xref: /trunk/main/sc/source/ui/unoobj/scdetect.cxx (revision fd9eb3f2)
1 /*************************************************************************
2  *
3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4  *
5  * Copyright 2000, 2010 Oracle and/or its affiliates.
6  *
7  * OpenOffice.org - a multi-platform office productivity suite
8  *
9  * This file is part of OpenOffice.org.
10  *
11  * OpenOffice.org is free software: you can redistribute it and/or modify
12  * it under the terms of the GNU Lesser General Public License version 3
13  * only, as published by the Free Software Foundation.
14  *
15  * OpenOffice.org is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18  * GNU Lesser General Public License version 3 for more details
19  * (a copy is included in the LICENSE file that accompanied this code).
20  *
21  * You should have received a copy of the GNU Lesser General Public License
22  * version 3 along with OpenOffice.org.  If not, see
23  * <http://www.openoffice.org/license.html>
24  * for a copy of the LGPLv3 License.
25  *
26  ************************************************************************/
27 
28 // MARKER(update_precomp.py): autogen include statement, do not remove
29 #include "precompiled_sc.hxx"
30 
31 #include "scdetect.hxx"
32 
33 #include <framework/interaction.hxx>
34 #include <com/sun/star/lang/XMultiServiceFactory.hpp>
35 #include <com/sun/star/beans/PropertyValue.hpp>
36 #include <com/sun/star/frame/XFrame.hpp>
37 #include <com/sun/star/frame/XModel.hpp>
38 #include <com/sun/star/awt/XWindow.hpp>
39 #include <com/sun/star/lang/XUnoTunnel.hpp>
40 #ifndef _UNOTOOLS_PROCESSFACTORY_HXX
41 #include <comphelper/processfactory.hxx>
42 #endif
43 #include <com/sun/star/beans/PropertyValue.hpp>
44 #include <com/sun/star/container/XNameAccess.hpp>
45 #include <com/sun/star/io/XInputStream.hpp>
46 #include <com/sun/star/task/XInteractionHandler.hpp>
47 #include <com/sun/star/lang/WrappedTargetRuntimeException.hpp>
48 #include <com/sun/star/ucb/CommandAbortedException.hpp>
49 #include <com/sun/star/ucb/InteractiveAppException.hpp>
50 #include <com/sun/star/ucb/XContent.hpp>
51 #include <com/sun/star/packages/zip/ZipIOException.hpp>
52 
53 
54 #include <framework/interaction.hxx>
55 
56 #ifndef _TOOLKIT_UNOHLP_HXX
57 #include <toolkit/helper/vclunohelper.hxx>
58 #endif
59 #include <ucbhelper/simpleinteractionrequest.hxx>
60 
61 #include <svtools/parhtml.hxx>
62 #include <rtl/ustring.h>
63 #include <rtl/logfile.hxx>
64 #include <svl/itemset.hxx>
65 #include <vcl/window.hxx>
66 #include <svl/eitem.hxx>
67 #include <svl/stritem.hxx>
68 #include <tools/urlobj.hxx>
69 #include <vos/mutex.hxx>
70 #include <svtools/sfxecode.hxx>
71 #include <svtools/ehdl.hxx>
72 #include <sot/storinfo.hxx>
73 #include <vcl/svapp.hxx>
74 #include <sfx2/sfxsids.hrc>
75 #include <sfx2/request.hxx>
76 #include <sfx2/docfile.hxx>
77 #include <sfx2/docfilt.hxx>
78 #include <sfx2/fcontnr.hxx>
79 #include <sfx2/app.hxx>
80 #include <sfx2/brokenpackageint.hxx>
81 #include <sot/storage.hxx>
82 
83 using namespace ::com::sun::star;
84 using namespace ::com::sun::star::uno;
85 using namespace ::com::sun::star::io;
86 using namespace ::com::sun::star::frame;
87 using namespace ::com::sun::star::task;
88 using namespace ::com::sun::star::beans;
89 using namespace ::com::sun::star::lang;
90 using namespace ::com::sun::star::ucb;
91 using ::rtl::OUString;
92 
93 ScFilterDetect::ScFilterDetect( const REFERENCE < ::com::sun::star::lang::XMultiServiceFactory >& /* xFactory */ )
94 {
95 }
96 
