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             wxTextAttrEx paraStyle(para->GetCombinedAttributes());
 102
 103             OutputParagraphFormatting(currentParaStyle, paraStyle, 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                     wxTextAttrEx charStyle(para->GetCombinedAttributes(obj->GetAttributes()));
 113                     BeginCharacterFormatting(currentCharStyle, charStyle, paraStyle, stream);
 114
 115                     str << textObj->GetText();
 116
 117                     EndCharacterFormatting(currentCharStyle, charStyle, paraStyle, stream);
 118                 }
 119
 120                 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 121                 if( image && !image->IsEmpty())
 122                     Image_to_Base64( image, stream );
 123
 124                 node2 = node2->GetNext();
 125             }
 126             str << wxT("\n");
 127             //OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream, false);
 128         }
 129         node = node->GetNext();
 130     }
 131
 132     str << wxT("</font></td></tr></table></body></html>\n");
 133
 134     return true;
 135 }
 136
 137 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttrEx& currentStyle, const wxTextAttrEx& thisStyle, const wxTextAttrEx& paraStyle, wxOutputStream& stream)
 138 {
 139     wxTextOutputStream str(stream);
 140
 141     //Is the item bulleted one?
 142     if( paraStyle.GetBulletStyle() != wxTEXT_ATTR_BULLET_STYLE_NONE )
 143     {
 144         //Is there any opened list?
 145         if( m_list )
 146         {
 147             //Yes there is
 148
 149             //Is the item among the previous ones
 150             //Is the item one of the previous list tag's child items
 151             if( (paraStyle.GetLeftIndent() == (m_indent + 100)) || (paraStyle.GetLeftIndent() < 100) )
 152                 str << wxT("<li>");//Yes it is
 153             else
 154             {
 155                 //No it isn't
 156
 157                 //So we should close the list tag
 158                 str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 159
 160                 //And renavigate to new list's horizontal position
 161                 NavigateToListPosition(paraStyle, str);
 162                 //Ok it's done
 163
 164                 //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 165                 wxString tag;
 166                 TypeOfList(paraStyle, tag);
 167                 str << tag << wxT("<li>");
 168             }
 169         }
 170         else
 171         {
 172             //No there isn't a list
 173
 174             //navigate to new list's horizontal position(indent)
 175             NavigateToListPosition(paraStyle, str);
 176
 177             //Get the appropriate tag, an ol for numerical values, an ul for dot, square etc.
 178             wxString tag;
 179             TypeOfList(paraStyle, tag);
 180             str << tag << wxT("<li>");
 181
 182             //Now we have a list, mark it.
 183             m_list = true;
 184         }
 185     }
 186     else if( m_list )
 187     {
 188         //The item is not bulleted and there is a list what should be closed now.
 189         //So close the list
 190
 191         str << (m_is_ul ? wxT("</ul>") : wxT("</ol>"));
 192         //And mark as there is no an opened list
 193         m_list = false;
 194     }
 195
 196     // does the item have an indentation ?
 197     if( paraStyle.GetLeftIndent() )
 198     {
 199         if( paraStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_NONE )
 200         {
 201             if( m_indent )
 202             {
 203                 if( (paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent()) == m_indent )
 204                 {
 205                     if( paraStyle.GetLeftSubIndent() < 0 )
 206                     {
 207                         str << SymbolicIndent(~paraStyle.GetLeftSubIndent());
 208                     }
 209                 }
 210                 else
 211                 {
 212                     if( paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent() > m_indent )
 213                     {
 214                         Indent(paraStyle, str);
 215                         m_indent = paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent();
 216                         m_indents.Add( m_indent );
 217                     }
 218                     else
 219                     {
 220                         int i = m_indents.size() - 1;
 221                         for(; i > -1; i--)
 222                         {
 223                             if( m_indent < (paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent()) )
 224                             {
 225                                 Indent(paraStyle, str);
 226                                 m_indent = paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent();
 227                                 m_indents.Add( m_indent );
 228
 229                                 break;
 230                             }
 231                             else if( m_indent == (paraStyle.GetLeftIndent() + paraStyle.GetLeftSubIndent()) )
 232                             {
 233                                 if( paraStyle.GetLeftSubIndent() < 0 )
 234                                 {
 235                                     str << SymbolicIndent(~paraStyle.GetLeftSubIndent());
 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(paraStyle, str);
 255                 m_indent = paraStyle.GetLeftIndent() + paraStyle.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().c_str());
 278     if( thisStyle.GetFont().GetPointSize() != currentStyle.GetFont().GetPointSize() )
 279         style += wxString::Format(wxT(" size=\"%ld\""), Pt_To_Size(thisStyle.GetFont().GetPointSize()) );
 280     if( thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 281         style += wxString::Format(wxT(" color=\"%s\""), thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
 282
 283     if( style.size() )
