src/richtext/richtexthtml.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        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
  23 #ifndef WX_PRECOMP
  24   #include "wx/wx.h"
  25 #endif
  26
  27 #include "wx/filename.h"
  28 #include "wx/wfstream.h"
  29 #include "wx/txtstrm.h"
  30
  31 IMPLEMENT_DYNAMIC_CLASS(wxRichTextHTMLHandler, wxRichTextFileHandler)
  32
  33 /// Can we handle this filename (if using files)? By default, checks the extension.
  34 bool wxRichTextHTMLHandler::CanHandle(const wxString& filename) const
  35 {
  36     wxString path, file, ext;
  37     wxSplitPath(filename, & path, & file, & ext);
  38
  39     return (ext.Lower() == wxT("html") || ext.Lower() == wxT("htm"));
  40 }
  41
  42
  43 #if wxUSE_STREAMS
  44 bool wxRichTextHTMLHandler::DoLoadFile(wxRichTextBuffer *WXUNUSED(buffer), wxInputStream& WXUNUSED(stream))
  45 {
  46     return false;
  47 }
  48
  49 /*
  50  * We need to output only _changes_ in character formatting.
  51  */
  52
  53 bool wxRichTextHTMLHandler::DoSaveFile(wxRichTextBuffer *buffer, wxOutputStream& stream)
  54 {
  55     buffer->Defragment();
  56
  57     wxTextOutputStream str(stream);
  58
  59     wxTextAttrEx currentParaStyle = buffer->GetAttributes();
  60     wxTextAttrEx currentCharStyle = buffer->GetAttributes();
  61
  62     str << wxT("<html><head></head><body>\n");
  63
  64     /*
  65     wxRichText may be support paper formats like a1/a2/a3/a4
  66     when this widget grown enough, i should turn back and support its new features
  67     but not yet
  68
  69       str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td>");
  70
  71         wxString left_indent = SymbolicIndent(currentParaStyle.GetLeftIndent());
  72         wxString right_indent = SymbolicIndent(currentParaStyle.GetRightIndent());
  73
  74           str << wxString::Format(wxT("%s</td><td></td><td>%s</td></tr><tr>"),
  75           left_indent.c_str(), //Document-Wide Left Indent
  76           right_indent.c_str()); //Document-Wide Right Indent
  77
  78             str << wxT("<td></td><td width=\"100%\">");
  79     */
  80
  81     str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"100%\">");
  82
  83     str << wxString::Format(wxT("<font face=\"%s\" size=\"%i\" color=\"#%02X%02X%02X\" >"),
  84         currentParaStyle.GetFont().GetFaceName(), Pt_To_Size( currentParaStyle.GetFont().GetPointSize() ),
  85         currentParaStyle.GetTextColour().Red(), currentParaStyle.GetTextColour().Green(),
  86         currentParaStyle.GetTextColour().Blue());
  87
  88     //wxString align = GetAlignment( currentParaStyle.GetAlignment() );
  89     //str << wxString::Format(wxT("<p align=\"%s\">"), align );
  90
  91     m_font = false;
  92     m_indent = 0;
  93     m_list = false;
  94
  95     wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
  96     while (node)
  97     {
  98         wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
  99         wxASSERT (para != NULL);
 100
 101         if (para)
 102         {
 103             OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream);
 104
 105             wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
 106             while (node2)
 107             {
 108                 wxRichTextObject* obj = node2->GetData();
 109                 wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
 110                 if (textObj && !textObj->IsEmpty())
 111                 {
 112                     BeginCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
 113
 114                     str << textObj->GetText();
 115
 116                     EndCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
 117                 }
 118
 119                 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 120                 if( image && !image->IsEmpty())
 121                     Image_to_Base64( image, stream );
 122
 123                 node2 = node2->GetNext();
 124             }
 125             //OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream, false);
 126         }
 127         node = node->GetNext();
 128     }
 129
 130     str << wxT("</font></td></tr></table></body></html>\n");
 131
 132     return true;
 133 }
 134
 135 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttrEx& currentStyle, const wxTextAttrEx& thisStyle, wxOutputStream& stream)
 136 {
 137     wxTextOutputStream str(stream);
 138
 139     //Is the item bulleted one?
 140     if( thisStyle.GetBulletStyle() != wxTEXT_ATTR_BULLET_STYLE_NONE )
 141     {
 142         //Is there any opened list?
