xref: /trunk/main/sfx2/source/bastyp/sfxhtml.cxx (revision 86e1cf34)
1 /**************************************************************
2  *
3  * Licensed to the Apache Software Foundation (ASF) under one
4  * or more contributor license agreements.  See the NOTICE file
5  * distributed with this work for additional information
6  * regarding copyright ownership.  The ASF licenses this file
7  * to you under the Apache License, Version 2.0 (the
8  * "License"); you may not use this file except in compliance
9  * with the License.  You may obtain a copy of the License at
10  *
11  *   http://www.apache.org/licenses/LICENSE-2.0
12  *
13  * Unless required by applicable law or agreed to in writing,
14  * software distributed under the License is distributed on an
15  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
16  * KIND, either express or implied.  See the License for the
17  * specific language governing permissions and limitations
18  * under the License.
19  *
20  *************************************************************/
21 
22 
23 
24 // MARKER(update_precomp.py): autogen include statement, do not remove
25 #include "precompiled_sfx2.hxx"
26 
27 #include <tools/urlobj.hxx>
28 
29 #include <sfx2/objsh.hxx>
30 #include <sfx2/docfile.hxx>
31 #include "openflag.hxx"
32 
33 #include <svtools/htmlkywd.hxx>
34 #include <svtools/htmltokn.h>
35 #include <svtools/imap.hxx>
36 #include <svtools/imapcirc.hxx>
37 #include <svtools/imapobj.hxx>
38 #include <svtools/imappoly.hxx>
39 #include <svtools/imaprect.hxx>
40 #ifndef _SVSTDARR_ULONGS_DECL
41 #define _SVSTDARR_ULONGS
42 #include <svl/svstdarr.hxx>
43 #endif
44 #include <svl/zforlist.hxx>
45 #include <rtl/tencinfo.h>
46 #include <tools/tenccvt.hxx>
47 
48 #include <sfx2/sfxhtml.hxx>
49 
50 #include <com/sun/star/beans/XPropertyContainer.hpp>
51 
52 
53 using namespace ::com::sun::star;
54 
55 
56 sal_Char __FAR_DATA sHTML_MIME_text[] = "text/";
57 sal_Char __FAR_DATA sHTML_MIME_application[] = "application/";
58 sal_Char __FAR_DATA sHTML_MIME_experimental[] = "x-";
59 
60 // <INPUT TYPE=xxx>
61 static HTMLOptionEnum __READONLY_DATA aAreaShapeOptEnums[] =
62 {
63 	{ OOO_STRING_SVTOOLS_HTML_SH_rect,		IMAP_OBJ_RECTANGLE	},
64 	{ OOO_STRING_SVTOOLS_HTML_SH_rectangle,	IMAP_OBJ_RECTANGLE	},
65 	{ OOO_STRING_SVTOOLS_HTML_SH_circ,		IMAP_OBJ_CIRCLE		},
66 	{ OOO_STRING_SVTOOLS_HTML_SH_circle,   	IMAP_OBJ_CIRCLE		},
67 	{ OOO_STRING_SVTOOLS_HTML_SH_poly,   		IMAP_OBJ_POLYGON	},
68 	{ OOO_STRING_SVTOOLS_HTML_SH_polygon,		IMAP_OBJ_POLYGON	},
69 	{ 0,					0					}
70 };
71 
SfxHTMLParser(SvStream & rStream,sal_Bool bIsNewDoc,SfxMedium * pMed)72 SfxHTMLParser::SfxHTMLParser( SvStream& rStream, sal_Bool bIsNewDoc,
73 							  SfxMedium *pMed ) :
74 	HTMLParser( rStream, bIsNewDoc ),
75 	pMedium( pMed ), pDLMedium( 0 ),
76 	nMetaTags( 0 )
77 {
78 	DBG_ASSERT( RTL_TEXTENCODING_DONTKNOW == GetSrcEncoding( ),
79 				"SfxHTMLParser::SfxHTMLParser: Wo kommt der ZS her?" );
80 	DBG_ASSERT( !IsSwitchToUCS2(),
81 				"SfxHTMLParser::SfxHTMLParser: Switch to UCS2?" );
82 
83 	// Although the real default encoding is ISO8859-1, we use MS-1252
84 	// als default encoding.
