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_nbsp = 0;
  51     m_EncConv = NULL;
  52     m_InputEnc = wxFONTENCODING_ISO8859_1;
  53     m_OutputEnc = wxFONTENCODING_DEFAULT;
  54 #endif
  55     m_lastWordCell = NULL;
  56
  57     {
  58         int i, j, k, l, m;
  59         for (i = 0; i < 2; i++)
  60             for (j = 0; j < 2; j++)
  61                 for (k = 0; k < 2; k++)
  62                     for (l = 0; l < 2; l++)
  63                         for (m = 0; m < 7; m++)
  64                         {
  65                             m_FontsTable[i][j][k][l][m] = NULL;
  66                             m_FontsFacesTable[i][j][k][l][m] = wxEmptyString;
  67 #if !wxUSE_UNICODE
  68                             m_FontsEncTable[i][j][k][l][m] = wxFONTENCODING_DEFAULT;
  69 #endif
  70                         }
  71
  72         SetFonts(wxEmptyString, wxEmptyString, NULL);
  73     }
  74
  75     // fill in wxHtmlParser's tables:
  76     wxList::compatibility_iterator node = m_Modules.GetFirst();
  77     while (node)
  78     {
  79         wxHtmlTagsModule *mod = (wxHtmlTagsModule*) node->GetData();
  80         mod->FillHandlersTable(this);
  81         node = node->GetNext();
  82     }
  83 }
  84
  85 wxHtmlWinParser::~wxHtmlWinParser()
  86 {
  87     int i, j, k, l, m;
  88
  89     for (i = 0; i < 2; i++)
  90         for (j = 0; j < 2; j++)
  91             for (k = 0; k < 2; k++)
  92                 for (l = 0; l < 2; l++)
  93                     for (m = 0; m < 7; m++)
  94                     {
  95                         if (m_FontsTable[i][j][k][l][m] != NULL)
  96                             delete m_FontsTable[i][j][k][l][m];
  97                     }
  98 #if !wxUSE_UNICODE
  99     delete m_EncConv;
 100 #endif
 101     delete[] m_tmpStrBuf;
 102 }
 103
 104 void wxHtmlWinParser::AddModule(wxHtmlTagsModule *module)
 105 {
 106     m_Modules.Append(module);
 107 }
 108
 109 void wxHtmlWinParser::RemoveModule(wxHtmlTagsModule *module)
 110 {
 111     m_Modules.DeleteObject(module);
 112 }
 113
 114 // build all HTML font sizes (1..7) from the given base size
 115 static void wxBuildFontSizes(int *sizes, int size)
 116 {
 117     // using a fixed factor (1.2, from CSS2) is a bad idea as explained at
 118     // http://www.w3.org/TR/CSS21/fonts.html#font-size-props but this is by far
 119     // simplest thing to do so still do it like this for now
 120     sizes[0] = int(size * 0.75); // exception to 1.2 rule, otherwise too small
 121     sizes[1] = int(size * 0.83);
 122     sizes[2] = size;
 123     sizes[3] = int(size * 1.2);
 124     sizes[4] = int(size * 1.44);
 125     sizes[5] = int(size * 1.73);
 126     sizes[6] = int(size * 2);
 127 }
 128
 129 static int wxGetDefaultHTMLFontSize()
 130 {
 131     // base the default font size on the size of the default system font but
 132     // also ensure that we have a font of reasonable size, otherwise small HTML
 133     // fonts are unreadable
 134     int size = wxNORMAL_FONT->GetPointSize();
 135     if ( size < 10 )
 136         size = 10;
 137     return size;
 138 }
 139
 140 void wxHtmlWinParser::SetFonts(const wxString& normal_face,
 141                                const wxString& fixed_face,
 142                                const int *sizes)
 143 {
 144     static int default_sizes[7] = { 0 };
 145     if ( !sizes )
 146     {
 147         if ( !default_sizes[0] )
 148             wxBuildFontSizes(default_sizes, wxGetDefaultHTMLFontSize());
 149
 150         sizes = default_sizes;
 151     }
 152
 153     int i, j, k, l, m;
 154
 155     for (i = 0; i < 7; i++)
 156         m_FontsSizes[i] = sizes[i];
 157
 158     m_FontFaceFixed = fixed_face;
 159     m_FontFaceNormal = normal_face;
 160
 161 #if !wxUSE_UNICODE
 162     SetInputEncoding(m_InputEnc);
