fixed encoding handling in ANSI mode
[wxWidgets.git] / src / html / winpars.cpp
1 /////////////////////////////////////////////////////////////////////////////
2 // Name: winpars.cpp
3 // Purpose: wxHtmlParser class (generic parser)
4 // Author: Vaclav Slavik
5 // RCS-ID: $Id$
6 // Copyright: (c) 1999 Vaclav Slavik
7 // Licence: wxWindows Licence
8 /////////////////////////////////////////////////////////////////////////////
9
10
11 #ifdef __GNUG__
12 #pragma implementation "winpars.h"
13 #endif
14
15 #include "wx/wxprec.h"
16
17 #include "wx/defs.h"
18 #if wxUSE_HTML && wxUSE_STREAMS
19
20 #ifdef __BORLANDC__
21 #pragma hdrstop
22 #endif
23
24 #ifndef WXPRECOMP
25 #include "wx/intl.h"
26 #include "wx/dc.h"
27 #endif
28
29 #include "wx/html/htmldefs.h"
30 #include "wx/html/winpars.h"
31 #include "wx/html/htmlwin.h"
32 #include "wx/fontmap.h"
33 #include "wx/log.h"
34
35
36 //-----------------------------------------------------------------------------
37 // wxHtmlWinParser
38 //-----------------------------------------------------------------------------
39
40
41 wxList wxHtmlWinParser::m_Modules;
42
43 wxHtmlWinParser::wxHtmlWinParser(wxHtmlWindow *wnd) : wxHtmlParser()
44 {
45 m_tmpStrBuf = NULL;
46 m_tmpStrBufSize = 0;
47 m_Window = wnd;
48 m_Container = NULL;
49 m_DC = NULL;
50 m_CharHeight = m_CharWidth = 0;
51 m_UseLink = FALSE;
52 #if !wxUSE_UNICODE
53 m_EncConv = NULL;
54 m_InputEnc = wxFONTENCODING_ISO8859_1;
55 m_OutputEnc = wxFONTENCODING_DEFAULT;
56 #endif
57
58 {
59 int i, j, k, l, m;
60 for (i = 0; i < 2; i++)
61 for (j = 0; j < 2; j++)
62 for (k = 0; k < 2; k++)
63 for (l = 0; l < 2; l++)
64 for (m = 0; m < 7; m++)
65 {
66 m_FontsTable[i][j][k][l][m] = NULL;
67 m_FontsFacesTable[i][j][k][l][m] = wxEmptyString;
68 #if !wxUSE_UNICODE
69 m_FontsEncTable[i][j][k][l][m] = wxFONTENCODING_DEFAULT;
70 #endif
71 }
72 #ifdef __WXMSW__
73 static int default_sizes[7] = {7, 8, 10, 12, 16, 22, 30};
74 #elif defined(__WXMAC__)
75 static int default_sizes[7] = {9, 12, 14, 18, 24, 30, 36};
76 #else
77 static int default_sizes[7] = {10, 12, 14, 16, 19, 24, 32};
78 #endif
79 SetFonts(wxT(""), wxT(""), default_sizes);
80 }
81
82 // fill in wxHtmlParser's tables:
83 wxNode *node = m_Modules.GetFirst();
84 while (node)
85 {
86 wxHtmlTagsModule *mod = (wxHtmlTagsModule*) node->GetData();
87 mod->FillHandlersTable(this);
88 node = node->GetNext();
89 }
90 }
91
92 wxHtmlWinParser::~wxHtmlWinParser()
93 {
94 int i, j, k, l, m;
95
96 for (i = 0; i < 2; i++)
97 for (j = 0; j < 2; j++)
98 for (k = 0; k < 2; k++)
99 for (l = 0; l < 2; l++)
100 for (m = 0; m < 7; m++)
101 {
102 if (m_FontsTable[i][j][k][l][m] != NULL)
103 delete m_FontsTable[i][j][k][l][m];
104 }
105 #if !wxUSE_UNICODE
106 delete m_EncConv;
107 #endif
108 delete[] m_tmpStrBuf;
109 }
110
111 void wxHtmlWinParser::AddModule(wxHtmlTagsModule *module)
112 {
113 m_Modules.Append(module);
114 }
115
116 void wxHtmlWinParser::RemoveModule(wxHtmlTagsModule *module)
117 {
118 m_Modules.DeleteObject(module);
119 }
120
121 void wxHtmlWinParser::SetFonts(wxString normal_face, wxString fixed_face, const int *sizes)
122 {
123 int i, j, k, l, m;
124
125 for (i = 0; i < 7; i++) m_FontsSizes[i] = sizes[i];
126 m_FontFaceFixed = fixed_face;
127 m_FontFaceNormal = normal_face;
128
129 #if !wxUSE_UNICODE
130 SetInputEncoding(m_InputEnc);
131 #endif
132
133 for (i = 0; i < 2; i++)
134 for (j = 0; j < 2; j++)
135 for (k = 0; k < 2; k++)
136 for (l = 0; l < 2; l++)