97 ScFilterDetect::~ScFilterDetect()
98 {
99 }
100 
101 static const sal_Char __FAR_DATA pFilterSc50[]		= "StarCalc 5.0";
102 static const sal_Char __FAR_DATA pFilterSc50Temp[]	= "StarCalc 5.0 Vorlage/Template";
103 static const sal_Char __FAR_DATA pFilterSc40[]		= "StarCalc 4.0";
104 static const sal_Char __FAR_DATA pFilterSc40Temp[]	= "StarCalc 4.0 Vorlage/Template";
105 static const sal_Char __FAR_DATA pFilterSc30[]		= "StarCalc 3.0";
106 static const sal_Char __FAR_DATA pFilterSc30Temp[]	= "StarCalc 3.0 Vorlage/Template";
107 static const sal_Char __FAR_DATA pFilterSc10[]		= "StarCalc 1.0";
108 static const sal_Char __FAR_DATA pFilterXML[]		= "StarOffice XML (Calc)";
109 static const sal_Char __FAR_DATA pFilterAscii[]		= "Text - txt - csv (StarCalc)";
110 static const sal_Char __FAR_DATA pFilterLotus[]		= "Lotus";
111 static const sal_Char __FAR_DATA pFilterQPro6[]		= "Quattro Pro 6.0";
112 static const sal_Char __FAR_DATA pFilterExcel4[]	= "MS Excel 4.0";
113 static const sal_Char __FAR_DATA pFilterEx4Temp[]	= "MS Excel 4.0 Vorlage/Template";
114 static const sal_Char __FAR_DATA pFilterExcel5[]	= "MS Excel 5.0/95";
115 static const sal_Char __FAR_DATA pFilterEx5Temp[]	= "MS Excel 5.0/95 Vorlage/Template";
116 static const sal_Char __FAR_DATA pFilterExcel95[]	= "MS Excel 95";
117 static const sal_Char __FAR_DATA pFilterEx95Temp[]	= "MS Excel 95 Vorlage/Template";
118 static const sal_Char __FAR_DATA pFilterExcel97[]	= "MS Excel 97";
119 static const sal_Char __FAR_DATA pFilterEx97Temp[]	= "MS Excel 97 Vorlage/Template";
120 static const sal_Char __FAR_DATA pFilterExcelXML[]  = "MS Excel 2003 XML";
121 static const sal_Char __FAR_DATA pFilterDBase[]		= "dBase";
122 static const sal_Char __FAR_DATA pFilterDif[]		= "DIF";
123 static const sal_Char __FAR_DATA pFilterSylk[]		= "SYLK";
124 static const sal_Char __FAR_DATA pFilterHtml[]		= "HTML (StarCalc)";
125 static const sal_Char __FAR_DATA pFilterHtmlWeb[]	= "calc_HTML_WebQuery";
126 static const sal_Char __FAR_DATA pFilterRtf[]		= "Rich Text Format (StarCalc)";
127 
128 
129 static sal_Bool lcl_MayBeAscii( SvStream& rStream )
130 {
131     // ASCII/CSV is considered possible if there are no null bytes, or a Byte
132     // Order Mark is present, or if, for Unicode UCS2/UTF-16, all null bytes
133     // are on either even or uneven byte positions.
134 
135 	rStream.Seek(STREAM_SEEK_TO_BEGIN);
136 
137     const size_t nBufSize = 2048;
138 	sal_uInt16 aBuffer[ nBufSize ];
139     sal_uInt8* pByte = reinterpret_cast<sal_uInt8*>(aBuffer);
140 	sal_uLong nBytesRead = rStream.Read( pByte, nBufSize*2);
141 
142 	if ( nBytesRead >= 2 && (aBuffer[0] == 0xfffe || aBuffer[0] == 0xfeff) )
143 	{
144         // Unicode BOM file may contain null bytes.
145 		return sal_True;
146 	}
147 
148 	const sal_uInt16* p = aBuffer;
149     sal_uInt16 nMask = 0xffff;
150     nBytesRead /= 2;
151 	while( nBytesRead-- && nMask )
152     {
153         sal_uInt16 nVal = *p++ & nMask;
154         if (!(nVal & 0x00ff))
155             nMask &= 0xff00;
156         if (!(nVal & 0xff00))
157             nMask &= 0x00ff;
158     }
159 
160 	return nMask != 0;
161 }
162 
163 static const SfxFilter* lcl_DetectExcelXML( SvStream& rStream, SfxFilterMatcher& rMatcher )
164 {
165     const SfxFilter* pFound = NULL;
166     rStream.Seek(STREAM_SEEK_TO_BEGIN);
167 
168     const size_t nBufSize = 4000;
169     sal_uInt8 aBuffer[ nBufSize ];
170     sal_uLong nBytesRead = rStream.Read( aBuffer, nBufSize );
171     sal_uLong nXMLStart = 0;
172 
173     // Skip UTF-8 BOM if present.
