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     wxSplitPath(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     wxTextAttrEx currentParaStyle = buffer->GetAttributes();
  83     wxTextAttrEx currentCharStyle = buffer->GetAttributes();
  84
  85     if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
  86         str << wxT("<html><head></head><body>\n");
  87
  88     str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"100%\">");
  89
  90     OutputFont(currentParaStyle, str);
  91
  92     m_font = false;
  93     m_inTable = false;
  94
  95     m_indents.Clear();
  96     m_listTypes.Clear();
  97
  98     wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
  99     while (node)
 100     {
 101         wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
 102         wxASSERT (para != NULL);
 103
 104         if (para)
 105         {
 106             wxTextAttrEx paraStyle(para->GetCombinedAttributes());
 107
 108             BeginParagraphFormatting(currentParaStyle, paraStyle, str);
 109
 110             wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
 111             while (node2)
 112             {
 113                 wxRichTextObject* obj = node2->GetData();
 114                 wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
 115                 if (textObj && !textObj->IsEmpty())
 116                 {
 117                     wxTextAttrEx charStyle(para->GetCombinedAttributes(obj->GetAttributes()));
 118                     BeginCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 119
 120                     wxString text = textObj->GetText();
 121
 122                     if (charStyle.HasTextEffects() && (charStyle.GetTextEffects() & wxTEXT_ATTR_EFFECT_CAPITALS))
 123                         text.MakeUpper();
 124
 125                     wxString toReplace = wxRichTextLineBreakChar;
 126                     text.Replace(toReplace, wxT("<br>"));
 127
 128                     str << text;
 129
 130                     EndCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 131                 }
 132
 133                 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 134                 if( image && !image->IsEmpty())
 135                     WriteImage( image, stream );
 136
 137                 node2 = node2->GetNext();
 138             }
 139
 140             EndParagraphFormatting(currentParaStyle, paraStyle, str);
 141
 142             str << wxT("\n");
 143         }
 144         node = node->GetNext();
 145     }
 146
 147     CloseLists(-1, str);
 148
 149     str << wxT("</font>");
 150
 151     str << wxT("</td></tr></table><p>");
 152
 153     if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
 154         str << wxT("</body></html>");
 155
 156     str << wxT("\n");
 157
 158     m_buffer = NULL;
 159
 160     return true;
 161 }
 162
 163 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttrEx& currentStyle, const wxTextAttrEx& thisStyle, const wxTextAttrEx& WXUNUSED(paraStyle), wxTextOutputStream& str)
 164 {
 165     wxString style;
 166
 167     // Is there any change in the font properties of the item?
 168     if (thisStyle.GetFont().GetFaceName() != currentStyle.GetFont().GetFaceName())
 169     {
 170         wxString faceName(thisStyle.GetFont().GetFaceName());
 171         style += wxString::Format(wxT(" face=\"%s\""), faceName.c_str());
 172     }
 173     if (thisStyle.GetFont().GetPointSize() != currentStyle.GetFont().GetPointSize())
 174         style += wxString::Format(wxT(" size=\"%ld\""), PtToSize(thisStyle.GetFont().GetPointSize()));
 175     if (thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 176     {
 177         wxString color(thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX));
 178         style += wxString::Format(wxT(" color=\"%s\""), color.c_str());
 179     }
 180
 181     if (style.size())
 182     {
 183         str << wxString::Format(wxT("<font %s >"), style.c_str());
 184         m_font = true;
 185     }
 186
 187     if (thisStyle.GetFont().GetWeight() == wxBOLD)
 188         str << wxT("<b>");
 189     if (thisStyle.GetFont().GetStyle() == wxITALIC)
 190         str << wxT("<i>");
 191     if (thisStyle.GetFont().GetUnderlined())
 192         str << wxT("<u>");
 193
 194     if (thisStyle.HasURL())
 195         str << wxT("<a href=\"") << thisStyle.GetURL() << wxT("\">");
 196 }
 197
 198 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, const wxTextAttrEx& WXUNUSED(paraStyle), wxTextOutputStream& stream)
 199 {
 200     if (thisStyle.HasURL())
 201         stream << wxT("</a>");
 202
 203     if (thisStyle.GetFont().GetUnderlined())
 204         stream << wxT("</u>");
 205     if (thisStyle.GetFont().GetStyle() == wxITALIC)
 206         stream << wxT("</i>");
 207     if (thisStyle.GetFont().GetWeight() == wxBOLD)
 208         stream << wxT("</b>");
 209
 210     if (m_font)
 211     {
 212         m_font = false;
 213         stream << wxT("</font>");
 214     }
 215 }
 216
 217 /// Begin paragraph formatting
 218 void wxRichTextHTMLHandler::BeginParagraphFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxTextOutputStream& str)
 219 {
 220     if (thisStyle.HasPageBreak())
 221     {
 222         str << wxT("</tr></td></table>");
