]> git.saurik.com Git - wxWidgets.git/blame - src/richtext/richtexthtml.cpp
added a cast to wxDecodeSurrogate() to fix wxMSW cross-compilation
[wxWidgets.git] / src / richtext / richtexthtml.cpp
CommitLineData
b71e9aa4
JS
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
31IMPLEMENT_DYNAMIC_CLASS(wxRichTextHTMLHandler, wxRichTextFileHandler)
32
33/// Can we handle this filename (if using files)? By default, checks the extension.
34bool 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
44bool 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
53bool wxRichTextHTMLHandler::DoSaveFile(wxRichTextBuffer *buffer, wxOutputStream& stream)
54{
55 buffer->Defragment();
2dec6761 56
b71e9aa4 57 wxTextOutputStream str(stream);
2dec6761 58
b71e9aa4
JS
59 wxTextAttrEx currentParaStyle = buffer->GetAttributes();
60 wxTextAttrEx currentCharStyle = buffer->GetAttributes();
2dec6761 61
b71e9aa4 62 str << wxT("<html><head></head><body>\n");
2dec6761
JS
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
d0c3476b 83 str << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\" color=\"#%02X%02X%02X\" >"),
21e354f1 84 currentParaStyle.GetFont().GetFaceName().c_str(), Pt_To_Size( currentParaStyle.GetFont().GetPointSize() ),
2dec6761
JS
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
b71e9aa4
JS
95 wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
96 while (node)
97 {
98 wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
99 wxASSERT (para != NULL);
2dec6761 100
b71e9aa4
JS
101 if (para)
102 {
2dec6761
JS
103 OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream);
104
b71e9aa4
JS
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 {
2dec6761
JS
112 BeginCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
113
b71e9aa4 114 str << textObj->GetText();
2dec6761
JS
115
116 EndCharacterFormatting(currentCharStyle, obj->GetAttributes(), stream);
b71e9aa4 117 }
2dec6761
JS
118
119 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
120 if( image && !image->IsEmpty())
121 Image_to_Base64( image, stream );
122
b71e9aa4
JS
123 node2 = node2->GetNext();
124 }
2dec6761 125 //OutputParagraphFormatting(currentParaStyle, para->GetAttributes(), stream, false);
b71e9aa4 126 }
b71e9aa4
JS
127 node = node->GetNext();
128 }
2dec6761
JS
129
130 str << wxT("</font></td></tr></table></body></html>\n");
131
b71e9aa4
JS
132 return true;
133}
134
2dec6761 135void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttrEx& currentStyle, const wxTextAttrEx& thisStyle, wxOutputStream& stream)
b71e9aa4
JS
136{
137 wxTextOutputStream str(stream);
2dec6761
JS
138
139 //Is the item bulleted one?
140 if( thisStyle.GetBulletStyle() != wxTEXT_ATTR_BULLET_STYLE_NONE )
b71e9aa4 141 {
2dec6761
JS
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);
21e354f1 165 str << tag << wxT("<li>");
2dec6761
JS
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);
21e354f1 178 str << tag << wxT("<li>");
2dec6761
JS
179
180 //Now we have a list, mark it.
181 m_list = true;
182 }
b71e9aa4 183 }
2dec6761
JS
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
b71e9aa4 188
2dec6761
JS
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() )
b71e9aa4 196 {
2dec6761
JS
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 {
21e354f1 205 str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
2dec6761
JS
206 }
207 }
208 else
209 {
210 if( thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent() > m_indent )
211 {
212 Indent(thisStyle, str);
213 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
214 m_indents.Add( m_indent );
215 }
216 else
217 {
218 int i = m_indents.size() - 1;
219 for(; i > -1; i--)
220 {
221 if( m_indent < (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
222 {
223 Indent(thisStyle, str);
224 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
225 m_indents.Add( m_indent );
226
227 break;
228 }
229 else if( m_indent == (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) )
230 {
231 if( thisStyle.GetLeftSubIndent() < 0 )
232 {
21e354f1 233 str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
2dec6761
JS
234 }
235 break;
236 }
237 else
238 {
239 str << wxT("</td></tr></table>");
240
241 m_indents.RemoveAt(i);
242
243 if( i < 1 ){m_indent=0; break;}
244 m_indent = m_indents[i-1];
245 }
246 }
247 }
248 }
249 }
250 else
251 {
252 Indent(thisStyle, str);
253 m_indent = thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent();
254 m_indents.Add( m_indent );
255 }
256 }
b71e9aa4 257 }
2dec6761 258 else if( m_indent )
b71e9aa4 259 {
2dec6761
JS
260 //The item is not indented and there is a table(s) what should be closed now.