174     // No need to handle UTF-16 etc (also rejected in XMLFilterDetect).
175     if ( nBytesRead >= 3 && aBuffer[0] == 0xEF && aBuffer[1] == 0xBB && aBuffer[2] == 0xBF )
176         nXMLStart = 3;
177 
178     if ( nBytesRead >= nXMLStart + 5 && rtl_compareMemory( aBuffer+nXMLStart, "<?xml", 5 ) == 0 )
179     {
180         // Be consistent with XMLFilterDetect service: Check for presence of "Workbook" in XML file.
181 
182         rtl::OString aTryStr( "Workbook" );
183         rtl::OString aFileString(reinterpret_cast<const sal_Char*>(aBuffer), nBytesRead);
184 
185         if (aFileString.indexOf(aTryStr) >= 0)
186             pFound = rMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcelXML) );
187     }
188 
189     return pFound;
190 }
191 
192 static sal_Bool lcl_MayBeDBase( SvStream& rStream )
193 {
194     // Look for dbf marker, see connectivity/source/inc/dbase/DTable.hxx
195     // DBFType for values.
196     const sal_uInt8 nValidMarks[] = {
197         0x03, 0x04, 0x05, 0x30, 0x43, 0xB3, 0x83, 0x8b, 0x8e, 0xf5 };
198     sal_uInt8 nMark;
199     rStream.Seek(STREAM_SEEK_TO_BEGIN);
200     rStream >> nMark;
201     bool bValidMark = false;
202     for (size_t i=0; i < sizeof(nValidMarks)/sizeof(nValidMarks[0]) && !bValidMark; ++i)
203     {
204         if (nValidMarks[i] == nMark)
205             bValidMark = true;
206     }
207     if ( !bValidMark )
208         return sal_False;
209 
210     const size_t nHeaderBlockSize = 32;
211     // Empty dbf is >= 32*2+1 bytes in size.
212     const size_t nEmptyDbf = nHeaderBlockSize * 2 + 1;
213 
214 	rStream.Seek(STREAM_SEEK_TO_END);
215 	sal_uLong nSize = rStream.Tell();
216 	if ( nSize < nEmptyDbf )
217 		return sal_False;
218 
219 	// length of header starts at 8
220 	rStream.Seek(8);
221 	sal_uInt16 nHeaderLen;
222 	rStream >> nHeaderLen;
223 
224 	if ( nHeaderLen < nEmptyDbf || nSize < nHeaderLen )
225 		return sal_False;
226 
227     // Last byte of header must be 0x0d, this is how it's specified.
228     // #i9581#,#i26407# but some applications don't follow the specification
229     // and pad the header with one byte 0x00 to reach an
230     // even boundary. Some (#i88577# ) even pad more or pad using a 0x1a ^Z
231     // control character (#i8857#). This results in:
232     // Last byte of header must be 0x0d on 32 bytes boundary.
233 	sal_uInt16 nBlocks = (nHeaderLen - 1) / nHeaderBlockSize;
234 	sal_uInt8 nEndFlag = 0;
235 	while ( nBlocks > 1 && nEndFlag != 0x0d ) {
236 		rStream.Seek( nBlocks-- * nHeaderBlockSize );
237 		rStream >> nEndFlag;
238 	}
239 
240 	return ( 0x0d == nEndFlag );
241 }
242 
243 #if 0
244 static sal_Bool lcl_IsAnyXMLFilter( const SfxFilter* pFilter )
245 {
246 	if ( !pFilter )
247 		return sal_False;
248 
249 	//	sal_True for XML file or template
250 	//	(template filter has no internal name -> allow configuration key names)
251 
252 	String aName(pFilter->GetFilterName());
253 	return aName.EqualsAscii(pFilterXML) ||
254 		   aName.EqualsAscii("calc_StarOffice_XML_Calc") ||
255 		   aName.EqualsAscii("calc_StarOffice_XML_Calc_Template");
256 }
257 #endif
258 
259 ::rtl::OUString SAL_CALL ScFilterDetect::detect( ::com::sun::star::uno::Sequence< ::com::sun::star::beans::PropertyValue >& lDescriptor ) throw( ::com::sun::star::uno::RuntimeException )
260 {
261     REFERENCE< XInputStream > xStream;
262     REFERENCE< XContent > xContent;
263     REFERENCE< XInteractionHandler > xInteraction;
264     String aURL;
265 	::rtl::OUString sTemp;
266     String aTypeName;            // a name describing the type (from MediaDescriptor, usually from flat detection)
267     String aPreselectedFilterName;      // a name describing the filter to use (from MediaDescriptor, usually from UI action)
268 
269 	::rtl::OUString aDocumentTitle; // interesting only if set in this method
270 
271 	// opening as template is done when a parameter tells to do so and a template filter can be detected
272     // (otherwise no valid filter would be found) or if the detected filter is a template filter and
273 	// there is no parameter that forbids to open as template
274 	sal_Bool bOpenAsTemplate = sal_False;
275     sal_Bool bWasReadOnly = sal_False, bReadOnly = sal_False;
276 
277 	sal_Bool bRepairPackage = sal_False;
278 	sal_Bool bRepairAllowed = sal_False;
279 
280 	// now some parameters that can already be in the array, but may be overwritten or new inserted here