 223         str << wxT("<div style=\"page-break-after:always\"></div>\n");
 224         str << wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"100%\">");
 225     }
 226
 227     if (thisStyle.HasLeftIndent() && thisStyle.GetLeftIndent() != 0)
 228     {
 229         if (thisStyle.HasBulletStyle())
 230         {
 231             int indent = thisStyle.GetLeftIndent();
 232
 233             // Close levels high than this
 234             CloseLists(indent, str);
 235
 236             if (m_indents.GetCount() > 0 && indent == m_indents.Last())
 237             {
 238                 // Same level, no need to start a new list
 239             }
 240             else if (m_indents.GetCount() == 0 || indent > m_indents.Last())
 241             {
 242                 m_indents.Add(indent);
 243
 244                 wxString tag;
 245                 int listType = TypeOfList(thisStyle, tag);
 246                 m_listTypes.Add(listType);
 247
 248                 wxString align = GetAlignment(thisStyle);
 249                 str << wxString::Format(wxT("<p align=\"%s\">"), align.c_str());
 250
 251                 str << tag;
 252             }
 253
 254             str << wxT("<li> ");
 255         }
 256         else
 257         {
 258             CloseLists(-1, str);
 259
 260             wxString align = GetAlignment(thisStyle);
 261             str << wxString::Format(wxT("<p align=\"%s\">"), align.c_str());
 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 }
 287
 288 /// End paragraph formatting
 289 void wxRichTextHTMLHandler::EndParagraphFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxTextOutputStream& stream)
 290 {
 291     if (m_inTable)
 292     {
 293         if (thisStyle.HasFont())
 294             stream << wxT("</font>");
 295
 296         stream << wxT("</td></tr></table>\n");
 297         m_inTable = false;
 298     }
 299 }
 300
 301 /// Closes lists to level (-1 means close all)
 302 void wxRichTextHTMLHandler::CloseLists(int level, wxTextOutputStream& str)
 303 {
 304     // Close levels high than this
 305     int i = m_indents.GetCount()-1;
 306     while (i >= 0)
 307     {
 308         int l = m_indents[i];
 309         if (l > level)
 310         {
 311             if (m_listTypes[i] == 0)
 312                 str << wxT("</ol>");
 313             else
 314                 str << wxT("</ul>");
 315             m_indents.RemoveAt(i);
 316             m_listTypes.RemoveAt(i);
 317         }
 318         else
 319             break;
 320         i --;
 321      }
 322 }
 323
 324 /// Output font tag
 325 void wxRichTextHTMLHandler::OutputFont(const wxTextAttrEx& style, wxTextOutputStream& stream)
 326 {
 327     if (style.HasFont())
 328     {
 329         stream << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\" color=\"%s\" >"),
 330                 style.GetFont().GetFaceName().c_str(), PtToSize(style.GetFont().GetPointSize()),
 331                 style.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
 332     }
 333 }
 334
 335 int wxRichTextHTMLHandler::TypeOfList( const wxTextAttrEx& thisStyle, wxString& tag )
 336 {
 337     // We can use number attribute of li tag but not all the browsers support it.
 338     // also wxHtmlWindow doesn't support type attribute.
 339
 340     bool m_is_ul = false;
 341     if (thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 342         tag = wxT("<ol type=\"1\">");
 343     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER)
 344         tag = wxT("<ol type=\"A\">");
 345     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER)
 346         tag = wxT("<ol type=\"a\">");
 347     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER)
 348         tag = wxT("<ol type=\"I\">");
 349     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER)
 350         tag = wxT("<ol type=\"i\">");
 351     else
 352     {
 353         tag = wxT("<ul>");
 354         m_is_ul = true;
 355     }
 356
 357     if (m_is_ul)
 358         return 1;
 359     else
 360         return 0;
 361 }
 362
 363 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttrEx& thisStyle )
 364 {
 365     switch( thisStyle.GetAlignment() )
 366     {
 367     case wxTEXT_ALIGNMENT_LEFT:
 368         return  wxT("left");
 369     case wxTEXT_ALIGNMENT_RIGHT:
 370         return wxT("right");
 371     case wxTEXT_ALIGNMENT_CENTER:
 372         return wxT("center");
 373     case wxTEXT_ALIGNMENT_JUSTIFIED:
 374         return wxT("justify");
 375     default:
 376         return wxT("left");
 377     }
 378 }
 379
 380 void wxRichTextHTMLHandler::WriteImage(wxRichTextImage* image, wxOutputStream& stream)
 381 {
 382     wxTextOutputStream str(stream);
 383
 384     str << wxT("<img src=\"");
 385
 386 #if wxUSE_FILESYSTEM
 387     if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 388     {
 389         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 390             image->LoadFromBlock();
 391         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 392             image->MakeBlock();
 393
 394         if (image->GetImage().Ok())
 395         {
 396             wxString ext(image->GetImageBlock().GetExtension());
 397             wxString tempFilename(wxString::Format(wxT("image%d.%s"), sm_fileCounter, ext));