 143         if( m_list )
 144         {
 145             //Yes there is
 146
 147             //Is the item among the previous ones
 148             //Is the item one of the previous list tag's child items
 149             if( (thisStyle.GetLeftIndent() == (m_indent + 100)) || (thisStyle.GetLeftIndent() < 100) )
 150                 str << wxT("<li>");//Yes it is
 151             else
 152             {
 153                 //No it isn't
 154
 155                 //So we should close the list tag
 156                 str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 157
 158                 //And renavigate to new list's horizontal position
 159                 NavigateToListPosition(thisStyle, str);
 160                 //Ok it's done
 161
 162                 //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 163                 wxString tag;
 164                 TypeOfList(thisStyle, tag);
 165                 str << wxString::Format(wxT("%s<li>"), tag);
 166             }
 167         }
 168         else
 169         {
 170             //No there isn't a list
 171
 172             //navigate to new list's horizontal position(indent)
 173             NavigateToListPosition(thisStyle, str);
 174
 175             //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 176             wxString tag;
 177             TypeOfList(thisStyle, tag);
 178             str << wxString::Format(wxT("%s<li>"), tag);
 179
 180             //Now we have a list, mark it.
 181             m_list = true;
 182         }
 183     }
 184     else if( m_list )
 185     {
 186         //The item is not bulleted and there is a list what should be closed now.
 187         //So close the list
 188
 189         str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 190         //And mark as there is no an opened list
 191         m_list = false;
 192     }
 193
 194     // does the item have an indentation ?
 195     if( thisStyle.GetLeftIndent() )
 196     {
 197         if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_NONE )
 198         {
 199             if( m_indent )
 200             {
 201                 if( (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) == m_indent )
 202                 {
 203                     if( thisStyle.GetLeftSubIndent() < 0 )
 204                     {
 205                         wxString symbolic_indent = SymbolicIndent(~thisStyle.GetLeftSubIndent());
 206                         str << wxString::Format(wxT("%s"), symbolic_indent);
 207                     }
 208                 }
 209                 else
 210                 {
 211                     if( thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent() > m_indent )
 212                     {
 213                         Indent(thisStyle, str);
 214                         m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 215                         m_indents.Add( m_indent );
 216                     }
 217                     else
 218                     {
 219                         int i = m_indents.size() - 1;
 220                         for(; i > -1; i--)
 221                         {
 222                             if( m_indent < (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
 223                             {
 224                                 Indent(thisStyle, str);
 225                                 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 226                                 m_indents.Add( m_indent );
 227
 228                                 break;
 229                             }
 230                             else if( m_indent == (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
 231                             {
 232                                 if( thisStyle.GetLeftSubIndent() < 0 )
 233                                 {
 234                                     wxString symbolic_indent = SymbolicIndent(~thisStyle.GetLeftSubIndent());
 235                                     str << wxString::Format(wxT("%s"), symbolic_indent);
 236                                 }
 237                                 break;
 238                             }
 239                             else
 240                             {
 241                                 str << wxT("</td></tr></table>");
 242
 243                                 m_indents.RemoveAt(i);
 244
 245                                 if( i < 1 ){m_indent=0; break;}
 246                                 m_indent = m_indents[i-1];
 247                             }
 248                         }
 249                     }
 250                 }
 251             }
 252             else
 253             {
 254                 Indent(thisStyle, str);
 255                 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 256                 m_indents.Add( m_indent );
 257             }
 258         }
 259     }
 260     else if( m_indent )
 261     {
 262         //The item is not indented and there is a table(s) what should be closed now.
 263
 264         //So close them
 265         for(unsigned int i = 0; i < m_indents.size(); i++ )
 266             str << wxT("</td></tr></table>");
 267
 268         m_indent = 0;
 269         m_indents.Clear();
 270     }
 271
 272
 273     wxString style;
 274
 275     //Is there any change on the font properties of the item
 276     if( thisStyle.GetFont().GetFaceName() != currentStyle.GetFont().GetFaceName() )
 277         style += wxString::Format(wxT(" face=\"%s\""), thisStyle.GetFont().GetFaceName());
 278     if( thisStyle.GetFont().GetPointSize() != currentStyle.GetFont().GetPointSize() )
 279         style += wxString::Format(wxT(" size=\"%i\""), Pt_To_Size(thisStyle.GetFont().GetPointSize()) );
 280     if( thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 281         style += wxString::Format(wxT(" color=\"#%02X%02X%02X\""), thisStyle.GetTextColour().Red(),
 282         thisStyle.GetTextColour().Green(), thisStyle.GetTextColour().Blue());
 283
 284     if( style.size() ){str << wxString::Format(wxT("<font %s >"), style); m_font = true;}
 285
 286     if( thisStyle.GetFont().GetWeight() == wxBOLD )
 287         str << wxT("<b>");
 288     if( thisStyle.GetFont().GetStyle() == wxITALIC )
 289         str << wxT("<i>");
 290     if( thisStyle.GetFont().GetUnderlined() )
 291         str << wxT("<u>");
 292 }
 293
 294 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxOutputStream& stream)
 295 {
 296     wxTextOutputStream str(stream);
 297
 298     if( thisStyle.GetFont().GetUnderlined() )
 299         str << wxT("</u>");
 300     if( thisStyle.GetFont().GetStyle() == wxITALIC )
 301         str << wxT("</i>");
 302     if( thisStyle.GetFont().GetWeight() == wxBOLD )
 303         str << wxT("</b>");
 304
 305     if( m_font )
 306     {
 307         m_font = false;
 308         str << wxT("</font>");
 309     }
 310 }
 311
 312 /// Output paragraph formatting
 313 void wxRichTextHTMLHandler::OutputParagraphFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxOutputStream& stream)
 314 {
 315     //If there is no opened list currently, insert a <p> after every paragraph
 316     if(!m_list)
 317     {
 318         wxTextOutputStream str(stream);
 319         wxString align = GetAlignment( thisStyle );
 320         str << wxString::Format(wxT("<p align=\"%s\">"), align);
 321     }
 322 }
 323
 324 void wxRichTextHTMLHandler::NavigateToListPosition(const wxTextAttrEx& thisStyle, wxTextOutputStream& str)
 325 {
 326     //indenting an item using an ul/ol tag is equal to inserting 5 x &nbsp; on its left side.