281 	// remember their indices in the case new values must be added to the array
282 	sal_Int32 nPropertyCount = lDescriptor.getLength();
283     sal_Int32 nIndexOfFilterName = -1;
284     sal_Int32 nIndexOfInputStream = -1;
285     sal_Int32 nIndexOfContent = -1;
286     sal_Int32 nIndexOfReadOnlyFlag = -1;
287     sal_Int32 nIndexOfTemplateFlag = -1;
288     sal_Int32 nIndexOfDocumentTitle = -1;
289     bool bFakeXLS = false;
290 
291     for( sal_Int32 nProperty=0; nProperty<nPropertyCount; ++nProperty )
292 	{
293         // extract properties
294         if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("URL")) )
295 		{
296 			lDescriptor[nProperty].Value >>= sTemp;
297 			aURL = sTemp;
298 		}
299         else if( !aURL.Len() && lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("FileName")) )
300 		{
301 			lDescriptor[nProperty].Value >>= sTemp;
302 			aURL = sTemp;
303 		}
304         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("TypeName")) )
305 		{
306 			lDescriptor[nProperty].Value >>= sTemp;
307             aTypeName = sTemp;
308 		}
309         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("FilterName")) )
310 		{
311 			lDescriptor[nProperty].Value >>= sTemp;
312             aPreselectedFilterName = sTemp;
313 
314             // if the preselected filter name is not correct, it must be erased after detection
315             // remember index of property to get access to it later
316             nIndexOfFilterName = nProperty;
317 		}
318         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("InputStream")) )
319             nIndexOfInputStream = nProperty;
320         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("ReadOnly")) )
321             nIndexOfReadOnlyFlag = nProperty;
322         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("UCBContent")) )
323             nIndexOfContent = nProperty;
324         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("AsTemplate")) )
325 		{
326 			lDescriptor[nProperty].Value >>= bOpenAsTemplate;
327             nIndexOfTemplateFlag = nProperty;
328 		}
329         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("InteractionHandler")) )
330             lDescriptor[nProperty].Value >>= xInteraction;
331         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("RepairPackage")) )
332             lDescriptor[nProperty].Value >>= bRepairPackage;
333         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("DocumentTitle")) )
334             nIndexOfDocumentTitle = nProperty;
335 	}
336 
337     // can't check the type for external filters, so set the "dont" flag accordingly
338     ::vos::OGuard aGuard( Application::GetSolarMutex() );
339     //SfxFilterFlags nMust = SFX_FILTER_IMPORT, nDont = SFX_FILTER_NOTINSTALLED;
340 
341     SfxAllItemSet *pSet = new SfxAllItemSet( SFX_APP()->GetPool() );
342     TransformParameters( SID_OPENDOC, lDescriptor, *pSet );
343     SFX_ITEMSET_ARG( pSet, pItem, SfxBoolItem, SID_DOC_READONLY, sal_False );
344 
345     bWasReadOnly = pItem && pItem->GetValue();
346 
347 	const SfxFilter* pFilter = 0;
348 	String aPrefix = String::CreateFromAscii( "private:factory/" );
349 	if( aURL.Match( aPrefix ) == aPrefix.Len() )
350 	{
351 		String aPattern( aPrefix );
352 		aPattern += String::CreateFromAscii("scalc");
353 		if ( aURL.Match( aPattern ) >= aPattern.Len() )
354 			pFilter = SfxFilter::GetDefaultFilterFromFactory( aURL );
355 	}
356 	else
357 	{
358 		// container for Calc filters
359 		SfxFilterMatcher aMatcher( String::CreateFromAscii("scalc") );
360 		if ( aPreselectedFilterName.Len() )
361 			pFilter = SfxFilter::GetFilterByName( aPreselectedFilterName );
362 		else if( aTypeName.Len() )
363 			pFilter = aMatcher.GetFilter4EA( aTypeName );
364 
365 	    // ctor of SfxMedium uses owner transition of ItemSet
366 	    SfxMedium aMedium( aURL, bWasReadOnly ? STREAM_STD_READ : STREAM_STD_READWRITE, sal_False, NULL, pSet );
367 	    aMedium.UseInteractionHandler( sal_True );
368 
369 	    sal_Bool bIsStorage = aMedium.IsStorage();
370 	    if ( aMedium.GetErrorCode() == ERRCODE_NONE )
371 	    {
372 	        // remember input stream and content and put them into the descriptor later
373 			// should be done here since later the medium can switch to a version
374 	        xStream.set(aMedium.GetInputStream());
375 			xContent.set(aMedium.GetContent());
376         	bReadOnly = aMedium.IsReadOnly();
377 
378 	        // maybe that IsStorage() already created an error!
