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
  23 #ifndef WX_PRECOMP
  24 #endif
  25
  26 #include "wx/filename.h"
  27 #include "wx/wfstream.h"
  28 #include "wx/txtstrm.h"
  29
  30 IMPLEMENT_DYNAMIC_CLASS(wxRichTextHTMLHandler, wxRichTextFileHandler)
  31
  32 /// Can we handle this filename (if using files)? By default, checks the extension.
  33 bool wxRichTextHTMLHandler::CanHandle(const wxString& filename) const
  34 {
  35     wxString path, file, ext;
  36     wxSplitPath(filename, & path, & file, & ext);
  37
  38     return (ext.Lower() == wxT("html") || ext.Lower() == wxT("htm"));
  39 }
  40
  41
  42 #if wxUSE_STREAMS
  43 bool wxRichTextHTMLHandler::DoLoadFile(wxRichTextBuffer *WXUNUSED(buffer), wxInputStream& WXUNUSED(stream))
  44 {
  45     return false;
  46 }
  47
  48 /*
  49  * We need to output only _changes_ in character formatting.
  50  */
  51
  52 bool wxRichTextHTMLHandler::DoSaveFile(wxRichTextBuffer *buffer, wxOutputStream& stream)
  53 {
  54     buffer->Defragment();
  55
  56     wxTextOutputStream str(stream);
  57
  58     wxTextAttrEx currentParaStyle = buffer->GetAttributes();
  59     wxTextAttrEx currentCharStyle = buffer->GetAttributes();
  60
  61     str << wxT("<html><head></head><body>\n");
  62
  63     /*
  64     wxRichText may be support paper formats like a1/a2/a3/a4
  65     when this widget grown enough, i should turn back and support its new features
  66     but not yet
  67
  68       str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td>");
  69
  70         wxString left_indent = SymbolicIndent(currentParaStyle.GetLeftIndent());
  71         wxString right_indent = SymbolicIndent(currentParaStyle.GetRightIndent());
  72
  73           str << wxString::Format(wxT("%s</td><td></td><td>%s</td></tr><tr>"),
  74           left_indent.c_str(), //Document-Wide Left Indent
  75           right_indent.c_str()); //Document-Wide Right Indent
  76
  77             str << wxT("<td></td><td width=\"100%\">");
  78     */
  79
  80     str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"100%\">");
  81
  82     str << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\" color=\"%s\" >"),
  83         currentParaStyle.GetFont().GetFaceName().c_str(), Pt_To_Size( currentParaStyle.GetFont().GetPointSize() ),
  84         currentParaStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
  85
  86     //wxString align = GetAlignment( currentParaStyle.GetAlignment() );
  87     //str << wxString::Format(wxT("<p align=\"%s\">"), align );
  88
  89     m_font = false;
  90     m_indent = 0;
  91     m_list = false;
  92
  93     wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
  94     while (node)
  95     {
  96         wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
  97         wxASSERT (para != NULL);
  98
  99         if (para)
 100         {
 101             OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream);
 102
 103             wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
 104             while (node2)
 105             {
 106                 wxRichTextObject* obj = node2->GetData();
 107                 wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
 108                 if (textObj && !textObj->IsEmpty())
 109                 {
 110                     BeginCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
 111
 112                     str << textObj->GetText();
 113
 114                     EndCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
 115                 }
 116
 117                 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 118                 if( image && !image->IsEmpty())
 119                     Image_to_Base64( image, stream );
 120
 121                 node2 = node2->GetNext();
 122             }
 123             str << wxT("\n");
 124             //OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream, false);
 125         }
 126         node = node->GetNext();
 127     }
 128
 129     str << wxT("</font></td></tr></table></body></html>\n");
 130
 131     return true;
 132 }
 133
 134 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttrEx& currentStyle, const wxTextAttrEx& thisStyle, wxOutputStream& stream)
 135 {
 136     wxTextOutputStream str(stream);
 137
 138     //Is the item bulleted one?
 139     if( thisStyle.GetBulletStyle() != wxTEXT_ATTR_BULLET_STYLE_NONE )
 140     {
 141         //Is there any opened list?
 142         if( m_list )
 143         {
 144             //Yes there is
 145
 146             //Is the item among the previous ones
 147             //Is the item one of the previous list tag's child items
 148             if( (thisStyle.GetLeftIndent() == (m_indent + 100)) || (thisStyle.GetLeftIndent() < 100) )
 149                 str << wxT("<li>");//Yes it is
 150             else
 151             {
 152                 //No it isn't
 153
 154                 //So we should close the list tag
 155                 str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 156
 157                 //And renavigate to new list's horizontal position
 158                 NavigateToListPosition(thisStyle, str);
 159                 //Ok it's done
 160
 161                 //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 162                 wxString tag;
 163                 TypeOfList(thisStyle, tag);
 164                 str << tag << wxT("<li>");
 165             }
 166         }
 167         else
 168         {
 169             //No there isn't a list
 170
 171             //navigate to new list's horizontal position(indent)
 172             NavigateToListPosition(thisStyle, str);
 173
 174             //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 175             wxString tag;
 176             TypeOfList(thisStyle, tag);
 177             str << tag << wxT("<li>");
 178
 179             //Now we have a list, mark it.
