src/richtext/richtexthtml.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        src/richtext/richtexthtml.cpp
   3 // Purpose:     HTML I/O for wxRichTextCtrl
   4 // Author:      Julian Smart
   5 // Modified by:
   6 // Created:     2005-09-30
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) Julian Smart
   9 // Licence:     wxWindows licence
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 // For compilers that support precompilation, includes "wx.h".
  13 #include "wx/wxprec.h"
  14
  15 #ifdef __BORLANDC__
  16     #pragma hdrstop
  17 #endif
  18
  19 #if wxUSE_RICHTEXT
  20
  21 #include "wx/richtext/richtexthtml.h"
  22 #include "wx/richtext/richtextstyles.h"
  23
  24 #ifndef WX_PRECOMP
  25 #endif
  26
  27 #include "wx/filename.h"
  28 #include "wx/wfstream.h"
  29 #include "wx/txtstrm.h"
  30
  31 #if wxUSE_FILESYSTEM
  32 #include "wx/filesys.h"
  33 #include "wx/fs_mem.h"
  34 #endif
  35
  36 IMPLEMENT_DYNAMIC_CLASS(wxRichTextHTMLHandler, wxRichTextFileHandler)
  37
  38 int wxRichTextHTMLHandler::sm_fileCounter = 1;
  39
  40 wxRichTextHTMLHandler::wxRichTextHTMLHandler(const wxString& name, const wxString& ext, int type)
  41     : wxRichTextFileHandler(name, ext, type), m_buffer(NULL), m_font(false), m_inTable(false)
  42 {
  43     m_fontSizeMapping.Add(8);
  44     m_fontSizeMapping.Add(10);
  45     m_fontSizeMapping.Add(13);
  46     m_fontSizeMapping.Add(17);
  47     m_fontSizeMapping.Add(22);
  48     m_fontSizeMapping.Add(30);
  49     m_fontSizeMapping.Add(100);
  50 }
  51
  52 /// Can we handle this filename (if using files)? By default, checks the extension.
  53 bool wxRichTextHTMLHandler::CanHandle(const wxString& filename) const
  54 {
  55     wxString path, file, ext;
  56     wxFileName::SplitPath(filename, & path, & file, & ext);
  57
  58     return (ext.Lower() == wxT("html") || ext.Lower() == wxT("htm"));
  59 }
  60
  61
  62 #if wxUSE_STREAMS
  63 bool wxRichTextHTMLHandler::DoLoadFile(wxRichTextBuffer *WXUNUSED(buffer), wxInputStream& WXUNUSED(stream))
  64 {
  65     return false;
  66 }
  67
  68 /*
  69  * We need to output only _changes_ in character formatting.
  70  */
  71
  72 bool wxRichTextHTMLHandler::DoSaveFile(wxRichTextBuffer *buffer, wxOutputStream& stream)
  73 {
  74     m_buffer = buffer;
  75
  76     ClearTemporaryImageLocations();
  77
  78     buffer->Defragment();
  79
  80     wxTextOutputStream str(stream);
  81
  82     wxTextAttr currentParaStyle = buffer->GetAttributes();
  83     wxTextAttr currentCharStyle = buffer->GetAttributes();
  84
  85     if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
  86         str << wxT("<html><head></head><body>\n");
  87
  88     OutputFont(currentParaStyle, str);
  89
  90     m_font = false;
  91     m_inTable = false;
  92
  93     m_indents.Clear();
  94     m_listTypes.Clear();
  95
  96     wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
  97     while (node)
  98     {
  99         wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
 100         wxASSERT (para != NULL);
 101
 102         if (para)
 103         {
 104             wxTextAttr paraStyle(para->GetCombinedAttributes());
 105
 106             BeginParagraphFormatting(currentParaStyle, paraStyle, str);
 107
 108             wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
 109             while (node2)
 110             {
 111                 wxRichTextObject* obj = node2->GetData();
 112                 wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
 113                 if (textObj && !textObj->IsEmpty())
 114                 {
 115                     wxTextAttr charStyle(para->GetCombinedAttributes(obj->GetAttributes()));
 116                     BeginCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 117
 118                     wxString text = textObj->GetText();
 119
 120                     if (charStyle.HasTextEffects() && (charStyle.GetTextEffects() & wxTEXT_ATTR_EFFECT_CAPITALS))
 121                         text.MakeUpper();
 122
 123                     wxString toReplace = wxRichTextLineBreakChar;
 124                     text.Replace(toReplace, wxT("<br>"));
 125
 126                     str << text;
 127
 128                     EndCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 129                 }
 130
 131                 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 132                 if( image && (!image->IsEmpty() || image->GetImageBlock().GetData()))
 133                     WriteImage( image, stream );
 134
 135                 node2 = node2->GetNext();
 136             }
 137
 138             EndParagraphFormatting(currentParaStyle, paraStyle, str);
 139
 140             str << wxT("\n");
 141         }
 142         node = node->GetNext();
 143     }
 144
 145     CloseLists(-1, str);
 146
 147     str << wxT("</font>");
 148
 149     if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
 150         str << wxT("</body></html>");
 151
 152     str << wxT("\n");
 153
 154     m_buffer = NULL;
 155
 156     return true;
 157 }
 158
 159 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttr& currentStyle, const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& str)
 160 {
 161     wxString style;
 162
 163     // Is there any change in the font properties of the item?