379 	        if ( bIsStorage )
380 			{
381                 uno::Reference < embed::XStorage > xStorage(aMedium.GetStorage( sal_False ));
382 				if ( aMedium.GetLastStorageCreationState() != ERRCODE_NONE )
383 				{
384 					// error during storage creation means _here_ that the medium
385 					// is broken, but we can not handle it in medium since unpossibility
386 					// to create a storage does not _always_ means that the medium is broken
387 					aMedium.SetError( aMedium.GetLastStorageCreationState(), ::rtl::OUString( RTL_CONSTASCII_USTRINGPARAM( OSL_LOG_PREFIX ) ) );
388 					if ( xInteraction.is() )
389 					{
390 						OUString empty;
391 						try
392 						{
393 							InteractiveAppException xException( empty,
394 															REFERENCE< XInterface >(),
395 															InteractionClassification_ERROR,
396 															aMedium.GetError() );
397 
398                             REFERENCE< XInteractionRequest > xRequest(
399 								new ucbhelper::SimpleInteractionRequest( makeAny( xException ),
400 																 	 ucbhelper::CONTINUATION_APPROVE ) );
401 							xInteraction->handle( xRequest );
402 						}
403 						catch ( Exception & ) {};
404 					}
405 				}
406                 else if ( xStorage.is() )
407 				{
408 					try
409 					{
410                         String aFilterName;
411                         if ( pFilter )
412                             aFilterName = pFilter->GetName();
413                         aTypeName = SfxFilter::GetTypeFromStorage( xStorage, pFilter ? pFilter->IsOwnTemplateFormat() : sal_False, &aFilterName );
414 					}
415 					catch( lang::WrappedTargetException& aWrap )
416 					{
417 						packages::zip::ZipIOException aZipException;
418 
419 						// repairing is done only if this type is requested from outside
420 						if ( ( aWrap.TargetException >>= aZipException ) && aTypeName.Len() )
421 						{
422 							if ( xInteraction.is() )
423 							{
424 								// the package is broken one
425        							aDocumentTitle = aMedium.GetURLObject().getName(
426 															INetURLObject::LAST_SEGMENT,
427 															true,
428 															INetURLObject::DECODE_WITH_CHARSET );
429 
430 								if ( !bRepairPackage )
431 								{
432                                     // ask the user whether he wants to try to repair
433                                     RequestPackageReparation aRequest( aDocumentTitle );
434                                     xInteraction->handle( aRequest.GetRequest() );
435                                     bRepairAllowed = aRequest.isApproved();
436 								}
437 
438 								if ( !bRepairAllowed )
439 								{
440 									// repair either not allowed or not successful
441                                     NotifyBrokenPackage aNotifyRequest( aDocumentTitle );
442                                     xInteraction->handle( aNotifyRequest.GetRequest() );
443 								}
444 							}
445 
446 							if ( !bRepairAllowed )
447 								aTypeName.Erase();
448 						}
449 					}
450 					catch( uno::RuntimeException& )
451 					{
452 						throw;
453 					}
454 					catch( uno::Exception& )
455 					{
456 						aTypeName.Erase();
457 					}
458 
459                    	if ( aTypeName.Len() )
460                        	pFilter = SfxFilterMatcher( String::CreateFromAscii("scalc") ).GetFilter4EA( aTypeName );
461 
462 				}
463 			}
464 			else
465 			{
466                 bool bIsXLS = false;
467                 SvStream* pStream = aMedium.GetInStream();
468                 const SfxFilter* pPreselectedFilter = pFilter;
469                 if ( pPreselectedFilter && pPreselectedFilter->GetName().SearchAscii("Excel") != STRING_NOTFOUND )
470                     bIsXLS = true;
471                 pFilter = 0;
472                 if ( pStream )
473                 {
474                     SotStorageRef aStorage = new SotStorage ( pStream, sal_False );
475                     if ( !aStorage->GetError() )
476                     {
477                         // Excel-5: detect through contained streams
478                         // there are some "excel" formats from 3rd party vendors that need to be distinguished
479                         String aStreamName(RTL_CONSTASCII_STRINGPARAM("Workbook"));
480                         sal_Bool bExcel97Stream = ( aStorage->IsStream( aStreamName ) );
481 
482                         aStreamName = String(RTL_CONSTASCII_STRINGPARAM("Book"));
483                         sal_Bool bExcel5Stream = ( aStorage->IsStream( aStreamName ) );
484                         if ( bExcel97Stream || bExcel5Stream )
485                         {
486                             if ( bExcel97Stream )