85 	SetSrcEncoding( GetExtendedCompatibilityTextEncoding(  RTL_TEXTENCODING_ISO_8859_1 ) );
86 
87 	// If the file starts with a BOM, switch to UCS2.
88 	SetSwitchToUCS2( sal_True );
89 }
90 
~SfxHTMLParser()91 __EXPORT SfxHTMLParser::~SfxHTMLParser()
92 {
93 	DBG_ASSERT( !pDLMedium, "Da ist ein File-Download stehengeblieben" );
94 	delete pDLMedium;
95 }
96 
ParseMapOptions(ImageMap * pImageMap,const HTMLOptions * pOptions)97 sal_Bool SfxHTMLParser::ParseMapOptions(ImageMap * pImageMap,
98 									const HTMLOptions * pOptions)
99 {
100 	DBG_ASSERT( pImageMap, "ParseMapOptions: keine Image-Map" );
101 	DBG_ASSERT( pOptions, "ParseMapOptions: keine Optionen" );
102 
103 	String aName;
104 
105 	for( sal_uInt16 i=pOptions->Count(); i; )
106 	{
107 		const HTMLOption *pOption = (*pOptions)[--i];
108 		switch( pOption->GetToken() )
109 		{
110 		case HTML_O_NAME:
111 			aName = pOption->GetString();
112 			break;
113 		}
114 	}
115 
116 	if( aName.Len() )
117 		pImageMap->SetName( aName );
118 
119 	return aName.Len() > 0;
120 }
121 
ParseAreaOptions(ImageMap * pImageMap,const String & rBaseURL,const HTMLOptions * pOptions,sal_uInt16 nEventMouseOver,sal_uInt16 nEventMouseOut)122 sal_Bool SfxHTMLParser::ParseAreaOptions(ImageMap * pImageMap, const String& rBaseURL,
123 									 const HTMLOptions * pOptions,
124 									 sal_uInt16 nEventMouseOver,
125 									 sal_uInt16 nEventMouseOut )
126 {
127 	DBG_ASSERT( pImageMap, "ParseAreaOptions: keine Image-Map" );
128 	DBG_ASSERT( pOptions, "ParseAreaOptions: keine Optionen" );
129 
130 	sal_uInt16 nShape = IMAP_OBJ_RECTANGLE;
131 	SvULongs aCoords;
132 	String aName, aHRef, aAlt, aTarget, sEmpty;
133 	sal_Bool bNoHRef = sal_False;
134 	SvxMacroTableDtor aMacroTbl;
135 
136 	for( sal_uInt16 i=pOptions->Count(); i; )
137 	{
138 		sal_uInt16 nEvent = 0;
139 		ScriptType eScrpType = STARBASIC;
140 		const HTMLOption *pOption = (*pOptions)[--i];
141 		switch( pOption->GetToken() )
142 		{
143 		case HTML_O_NAME:
144 			aName = pOption->GetString();
145 			break;
146 		case HTML_O_SHAPE:
147 			pOption->GetEnum( nShape, aAreaShapeOptEnums );
148 			break;
149 		case HTML_O_COORDS:
150 			pOption->GetNumbers( aCoords, sal_True );
151 			break;
152 		case HTML_O_HREF:
153             aHRef = INetURLObject::GetAbsURL( rBaseURL, pOption->GetString() );
154 			break;
155 		case HTML_O_NOHREF:
156 			bNoHRef = sal_True;
157 			break;
158 		case HTML_O_ALT:
159 			aAlt = pOption->GetString();
160 			break;
161 		case HTML_O_TARGET:
162 			aTarget = pOption->GetString();
163 			break;
164 
165 		case HTML_O_ONMOUSEOVER:
166 			eScrpType = JAVASCRIPT;
167 		case HTML_O_SDONMOUSEOVER:
168 			nEvent = nEventMouseOver;
169 			goto IMAPOBJ_SETEVENT;
170 
171 		case HTML_O_ONMOUSEOUT:
172 			eScrpType = JAVASCRIPT;
173 		case HTML_O_SDONMOUSEOUT:
174 			nEvent = nEventMouseOut;
175 			goto IMAPOBJ_SETEVENT;
176 IMAPOBJ_SETEVENT:
177 			if( nEvent )
178 			{
179 				String sTmp( pOption->GetString() );
180 				if( sTmp.Len() )
181 				{
182 					sTmp.ConvertLineEnd();
183 					aMacroTbl.Insert( nEvent,
184 						new SvxMacro( sTmp, sEmpty, eScrpType ));
185 				}
186 			}
187 			break;
188 		}
189 	}
190 
191 	if( bNoHRef )
192 		aHRef.Erase();
193 
194 	sal_Bool bNewArea = sal_True;
195 	switch( nShape )
196 	{
197 	case IMAP_OBJ_RECTANGLE:
198 		if( aCoords.Count() >=4 )
199 		{
200 			Rectangle aRec( aCoords[0], aCoords[1],
201 							aCoords[2], aCoords[3] );
202 			IMapRectangleObject aMapRObj( aRec, aHRef, aAlt, String(), aTarget, aName,
203 										  !bNoHRef );
204 			if( aMacroTbl.Count() )
205 				aMapRObj.SetMacroTable( aMacroTbl );
206 			pImageMap->InsertIMapObject( aMapRObj );
207 		}
208 		break;
209 	case IMAP_OBJ_CIRCLE:
210 		if( aCoords.Count() >=3 )
211 		{
212 			Point aPoint( aCoords[0], aCoords[1] );
213 			IMapCircleObject aMapCObj( aPoint, aCoords[2],aHRef, aAlt, String(),
214 									   aTarget, aName, !bNoHRef );
215 			if( aMacroTbl.Count() )
216 				aMapCObj.SetMacroTable( aMacroTbl );
217 			pImageMap->InsertIMapObject( aMapCObj );
218 		}
219 		break;
220 	case IMAP_OBJ_POLYGON:
221 		if( aCoords.Count() >=6 )
222 		{
223 			sal_uInt16 nCount = aCoords.Count() / 2;
224 			Polygon aPoly( nCount );
225 			for( sal_uInt16 i=0; i<nCount; i++ )
226 				aPoly[i] = Point( aCoords[2*i], aCoords[2*i+1] );
227 			IMapPolygonObject aMapPObj( aPoly, aHRef, aAlt, String(), aTarget, aName,
228 										!bNoHRef );
229 			if( aMacroTbl.Count() )
230 				aMapPObj.SetMacroTable( aMacroTbl );
231 			pImageMap->InsertIMapObject( aMapPObj );
232 		}
233 		break;
234 	default:
235 		bNewArea = sal_False;
236 	}
237 
238 	return bNewArea;
239 }
240 
241 
StartFileDownload(const String & rURL,int nToken,SfxObjectShell * pSh)242 void SfxHTMLParser::StartFileDownload( const String& rURL, int nToken,
243 									   SfxObjectShell *pSh )
244 {
245 	DBG_ASSERT( !pDLMedium, "StartFileDwonload bei aktivem Download" );
246 	if( pDLMedium )
247 		return;
248 
249 	pDLMedium = new SfxMedium( rURL, SFX_STREAM_READONLY, sal_False );
250 	if( pSh )
251 	{
252 		// Medium registrieren, damit abgebrochen werden kann
253 		pSh->RegisterTransfer( *pDLMedium );
254 
255 		// Target-Frame uebertragen, damit auch javascript:-URLs
256 		// "geladen" werden koennen.
257 		//const SfxMedium *pShMedium = pSh->GetMedium();
258 		//if( pShMedium )
259 		//	pDLMedium->SetLoadTargetFrame( pShMedium->GetLoadTargetFrame() );
260 	}
261 
262 	// Download anstossen (Achtung: Kann auch synchron sein).
