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 #if wxUSE_UNICODE
  81     wxCSConv* customEncoding = NULL;
  82     wxMBConv* conv = NULL;
  83     if (!GetEncoding().IsEmpty())
  84     {
  85         customEncoding = new wxCSConv(GetEncoding());
  86         if (!customEncoding->IsOk())
  87         {
  88             delete customEncoding;
  89             customEncoding = NULL;
  90         }
  91     }
  92     if (customEncoding)
  93         conv = customEncoding;
  94     else
  95         conv = & wxConvUTF8;
  96 #endif
  97
  98     {
  99 #if wxUSE_UNICODE
 100         wxTextOutputStream str(stream, wxEOL_NATIVE, *conv);
 101 #else
 102         wxTextOutputStream str(stream, wxEOL_NATIVE);
 103 #endif
 104
 105         wxTextAttr currentParaStyle = buffer->GetAttributes();
 106         wxTextAttr currentCharStyle = buffer->GetAttributes();
 107
 108         if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
 109             str << wxT("<html><head></head><body>\n");
 110
 111         OutputFont(currentParaStyle, str);
 112
 113         m_font = false;
 114         m_inTable = false;
 115
 116         m_indents.Clear();
 117         m_listTypes.Clear();
 118
 119         wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
 120         while (node)
 121         {
 122             wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
 123             wxASSERT (para != NULL);
 124
 125             if (para)
 126             {
 127                 wxTextAttr paraStyle(para->GetCombinedAttributes());
 128
 129                 BeginParagraphFormatting(currentParaStyle, paraStyle, str);
 130
 131                 wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
 132                 while (node2)
 133                 {
 134                     wxRichTextObject* obj = node2->GetData();
 135                     wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
 136                     if (textObj && !textObj->IsEmpty())
 137                     {
 138                         wxTextAttr charStyle(para->GetCombinedAttributes(obj->GetAttributes()));
 139                         BeginCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 140
 141                         wxString text = textObj->GetText();
 142
 143                         if (charStyle.HasTextEffects() && (charStyle.GetTextEffects() & wxTEXT_ATTR_EFFECT_CAPITALS))
 144                             text.MakeUpper();
 145
 146                         wxString toReplace = wxRichTextLineBreakChar;
 147                         text.Replace(toReplace, wxT("<br>"));
 148
 149                         str << text;
 150
 151                         EndCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
 152                     }
 153
 154                     wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
 155                     if( image && (!image->IsEmpty() || image->GetImageBlock().GetData()))
 156                         WriteImage( image, stream );
 157
 158                     node2 = node2->GetNext();
 159                 }
 160
 161                 EndParagraphFormatting(currentParaStyle, paraStyle, str);
 162
 163                 str << wxT("\n");
 164             }
 165             node = node->GetNext();
 166         }
 167
 168         CloseLists(-1, str);
 169
 170         str << wxT("</font>");
 171
 172         if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
 173             str << wxT("</body></html>");
 174
 175         str << wxT("\n");
 176     }
 177
 178 #if wxUSE_UNICODE
 179     if (customEncoding)
 180         delete customEncoding;
 181 #endif
 182
 183     m_buffer = NULL;
 184
 185     return true;
 186 }
 187
 188 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttr& currentStyle, const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& str)
 189 {
 190     wxString style;
 191
 192     // Is there any change in the font properties of the item?
