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     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     int flags = wxFS_READ;
 336     if (type == wxHTML_URL_IMAGE)
 337         flags |= wxFS_SEEKABLE;
 338
 339     return GetFS()->OpenFile(myurl, flags);
 340 }
 341
 342 void wxHtmlWinParser::AddText(const wxChar* txt)
 343 {
 344     size_t i = 0,
 345            x,
 346            lng = wxStrlen(txt);
 347     register wxChar d;
 348     int templen = 0;
 349     wxChar nbsp = GetEntitiesParser()->GetCharForCode(160 /* nbsp */);
 350
 351     if (lng+1 > m_tmpStrBufSize)
 352     {
 353         delete[] m_tmpStrBuf;
 354         m_tmpStrBuf = new wxChar[lng+1];
 355         m_tmpStrBufSize = lng+1;
 356     }
 357     wxChar *temp = m_tmpStrBuf;
 358
 359     if (m_tmpLastWasSpace)
 360     {
 361         while ((i < lng) &&
 362                ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) || (txt[i] == wxT(' ')) ||
 363                 (txt[i] == wxT('\t')))) i++;
 364     }
 365
 366     while (i < lng)
 367     {
 368         x = 0;
 369         d = temp[templen++] = txt[i];
 370         if ((d == wxT('\n')) || (d == wxT('\r')) || (d == wxT(' ')) || (d == wxT('\t')))
 371         {
 372             i++, x++;
 373             while ((i < lng) && ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) ||
 374                                  (txt[i] == wxT(' ')) || (txt[i] == wxT('\t')))) i++, x++;
 375         }
 376         else i++;
 377
 378         if (x)
 379         {
 380             temp[templen-1] = wxT(' ');
 381             DoAddText(temp, templen, nbsp);
 382             m_tmpLastWasSpace = true;
 383         }
 384     }
 385
 386     if (templen && (templen > 1 || temp[0] != wxT(' ')))
 387     {
 388         DoAddText(temp, templen, nbsp);
 389         m_tmpLastWasSpace = false;
 390     }
 391 }
 392
 393 void wxHtmlWinParser::DoAddText(wxChar *temp, int& templen, wxChar nbsp)
 394 {
 395     temp[templen] = 0;
 396     templen = 0;
 397 #if !wxUSE_UNICODE
 398     if (m_EncConv)
 399         m_EncConv->Convert(temp);
 400 #endif
 401     size_t len = wxStrlen(temp);
 402     for (size_t j = 0; j < len; j++)
 403     {
 404         if (temp[j] == nbsp)
 405             temp[j] = wxT(' ');
 406     }
 407
 408     wxHtmlCell *c = new wxHtmlWordCell(temp, *(GetDC()));
 409
 410     ApplyStateToCell(c);
 411
 412     m_Container->InsertCell(c);
 413     ((wxHtmlWordCell*)c)->SetPreviousWord(m_lastWordCell);
 414     m_lastWordCell = (wxHtmlWordCell*)c;
 415 }
 416
 417
 418
 419 wxHtmlContainerCell* wxHtmlWinParser::OpenContainer()
 420 {
 421     m_Container = new wxHtmlContainerCell(m_Container);
 422     m_Container->SetAlignHor(m_Align);
 423     m_tmpLastWasSpace = true;
 424         /* to avoid space being first character in paragraph */
 425     return m_Container;
 426 }
 427
 428
 429
 430 wxHtmlContainerCell* wxHtmlWinParser::SetContainer(wxHtmlContainerCell *c)
 431 {
 432     m_tmpLastWasSpace = true;
 433         /* to avoid space being first character in paragraph */
 434     return m_Container = c;
 435 }
 436
 437
 438
 439 wxHtmlContainerCell* wxHtmlWinParser::CloseContainer()
 440 {
 441     m_Container = m_Container->GetParent();
 442     return m_Container;
 443 }
 444
 445
 446 void wxHtmlWinParser::SetFontSize(int s)
 447 {
 448     if (s < 1) s = 1;
 449     else if (s > 7) s = 7;
 450     m_FontSize = s;
 451 }
 452
 453
 454
 455 wxFont* wxHtmlWinParser::CreateCurrentFont()
 456 {
 457     int fb = GetFontBold(),
 458         fi = GetFontItalic(),
 459         fu = GetFontUnderlined(),
 460         ff = GetFontFixed(),
 461         fs = GetFontSize() - 1 /*remap from <1;7> to <0;6>*/ ;
 462
 463     wxString face = ff ? m_FontFaceFixed : m_FontFaceNormal;
 464     wxString *faceptr = &(m_FontsFacesTable[fb][fi][fu][ff][fs]);
 465     wxFont **fontptr = &(m_FontsTable[fb][fi][fu][ff][fs]);
 466 #if !wxUSE_UNICODE
 467     wxFontEncoding *encptr = &(m_FontsEncTable[fb][fi][fu][ff][fs]);
 468 #endif
 469
 470     if (*fontptr != NULL && (*faceptr != face
 471 #if !wxUSE_UNICODE
 472                              || *encptr != m_OutputEnc
 473 #endif
 474                             ))
 475     {
 476         delete *fontptr;
 477         *fontptr = NULL;
 478     }
 479
 480     if (*fontptr == NULL)
 481     {
 482         *faceptr = face;
 483         *fontptr = new wxFont(
 484                        (int) (m_FontsSizes[fs] * m_PixelScale),
 485                        ff ? wxMODERN : wxSWISS,
 486                        fi ? wxITALIC : wxNORMAL,