263     if ( sal_True /*pMedium->GetDoneLink() == Link()*/ )
264         pDLMedium->DownLoad();
265     else
266     {
267         // Downloading-Flag auf sal_True setzen. Es werden dann auch
268         // Data-Available-Links, wenn wir in den Pending-Staus gelangen.
269         SetDownloadingFile( sal_True );
270         pDLMedium->DownLoad( STATIC_LINK( this, SfxHTMLParser, FileDownloadDone ) );
271 
272         // Wenn das Dowsnloading-Flag noch gesetzt ist erfolgt der Download
273         // asynchron. Wir gehen dann in den Pedning-Staus und warten dort.
274         // Solange sind alle Aufrufe des Data-Avaialble-Link gesperrt.
275         if( IsDownloadingFile() )
276         {
277             // Den aktuellen Zustand einfrieren und in den Pending-Status gehen.
278             // Wenn der Download beendet oder abgebrochen wurde, wird ueber
279             // NewDataRead ein Continue mit dem uebergeben Token angesteossen.
280             SaveState( nToken );
281             eState = SVPAR_PENDING;
282         }
283     }
284 }
285 
GetFileDownloadMIME(String & rMIME)286 sal_Bool SfxHTMLParser::GetFileDownloadMIME( String& rMIME )
287 {
288 	return pDLMedium && pDLMedium->GetErrorCode()==0 &&
289 		   pDLMedium->GetMIMEAndRedirect(rMIME)==0;
290 }
291 
FinishFileDownload(String & rStr)292 sal_Bool SfxHTMLParser::FinishFileDownload( String& rStr )
293 {
294 	String aStr;
295 
296 	sal_Bool bOK = pDLMedium && pDLMedium->GetErrorCode()==0;
297 	if( bOK )
298 	{
299 		SvStream* pStream = pDLMedium->GetInStream();
300 		DBG_ASSERT( pStream, "Kein In-Stream vom Medium erhalten" );
301 
302 		SvMemoryStream aStream;
303 		if( pStream )	// HACK wegen #65563#
304 			aStream << *pStream;
305 
306 		aStream.Seek( STREAM_SEEK_TO_END );
307 		DBG_ASSERT( aStream.Tell() < STRING_MAXLEN,
308 					"File zu lang fuer einen String, Ende abgeschnitten" );
309 		xub_StrLen nLen = aStream.Tell() < STRING_MAXLEN
310 						? (xub_StrLen)aStream.Tell()
311 						: STRING_MAXLEN;
312 
313         // TODO: untested!!!
314 		rtl_TextEncoding eEnc =
315 			GetExtendedCompatibilityTextEncoding( RTL_TEXTENCODING_ISO_8859_1 );
316 		String sMime;
317 		if( pDLMedium->GetMIMEAndRedirect( sMime ) == 0 )
318 		{
319 			rtl_TextEncoding eMimeEnc = GetEncodingByMIME( sMime );
320 			if( RTL_TEXTENCODING_DONTKNOW != eMimeEnc )
321 				eEnc = eMimeEnc;
322 		}
323 
324 		ByteString sBuffer;
325 		sal_Char* pBuffer = sBuffer.AllocBuffer(nLen);
326 		aStream.Seek( 0 );
327 		aStream.Read((void*)pBuffer, nLen);
328 		rStr = String( pBuffer, RTL_TEXTENCODING_UTF8);
329 	}
330 
331 	delete pDLMedium;
332 	pDLMedium = 0;
333 
334 	return bOK;
335 }
336 
IMPL_STATIC_LINK(SfxHTMLParser,FileDownloadDone,void *,EMPTYARG)337 IMPL_STATIC_LINK( SfxHTMLParser, FileDownloadDone, void*, EMPTYARG )
338 {
339 	// Der Download ist jetzt abgeschlossen. Ausserdem muss/darf der
340 	// Data-Available-Link wieder durchgelassen werden.
341 	pThis->SetDownloadingFile( sal_False );
342 
343 	// ... und einmal aufrufen, damit weitergelesen wird.