 193     if (thisStyle.GetFontFaceName() != currentStyle.GetFontFaceName())
 194     {
 195         wxString faceName(thisStyle.GetFontFaceName());
 196         style += wxString::Format(wxT(" face=\"%s\""), faceName.c_str());
 197     }
 198     if (thisStyle.GetFontSize() != currentStyle.GetFontSize())
 199         style += wxString::Format(wxT(" size=\"%ld\""), PtToSize(thisStyle.GetFontSize()));
 200     if (thisStyle.GetTextColour() != currentStyle.GetTextColour() )
 201     {
 202         wxString color(thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX));
 203         style += wxString::Format(wxT(" color=\"%s\""), color.c_str());
 204     }
 205
 206     if (style.size())
 207     {
 208         str << wxString::Format(wxT("<font %s >"), style.c_str());
 209         m_font = true;
 210     }
 211
 212     if (thisStyle.GetFontWeight() == wxBOLD)
 213         str << wxT("<b>");
 214     if (thisStyle.GetFontStyle() == wxITALIC)
 215         str << wxT("<i>");
 216     if (thisStyle.GetFontUnderlined())
 217         str << wxT("<u>");
 218
 219     if (thisStyle.HasURL())
 220         str << wxT("<a href=\"") << thisStyle.GetURL() << wxT("\">");
 221 }
 222
 223 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& stream)
 224 {
 225     if (thisStyle.HasURL())
 226         stream << wxT("</a>");
 227
 228     if (thisStyle.GetFontUnderlined())
 229         stream << wxT("</u>");
 230     if (thisStyle.GetFontStyle() == wxITALIC)
 231         stream << wxT("</i>");
 232     if (thisStyle.GetFontWeight() == wxBOLD)
 233         stream << wxT("</b>");
 234
 235     if (m_font)
 236     {
 237         m_font = false;
 238         stream << wxT("</font>");
 239     }
 240 }
 241
 242 /// Begin paragraph formatting
 243 void wxRichTextHTMLHandler::BeginParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& str)
 244 {
 245     if (thisStyle.HasPageBreak())
 246     {
 247         str << wxT("<div style=\"page-break-after:always\"></div>\n");
 248     }
 249
 250     if (thisStyle.HasLeftIndent() && thisStyle.GetLeftIndent() != 0)
 251     {
 252         if (thisStyle.HasBulletStyle())
 253         {
 254             int indent = thisStyle.GetLeftIndent();
 255
 256             // Close levels high than this
 257             CloseLists(indent, str);
 258
 259             if (m_indents.GetCount() > 0 && indent == m_indents.Last())
 260             {
 261                 // Same level, no need to start a new list
 262             }
 263             else if (m_indents.GetCount() == 0 || indent > m_indents.Last())
 264             {
 265                 m_indents.Add(indent);
 266
 267                 wxString tag;
 268                 int listType = TypeOfList(thisStyle, tag);
 269                 m_listTypes.Add(listType);
 270
 271                 // wxHTML needs an extra <p> before a list when using <p> ... </p> in previous paragraphs.
 272                 // TODO: pass a flag that indicates we're using wxHTML.
 273                 str << wxT("<p>\n");
 274
 275                 str << tag;
 276             }
 277
 278             str << wxT("<li> ");
 279         }
 280         else
 281         {
 282             CloseLists(-1, str);
 283
 284             wxString align = GetAlignment(thisStyle);
 285             str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
 286
 287             if (thisStyle.HasParagraphSpacingAfter() && thisStyle.GetParagraphSpacingAfter() == 0)
 288                 str << wxT(" style=\"margin: 0px;\"");
 289
 290             str << wxT(">");
 291
 292             // Use a table
 293             int indentTenthsMM = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
 294             // TODO: convert to pixels
 295             int indentPixels = indentTenthsMM/4;
 296             str << wxString::Format(wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"%d\"></td><td>"), indentPixels);
 297
 298             OutputFont(thisStyle, str);
 299
 300             if (thisStyle.GetLeftSubIndent() < 0)
 301             {
 302                 str << SymbolicIndent( - thisStyle.GetLeftSubIndent());
 303             }
 304
 305             m_inTable = true;
 306         }
 307     }
 308     else
 309     {
 310         CloseLists(-1, str);
 311
 312         wxString align = GetAlignment(thisStyle);
 313         str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
 314
 315         if (thisStyle.HasParagraphSpacingAfter() && thisStyle.GetParagraphSpacingAfter() == 0)
 316             str << wxT(" style=\"margin: 0px;\"");
 317
 318         str << wxT(">");
 319     }
 320 }
 321