137 for (m = 0; m < 7; m++) {
138 if (m_FontsTable[i][j][k][l][m] != NULL)
139 {
140 delete m_FontsTable[i][j][k][l][m];
141 m_FontsTable[i][j][k][l][m] = NULL;
142 }
143 }
144 }
145
146 void wxHtmlWinParser::InitParser(const wxString& source)
147 {
148 wxHtmlParser::InitParser(source);
149 wxASSERT_MSG(m_DC != NULL, wxT("no DC assigned to wxHtmlWinParser!!"));
150
151 m_FontBold = m_FontItalic = m_FontUnderlined = m_FontFixed = FALSE;
152 m_FontSize = 3; //default one
153 CreateCurrentFont(); // we're selecting default font into
154 m_DC->GetTextExtent( wxT("H"), &m_CharWidth, &m_CharHeight);
155 /* NOTE : we're not using GetCharWidth/Height() because
156 of differences under X and win
157 */
158
159 m_UseLink = FALSE;
160 m_Link = wxHtmlLinkInfo( wxT(""), wxT("") );
161 m_LinkColor.Set(0, 0, 0xFF);
162 m_ActualColor.Set(0, 0, 0);
163 m_Align = wxHTML_ALIGN_LEFT;
164 m_tmpLastWasSpace = FALSE;
165
166 OpenContainer();
167 OpenContainer();
168
169 #if !wxUSE_UNICODE
170 wxString charset = ExtractCharsetInformation(source);
171 if (!charset.empty())
172 {
173 wxFontEncoding enc = wxFontMapper::Get()->CharsetToEncoding(charset);
174 if (enc != wxFONTENCODING_SYSTEM)
175 SetInputEncoding(enc);
176 }
177 #endif
178
179 m_Container->InsertCell(new wxHtmlColourCell(m_ActualColor));
180 m_Container->InsertCell(new wxHtmlFontCell(CreateCurrentFont()));
181 }
182
183 void wxHtmlWinParser::DoneParser()
184 {
185 m_Container = NULL;
186 #if !wxUSE_UNICODE
187 SetInputEncoding(wxFONTENCODING_ISO8859_1); // for next call
188 #endif
189 wxHtmlParser::DoneParser();
190 }
191
192 wxObject* wxHtmlWinParser::GetProduct()
193 {
194 wxHtmlContainerCell *top;
195
196 CloseContainer();
197 OpenContainer();
198
199 top = m_Container;
200 while (top->GetParent()) top = top->GetParent();
201 return top;
202 }
203
204 wxFSFile *wxHtmlWinParser::OpenURL(wxHtmlURLType type,
205 const wxString& url) const
206 {
207 // FIXME - normalize the URL to full path before passing to
208 // OnOpeningURL!!
209 if ( m_Window )
210 {
211 wxString myurl(url);
212 wxHtmlOpeningStatus status;
213 for (;;)
214 {
215 wxString redirect;
216 status = m_Window->OnOpeningURL(type, myurl, &redirect);
217 if ( status != wxHTML_REDIRECT )
218 break;
219
220 myurl = redirect;
221 }
222
223 if ( status == wxHTML_BLOCK )
224 return NULL;
225
226 return GetFS()->OpenFile(myurl);
227 }
228
229 return wxHtmlParser::OpenURL(type, url);
230 }
231
232 void wxHtmlWinParser::AddText(const wxChar* txt)
233 {
234 wxHtmlCell *c;
235 size_t i = 0,
236 x,
237 lng = wxStrlen(txt);
238 register wxChar d;
239 int templen = 0;
240 wxChar nbsp = GetEntitiesParser()->GetCharForCode(160 /* nbsp */);
241
242 if (lng+1 > m_tmpStrBufSize)
243 {
244 delete[] m_tmpStrBuf;
245 m_tmpStrBuf = new wxChar[lng+1];
246 m_tmpStrBufSize = lng+1;
247 }
248 wxChar *temp = m_tmpStrBuf;
249
250 if (m_tmpLastWasSpace)
251 {
252 while ((i < lng) &&
253 ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) || (txt[i] == wxT(' ')) ||
254 (txt[i] == wxT('\t')))) i++;
255 }
256
257 while (i < lng)
258 {
259 x = 0;
260 d = temp[templen++] = txt[i];
261 if ((d == wxT('\n')) || (d == wxT('\r')) || (d == wxT(' ')) || (d == wxT('\t')))
262 {
263 i++, x++;
264 while ((i < lng) && ((txt[i] == wxT('\n')) || (txt[i] == wxT('\r')) ||
265 (txt[i] == wxT(' ')) || (txt[i] == wxT('\t')))) i++, x++;
266 }
267 else i++;
268
269 if (x)
270 {
271 temp[templen-1] = wxT(' ');
272 temp[templen] = 0;
273 #if 0 // VS - WHY was this here?!