344 	pThis->CallAsyncCallLink();
345 
346 	return 0;
347 }
348 
GetScriptType_Impl(SvKeyValueIterator * pHTTPHeader)349 void SfxHTMLParser::GetScriptType_Impl( SvKeyValueIterator *pHTTPHeader )
350 {
351 	aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_JAVASCRIPT);
352 	eScriptType = JAVASCRIPT;
353 	if( pHTTPHeader )
354 	{
355 		SvKeyValue aKV;
356 		for( sal_Bool bCont = pHTTPHeader->GetFirst( aKV ); bCont;
357 			 bCont = pHTTPHeader->GetNext( aKV ) )
358 		{
359 			if( aKV.GetKey().EqualsIgnoreCaseAscii(
360 									OOO_STRING_SVTOOLS_HTML_META_content_script_type ) )
361 			{
362 				if( aKV.GetValue().Len() )
363 				{
364 					String aTmp( aKV.GetValue() );
365 					if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_text, 0, 5 ) )
366 						aTmp.Erase( 0, 5 );
367 					else if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_application,
368 														 0, 12 ) )
369 						aTmp.Erase( 0, 12 );
370 					else
371 						break;
372 
373 					if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_experimental, 0,
374 													2 ) )
375 					{
376 						aTmp.Erase( 0, 2 );
377 					}
378 
379 					if( aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_starbasic ) )
380 					{
381 						eScriptType = STARBASIC;
382 						aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_STARBASIC);
383 					}
384 					if( !aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_javascript ) )
385 					{
386 						eScriptType = EXTENDED_STYPE;
387 						aScriptType = aTmp;
388 					}
389 				}
390 				break;
391 			}
392 		}
393 	}
394 }
395 
GetScriptType(SvKeyValueIterator * pHTTPHeader) const396 ScriptType SfxHTMLParser::GetScriptType( SvKeyValueIterator *pHTTPHeader ) const
397 {
398 	if( !aScriptType.Len() )
399 		((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader );
400 
401 	return eScriptType;
402 }
403 
GetScriptTypeString(SvKeyValueIterator * pHTTPHeader) const404 const String& SfxHTMLParser::GetScriptTypeString(
405 									SvKeyValueIterator *pHTTPHeader ) const
406 {
407 	if( !aScriptType.Len() )
408 		((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader );
409 
410 	return aScriptType;
411 }
412 
GetTableDataOptionsValNum(sal_uInt32 & nNumForm,LanguageType & eNumLang,const String & aValStr,const String & aNumStr,SvNumberFormatter & rFormatter)413 double SfxHTMLParser::GetTableDataOptionsValNum( sal_uInt32& nNumForm,
414 		LanguageType& eNumLang, const String& aValStr, const String& aNumStr,
415 		SvNumberFormatter& rFormatter )
416 {
417 	LanguageType eParseLang = (LanguageType )aNumStr.ToInt32();
418 	sal_uInt32 nParseForm =
419 		rFormatter.GetFormatForLanguageIfBuiltIn( 0, eParseLang );
420 	double fVal;
421 	rFormatter.IsNumberFormat( aValStr, nParseForm, fVal );
422 	if ( aNumStr.GetTokenCount( ';' ) > 2 )
423 	{
424 		eNumLang = (LanguageType)aNumStr.GetToken( 1, ';' ).ToInt32();
425 		xub_StrLen nPos = aNumStr.Search( ';' );
426 		nPos = aNumStr.Search( ';', nPos + 1 );
427 		String aFormat( aNumStr.Copy( nPos + 1 ) );
428 		xub_StrLen nCheckPos;
429 		short nType;
430 		if ( eNumLang != LANGUAGE_SYSTEM )
431 			rFormatter.PutEntry( aFormat, nCheckPos, nType, nNumForm, eNumLang );
432 		else
433 			rFormatter.PutandConvertEntry( aFormat, nCheckPos, nType, nNumForm,
434 				eParseLang, eNumLang );
435 	}
436 	else
437 	{
438 		eNumLang = LANGUAGE_SYSTEM;
439 		nNumForm = rFormatter.GetFormatForLanguageIfBuiltIn( 0, eNumLang );
440 	}
441 	return fVal;
442 }
443 
444