFONTENCODING_DEFAULT)
 528         SetInputEncoding(m_InputEnc);
 529 #endif
 530 }
 531
 532 void wxHtmlWinParser::ApplyStateToCell(wxHtmlCell *cell)
 533 {
 534     // set the link:
 535     if (m_UseLink)
 536         cell->SetLink(GetLink());
 537
 538     // apply current script mode settings:
 539     cell->SetScriptMode(GetScriptMode(), GetScriptBaseline());
 540 }
 541
 542
 543 #if !wxUSE_UNICODE
 544 void wxHtmlWinParser::SetInputEncoding(wxFontEncoding enc)
 545 {
 546     m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 547     if (m_EncConv)
 548     {
 549         delete m_EncConv;
 550         m_EncConv = NULL;
 551     }
 552
 553     if (enc == wxFONTENCODING_DEFAULT) return;
 554
 555     wxFontEncoding altfix, altnorm;
 556     bool availfix, availnorm;
 557
 558     // exact match?
 559     availnorm = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceNormal);
 560     availfix = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceFixed);
 561     if (availnorm && availfix)
 562         m_OutputEnc = enc;
 563
 564     // alternatives?
 565     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false) &&
 566              wxFontMapper::Get()->GetAltForEncoding(enc, &altfix, m_FontFaceFixed, false) &&
 567              altnorm == altfix)
 568         m_OutputEnc = altnorm;
 569
 570     // at least normal face?
 571     else if (availnorm)
 572         m_OutputEnc = enc;
 573     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false))
 574         m_OutputEnc = altnorm;
 575
 576     else
 577     {
 578 #ifndef __WXMAC__
 579         // okay, let's convert to ISO_8859-1, available always
 580         m_OutputEnc = wxFONTENCODING_DEFAULT;
 581 #else
 582         m_OutputEnc = wxLocale::GetSystemEncoding() ;
 583 #endif
 584     }
 585
 586     m_InputEnc = enc;
 587     if (m_OutputEnc == wxFONTENCODING_DEFAULT)
 588         GetEntitiesParser()->SetEncoding(wxFONTENCODING_SYSTEM);
 589     else
 590         GetEntitiesParser()->SetEncoding(m_OutputEnc);
 591
 592     if (m_InputEnc == m_OutputEnc) return;
 593
 594     m_EncConv = new wxEncodingConverter();
 595     if (!m_EncConv->Init(m_InputEnc,
 596                            (m_OutputEnc == wxFONTENCODING_DEFAULT) ?
 597                                       wxFONTENCODING_ISO8859_1 : m_OutputEnc,
 598                            wxCONVERT_SUBSTITUTE))
 599     { // total failure :-(
 600         wxLogError(_("Failed to display HTML document in %s encoding"),
 601                    wxFontMapper::GetEncodingName(enc).c_str());
 602         m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 603         delete m_EncConv;
 604         m_EncConv = NULL;
 605     }
 606 }
 607 #endif
 608
 609
 610
 611
 612 //-----------------------------------------------------------------------------
 613 // wxHtmlWinTagHandler
 614 //-----------------------------------------------------------------------------
 615
 616 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinTagHandler, wxHtmlTagHandler)
 617
 618 //-----------------------------------------------------------------------------
 619 // wxHtmlTagsModule
 620 //-----------------------------------------------------------------------------
 621
 622 // NB: This is *NOT* winpars.cpp's initialization and shutdown code!!
 623 //     This module is an ancestor for tag handlers modules defined
 624 //     in m_*.cpp files with TAGS_MODULE_BEGIN...TAGS_MODULE_END construct.
 625 //
 626 //     Do not add any winpars.cpp shutdown or initialization code to it,
 627 //     create a new module instead!
 628
 629 IMPLEMENT_DYNAMIC_CLASS(wxHtmlTagsModule, wxModule)
 630
 631 bool wxHtmlTagsModule::OnInit()
 632 {
 633     wxHtmlWinParser::AddModule(this);
 634     return true;
 635 }
 636
 637 void wxHtmlTagsModule::OnExit()
 638 {
 639     wxHtmlWinParser::RemoveModule(this);
 640 }
 641
 642 #endif