 327     //so we should start from 100 point left
 328
 329     //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
 330     //of the item, horizontally?
 331     if( m_indent + 100 < thisStyle.GetLeftIndent() )
 332     {
 333         //yes it is
 334         LIndent(thisStyle, str);
 335         m_indent = thisStyle.GetLeftIndent() - 100;
 336         m_indents.Add( m_indent );
 337         return;
 338     }
 339     //No it isn't
 340
 341     int i = m_indents.size() - 1;
 342     for(; i > -1; i--)
 343     {
 344         //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
 345         //of the item ?
 346         if( m_indent + 100 < thisStyle.GetLeftIndent() )
 347         {
 348             //Yes it is
 349             LIndent(thisStyle, str);
 350             m_indent = thisStyle.GetLeftIndent() - 100;
 351             m_indents.Add( m_indent );
 352             break;
 353         }
 354         else if( m_indent + 100 == thisStyle.GetLeftIndent() )
 355             break;//exact match
 356         else
 357         {
 358             //No it is not, the second td's left wall of the current indentaion table is at the
 359             //right side of the current item horizontally, so close it.
 360             str << wxT("</td></tr></table>");
 361
 362             m_indents.RemoveAt(i);
 363
 364             if( i < 1 ){m_indent=0; break;}
 365             m_indent = m_indents[i-1];
 366         }
 367     }
 368 }
 369 void wxRichTextHTMLHandler::Indent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
 370 {
 371     //As a five year experienced web developer i assure you there is no way to indent an item
 372     //in html way, but we can use tables.
 373
 374
 375
 376     //Item -> "Hello world"
 377     //Its Left Indentation -> 100
 378     //Its Left Sub-Indentation ->40
 379     //A typical indentation-table for the item will be construct as the following
 380
 381     //3 x nbsp = 60
 382     //2 x nbsp = 40
 383     //LSI = Left Sub Indent
 384     //LI = Left Indent - LSI
 385     //
 386     //-------------------------------------------
 387     //|&nbsp;&nbsp;nbsp;|nbsp;nbsp;Hello World  |
 388     //|            |        |    |                  |
 389     //|        V        |    V                  |
 390     //|      --LI--     | --LSI--               |
 391     //-------------------------------------------
 392
 393     str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
 394
 395     wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) - m_indent );
 396     str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent );
 397     str << wxT("<td width=\"100%\">");
 398
 399     if( thisStyle.GetLeftSubIndent() < 0 )
 400     {
 401         symbolic_indent = SymbolicIndent(~thisStyle.GetLeftSubIndent());
 402         str << wxString::Format(wxT("%s"), symbolic_indent);
 403     }
 404 }
 405
 406 void wxRichTextHTMLHandler::LIndent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
 407 {
 408     //Code:
 409     //r.BeginNumberedBullet(1, 200, 60);
 410     //r.Newline();
 411     //r.WriteText(wxT("first item"));
 412     //r.EndNumberedBullet();
 413     //r.BeginNumberedBullet(2, 200, 60);
 414     //r.Newline();
 415     //r.WriteText(wxT("second item."));
 416     //r.EndNumberedBullet();
 417     //
 418     //A typical indentation-table for the item will be construct as the following
 419
 420     //1 x nbsp = 20 point
 421     //ULI -> 100pt (UL/OL tag indents its sub element by 100 point)
 422     //<--------- 100 pt ---------->|
 423     //------------------------------------------------------
 424     //|&nbsp;&nbsp;nbsp;&nbsp;nbsp;|<ul>                   |
 425     //|                                |<-ULI-><li>first item  |
 426     //|                            |<-ULI-><li>second item |
 427     //|                            |</ul>                  |
 428     //------------------------------------------------------
 429     //                             |<-100->|
 430
 431
 432     str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
 433
 434     wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() - m_indent) - 100);
 435     str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent );
 436     str << wxT("<td width=\"100%\">");
 437 }
 438
 439 void wxRichTextHTMLHandler::TypeOfList( const wxTextAttrEx& thisStyle, wxString& tag )
 440 {
 441     //We can use number attribute of li tag but not all the browsers support it.