 398             wxMemoryFSHandler::AddFile(tempFilename, image->GetImage(), image->GetImageBlock().GetImageType());
 399
 400             m_imageLocations.Add(tempFilename);
 401
 402             str << wxT("memory:") << tempFilename;
 403         }
 404         else
 405             str << wxT("memory:?");
 406
 407         sm_fileCounter ++;
 408     }
 409     else if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 410     {
 411         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 412             image->LoadFromBlock();
 413         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 414             image->MakeBlock();
 415
 416         if (image->GetImage().Ok())
 417         {
 418             wxString tempDir(GetTempDir());
 419             if (tempDir.IsEmpty())
 420                 tempDir = wxFileName::GetTempDir();
 421
 422             wxString ext(image->GetImageBlock().GetExtension());
 423             wxString tempFilename(wxString::Format(wxT("%s/image%d.%s"), tempDir, sm_fileCounter, ext));
 424             image->GetImageBlock().Write(tempFilename);
 425
 426             m_imageLocations.Add(tempFilename);
 427
 428             str << wxFileSystem::FileNameToURL(tempFilename);
 429         }
 430         else
 431             str << wxT("file:?");
 432
 433         sm_fileCounter ++;
 434     }
 435     else // if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_BASE64) // this is implied
 436 #endif
 437     {
 438         str << wxT("data:");
 439         str << GetMimeType(image->GetImageBlock().GetImageType());
 440         str << wxT(";base64,");
 441
 442         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 443             image->MakeBlock();
 444
 445         wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 446         str << data;
 447
 448         delete[] data;
 449     }
 450
 451     str << wxT("\" />");
 452 }
 453
 454 long wxRichTextHTMLHandler::PtToSize(long size)
 455 {
 456     int i;
 457     int len = m_fontSizeMapping.GetCount();
 458     for (i = 0; i < len; i++)
 459         if (size <= m_fontSizeMapping[i])
 460             return i+1;
 461     return 7;
 462 }
 463
 464 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 465 {
 466     wxString in;
 467     for(;indent > 0; indent -= 20)
 468         in.Append( wxT("&nbsp;") );
 469     return in;
 470 }
 471
 472 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 473 {
 474     switch(imageType)
 475     {
 476     case wxBITMAP_TYPE_BMP:
 477         return wxT("image/bmp");
 478     case wxBITMAP_TYPE_TIF:
 479         return wxT("image/tiff");
 480     case wxBITMAP_TYPE_GIF:
 481         return wxT("image/gif");
 482     case wxBITMAP_TYPE_PNG:
 483         return wxT("image/png");
 484     case wxBITMAP_TYPE_JPEG:
 485         return wxT("image/jpeg");
 486     default:
 487         return wxT("image/unknown");
 488     }
 489 }
 490
 491 // exim-style base64 encoder
 492 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 493 {
 494     // elements of enc64 array must be 8 bit values
 495     // otherwise encoder will fail
 496     // hmmm.. Does wxT macro define a char as 16 bit value
 497     // when compiling with UNICODE option?
 498     static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 499     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 500     wxChar* p = output;
 501
 502     while( in_len-- > 0 )
 503     {
 504         register wxChar a, b;
 505
 506         a = *input++;
 507
 508         *p++ = enc64[ (a >> 2) & 0x3f ];
 509
 510         if( in_len-- == 0 )
 511         {
 512             *p++ = enc64[ (a << 4 ) & 0x30 ];
 513             *p++ = '=';
 514             *p++ = '=';
 515             break;
 516         }
 517
 518         b = *input++;
 519
 520         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 521
 522         if( in_len-- == 0 )
 523         {
 524             *p++ = enc64[ (b << 2) & 0x3f ];
 525             *p++ = '=';
 526             break;
 527         }
 528
 529         a = *input++;
 530
 531         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 532
 533         *p++ = enc64[ a & 0x3f ];
 534     }
 535     *p = 0;
 536
 537     return output;
 538 }
 539 #endif
 540 // wxUSE_STREAMS
 541
 542 /// Delete the in-memory or temporary files generated by the last operation
 543 bool wxRichTextHTMLHandler::DeleteTemporaryImages()
 544 {
 545     return DeleteTemporaryImages(GetFlags(), m_imageLocations);
 546 }
 547
 548 /// Delete the in-memory or temporary files generated by the last operation
 549 bool wxRichTextHTMLHandler::DeleteTemporaryImages(int flags, const wxArrayString& imageLocations)
 550 {
 551     size_t i;
 552     for (i = 0; i < imageLocations.GetCount(); i++)
 553     {
 554         wxString location = imageLocations[i];
 555
 556         if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 557         {
 558 #if wxUSE_FILESYSTEM
 559             wxMemoryFSHandler::RemoveFile(location);
 560 #endif
 561         }
 562         else if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 563         {
 564             if (wxFileExists(location))
 565                 wxRemoveFile(location);
 566         }
 567     }
 568
 569     return true;
 570 }
 571
 572
 573 #endif
 574 // wxUSE_RICHTEXT
 575