 322 /// End paragraph formatting
 323 void wxRichTextHTMLHandler::EndParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& stream)
 324 {
 325     if (m_inTable)
 326     {
 327         if (thisStyle.HasFont())
 328             stream << wxT("</font>");
 329
 330         stream << wxT("</td></tr></table></p>\n");
 331         m_inTable = false;
 332     }
 333     else if (!thisStyle.HasBulletStyle())
 334         stream << wxT("</p>\n");
 335 }
 336
 337 /// Closes lists to level (-1 means close all)
 338 void wxRichTextHTMLHandler::CloseLists(int level, wxTextOutputStream& str)
 339 {
 340     // Close levels high than this
 341     int i = m_indents.GetCount()-1;
 342     while (i >= 0)
 343     {
 344         int l = m_indents[i];
 345         if (l > level)
 346         {
 347             if (m_listTypes[i] == 0)
 348                 str << wxT("</ol>");
 349             else
 350                 str << wxT("</ul>");
 351             m_indents.RemoveAt(i);
 352             m_listTypes.RemoveAt(i);
 353         }
 354         else
 355             break;
 356         i --;
 357      }
 358 }
 359
 360 /// Output font tag
 361 void wxRichTextHTMLHandler::OutputFont(const wxTextAttr& style, wxTextOutputStream& stream)
 362 {
 363     if (style.HasFont())
 364     {
 365         stream << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\""), style.GetFontFaceName().c_str(), PtToSize(style.GetFontSize()));
 366         if (style.HasTextColour())
 367             stream << wxString::Format(wxT(" color=\"%s\""), style.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
 368         stream << wxT(" >");
 369     }
 370 }
 371
 372 int wxRichTextHTMLHandler::TypeOfList( const wxTextAttr& thisStyle, wxString& tag )
 373 {
 374     // We can use number attribute of li tag but not all the browsers support it.
 375     // also wxHtmlWindow doesn't support type attribute.
 376
 377     bool m_is_ul = false;
 378     if (thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
 379         tag = wxT("<ol type=\"1\">");
 380     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER)
 381         tag = wxT("<ol type=\"A\">");
 382     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER)
 383         tag = wxT("<ol type=\"a\">");
 384     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER)
 385         tag = wxT("<ol type=\"I\">");
 386     else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER)
 387         tag = wxT("<ol type=\"i\">");
 388     else
 389     {
 390         tag = wxT("<ul>");
 391         m_is_ul = true;
 392     }
 393
 394     if (m_is_ul)
 395         return 1;
 396     else
 397         return 0;
 398 }
 399
 400 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttr& thisStyle )
 401 {
 402     switch( thisStyle.GetAlignment() )
 403     {
 404     case wxTEXT_ALIGNMENT_LEFT:
 405         return  wxT("left");
 406     case wxTEXT_ALIGNMENT_RIGHT:
 407         return wxT("right");
 408     case wxTEXT_ALIGNMENT_CENTER:
 409         return wxT("center");
 410     case wxTEXT_ALIGNMENT_JUSTIFIED:
 411         return wxT("justify");
 412     default:
 413         return wxT("left");
 414     }
 415 }
 416
 417 void wxRichTextHTMLHandler::WriteImage(wxRichTextImage* image, wxOutputStream& stream)
 418 {
 419     wxTextOutputStream str(stream);
 420
 421     str << wxT("<img src=\"");
 422
 423 #if wxUSE_FILESYSTEM
 424     if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 425     {
 426         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 427             image->LoadFromBlock();
 428         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 429             image->MakeBlock();
 430
 431         if (image->GetImage().Ok())
 432         {
 433             wxString ext(image->GetImageBlock().GetExtension());
 434             wxString tempFilename(wxString::Format(wxT("image%d.%s"), sm_fileCounter, ext));
 435             wxMemoryFSHandler::AddFile(tempFilename, image->GetImage(), image->GetImageBlock().GetImageType());
 436
 437             m_imageLocations.Add(tempFilename);
 438
 439             str << wxT("memory:") << tempFilename;
 440         }
 441         else
 442             str << wxT("memory:?");
 443
 444         sm_fileCounter ++;
 445     }
 446     else if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 447     {
 448         if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
 449             image->LoadFromBlock();
 450         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 451             image->MakeBlock();
 452
 453         if (image->GetImage().Ok())
 454         {