 163 #endif
 164
 165     for (i = 0; i < 2; i++)
 166         for (j = 0; j < 2; j++)
 167             for (k = 0; k < 2; k++)
 168                 for (l = 0; l < 2; l++)
 169                     for (m = 0; m < 7; m++) {
 170                         if (m_FontsTable[i][j][k][l][m] != NULL)
 171                         {
 172                             delete m_FontsTable[i][j][k][l][m];
 173                             m_FontsTable[i][j][k][l][m] = NULL;
 174                         }
 175                     }
 176 }
 177
 178 void wxHtmlWinParser::SetStandardFonts(int size,
 179                                        const wxString& normal_face,
 180                                        const wxString& fixed_face)
 181 {
 182     if (size == -1)
 183         size = wxGetDefaultHTMLFontSize();
 184
 185     int f_sizes[7];
 186     wxBuildFontSizes(f_sizes, size);
 187
 188     wxString normal = normal_face;
 189     if ( normal.empty() )
 190         normal = wxNORMAL_FONT->GetFaceName();
 191
 192     SetFonts(normal, fixed_face, f_sizes);
 193 }
 194
 195 void wxHtmlWinParser::InitParser(const wxString& source)
 196 {
 197     wxHtmlParser::InitParser(source);
 198     wxASSERT_MSG(m_DC != NULL, wxT("no DC assigned to wxHtmlWinParser!!"));
 199
 200     m_FontBold = m_FontItalic = m_FontUnderlined = m_FontFixed = FALSE;
 201     m_FontSize = 3; //default one
 202     CreateCurrentFont();           // we're selecting default font into
 203
 204     // we're not using GetCharWidth/Height() because of
 205     // differences under X and win
 206     wxCoord w,h;
 207     m_DC->GetTextExtent( wxT("H"), &w, &h);
 208     m_CharWidth = w;
 209     m_CharHeight = h;
 210
 211     m_UseLink = false;
 212     m_Link = wxHtmlLinkInfo( wxEmptyString );
 213     m_LinkColor.Set(0, 0, 0xFF);
 214     m_ActualColor.Set(0, 0, 0);
 215     m_Align = wxHTML_ALIGN_LEFT;
 216     m_ScriptMode = wxHTML_SCRIPT_NORMAL;
 217     m_ScriptBaseline = 0;
 218     m_tmpLastWasSpace = false;
 219     m_lastWordCell = NULL;
 220
 221     // open the toplevel container that contains everything else and that
 222     // is never closed (this makes parser's life easier):
 223     OpenContainer();
 224
 225     // then open the first container into which page's content will go:
 226     OpenContainer();
 227
 228 #if !wxUSE_UNICODE
 229     wxString charset = ExtractCharsetInformation(source);
 230     if (!charset.empty())
 231     {
 232         wxFontEncoding enc = wxFontMapper::Get()->CharsetToEncoding(charset);
 233         if (enc != wxFONTENCODING_SYSTEM)
 234           SetInputEncoding(enc);
 235     }
 236 #endif
 237
 238     m_Container->InsertCell(new wxHtmlColourCell(m_ActualColor));
 239     wxColour windowColour = wxSystemSettings::GetColour(wxSYS_COLOUR_WINDOW) ;
 240
 241     m_Container->InsertCell
 242                  (
 243                    new wxHtmlColourCell
 244                        (
 245                          m_windowInterface
 246                             ? m_windowInterface->GetHTMLBackgroundColour()
 247                             : windowColour,
 248                          wxHTML_CLR_BACKGROUND
 249                        )
 250                   );
 251
 252     m_Container->InsertCell(new wxHtmlFontCell(CreateCurrentFont()));
 253 }
 254
 255 void wxHtmlWinParser::DoneParser()
 256 {
 257     m_Container = NULL;
 258 #if !wxUSE_UNICODE
 259     SetInputEncoding(wxFONTENCODING_ISO8859_1); // for next call
 260 #endif
 261     wxHtmlParser::DoneParser();
 262 }
 263
 264 #if WXWIN_COMPATIBILITY_2_6
 265 wxHtmlWindow *wxHtmlWinParser::GetWindow()
 266 {
 267     if (!m_windowInterface)
 268         return NULL;
 269     return wxDynamicCast(m_windowInterface->GetHTMLWindow(), wxHtmlWindow);
 270 }