 180             m_list = true;
 181         }
 182     }
 183     else if( m_list )
 184     {
 185         //The item is not bulleted and there is a list what should be closed now.
 186         //So close the list
 187
 188         str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 189         //And mark as there is no an opened list
 190         m_list = false;
 191     }
 192
 193     // does the item have an indentation ?
 194     if( thisStyle.GetLeftIndent() )
 195     {
 196         if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_NONE )
 197         {
 198             if( m_indent )
 199             {
 200                 if( (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) == m_indent )
 201                 {
 202                     if( thisStyle.GetLeftSubIndent() < 0 )
 203                     {
 204                         str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
 205                     }
 206                 }
 207                 else
 208                 {
 209                     if( thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent() > m_indent )
 210                     {
 211                         Indent(thisStyle, str);
 212                         m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 213                         m_indents.Add( m_indent );
 214                     }
 215                     else
 216                     {
 217                         int i = m_indents.size() - 1;
 218                         for(; i > -1; i--)
 219                         {
 220                             if( m_indent < (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
 221                             {
 222                                 Indent(thisStyle, str);
 223                                 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 224                                 m_indents.Add( m_indent );
 225
 226                                 break;
 227                             }
 228                             else if( m_indent == (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
 229                             {
 230                                 if( thisStyle.GetLeftSubIndent() < 0 )
 231                                 {
 232                                     str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
 233                                 }
 234                                 break;
 235                             }
 236                             else
 237                             {
 238                                 str << wxT("</td></tr></table>");
 239
 240                                 m_indents.RemoveAt(i);
 241
 242                                 if( i < 1 ){m_indent=0; break;}
 243                                 m_indent = m_indents[i-1];
 244                             }
 245                         }
 246                     }
 247                 }
 248             }
 249             else
 250             {
 251                 Indent(thisStyle, str);
 252                 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 253                 m_indents.Add( m_indent );
 254             }
 255         }
 256     }
 257     else if( m_indent )
 258     {
 259         //The item is not indented and there is a table(s) what should be closed now.
 260
 261         //So close them
 262         for(unsigned int i = 0; i < m_indents.size(); i++ )
 263             str << wxT("</td></tr></table>");
 264
 265         m_indent = 0;
 266         m_indents.Clear();
 267     }
 268
 269
 270     wxString style;
 271
 272     //Is there any change on the font properties of the item
 273     if( thisStyle.GetFont().GetFaceName() != currentStyle.GetFont().GetFaceName() )
 274         style += wxString::Format(wxT(" face=\"%s\""), thisStyle.GetFont().GetFaceName().c_str());
 275     if( thisStyle.GetFont().GetPointSize() != currentStyle.GetFont().GetPointSize() )
 276         style += wxString::Format(wxT(" size=\"%ld\""), Pt_To_Size(thisStyle.GetFont().GetPointSize()) );
 277     if( thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 278         style += wxString::Format(wxT(" color=\"%s\""), thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
 279
 280     if( style.size() )
 281     {
 282         str << wxString::Format(wxT("<font %s >"), style.c_str());
 283         m_font = true;
 284     }
 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.c_str());
 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.c_str() );
 397     str << wxT("<td width=\"100%\">");
 398
 399     if( thisStyle.GetLeftSubIndent() < 0 )
 400     {
 401         str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
 402     }
 403 }
 404
 405 void wxRichTextHTMLHandler::LIndent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
 406 {
 407     //Code:
 408     //r.BeginNumberedBullet(1, 200, 60);
 409     //r.Newline();
 410     //r.WriteText(wxT("first item"));
 411     //r.EndNumberedBullet();
 412     //r.BeginNumberedBullet(2, 200, 60);
 413     //r.Newline();
 414     //r.WriteText(wxT("second item."));
 415     //r.EndNumberedBullet();
 416     //
 417     //A typical indentation-table for the item will be construct as the following
 418
 419     //1 x nbsp = 20 point
 420     //ULI -> 100pt (UL/OL tag indents its sub element by 100 point)
 421     //<--------- 100 pt ---------->|
 422     //------------------------------------------------------
 423     //|&nbsp;&nbsp;nbsp;&nbsp;nbsp;|<ul>                   |
 424     //|                            |<-ULI-><li>first item  |
 425     //|                            |<-ULI-><li>second item |
 426     //|                            |</ul>                  |
 427     //------------------------------------------------------
 428     //                             |<-100->|
 429
 430
 431     str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
 432
 433     wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() - m_indent) - 100);
 434     str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent.c_str() );
 435     str << wxT("<td width=\"100%\">");
 436 }
 437
 438 void wxRichTextHTMLHandler::TypeOfList( const wxTextAttrEx& thisStyle, wxString& tag )
 439 {
 440     //We can use number attribute of li tag but not all the browsers support it.