274 if (templen == 1) continue;
275 #endif
276 templen = 0;
277 #if !wxUSE_UNICODE
278 if (m_EncConv)
279 m_EncConv->Convert(temp);
280 #endif
281 size_t len = wxStrlen(temp);
282 for (size_t j = 0; j < len; j++)
283 if (temp[j] == nbsp)
284 temp[j] = wxT(' ');
285 c = new wxHtmlWordCell(temp, *(GetDC()));
286 if (m_UseLink)
287 c->SetLink(m_Link);
288 m_Container->InsertCell(c);
289 m_tmpLastWasSpace = TRUE;
290 }
291 }
292
293 if (templen && (templen > 1 || temp[0] != wxT(' ')))
294 {
295 temp[templen] = 0;
296 #if !wxUSE_UNICODE
297 if (m_EncConv)
298 m_EncConv->Convert(temp);
299 #endif
300 size_t len = wxStrlen(temp);
301 for (size_t j = 0; j < len; j++)
302 if (temp[j] == nbsp)
303 temp[j] = wxT(' ');
304 c = new wxHtmlWordCell(temp, *(GetDC()));
305 if (m_UseLink)
306 c->SetLink(m_Link);
307 m_Container->InsertCell(c);
308 m_tmpLastWasSpace = FALSE;
309 }
310 }
311
312
313
314 wxHtmlContainerCell* wxHtmlWinParser::OpenContainer()
315 {
316 m_Container = new wxHtmlContainerCell(m_Container);
317 m_Container->SetAlignHor(m_Align);
318 m_tmpLastWasSpace = TRUE;
319 /* to avoid space being first character in paragraph */
320 return m_Container;
321 }
322
323
324
325 wxHtmlContainerCell* wxHtmlWinParser::SetContainer(wxHtmlContainerCell *c)
326 {
327 m_tmpLastWasSpace = TRUE;
328 /* to avoid space being first character in paragraph */
329 return m_Container = c;
330 }
331
332
333
334 wxHtmlContainerCell* wxHtmlWinParser::CloseContainer()
335 {
336 m_Container = m_Container->GetParent();
337 return m_Container;
338 }
339
340
341 void wxHtmlWinParser::SetFontSize(int s)
342 {
343 if (s < 1) s = 1;
344 else if (s > 7) s = 7;
345 m_FontSize = s;
346 }
347
348
349
350 wxFont* wxHtmlWinParser::CreateCurrentFont()
351 {
352 int fb = GetFontBold(),
353 fi = GetFontItalic(),
354 fu = GetFontUnderlined(),
355 ff = GetFontFixed(),
356 fs = GetFontSize() - 1 /*remap from <1;7> to <0;6>*/ ;
357
358 wxString face = ff ? m_FontFaceFixed : m_FontFaceNormal;
359 wxString *faceptr = &(m_FontsFacesTable[fb][fi][fu][ff][fs]);
360 wxFont **fontptr = &(m_FontsTable[fb][fi][fu][ff][fs]);
361 #if !wxUSE_UNICODE
362 wxFontEncoding *encptr = &(m_FontsEncTable[fb][fi][fu][ff][fs]);
363 #endif
364
365 if (*fontptr != NULL && (*faceptr != face
366 #if !wxUSE_UNICODE
367 || *encptr != m_OutputEnc
368 #endif
369 ))
370 {
371 delete *fontptr;
372 *fontptr = NULL;
373 }
374
375 if (*fontptr == NULL)
376 {
377 *faceptr = face;
378 *fontptr = new wxFont(
379 (int) (m_FontsSizes[fs] * m_PixelScale),
380 ff ? wxMODERN : wxSWISS,
381 fi ? wxITALIC : wxNORMAL,
382 fb ? wxBOLD : wxNORMAL,
383 fu ? TRUE : FALSE, face
384 #if wxUSE_UNICODE
385 );
386 #else
387 , m_OutputEnc);
388 *encptr = m_OutputEnc;
389 #endif
390 }
391 m_DC->SetFont(**fontptr);
392 return (*fontptr);
393 }
394
395
396
397 void wxHtmlWinParser::SetLink(const wxHtmlLinkInfo& link)
398 {
399 m_Link = link;
400 m_UseLink = (link.GetHref() != wxEmptyString);
401 }
402
403
404 void wxHtmlWinParser::SetFontFace(const wxString& face)
405 {
406 if (GetFontFixed()) m_FontFaceFixed = face;