 271 #endif
 272
 273 wxObject* wxHtmlWinParser::GetProduct()
 274 {
 275     wxHtmlContainerCell *top;
 276
 277     CloseContainer();
 278     OpenContainer();
 279
 280     top = m_Container;
 281     while (top->GetParent()) top = top->GetParent();
 282     top->RemoveExtraSpacing(true, true);
 283
 284     return top;
 285 }
 286
 287 wxFSFile *wxHtmlWinParser::OpenURL(wxHtmlURLType type,
 288                                    const wxString& url) const
 289 {
 290     if ( !m_windowInterface )
 291         return wxHtmlParser::OpenURL(type, url);
 292
 293     wxString myurl(url);
 294     wxHtmlOpeningStatus status;
 295     for (;;)
 296     {
 297         wxString myfullurl(myurl);
 298
 299         // consider url as absolute path first
 300         wxURI current(myurl);
 301         myfullurl = current.BuildUnescapedURI();
 302
 303         // if not absolute then ...
 304         if( current.IsReference() )
 305         {
 306             wxString basepath = GetFS()->GetPath();
 307             wxURI base(basepath);
 308
 309             // ... try to apply base path if valid ...
 310             if( !base.IsReference() )
 311             {
 312                 wxURI path(myfullurl);
 313                 path.Resolve( base );
 314                 myfullurl = path.BuildUnescapedURI();
 315             }
 316             else
 317             {
 318                 // ... or force such addition if not included already
 319                 if( !current.GetPath().Contains(base.GetPath()) )
 320                 {
 321                     basepath += myurl;
 322                     wxURI connected( basepath );
 323                     myfullurl = connected.BuildUnescapedURI();
 324                 }
 325             }
 326         }
 327
 328         wxString redirect;
 329         status = m_windowInterface->OnHTMLOpeningURL(type, myfullurl, &redirect);
 330         if ( status != wxHTML_REDIRECT )
 331             break;
 332
 333         myurl = redirect;
 334     }
 335
 336     if ( status == wxHTML_BLOCK )
 337         return NULL;
 338
 339     int flags = wxFS_READ;
 340     if (type == wxHTML_URL_IMAGE)
 341         flags |= wxFS_SEEKABLE;
 342
 343     return GetFS()->OpenFile(myurl, flags);
 344 }
 345
 346 void wxHtmlWinParser::AddText(const wxString& txt)
 347 {
 348     register wxChar d;
 349     int templen = 0;
 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);
 395             m_tmpLastWasSpace = true;
 396         }
 397     }
 398
 399     if (templen && (templen > 1 || temp[0] != wxT(' ')))
 400     {
 401         DoAddText(temp, templen);
 402         m_tmpLastWasSpace = false;
 403     }
 404 }
 405
 406 void wxHtmlWinParser::DoAddText(wxChar *temp, int& templen)
 407 {
 408     #define NBSP_UNICODE_VALUE 160
 409 #if !wxUSE_UNICODE
 410     if ( m_nbsp == 0 )
 411         m_nbsp = GetEntitiesParser()->GetCharForCode(NBSP_UNICODE_VALUE);
 412     #define CUR_NBSP_VALUE m_nbsp
 413 #else
 414     #define CUR_NBSP_VALUE NBSP_UNICODE_VALUE
 415 #endif
 416
 417     temp[templen] = 0;
 418     templen = 0;
 419 #if !wxUSE_UNICODE
 420     if (m_EncConv)
 421         m_EncConv->Convert(temp);
 422 #endif
 423     size_t len = wxStrlen(temp);
 424     for (size_t j = 0; j < len; j++)
 425     {
 426         if (temp[j] == CUR_NBSP_VALUE)
 427             temp[j] = wxT(' ');
 428     }
 429
 430     wxHtmlCell *c = new wxHtmlWordCell(temp, *(GetDC()));
 431
 432     ApplyStateToCell(c);
 433
 434     m_Container->InsertCell(c);
 435     ((wxHtmlWordCell*)c)->SetPreviousWord(m_lastWordCell);
 436     m_lastWordCell = (wxHtmlWordCell*)c;
 437 }
 438
 439
 440
 441 wxHtmlContainerCell* wxHtmlWinParser::OpenContainer()
 442 {
 443     m_Container = new wxHtmlContainerCell(m_Container);
 444     m_Container->SetAlignHor(m_Align);