 164     if (thisStyle.GetFontFaceName() != currentStyle.GetFontFaceName())
 165     {
 166         wxString faceName(thisStyle.GetFontFaceName());
 167         style += wxString::Format(wxT(" face=\"%s\""), faceName.c_str());
 168     }
 169     if (thisStyle.GetFontSize() != currentStyle.GetFontSize())
 170         style += wxString::Format(wxT(" size=\"%ld\""), PtToSize(thisStyle.GetFontSize()));
 171     if (thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 172     {
 173         wxString color(thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX));
 174         style += wxString::Format(wxT(" color=\"%s\""), color.c_str());
 175     }
 176
 177     if (style.size())
 178     {
 179         str << wxString::Format(wxT("<font %s >"), style.c_str());
 180         m_font = true;
 181     }
 182
 183     if (thisStyle.GetFontWeight() == wxBOLD)
 184         str << wxT("<b>");
 185     if (thisStyle.GetFontStyle() == wxITALIC)
 186         str << wxT("<i>");
 187     if (thisStyle.GetFontUnderlined())
 188         str << wxT("<u>");
 189
 190     if (thisStyle.HasURL())
 191         str << wxT("<a href=\"") << thisStyle.GetURL() << wxT("\">");
 192 }
 193
 194 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& stream)
 195 {
 196     if (thisStyle.HasURL())
 197         stream << wxT("</a>");
 198
 199     if (thisStyle.GetFontUnderlined())
 200         stream << wxT("</u>");
 201     if (thisStyle.GetFontStyle() == wxITALIC)
 202         stream << wxT("</i>");
 203     if (thisStyle.GetFontWeight() == wxBOLD)
 204         stream << wxT("</b>");
 205
 206     if (m_font)
 207     {
 208         m_font = false;
 209         stream << wxT("</font>");
 210     }
 211 }
 212
 213 /// Begin paragraph formatting
 214 void wxRichTextHTMLHandler::BeginParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& str)
 215 {
 216     if (thisStyle.HasPageBreak())
 217     {
 218         str << wxT("<div style=\"page-break-after:always\"></div>\n");
 219     }
 220
 221     if (thisStyle.HasLeftIndent() && thisStyle.GetLeftIndent() != 0)
 222     {
 223         if (thisStyle.HasBulletStyle())
 224         {
 225             int indent = thisStyle.GetLeftIndent();
 226
 227             // Close levels high than this
 228             CloseLists(indent, str);
 229
 230             if (m_indents.GetCount() > 0 && indent == m_indents.Last())
 231             {
 232                 // Same level, no need to start a new list
 233             }
 234             else if (m_indents.GetCount() == 0 || indent > m_indents.Last())
 235             {
 236                 m_indents.Add(indent);
 237
 238                 wxString tag;
 239                 int listType = TypeOfList(thisStyle, tag);
 240                 m_listTypes.Add(listType);
 241
 242                 // wxHTML needs an extra <p> before a list when using <p> ... </p> in previous paragraphs.
 243                 // TODO: pass a flag that indicates we're using wxHTML.