407 else m_FontFaceNormal = face;
408
409 #if !wxUSE_UNICODE
410 if (m_InputEnc != wxFONTENCODING_DEFAULT)
411 SetInputEncoding(m_InputEnc);
412 #endif
413 }
414
415
416
417 #if !wxUSE_UNICODE
418 void wxHtmlWinParser::SetInputEncoding(wxFontEncoding enc)
419 {
420 m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
421 if (m_EncConv)
422 {
423 delete m_EncConv;
424 m_EncConv = NULL;
425 }
426
427 if (enc == wxFONTENCODING_DEFAULT) return;
428
429 wxFontEncoding altfix, altnorm;
430 bool availfix, availnorm;
431
432 // exact match?
433 availnorm = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceNormal);
434 availfix = wxFontMapper::Get()->IsEncodingAvailable(enc, m_FontFaceFixed);
435 if (availnorm && availfix)
436 m_OutputEnc = enc;
437
438 // alternatives?
439 else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, FALSE) &&
440 wxFontMapper::Get()->GetAltForEncoding(enc, &altfix, m_FontFaceFixed, FALSE) &&
441 altnorm == altfix)
442 m_OutputEnc = altnorm;
443
444 // at least normal face?
445 else if (availnorm)
446 m_OutputEnc = enc;
447 else if (wxFontMapper::Get()->GetAltForEncoding(enc, &altnorm, m_FontFaceNormal, FALSE))
448 m_OutputEnc = altnorm;
449
450 // okay, let convert to ISO_8859-1, available always
451 else
452 m_OutputEnc = wxFONTENCODING_DEFAULT;
453
454 m_InputEnc = enc;
455 if (m_OutputEnc == wxFONTENCODING_DEFAULT)
456 GetEntitiesParser()->SetEncoding(wxFONTENCODING_SYSTEM);
457 else
458 GetEntitiesParser()->SetEncoding(m_OutputEnc);
459
460 if (m_InputEnc == m_OutputEnc) return;
461
462 m_EncConv = new wxEncodingConverter();
463 if (!m_EncConv->Init(m_InputEnc,
464 (m_OutputEnc == wxFONTENCODING_DEFAULT) ?
465 wxFONTENCODING_ISO8859_1 : m_OutputEnc,
466 wxCONVERT_SUBSTITUTE))
467 { // total failture :-(
468 wxLogError(_("Failed to display HTML document in %s encoding"),
469 wxFontMapper::GetEncodingName(enc).c_str());
470 m_InputEnc = m_OutputEnc = wxFONTENCODING_DEFAULT;
471 delete m_EncConv;
472 m_EncConv = NULL;
473 }
474 }
475 #endif
476
477
478
479
480 //-----------------------------------------------------------------------------
481 // wxHtmlWinTagHandler
482 //-----------------------------------------------------------------------------
483
484 IMPLEMENT_ABSTRACT_CLASS(wxHtmlWinTagHandler, wxHtmlTagHandler)
485
486 //-----------------------------------------------------------------------------
487 // wxHtmlTagsModule
488 //-----------------------------------------------------------------------------
489
490 // NB: This is *NOT* winpars.cpp's initialization and shutdown code!!
491 // This module is an ancestor for tag handlers modules defined
492 // in m_*.cpp files with TAGS_MODULE_BEGIN...TAGS_MODULE_END construct.
493 //
494 // Do not add any winpars.cpp shutdown or initialization code to it,
495 // create a new module instead!
496
497 IMPLEMENT_DYNAMIC_CLASS(wxHtmlTagsModule, wxModule)
498
499 bool wxHtmlTagsModule::OnInit()
500 {
501 wxHtmlWinParser::AddModule(this);
502 return TRUE;
503 }
504
505 void wxHtmlTagsModule::OnExit()
506 {
507 wxHtmlWinParser::RemoveModule(this);
508 }
509
510 #endif
511