 442     //also wxHtmlWindow doesn't support type attribute.
 443
 444     m_is_ul = false;
 445     if( thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 446         tag = wxT("<ol type=\"1\">");
 447     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER )
 448         tag = wxT("<ol type=\"A\">");
 449     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER )
 450         tag = wxT("<ol type=\"a\">");
 451     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER )
 452         tag = wxT("<ol type=\"I\">");
 453     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER )
 454         tag = wxT("<ol type=\"i\">");
 455     else
 456     {
 457         tag = wxT("<ul>");
 458         m_is_ul = true;
 459     }
 460 }
 461
 462 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttrEx& thisStyle )
 463 {
 464     switch( thisStyle.GetAlignment() )
 465     {
 466     case wxTEXT_ALIGNMENT_LEFT:
 467         return  wxT("left");
 468     case wxTEXT_ALIGNMENT_RIGHT:
 469         return wxT("right");
 470     case wxTEXT_ALIGNMENT_CENTER:
 471         return wxT("center");
 472     case wxTEXT_ALIGNMENT_JUSTIFIED:
 473         return wxT("justify");
 474     default:
 475         return wxT("left");
 476     }
 477 }
 478
 479 void wxRichTextHTMLHandler::Image_to_Base64(wxRichTextImage* image, wxOutputStream& stream)
 480 {
 481     wxTextOutputStream str(stream);
 482
 483     str << wxT("<img src=\"");
 484     str << wxT("data:");
 485     str << GetMimeType(image->GetImageBlock().GetImageType());
 486     str << wxT(";base64,");
 487
 488     wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 489     str << data;
 490
 491     delete[] data;
 492
 493     str << wxT("\" />");
 494 }
 495
 496 long wxRichTextHTMLHandler::Pt_To_Size(long size)
 497 {
 498     //return most approximate size
 499     if(size < 9 ) return 1;
 500     else if( size < 11 ) return 2;
 501     else if( size < 14 ) return 3;
 502     else if( size < 18 ) return 4;
 503     else if( size < 23 ) return 5;
 504     else if( size < 30 ) return 6;
 505     else return 7;
 506 }
 507
 508 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 509 {
 510     wxString in;
 511     for(;indent > 0; indent -= 20)
 512         in.Append( wxT("&nbsp;") );
 513     return in;
 514 }
 515
 516 wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 517 {
 518     switch(imageType)
 519     {
 520     case wxBITMAP_TYPE_BMP:
 521         return wxT("image/bmp");
 522     case wxBITMAP_TYPE_TIF:
 523         return wxT("image/tiff");
 524     case wxBITMAP_TYPE_GIF:
 525         return wxT("image/gif");
 526     case wxBITMAP_TYPE_PNG:
 527         return wxT("image/png");
 528     case wxBITMAP_TYPE_JPEG:
 529         return wxT("image/jpeg");
 530     default:
 531         return wxT("image/unknown");
 532     }
 533 }
 534
 535 //exim-style base64 encoder
 536 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 537 {
 538     //elements of enc64 array must be 8 bit values
 539     //otherwise encoder will fail
 540     //hmmm.. Does wxT macro define a char as 16 bit value
 541     //when compiling with UNICODE option?
 542     const static wxChar* enc64 = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 543     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 544     wxChar* p = output;
 545
 546     while( in_len-- > 0 )
 547     {
 548         register wxChar a, b;
 549
 550         a = *input++;
 551
 552         *p++ = enc64[ (a >> 2) & 0x3f ];
 553
 554         if( in_len-- <= 0 )
 555         {
 556             *p++ = enc64[ (a << 4 ) & 0x30 ];
 557             *p++ = '=';
 558             *p++ = '=';
 559             break;
 560         }
 561
 562         b = *input++;
 563
 564         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 565
 566         if( in_len-- <= 0 )
 567         {
 568             *p++ = enc64[ (b << 2) & 0x3f ];
 569             *p++ = '=';
 570             break;
 571         }
 572
 573         a = *input++;
 574
 575         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 576
 577         *p++ = enc64[ a & 0x3f ];
 578     }
 579     *p = 0;
 580
 581     return output;
 582 }
 583 #endif
 584 // wxUSE_STREAMS
 585
 586 #endif
 587 // wxUSE_RICHTEXT
 588