 441     //also wxHtmlWindow doesn't support type attribute.
 442
 443     m_is_ul = false;
 444     if( thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 445         tag = wxT("<ol type=\"1\">");
 446     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER )
 447         tag = wxT("<ol type=\"A\">");
 448     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER )
 449         tag = wxT("<ol type=\"a\">");
 450     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER )
 451         tag = wxT("<ol type=\"I\">");
 452     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER )
 453         tag = wxT("<ol type=\"i\">");
 454     else
 455     {
 456         tag = wxT("<ul>");
 457         m_is_ul = true;
 458     }
 459 }
 460
 461 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttrEx& thisStyle )
 462 {
 463     switch( thisStyle.GetAlignment() )
 464     {
 465     case wxTEXT_ALIGNMENT_LEFT:
 466         return  wxT("left");
 467     case wxTEXT_ALIGNMENT_RIGHT:
 468         return wxT("right");
 469     case wxTEXT_ALIGNMENT_CENTER:
 470         return wxT("center");
 471     case wxTEXT_ALIGNMENT_JUSTIFIED:
 472         return wxT("justify");
 473     default:
 474         return wxT("left");
 475     }
 476 }
 477
 478 void wxRichTextHTMLHandler::Image_to_Base64(wxRichTextImage* image, wxOutputStream& stream)
 479 {
 480     wxTextOutputStream str(stream);
 481
 482     str << wxT("<img src=\"");
 483     str << wxT("data:");
 484     str << GetMimeType(image->GetImageBlock().GetImageType());
 485     str << wxT(";base64,");
 486
 487     if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 488         image->MakeBlock();
 489
 490     wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 491     str << data;
 492
 493     delete[] data;
 494
 495     str << wxT("\" />");
 496 }
 497
 498 long wxRichTextHTMLHandler::Pt_To_Size(long size)
 499 {
 500     //return most approximate size
 501     if(size < 9 ) return 1;
 502     else if( size < 11 ) return 2;
 503     else if( size < 14 ) return 3;
 504     else if( size < 18 ) return 4;
 505     else if( size < 23 ) return 5;
 506     else if( size < 30 ) return 6;
 507     else return 7;
 508 }
 509
 510 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 511 {
 512     wxString in;
 513     for(;indent > 0; indent -= 20)
 514         in.Append( wxT("&nbsp;") );
 515     return in;
 516 }
 517
 518 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 519 {
 520     switch(imageType)
 521     {
 522     case wxBITMAP_TYPE_BMP:
 523         return wxT("image/bmp");
 524     case wxBITMAP_TYPE_TIF:
 525         return wxT("image/tiff");
 526     case wxBITMAP_TYPE_GIF:
 527         return wxT("image/gif");
 528     case wxBITMAP_TYPE_PNG:
 529         return wxT("image/png");
 530     case wxBITMAP_TYPE_JPEG:
 531         return wxT("image/jpeg");
 532     default:
 533         return wxT("image/unknown");
 534     }
 535 }
 536
 537 //exim-style base64 encoder
 538 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 539 {
 540     //elements of enc64 array must be 8 bit values
 541     //otherwise encoder will fail
 542     //hmmm.. Does wxT macro define a char as 16 bit value
 543     //when compiling with UNICODE option?
 544     static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 545     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 546     wxChar* p = output;
 547
 548     while( in_len-- > 0 )
 549     {
 550         register wxChar a, b;
 551
 552         a = *input++;
 553
 554         *p++ = enc64[ (a >> 2) & 0x3f ];
 555
 556         if( in_len-- <= 0 )
 557         {
 558             *p++ = enc64[ (a << 4 ) & 0x30 ];
 559             *p++ = '=';
 560             *p++ = '=';
 561             break;
 562         }
 563
 564         b = *input++;
 565
 566         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 567
 568         if( in_len-- <= 0 )
 569         {
 570             *p++ = enc64[ (b << 2) & 0x3f ];
 571             *p++ = '=';
 572             break;
 573         }
 574
 575         a = *input++;
 576
 577         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 578
 579         *p++ = enc64[ a & 0x3f ];
 580     }
 581     *p = 0;
 582
 583     return output;
 584 }
 585 #endif
 586 // wxUSE_STREAMS
 587
 588 #endif
 589 // wxUSE_RICHTEXT