 487                        fb ? wxBOLD : wxNORMAL,
 488                        fu ? true : false, face
 489 #if wxUSE_UNICODE
 490                        );
 491 #else
 492                        , m_OutputEnc);
 493         *encptr = m_OutputEnc;
 494 #endif
 495     }
 496     m_DC->SetFont(**fontptr);
 497     return (*fontptr);
 498 }
 499
 500
 501
 502 void wxHtmlWinParser::SetLink(const wxHtmlLinkInfo& link)
 503 {
 504     m_Link = link;
 505     m_UseLink = (link.GetHref() != wxEmptyString);
 506 }
 507
 508 void wxHtmlWinParser::SetFontFace(const wxString& face)
 509 {
 510     if (GetFontFixed()) m_FontFaceFixed = face;
 511     else m_FontFaceNormal = face;
 512
 513 #if !wxUSE_UNICODE
 514     if (m_InputEnc != wxFONTENCODING_DEFAULT)
 515         SetInputEncoding(m_InputEnc);
 516 #endif
 517 }
 518
 519 void wxHtmlWinParser::ApplyStateToCell(wxHtmlCell *cell)
 520 {
 521     // set the link:
 522     if (m_UseLink)
 523         cell->SetLink(GetLink());
 524
 525     // apply current script mode settings:
 526     cell->SetScriptMode(GetScriptMode(), GetScriptBaseline());
 527 }
 528
 529
 530 #if !wxUSE_UNICODE
 531 void wxHtmlWinParser::SetInputEncoding(wxFontEncoding enc)
 532 {
 533     m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 534     if (m_EncConv)
 535     {
 536         delete m_EncConv;
 537         m_EncConv = NULL;
 538     }
 539
 540     if (enc == wxFONTENCODING_DEFAULT) return;
 541
 542     wxFontEncoding altfix, altnorm;
 543     bool availfix, availnorm;
 544
 545     // exact match?
 546     availnorm = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceNormal);
 547     availfix = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceFixed);
 548     if (availnorm && availfix)
 549         m_OutputEnc = enc;
 550
 551     // alternatives?
 552     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false) &&
 553              wxFontMapper::Get()->GetAltForEncoding(enc, &altfix, m_FontFaceFixed, false) &&
 554              altnorm == altfix)
 555         m_OutputEnc = altnorm;
 556
 557     // at least normal face?
 558     else if (availnorm)
 559         m_OutputEnc = enc;
 560     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false))
 561         m_OutputEnc = altnorm;
 562
 563     else
 564     {
 565 #ifndef __WXMAC__
 566         // okay, let's convert to ISO_8859-1, available always
 567         m_OutputEnc = wxFONTENCODING_DEFAULT;
 568 #else
 569         m_OutputEnc = wxLocale::GetSystemEncoding() ;
 570 #endif
 571     }
 572
 573     m_InputEnc = enc;
 574     if (m_OutputEnc == wxFONTENCODING_DEFAULT)
 575         GetEntitiesParser()->SetEncoding(wxFONTENCODING_SYSTEM);
 576     else
 577         GetEntitiesParser()->SetEncoding(m_OutputEnc);
 578
 579     if (m_InputEnc == m_OutputEnc) return;
 580
 581     m_EncConv = new wxEncodingConverter();
 582     if (!m_EncConv->Init(m_InputEnc,
 583                            (m_OutputEnc == wxFONTENCODING_DEFAULT) ?
 584                                       wxFONTENCODING_ISO8859_1 : m_OutputEnc,
 585                            wxCONVERT_SUBSTITUTE))
 586     { // total failure :-(
 587         wxLogError(_("Failed to display HTML document in %s encoding"),
 588                    wxFontMapper::GetEncodingName(enc).c_str());
 589         m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 590         delete m_EncConv;
 591         m_EncConv = NULL;
 592     }
 593 }
 594 #endif
 595
 596
 597
 598
 599 //-----------------------------------------------------------------------------
 600 // wxHtmlWinTagHandler
 601 //-----------------------------------------------------------------------------
 602
 603 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinTagHandler, wxHtmlTagHandler)
 604
 605 //-----------------------------------------------------------------------------
 606 // wxHtmlTagsModule
 607 //-----------------------------------------------------------------------------
 608
 609 // NB: This is *NOT* winpars.cpp's initialization and shutdown code!!
 610 //     This module is an ancestor for tag handlers modules defined
 611 //     in m_*.cpp files with TAGS_MODULE_BEGIN...TAGS_MODULE_END construct.
 612 //
 613 //     Do not add any winpars.cpp shutdown or initialization code to it,
 614 //     create a new module instead!
 615
 616 IMPLEMENT_DYNAMIC_CLASS(wxHtmlTagsModule, wxModule)
 617
 618 bool wxHtmlTagsModule::OnInit()
 619 {
 620     wxHtmlWinParser::AddModule(this);
 621     return true;
 622 }
 623
 624 void wxHtmlTagsModule::OnExit()
 625 {
 626     wxHtmlWinParser::RemoveModule(this);
 627 }
 628
 629 #endif