 244                 str << wxT("<p>\n");
 245
 246                 str << tag;
 247             }
 248
 249             str << wxT("<li> ");
 250         }
 251         else
 252         {
 253             CloseLists(-1, str);
 254
 255             wxString align = GetAlignment(thisStyle);
 256             str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
 257
 258             if (thisStyle.HasParagraphSpacingAfter() && thisStyle.GetParagraphSpacingAfter() == 0)
 259                 str << wxT(" style=\"line-height: 0px\"");
 260
 261             str << wxT(">");
 262
 263             // Use a table
 264             int indentTenthsMM = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 265             // TODO: convert to pixels
 266             int indentPixels = indentTenthsMM/4;
 267             str << wxString::Format(wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"%d\"></td><td>"), indentPixels);
 268
 269             OutputFont(thisStyle, str);
 270
 271             if (thisStyle.GetLeftSubIndent() < 0)
 272             {
 273                 str << SymbolicIndent( - thisStyle.GetLeftSubIndent());
 274             }
 275
 276             m_inTable = true;
 277         }
 278     }
 279     else
 280     {
 281         CloseLists(-1, str);
 282
 283         wxString align = GetAlignment(thisStyle);
 284         str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
 285
 286         if (thisStyle.HasParagraphSpacingAfter() && thisStyle.GetParagraphSpacingAfter() == 0)
 287             str << wxT(" style=\"line-height: 0px\"");
 288
 289         str << wxT(">");
 290     }
 291 }
 292
 293 /// End paragraph formatting
 294 void wxRichTextHTMLHandler::EndParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& stream)
 295 {
 296     if (m_inTable)
 297     {
 298         if (thisStyle.HasFont())
 299             stream << wxT("</font>");
 300
 301         stream << wxT("</td></tr></table></p>\n");
 302         m_inTable = false;
 303     }
 304     else if (!thisStyle.HasBulletStyle())
 305         stream << wxT("</p>\n");
 306 }
 307
 308 /// Closes lists to level (-1 means close all)
 309 void wxRichTextHTMLHandler::CloseLists(int level, wxTextOutputStream& str)
 310 {
 311     // Close levels high than this
 312     int i = m_indents.GetCount()-1;
 313     while (i >= 0)
 314     {
 315         int l = m_indents[i];
 316         if (l > level)
 317         {
 318             if (m_listTypes[i] == 0)
 319                 str << wxT("</ol>");
 320             else
 321                 str << wxT("</ul>");
 322             m_indents.RemoveAt(i);
 323             m_listTypes.RemoveAt(i);
 324         }
 325         else
 326             break;
 327         i --;
 328      }
 329 }
 330
 331 /// Output font tag
 332 void wxRichTextHTMLHandler::OutputFont(const wxTextAttr& style, wxTextOutputStream& stream)
 333 {
 334     if (style.HasFont())
 335     {
 336         stream << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\""), style.GetFontFaceName().c_str(), PtToSize(style.GetFontSize()));
 337         if (style.HasTextColour())
 338             stream << wxString::Format(wxT(" color=\"%s\""), style.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
 339         stream << wxT(" >");
 340     }
 341 }
 342
 343 int wxRichTextHTMLHandler::TypeOfList( const wxTextAttr& thisStyle, wxString& tag )
 344 {
 345     // We can use number attribute of li tag but not all the browsers support it.
 346     // also wxHtmlWindow doesn't support type attribute.
 347
 348     bool m_is_ul = false;
 349     if (thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 350         tag = wxT("<ol type=\"1\">");
 351     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER)
 352         tag = wxT("<ol type=\"A\">");
 353     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER)
 354         tag = wxT("<ol type=\"a\">");
 355     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER)
 356         tag = wxT("<ol type=\"I\">");
 357     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER)
 358         tag = wxT("<ol type=\"i\">");
 359     else
 360     {
 361         tag = wxT("<ul>");
 362         m_is_ul = true;
 363     }
 364
 365     if (m_is_ul)
 366         return 1;
 367     else
 368         return 0;
 369 }
 370
 371 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttr& thisStyle )
 372 {
 373     switch( thisStyle.GetAlignment() )
 374     {
 375     case wxTEXT_ALIGNMENT_LEFT:
 376         return  wxT("left");
 377     case wxTEXT_ALIGNMENT_RIGHT:
 378         return wxT("right");
 379     case wxTEXT_ALIGNMENT_CENTER:
 380         return wxT("center");
 381     case wxTEXT_ALIGNMENT_JUSTIFIED:
 382         return wxT("justify");
 383     default:
 384         return wxT("left");
 385     }
 386 }
 387
 388 void wxRichTextHTMLHandler::WriteImage(wxRichTextImage* image, wxOutputStream& stream)
 389 {
 390     wxTextOutputStream str(stream);
 391
 392     str << wxT("<img src=\"");
 393
 394 #if wxUSE_FILESYSTEM
 395     if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 396     {
 397         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 398             image->LoadFromBlock();
 399         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 400             image->MakeBlock();
 401
 402         if (image->GetImage().Ok())
 403         {
 404             wxString ext(image->GetImageBlock().GetExtension());