487                             {
488                                 String aOldName;
489                                 sal_Bool bIsCalcFilter = sal_True;
490                                 if ( pPreselectedFilter )
491                                 {
492                                     // cross filter; now this should be a type detection only, not a filter detection
493                                     // we can simulate it by preserving the preselected filter if the type matches
494                                     // example: Excel filters for Writer
495                                     aOldName = pPreselectedFilter->GetFilterName();
496                                     bIsCalcFilter = pPreselectedFilter->GetServiceName().EqualsAscii("com.sun.star.sheet.SpreadsheetDocument");
497                                 }
498 
499                                 if ( aOldName.EqualsAscii(pFilterEx97Temp) || !bIsCalcFilter )
500                                 {
501                                     //  Excel 97 template selected -> keep selection
502                                 }
503                                 else if ( bExcel5Stream &&
504                                             ( aOldName.EqualsAscii(pFilterExcel5) || aOldName.EqualsAscii(pFilterEx5Temp) ||
505                                             aOldName.EqualsAscii(pFilterExcel95) || aOldName.EqualsAscii(pFilterEx95Temp) ) )
506                                 {
507                                     //  dual format file and Excel 5 selected -> keep selection
508                                 }
509                                 else
510                                 {
511                                     //  else use Excel 97 filter
512                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcel97) );
513                                 }
514                             }
515                             else if ( bExcel5Stream )
516                             {
517                                 String aOldName;
518                                 sal_Bool bIsCalcFilter = sal_True;
519                                 if ( pPreselectedFilter )
520                                 {
521                                     // cross filter; now this should be a type detection only, not a filter detection
522                                     // we can simulate it by preserving the preselected filter if the type matches
523                                     // example: Excel filters for Writer
524                                     aOldName = pPreselectedFilter->GetFilterName();
525                                     bIsCalcFilter = pPreselectedFilter->GetServiceName().EqualsAscii("com.sun.star.sheet.SpreadsheetDocument");
526                                 }
527 
528                                 if ( aOldName.EqualsAscii(pFilterExcel95) || aOldName.EqualsAscii(pFilterEx95Temp) ||
529                                         aOldName.EqualsAscii(pFilterEx5Temp) || !bIsCalcFilter )
530                                 {
531                                     //  Excel 95 oder Vorlage (5 oder 95) eingestellt -> auch gut
532                                 }
533                                 else if ( aOldName.EqualsAscii(pFilterEx97Temp) )
534                                 {
535                                     // #101923# auto detection has found template -> return Excel5 template
536                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterEx5Temp) );
537                                 }
538                                 else
539                                 {
540                                     //  sonst wird als Excel 5-Datei erkannt
541                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcel5) );
542                                 }
543                             }
544                         }
545                     }
546                     else
547                     {
548                         SvStream &rStr = *pStream;
549 
550                         // Tabelle mit Suchmustern
551                         // Bedeutung der Sequenzen
552                         // 0x00??: genau Byte 0x?? muss an dieser Stelle stehen
553                         // 0x0100: ein Byte ueberlesen (don't care)
554                         // 0x02nn: ein Byte aus 0xnn Alternativen folgt
555                         // 0x8000: Erkennung abgeschlossen
556                         //
557 
558         #define M_DC        0x0100
559         #define M_ALT(ANZ)  (0x0200+(ANZ))
560         #define M_ENDE      0x8000
561 
562                         static const sal_uInt16 pLotus[] =      // Lotus 1/1A/2