 455             wxString tempDir(GetTempDir());
 456             if (tempDir.IsEmpty())
 457                 tempDir = wxFileName::GetTempDir();
 458
 459             wxString ext(image->GetImageBlock().GetExtension());
 460             wxString tempFilename(wxString::Format(wxT("%s/image%d.%s"), tempDir, sm_fileCounter, ext));
 461             image->GetImageBlock().Write(tempFilename);
 462
 463             m_imageLocations.Add(tempFilename);
 464
 465             str << wxFileSystem::FileNameToURL(tempFilename);
 466         }
 467         else
 468             str << wxT("file:?");
 469
 470         sm_fileCounter ++;
 471     }
 472     else // if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_BASE64) // this is implied
 473 #endif
 474     {
 475         str << wxT("data:");
 476         str << GetMimeType(image->GetImageBlock().GetImageType());
 477         str << wxT(";base64,");
 478
 479         if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
 480             image->MakeBlock();
 481
 482         wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
 483         str << data;
 484
 485         delete[] data;
 486     }
 487
 488     str << wxT("\" />");
 489 }
 490
 491 long wxRichTextHTMLHandler::PtToSize(long size)
 492 {
 493     int i;
 494     int len = m_fontSizeMapping.GetCount();
 495     for (i = 0; i < len; i++)
 496         if (size <= m_fontSizeMapping[i])
 497             return i+1;
 498     return 7;
 499 }
 500
 501 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
 502 {
 503     wxString in;
 504     for(;indent > 0; indent -= 20)
 505         in.Append( wxT("&nbsp;") );
 506     return in;
 507 }
 508
 509 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
 510 {
 511     switch(imageType)
 512     {
 513     case wxBITMAP_TYPE_BMP:
 514         return wxT("image/bmp");
 515     case wxBITMAP_TYPE_TIF:
 516         return wxT("image/tiff");
 517     case wxBITMAP_TYPE_GIF:
 518         return wxT("image/gif");
 519     case wxBITMAP_TYPE_PNG:
 520         return wxT("image/png");
 521     case wxBITMAP_TYPE_JPEG:
 522         return wxT("image/jpeg");
 523     default:
 524         return wxT("image/unknown");
 525     }
 526 }
 527
 528 // exim-style base64 encoder
 529 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
 530 {
 531     // elements of enc64 array must be 8 bit values
 532     // otherwise encoder will fail
 533     // hmmm.. Does wxT macro define a char as 16 bit value
 534     // when compiling with UNICODE option?
 535     static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 536     wxChar* output = new wxChar[4*((in_len+2)/3)+1];
 537     wxChar* p = output;
 538
 539     while( in_len-- > 0 )
 540     {
 541         register wxChar a, b;
 542
 543         a = *input++;
 544
 545         *p++ = enc64[ (a >> 2) & 0x3f ];
 546
 547         if( in_len-- == 0 )
 548         {
 549             *p++ = enc64[ (a << 4 ) & 0x30 ];
 550             *p++ = '=';
 551             *p++ = '=';
 552             break;
 553         }
 554
 555         b = *input++;
 556
 557         *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
 558
 559         if( in_len-- == 0 )
 560         {
 561             *p++ = enc64[ (b << 2) & 0x3f ];
 562             *p++ = '=';
 563             break;
 564         }
 565
 566         a = *input++;
 567
 568         *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
 569
 570         *p++ = enc64[ a & 0x3f ];
 571     }
 572     *p = 0;
 573
 574     return output;
 575 }
 576 #endif
 577 // wxUSE_STREAMS
 578
 579 /// Delete the in-memory or temporary files generated by the last operation
 580 bool wxRichTextHTMLHandler::DeleteTemporaryImages()
 581 {
 582     return DeleteTemporaryImages(GetFlags(), m_imageLocations);
 583 }
 584
 585 /// Delete the in-memory or temporary files generated by the last operation
 586 bool wxRichTextHTMLHandler::DeleteTemporaryImages(int flags, const wxArrayString& imageLocations)
 587 {
 588     size_t i;
 589     for (i = 0; i < imageLocations.GetCount(); i++)
 590     {
 591         wxString location = imageLocations[i];
 592
 593         if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
 594         {
 595 #if wxUSE_FILESYSTEM
 596             wxMemoryFSHandler::RemoveFile(location);
 597 #endif
 598         }
 599         else if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
 600         {
 601             if (wxFileExists(location))
 602                 wxRemoveFile(location);
 603         }
 604     }
 605
 606     return true;
 607 }
 608
 609
 610 #endif
 611 // wxUSE_RICHTEXT
 612