 405             wxString tempFilename(wxString::Format(wxT("image%d.%s"), sm_fileCounter, ext));
 406             wxMemoryFSHandler::AddFile(tempFilename, image->GetImage(), image->GetImageBlock().GetImageType());
 407
 408             m_imageLocations.Add(tempFilename);
 409
 410             str << wxT("memory:") << tempFilename;
 411         }
 412         else
 413             str << wxT("memory:?");
 414
 415         sm_fileCounter ++;
 416     }
 417     else if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 418     {
 419         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 420             image->LoadFromBlock();
 421         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 422             image->MakeBlock();
 423
 424         if (image->GetImage().Ok())
 425         {
 426             wxString tempDir(GetTempDir());
 427             if (tempDir.IsEmpty())
 428                 tempDir = wxFileName::GetTempDir();
 429
 430             wxString ext(image->GetImageBlock().GetExtension());
 431             wxString tempFilename(wxString::Format(wxT("%s/image%d.%s"), tempDir, sm_fileCounter, ext));
 432             image->GetImageBlock().Write(tempFilename);
 433
 434             m_imageLocations.Add(tempFilename);
 435
 436             str << wxFileSystem::FileNameToURL(tempFilename);
 437         }
 438         else
 439             str << wxT("file:?");
 440
 441         sm_fileCounter ++;
 442     }
 443     else // if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_BASE64) // this is implied
 444 #endif
 445     {
 446         str << wxT("data:");
 447         str << GetMimeType(image->GetImageBlock().GetImageType());
 448         str << wxT(";base64,");
 449
 450         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 451             image->MakeBlock();
 452
 453         wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 454         str << data;
 455
 456         delete[] data;
 457     }
 458
 459     str << wxT("\" />");
 460 }
 461
 462 long wxRichTextHTMLHandler::PtToSize(long size)
 463 {
 464     int i;
 465     int len = m_fontSizeMapping.GetCount();
 466     for (i = 0; i < len; i++)
 467         if (size <= m_fontSizeMapping[i])
 468             return i+1;
 469     return 7;
 470 }
 471
 472 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 473 {
 474     wxString in;
 475     for(;indent > 0; indent -= 20)
 476         in.Append( wxT("&nbsp;") );
 477     return in;
 478 }
 479
 480 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 481 {
 482     switch(imageType)
 483     {
 484     case wxBITMAP_TYPE_BMP:
 485         return wxT("image/bmp");
 486     case wxBITMAP_TYPE_TIF:
 487         return wxT("image/tiff");
 488     case wxBITMAP_TYPE_GIF:
 489         return wxT("image/gif");
 490     case wxBITMAP_TYPE_PNG:
 491         return wxT("image/png");
 492     case wxBITMAP_TYPE_JPEG:
 493         return wxT("image/jpeg");
 494     default:
 495         return wxT("image/unknown");
 496     }
 497 }
 498
 499 // exim-style base64 encoder
 500 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 501 {
 502     // elements of enc64 array must be 8 bit values
 503     // otherwise encoder will fail
 504     // hmmm.. Does wxT macro define a char as 16 bit value
 505     // when compiling with UNICODE option?
 506     static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 507     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 508     wxChar* p = output;
 509
 510     while( in_len-- > 0 )
 511     {
 512         register wxChar a, b;
 513
 514         a = *input++;
 515
 516         *p++ = enc64[ (a >> 2) & 0x3f ];
 517
 518         if( in_len-- == 0 )
 519         {
 520             *p++ = enc64[ (a << 4 ) & 0x30 ];
 521             *p++ = '=';
 522             *p++ = '=';
 523             break;
 524         }
 525
 526         b = *input++;
 527
 528         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 529
 530         if( in_len-- == 0 )
 531         {
 532             *p++ = enc64[ (b << 2) & 0x3f ];
 533             *p++ = '=';
 534             break;
 535         }
 536
 537         a = *input++;
 538
 539         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 540
 541         *p++ = enc64[ a & 0x3f ];
 542     }
 543     *p = 0;
 544
 545     return output;
 546 }
 547 #endif
 548 // wxUSE_STREAMS
 549
 550 /// Delete the in-memory or temporary files generated by the last operation
 551 bool wxRichTextHTMLHandler::DeleteTemporaryImages()
 552 {
 553     return DeleteTemporaryImages(GetFlags(), m_imageLocations);
 554 }
 555
 556 /// Delete the in-memory or temporary files generated by the last operation
 557 bool wxRichTextHTMLHandler::DeleteTemporaryImages(int flags, const wxArrayString& imageLocations)
 558 {
 559     size_t i;
 560     for (i = 0; i < imageLocations.GetCount(); i++)
 561     {
 562         wxString location = imageLocations[i];
 563
 564         if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 565         {
 566 #if wxUSE_FILESYSTEM
 567             wxMemoryFSHandler::RemoveFile(location);
 568 #endif
 569         }
 570         else if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 571         {
 572             if (wxFileExists(location))
 573                 wxRemoveFile(location);
 574         }
 575     }
 576
 577     return true;
 578 }
 579
 580
 581 #endif
 582 // wxUSE_RICHTEXT
 583