563                             { 0x0000, 0x0000, 0x0002, 0x0000,
564                             M_ALT(2), 0x0004, 0x0006,
565                             0x0004, M_ENDE };
566 
567                         static const sal_uInt16 pLotusNew[] =   // Lotus >= 9.7
568                             { 0x0000, 0x0000, M_DC, 0x0000,     // Rec# + Len (0x1a)
569                               M_ALT(3), 0x0003, 0x0004, 0x0005, // File Revision Code 97->ME
570                               0x0010, 0x0004, 0x0000, 0x0000,
571                               M_ENDE };
572 
573                         static const sal_uInt16 pExcel1[] =     // Excel BIFF2, BIFF3, BIFF4
574                             {   0x09,                                   // lobyte of BOF rec ID (0x0009, 0x0209, 0x0409)
575                                 M_ALT(3), 0x00, 0x02, 0x04,             // hibyte of BOF rec ID (0x0009, 0x0209, 0x0409)
576                                 M_ALT(3), 4, 6, 8,                      // lobyte of BOF rec size (4, 6, 8, 16)
577                                 0x00,                                   // hibyte of BOF rec size (4, 6, 8, 16)
578                                 M_DC, M_DC,                             // any version
579                                 M_ALT(3), 0x10, 0x20, 0x40,             // lobyte of data type (0x0010, 0x0020, 0x0040)
580                                 0x00,                                   // hibyte of data type (0x0010, 0x0020, 0x0040)
581                                 M_ENDE };
582 
583                         static const sal_uInt16 pExcel2[] =     // Excel BIFF4 Workspace
584                             {   0x09,                                   // lobyte of BOF rec ID (0x0409)
585                                 0x04,                                   // hibyte of BOF rec ID (0x0409)
586                                 M_ALT(3), 4, 6, 8,                      // lobyte of BOF rec size (4, 6, 8, 16)
587                                 0x00,                                   // hibyte of BOF rec size (4, 6, 8, 16)
588                                 M_DC, M_DC,                             // any version
589                                 0x00,                                   // lobyte of data type (0x0100)
590                                 0x01,                                   // hibyte of data type (0x0100)
591                                 M_ENDE };
592 
593                         static const sal_uInt16 pExcel3[] =     // #i23425# Excel BIFF5, BIFF7, BIFF8 (simple book stream)
594                             {   0x09,                                   // lobyte of BOF rec ID (0x0809)
595                                 0x08,                                   // hibyte of BOF rec ID (0x0809)
596                                 M_ALT(4), 4, 6, 8, 16,                  // lobyte of BOF rec size
597                                 0x00,                                   // hibyte of BOF rec size
598                                 M_DC, M_DC,                             // any version
599                                 M_ALT(5), 0x05, 0x06, 0x10, 0x20, 0x40, // lobyte of data type
600                                 0x00,                                   // hibyte of data type
601                                 M_ENDE };
602 
603                         static const sal_uInt16 pSc10[] =       // StarCalc 1.0 Dokumente
604                             { 'B', 'l', 'a', 'i', 's', 'e', '-', 'T', 'a', 'b', 'e', 'l', 'l',
605                             'e', 0x000A, 0x000D, 0x0000,    // Sc10CopyRight[16]
606                             M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC,
607                             M_DC, M_DC,                   // Sc10CopyRight[29]
608                             M_ALT(2), 0x0065, 0x0066,     // Versionsnummer 101 oder 102
609                             0x0000,
610                             M_ENDE };
611 
612                         static const sal_uInt16 pLotus2[] =     // Lotus >3
613                             { 0x0000, 0x0000, 0x001A, 0x0000,   // Rec# + Len (26)
614                             M_ALT(2), 0x0000, 0x0002,         // File Revision Code
615                             0x0010,
616                             0x0004, 0x0000,                   // File Revision Subcode
617                             M_ENDE };
618 
619                         static const sal_uInt16 pQPro[] =
620                                { 0x0000, 0x0000, 0x0002, 0x0000,
621                                  M_ALT(4), 0x0001, 0x0002, // WB1, WB2
622                                  0x0006, 0x0007,           // QPro 6/7 (?)