 445     m_tmpLastWasSpace = true;
 446         /* to avoid space being first character in paragraph */
 447     return m_Container;
 448 }
 449
 450
 451
 452 wxHtmlContainerCell* wxHtmlWinParser::SetContainer(wxHtmlContainerCell *c)
 453 {
 454     m_tmpLastWasSpace = true;
 455         /* to avoid space being first character in paragraph */
 456     return m_Container = c;
 457 }
 458
 459
 460
 461 wxHtmlContainerCell* wxHtmlWinParser::CloseContainer()
 462 {
 463     m_Container = m_Container->GetParent();
 464     return m_Container;
 465 }
 466
 467
 468 void wxHtmlWinParser::SetFontSize(int s)
 469 {
 470     if (s < 1)
 471         s = 1;
 472     else if (s > 7)
 473         s = 7;
 474     m_FontSize = s;
 475 }
 476
 477
 478
 479 wxFont* wxHtmlWinParser::CreateCurrentFont()
 480 {
 481     int fb = GetFontBold(),
 482         fi = GetFontItalic(),
 483         fu = GetFontUnderlined(),
 484         ff = GetFontFixed(),
 485         fs = GetFontSize() - 1 /*remap from <1;7> to <0;6>*/ ;
 486
 487     wxString face = ff ? m_FontFaceFixed : m_FontFaceNormal;
 488     wxString *faceptr = &(m_FontsFacesTable[fb][fi][fu][ff][fs]);
 489     wxFont **fontptr = &(m_FontsTable[fb][fi][fu][ff][fs]);
 490 #if !wxUSE_UNICODE
 491     wxFontEncoding *encptr = &(m_FontsEncTable[fb][fi][fu][ff][fs]);
 492 #endif
 493
 494     if (*fontptr != NULL && (*faceptr != face
 495 #if !wxUSE_UNICODE
 496                              || *encptr != m_OutputEnc
 497 #endif
 498                             ))
 499     {
 500         delete *fontptr;
 501         *fontptr = NULL;
 502     }
 503
 504     if (*fontptr == NULL)
 505     {
 506         *faceptr = face;
 507         *fontptr = new wxFont(
 508                        (int) (m_FontsSizes[fs] * m_PixelScale),
 509                        ff ? wxMODERN : wxSWISS,
 510                        fi ? wxITALIC : wxNORMAL,
 511                        fb ? wxBOLD : wxNORMAL,
 512                        fu ? true : false, face
 513 #if wxUSE_UNICODE
 514                        );
 515 #else
 516                        , m_OutputEnc);
 517         *encptr = m_OutputEnc;
 518 #endif
 519     }
 520     m_DC->SetFont(**fontptr);
 521     return (*fontptr);
 522 }
 523
 524
 525
 526 void wxHtmlWinParser::SetLink(const wxHtmlLinkInfo& link)
 527 {
 528     m_Link = link;
 529     m_UseLink = (link.GetHref() != wxEmptyString);
 530 }
 531
 532 void wxHtmlWinParser::SetFontFace(const wxString& face)
 533 {
 534     if (GetFontFixed())
 535         m_FontFaceFixed = face;
 536     else
 537         m_FontFaceNormal = face;
 538
 539 #if !wxUSE_UNICODE
 540     if (m_InputEnc != wxFONTENCODING_DEFAULT)
 541         SetInputEncoding(m_InputEnc);
 542 #endif
 543 }
 544
 545 void wxHtmlWinParser::ApplyStateToCell(wxHtmlCell *cell)
 546 {
 547     // set the link:
 548     if (m_UseLink)
 549         cell->SetLink(GetLink());
 550
 551     // apply current script mode settings:
 552     cell->SetScriptMode(GetScriptMode(), GetScriptBaseline());
 553 }
 554
 555
 556 #if !wxUSE_UNICODE
 557 void wxHtmlWinParser::SetInputEncoding(wxFontEncoding enc)
 558 {
 559     // the character used for non-breakable space may change:
 560     m_nbsp = 0;
 561
 562     m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 563     if (m_EncConv)
 564     {
 565         delete m_EncConv;
 566         m_EncConv = NULL;
 567     }
 568
 569     if (enc == wxFONTENCODING_DEFAULT)
 570         return;
 571
 572     wxFontEncoding altfix, altnorm;
 573     bool availfix, availnorm;
 574
 575     availnorm = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceNormal);
 576     availfix = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceFixed);
 577
 578     if (availnorm && availfix)
 579     {
 580         // exact match?