261
262 //So close them
263 for(unsigned int i = 0; i < m_indents.size(); i++ )
264 str << wxT("</td></tr></table>");
265
266 m_indent = 0;
267 m_indents.Clear();
b71e9aa4 268 }
2dec6761
JS
269
270
271 wxString style;
272
273 //Is there any change on the font properties of the item
274 if( thisStyle.GetFont().GetFaceName() != currentStyle.GetFont().GetFaceName() )
21e354f1 275 style += wxString::Format(wxT(" face=\"%s\""), thisStyle.GetFont().GetFaceName().c_str());
2dec6761 276 if( thisStyle.GetFont().GetPointSize() != currentStyle.GetFont().GetPointSize() )
d0c3476b 277 style += wxString::Format(wxT(" size=\"%ld\""), Pt_To_Size(thisStyle.GetFont().GetPointSize()) );
2dec6761
JS
278 if( thisStyle.GetTextColour() != currentStyle.GetTextColour() )
279 style += wxString::Format(wxT(" color=\"#%02X%02X%02X\""), thisStyle.GetTextColour().Red(),
280 thisStyle.GetTextColour().Green(), thisStyle.GetTextColour().Blue());
281
21e354f1
JS
282 if( style.size() )
283 {
284 str << wxString::Format(wxT("<font %s >"), style.c_str());
285 m_font = true;
286 }
2dec6761
JS
287
288 if( thisStyle.GetFont().GetWeight() == wxBOLD )
289 str << wxT("<b>");
290 if( thisStyle.GetFont().GetStyle() == wxITALIC )
291 str << wxT("<i>");
292 if( thisStyle.GetFont().GetUnderlined() )
293 str << wxT("<u>");
b71e9aa4
JS
294}
295
2dec6761 296void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxOutputStream& stream)
b71e9aa4 297{
b71e9aa4 298 wxTextOutputStream str(stream);
2dec6761
JS
299
300 if( thisStyle.GetFont().GetUnderlined() )
301 str << wxT("</u>");
302 if( thisStyle.GetFont().GetStyle() == wxITALIC )
303 str << wxT("</i>");
304 if( thisStyle.GetFont().GetWeight() == wxBOLD )
305 str << wxT("</b>");
306
307 if( m_font )
308 {
309 m_font = false;
310 str << wxT("</font>");
311 }
312}
b71e9aa4 313
2dec6761
JS
314/// Output paragraph formatting
315void wxRichTextHTMLHandler::OutputParagraphFormatting(const wxTextAttrEx& WXUNUSED(currentStyle), const wxTextAttrEx& thisStyle, wxOutputStream& stream)
316{
317 //If there is no opened list currently, insert a <p> after every paragraph
318 if(!m_list)
b71e9aa4 319 {
2dec6761
JS
320 wxTextOutputStream str(stream);
321 wxString align = GetAlignment( thisStyle );
21e354f1 322 str << wxString::Format(wxT("<p align=\"%s\">"), align.c_str());
b71e9aa4 323 }
2dec6761 324}
b71e9aa4 325
2dec6761
JS
326void wxRichTextHTMLHandler::NavigateToListPosition(const wxTextAttrEx& thisStyle, wxTextOutputStream& str)
327{
328 //indenting an item using an ul/ol tag is equal to inserting 5 x &nbsp; on its left side.
329 //so we should start from 100 point left
330
331 //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
332 //of the item, horizontally?