 284     {
 285         str << wxString::Format(wxT("<font %s >"), style.c_str());
 286         m_font = true;
 287     }
 288
 289     if( thisStyle.GetFont().GetWeight() == wxBOLD )
 290         str << wxT("<b>");
 291     if( thisStyle.GetFont().GetStyle() == wxITALIC )
 292         str << wxT("<i>");
 293     if( thisStyle.GetFont().GetUnderlined() )
 294         str << wxT("<u>");
 295 }
 296
 297 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, const wxTextAttrEx& WXUNUSED(paraStyle), wxOutputStream& stream)
 298 {
 299     wxTextOutputStream str(stream);
 300
 301     if( thisStyle.GetFont().GetUnderlined() )
 302         str << wxT("</u>");
 303     if( thisStyle.GetFont().GetStyle() == wxITALIC )
 304         str << wxT("</i>");
 305     if( thisStyle.GetFont().GetWeight() == wxBOLD )
 306         str << wxT("</b>");
 307
 308     if( m_font )
 309     {
 310         m_font = false;
 311         str << wxT("</font>");
 312     }
 313 }
 314
 315 /// Output paragraph formatting
 316 void wxRichTextHTMLHandler::OutputParagraphFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxOutputStream& stream)
 317 {
 318     //If there is no opened list currently, insert a <p> after every paragraph
 319     if(!m_list)
 320     {
 321         wxTextOutputStream str(stream);
 322         wxString align = GetAlignment( thisStyle );
 323         str << wxString::Format(wxT("<p align=\"%s\">"), align.c_str());
 324     }
 325 }
 326
 327 void wxRichTextHTMLHandler::NavigateToListPosition(const wxTextAttrEx& thisStyle, wxTextOutputStream& str)
 328 {
 329     //indenting an item using an ul/ol tag is equal to inserting 5 x &nbsp; on its left side.
 330     //so we should start from 100 point left
 331
 332     //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
 333     //of the item, horizontally?
 334     if( m_indent + 100 < thisStyle.GetLeftIndent() )
 335     {
 336         //yes it is
 337         LIndent(thisStyle, str);
 338         m_indent = thisStyle.GetLeftIndent() - 100;
 339         m_indents.Add( m_indent );
 340         return;
 341     }
 342     //No it isn't
 343
 344     int i = m_indents.size() - 1;
 345     for(; i > -1; i--)
 346     {
 347         //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
 348         //of the item ?
 349         if( m_indent + 100 < thisStyle.GetLeftIndent() )
 350         {
 351             //Yes it is
 352             LIndent(thisStyle, str);
 353             m_indent = thisStyle.GetLeftIndent() - 100;
 354             m_indents.Add( m_indent );
 355             break;
 356         }
 357         else if( m_indent + 100 == thisStyle.GetLeftIndent() )
 358             break;//exact match
 359         else
 360         {
 361             //No it is not, the second td's left wall of the current indentaion table is at the
 362             //right side of the current item horizontally, so close it.
 363             str << wxT("</td></tr></table>");
 364
 365             m_indents.RemoveAt(i);
 366
 367             if( i < 1 ){m_indent=0; break;}
 368             m_indent = m_indents[i-1];
 369         }
 370     }
 371 }
 372 void wxRichTextHTMLHandler::Indent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
 373 {
 374     //As a five year experienced web developer i assure you there is no way to indent an item
 375     //in html way, but we can use tables.
 376
 377
 378
 379     //Item -> "Hello world"
 380     //Its Left Indentation -> 100
 381     //Its Left Sub-Indentation ->40
 382     //A typical indentation-table for the item will be construct as the following
 383
 384     //3 x nbsp = 60
 385     //2 x nbsp = 40
 386     //LSI = Left Sub Indent
 387     //LI = Left Indent - LSI
 388     //
 389     //-------------------------------------------
 390     //|&nbsp;&nbsp;nbsp;|nbsp;nbsp;Hello World  |
 391     //|      |          |    |                  |
 392     //|        V        |    V                  |
 393     //|      --LI--     | --LSI--               |
 394     //-------------------------------------------
 395
 396     str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
 397
 398     wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) - m_indent );
 399     str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent.c_str() );
 400     str << wxT("<td width=\"100%\">");
 401
 402     if( thisStyle.GetLeftSubIndent() < 0 )
 403     {
 404         str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
 405     }
 406 }
 407
 408 void wxRichTextHTMLHandler::LIndent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
 409 {
 410     //Code:
 411     //r.BeginNumberedBullet(1, 200, 60);
 412     //r.Newline();
 413     //r.WriteText(wxT("first item"));
 414     //r.EndNumberedBullet();
 415     //r.BeginNumberedBullet(2, 200, 60);
 416     //r.Newline();
 417     //r.WriteText(wxT("second item."));
 418     //r.EndNumberedBullet();
 419     //
 420     //A typical indentation-table for the item will be construct as the following
 421
 422     //1 x nbsp = 20 point
 423     //ULI -> 100pt (UL/OL tag indents its sub element by 100 point)
 424     //<--------- 100 pt ---------->|
 425     //------------------------------------------------------
 426     //|&nbsp;&nbsp;nbsp;&nbsp;nbsp;|<ul>                   |
 427     //|                            |<-ULI-><li>first item  |
 428     //|                            |<-ULI-><li>second item |
 429     //|                            |</ul>                  |
 430     //------------------------------------------------------
 431     //                             |<-100->|
 432
 433
 434     str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
 435
 436     wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() - m_indent) - 100);
 437     str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent.c_str() );
 438     str << wxT("<td width=\"100%\">");
 439 }
 440
 441 void wxRichTextHTMLHandler::TypeOfList( const wxTextAttrEx& thisStyle, wxString& tag )
 442 {
 443     //We can use number attribute of li tag but not all the browsers support it.