 581         m_OutputEnc = enc;
 582     }
 583
 584     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false) &&
 585              wxFontMapper::Get()->GetAltForEncoding(enc, &altfix, m_FontFaceFixed, false) &&
 586              altnorm == altfix)
 587     {
 588         // alternatives?
 589         m_OutputEnc = altnorm;
 590     }
 591     else if (availnorm)
 592     {
 593         // at least normal face?
 594         m_OutputEnc = enc;
 595     }
 596     else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, false))
 597     {
 598         m_OutputEnc = altnorm;
 599     }
 600     else
 601     {
 602 #ifndef __WXMAC__
 603         // okay, let's convert to ISO_8859-1, available always
 604         m_OutputEnc = wxFONTENCODING_DEFAULT;
 605 #else
 606         m_OutputEnc = wxLocale::GetSystemEncoding() ;
 607 #endif
 608     }
 609
 610     m_InputEnc = enc;
 611     if (m_OutputEnc == wxFONTENCODING_DEFAULT)
 612     {
 613         GetEntitiesParser()->SetEncoding(wxFONTENCODING_SYSTEM);
 614     }
 615     else
 616     {
 617         GetEntitiesParser()->SetEncoding(m_OutputEnc);
 618     }
 619
 620     if (m_InputEnc == m_OutputEnc)
 621         return;
 622
 623     m_EncConv = new wxEncodingConverter();
 624     if (!m_EncConv->Init(m_InputEnc,
 625                            (m_OutputEnc == wxFONTENCODING_DEFAULT) ?
 626                                       wxFONTENCODING_ISO8859_1 : m_OutputEnc,
 627                            wxCONVERT_SUBSTITUTE))
 628     { // total failure :-(
 629         wxLogError(_("Failed to display HTML document in %s encoding"),
 630                    wxFontMapper::GetEncodingName(enc).c_str());
 631         m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
 632         delete m_EncConv;
 633         m_EncConv = NULL;
 634     }
 635 }
 636 #endif
 637
 638
 639
 640
 641 //-----------------------------------------------------------------------------
 642 // wxHtmlWinTagHandler
 643 //-----------------------------------------------------------------------------
 644
 645 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinTagHandler, wxHtmlTagHandler)
 646
 647 //-----------------------------------------------------------------------------
 648 // wxHtmlTagsModule
 649 //-----------------------------------------------------------------------------
 650
 651 // NB: This is *NOT* winpars.cpp's initialization and shutdown code!!
 652 //     This module is an ancestor for tag handlers modules defined
 653 //     in m_*.cpp files with TAGS_MODULE_BEGIN...TAGS_MODULE_END construct.
 654 //
 655 //     Do not add any winpars.cpp shutdown or initialization code to it,
 656 //     create a new module instead!
 657
 658 IMPLEMENT_DYNAMIC_CLASS(wxHtmlTagsModule, wxModule)
 659
 660 bool wxHtmlTagsModule::OnInit()
 661 {
 662     wxHtmlWinParser::AddModule(this);
 663     return true;
 664 }
 665
 666 void wxHtmlTagsModule::OnExit()
 667 {
 668     wxHtmlWinParser::RemoveModule(this);
 669 }
 670
 671 #endif