333 if( m_indent + 100 < thisStyle.GetLeftIndent() )
b71e9aa4 334 {
2dec6761
JS
335 //yes it is
336 LIndent(thisStyle, str);
337 m_indent = thisStyle.GetLeftIndent() - 100;
338 m_indents.Add( m_indent );
339 return;
b71e9aa4 340 }
2dec6761
JS
341 //No it isn't
342
343 int i = m_indents.size() - 1;
344 for(; i > -1; i--)
345 {
346 //Is the second td's left wall of the current indentaion table at the 100+ point-left-side
347 //of the item ?
348 if( m_indent + 100 < thisStyle.GetLeftIndent() )
349 {
350 //Yes it is
351 LIndent(thisStyle, str);
352 m_indent = thisStyle.GetLeftIndent() - 100;
353 m_indents.Add( m_indent );
354 break;
355 }
356 else if( m_indent + 100 == thisStyle.GetLeftIndent() )
357 break;//exact match
358 else
359 {
360 //No it is not, the second td's left wall of the current indentaion table is at the
361 //right side of the current item horizontally, so close it.
362 str << wxT("</td></tr></table>");
363
364 m_indents.RemoveAt(i);
365
366 if( i < 1 ){m_indent=0; break;}
367 m_indent = m_indents[i-1];
368 }
369 }
370}
371void wxRichTextHTMLHandler::Indent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
372{
373 //As a five year experienced web developer i assure you there is no way to indent an item
374 //in html way, but we can use tables.
375
376
377
378 //Item -> "Hello world"
379 //Its Left Indentation -> 100
380 //Its Left Sub-Indentation ->40
381 //A typical indentation-table for the item will be construct as the following
382
383 //3 x nbsp = 60
384 //2 x nbsp = 40
385 //LSI = Left Sub Indent
386 //LI = Left Indent - LSI
387 //
388 //-------------------------------------------
389 //|&nbsp;&nbsp;nbsp;|nbsp;nbsp;Hello World |
390 //| | | | |
391 //| V | V |
392 //| --LI-- | --LSI-- |
393 //-------------------------------------------
394
395 str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
396
397 wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent()) - m_indent );
21e354f1 398 str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent.c_str() );
2dec6761
JS
399 str << wxT("<td width=\"100%\">");
400
401 if( thisStyle.GetLeftSubIndent() < 0 )
402 {
21e354f1 403 str << SymbolicIndent(~thisStyle.GetLeftSubIndent());
2dec6761
JS
404 }
405}
406
407void wxRichTextHTMLHandler::LIndent( const wxTextAttrEx& thisStyle, wxTextOutputStream& str )
408{
409 //Code:
410 //r.BeginNumberedBullet(1, 200, 60);
411 //r.Newline();
412 //r.WriteText(wxT("first item"));
413 //r.EndNumberedBullet();
414 //r.BeginNumberedBullet(2, 200, 60);
415 //r.Newline();
416 //r.WriteText(wxT("second item."));
417 //r.EndNumberedBullet();
418 //
419 //A typical indentation-table for the item will be construct as the following
420
421 //1 x nbsp = 20 point
422 //ULI -> 100pt (UL/OL tag indents its sub element by 100 point)
423 //<--------- 100 pt ---------->|
424 //------------------------------------------------------
425 //|&nbsp;&nbsp;nbsp;&nbsp;nbsp;|<ul> |
426 //| |<-ULI-><li>first item |
427 //| |<-ULI-><li>second item |
428 //| |</ul> |
429 //------------------------------------------------------
430 // |<-100->|
431
432
433 str << wxT("<table width=\"100%\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\"><tr>");
434
435 wxString symbolic_indent = SymbolicIndent( (thisStyle.GetLeftIndent() - m_indent) - 100);
21e354f1 436 str << wxString::Format( wxT("<td>%s</td>"), symbolic_indent.c_str() );
2dec6761
JS
437 str << wxT("<td width=\"100%\">");
438}
439
440void wxRichTextHTMLHandler::TypeOfList( const wxTextAttrEx& thisStyle, wxString& tag )
441{
442 //We can use number attribute of li tag but not all the browsers support it.
443 //also wxHtmlWindow doesn't support type attribute.