 444     //also wxHtmlWindow doesn't support type attribute.
 445
 446     m_is_ul = false;
 447     if( thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 448         tag = wxT("<ol type=\"1\">");
 449     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER )
 450         tag = wxT("<ol type=\"A\">");
 451     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER )
 452         tag = wxT("<ol type=\"a\">");
 453     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER )
 454         tag = wxT("<ol type=\"I\">");
 455     else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER )
 456         tag = wxT("<ol type=\"i\">");
 457     else
 458     {
 459         tag = wxT("<ul>");
 460         m_is_ul = true;
 461     }
 462 }
 463
 464 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttrEx& thisStyle )
 465 {
 466     switch( thisStyle.GetAlignment() )
 467     {
 468     case wxTEXT_ALIGNMENT_LEFT:
 469         return  wxT("left");
 470     case wxTEXT_ALIGNMENT_RIGHT:
 471         return wxT("right");
 472     case wxTEXT_ALIGNMENT_CENTER:
 473         return wxT("center");
 474     case wxTEXT_ALIGNMENT_JUSTIFIED:
 475         return wxT("justify");
 476     default:
 477         return wxT("left");
 478     }
 479 }
 480
 481 void wxRichTextHTMLHandler::Image_to_Base64(wxRichTextImage* image, wxOutputStream& stream)
 482 {
 483     wxTextOutputStream str(stream);
 484
 485     str << wxT("<img src=\"");
 486     str << wxT("data:");
 487     str << GetMimeType(image->GetImageBlock().GetImageType());
 488     str << wxT(";base64,");
 489
 490     if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 491         image->MakeBlock();
 492
 493     wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 494     str << data;
 495
 496     delete[] data;
 497
 498     str << wxT("\" />");
 499 }
 500
 501 long wxRichTextHTMLHandler::Pt_To_Size(long size)
 502 {
 503     //return most approximate size
 504     if(size < 9 ) return 1;
 505     else if( size < 11 ) return 2;
 506     else if( size < 14 ) return 3;
 507     else if( size < 18 ) return 4;
 508     else if( size < 23 ) return 5;
 509     else if( size < 30 ) return 6;
 510     else return 7;
 511 }
 512
 513 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 514 {
 515     wxString in;
 516     for(;indent > 0; indent -= 20)
 517         in.Append( wxT("&nbsp;") );
 518     return in;
 519 }
 520
 521 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 522 {
 523     switch(imageType)
 524     {
 525     case wxBITMAP_TYPE_BMP:
 526         return wxT("image/bmp");
 527     case wxBITMAP_TYPE_TIF:
 528         return wxT("image/tiff");
 529     case wxBITMAP_TYPE_GIF:
 530         return wxT("image/gif");
 531     case wxBITMAP_TYPE_PNG:
 532         return wxT("image/png");
 533     case wxBITMAP_TYPE_JPEG:
 534         return wxT("image/jpeg");
 535     default:
 536         return wxT("image/unknown");
 537     }
 538 }
 539
 540 //exim-style base64 encoder
 541 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 542 {
 543     //elements of enc64 array must be 8 bit values
 544     //otherwise encoder will fail
 545     //hmmm.. Does wxT macro define a char as 16 bit value
 546     //when compiling with UNICODE option?
 547     static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 548     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 549     wxChar* p = output;
 550
 551     while( in_len-- > 0 )
 552     {
 553         register wxChar a, b;
 554
 555         a = *input++;
 556
 557         *p++ = enc64[ (a >> 2) & 0x3f ];
 558
 559         if( in_len-- <= 0 )
 560         {
 561             *p++ = enc64[ (a << 4 ) & 0x30 ];
 562             *p++ = '=';
 563             *p++ = '=';
 564             break;
 565         }
 566
 567         b = *input++;
 568
 569         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 570
 571         if( in_len-- <= 0 )
 572         {
 573             *p++ = enc64[ (b << 2) & 0x3f ];
 574             *p++ = '=';
 575             break;
 576         }
 577
 578         a = *input++;
 579
 580         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 581
 582         *p++ = enc64[ a & 0x3f ];
 583     }
 584     *p = 0;
 585
 586     return output;
 587 }
 588 #endif
 589 // wxUSE_STREAMS
 590
 591 #endif
 592 // wxUSE_RICHTEXT