444
445 m_is_ul = false;
446 if( thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
447 tag = wxT("<ol type=\"1\">");
448 else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER )
449 tag = wxT("<ol type=\"A\">");
450 else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER )
451 tag = wxT("<ol type=\"a\">");
452 else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER )
453 tag = wxT("<ol type=\"I\">");
454 else if( thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER )
455 tag = wxT("<ol type=\"i\">");
b71e9aa4
JS
456 else
457 {
2dec6761
JS
458 tag = wxT("<ul>");
459 m_is_ul = true;
b71e9aa4
JS
460 }
461}
462
2dec6761
JS
463wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttrEx& thisStyle )
464{
465 switch( thisStyle.GetAlignment() )
466 {
467 case wxTEXT_ALIGNMENT_LEFT:
468 return wxT("left");
469 case wxTEXT_ALIGNMENT_RIGHT:
470 return wxT("right");
471 case wxTEXT_ALIGNMENT_CENTER:
472 return wxT("center");
473 case wxTEXT_ALIGNMENT_JUSTIFIED:
474 return wxT("justify");
475 default:
476 return wxT("left");
477 }
478}
479
480void wxRichTextHTMLHandler::Image_to_Base64(wxRichTextImage* image, wxOutputStream& stream)
481{
482 wxTextOutputStream str(stream);
483
484 str << wxT("<img src=\"");
485 str << wxT("data:");
486 str << GetMimeType(image->GetImageBlock().GetImageType());
487 str << wxT(";base64,");
488
489 wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
490 str << data;
491
492 delete[] data;
493
494 str << wxT("\" />");
495}
496
497long wxRichTextHTMLHandler::Pt_To_Size(long size)
498{
499 //return most approximate size
500 if(size < 9 ) return 1;
501 else if( size < 11 ) return 2;
502 else if( size < 14 ) return 3;
503 else if( size < 18 ) return 4;
504 else if( size < 23 ) return 5;
505 else if( size < 30 ) return 6;
506 else return 7;
507}
508
509wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
510{
511 wxString in;
512 for(;indent > 0; indent -= 20)
513 in.Append( wxT("&nbsp;") );
514 return in;
515}
516
21e354f1 517const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
2dec6761
JS
518{
519 switch(imageType)
520 {
521 case wxBITMAP_TYPE_BMP:
522 return wxT("image/bmp");
523 case wxBITMAP_TYPE_TIF:
524 return wxT("image/tiff");
525 case wxBITMAP_TYPE_GIF:
526 return wxT("image/gif");
527 case wxBITMAP_TYPE_PNG:
528 return wxT("image/png");
529 case wxBITMAP_TYPE_JPEG:
530 return wxT("image/jpeg");
531 default:
532 return wxT("image/unknown");
533 }
534}
535
536//exim-style base64 encoder
537wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
538{
539 //elements of enc64 array must be 8 bit values
540 //otherwise encoder will fail
541 //hmmm.. Does wxT macro define a char as 16 bit value
542 //when compiling with UNICODE option?
21e354f1 543 static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
2dec6761
JS
544 wxChar* output = new wxChar[4*((in_len+2)/3)+1];
545 wxChar* p = output;
546
547 while( in_len-- > 0 )
548 {
549 register wxChar a, b;
550
551 a = *input++;
552
553 *p++ = enc64[ (a >> 2) & 0x3f ];
554
555 if( in_len-- <= 0 )
556 {
557 *p++ = enc64[ (a << 4 ) & 0x30 ];
558 *p++ = '=';
559 *p++ = '=';
560 break;
561 }
562
563 b = *input++;
564
565 *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
566
567 if( in_len-- <= 0 )
568 {
569 *p++ = enc64[ (b << 2) & 0x3f ];
570 *p++ = '=';
571 break;
572 }
573
574 a = *input++;
575
576 *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
577
578 *p++ = enc64[ a & 0x3f ];
579 }
580 *p = 0;
581
582 return output;
583}
b71e9aa4 584#endif
2dec6761 585// wxUSE_STREAMS
b71e9aa4
JS
586
587#endif
2dec6761
JS
588// wxUSE_RICHTEXT
589