623                                  0x0010,
624                                  M_ENDE };
625 
626                         static const sal_uInt16 pDIF1[] =       // DIF mit CR-LF
627                             {
628                             'T', 'A', 'B', 'L', 'E',
629                             M_DC, M_DC,
630                             '0', ',', '1',
631                             M_DC, M_DC,
632                             '\"',
633                             M_ENDE };
634 
635                         static const sal_uInt16 pDIF2[] =       // DIF mit CR oder LF
636                             {
637                             'T', 'A', 'B', 'L', 'E',
638                             M_DC,
639                             '0', ',', '1',
640                             M_DC,
641                             '\"',
642                             M_ENDE };
643 
644                         static const sal_uInt16 pSylk[] =       // Sylk
645                             {
646                             'I', 'D', ';',
647                             M_ALT(3), 'P', 'N', 'E',        // 'P' plus undocumented Excel extensions 'N' and 'E'
648                             M_ENDE };
649 
650                         static const sal_uInt16 *ppFilterPatterns[] =      // Arrays mit Suchmustern
651                             {
652                             pLotus,
653                             pExcel1,
654                             pExcel2,
655                             pExcel3,
656                             pSc10,
657                             pDIF1,
658                             pDIF2,
659                             pSylk,
660                             pLotusNew,
661                             pLotus2,
662                             pQPro
663                             };
664                         const sal_uInt16 nFilterCount = sizeof(ppFilterPatterns) / sizeof(ppFilterPatterns[0]);
665 
666                         static const sal_Char* const pFilterName[] =     // zugehoerige Filter
667                             {
668                             pFilterLotus,
669                             pFilterExcel4,
670                             pFilterExcel4,
671                             pFilterExcel4,
672                             pFilterSc10,
673                             pFilterDif,
674                             pFilterDif,
675                             pFilterSylk,
676                             pFilterLotus,
677                             pFilterLotus,
678                             pFilterQPro6
679                             };
680 
681                         // const sal_uInt16 nByteMask = 0xFF;
682 
683                         // suchen Sie jetzt!
684                         // ... realisiert ueber 'Mustererkennung'
685 
686                         sal_uInt8            nAkt;
687                         sal_Bool            bSync;          // Datei und Muster stimmen ueberein
688                         sal_uInt16          nFilter;        // Zaehler ueber alle Filter
689                         const sal_uInt16    *pSearch;       // aktuelles Musterwort
690 
691                         for ( nFilter = 0 ; nFilter < nFilterCount ; nFilter++ )
692                         {
693                             rStr.Seek( 0 ); // am Anfang war alles Uebel...
694                             rStr >> nAkt;
695                             pSearch = ppFilterPatterns[ nFilter ];
696                             bSync = sal_True;
697                             while( !rStr.IsEof() && bSync )
698                             {
699                                 register sal_uInt16 nMuster = *pSearch;
700 
701                                 if( nMuster < 0x0100 )
702                                 { //                                direkter Byte-Vergleich
703                                     if( ( sal_uInt8 ) nMuster != nAkt )
704                                         bSync = sal_False;
705                                 }
706                                 else if( nMuster & M_DC )
707                                 { //                                             don't care
708                                 }
709                                 else if( nMuster & M_ALT(0) )
710                                 { //                                      alternative Bytes
711                                     sal_uInt8 nAnzAlt = ( sal_uInt8 ) nMuster;
712                                     bSync = sal_False;          // zunaechst unsynchron
713                                     while( nAnzAlt > 0 )
714                                     {
715                                         pSearch++;
716                                         if( ( sal_uInt8 ) *pSearch == nAkt )
717                                             bSync = sal_True;   // jetzt erst Synchronisierung
718                                         nAnzAlt--;
719                                     }
720                                 }
721                                 else if( nMuster & M_ENDE )
722                                 { //                                        Format detected
723                                     if ( pFilterName[nFilter] == pFilterExcel4 && pPreselectedFilter &&
724                                         ( (pPreselectedFilter)->GetFilterName().EqualsAscii(pFilterEx4Temp) || pPreselectedFilter->GetTypeName().EqualsAscii("calc_MS_Excel_40") ) )
725                                     {
726                                         //  Excel 4 erkannt, Excel 4 Vorlage eingestellt -> auch gut
727                                         // oder Excel 4 Filter anderer Applikation (simulated type detection!)
728                                     }
729                                     else
730                                     {   // gefundenen Filter einstellen
731                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterName[ nFilter ]) );
732                                     }
733                                     bSync = sal_False;              // leave inner loop
734                                     nFilter = nFilterCount;     // leave outer loop
735                                 }
736                                 else
737                                 { //                                         Tabellenfehler
738                                     DBG_ERROR( "-ScApplication::DetectFilter(): Fehler in Mustertabelle");
739                                 }
740 
741                                 pSearch++;
742                                 rStr >> nAkt;
743                             }
744                         }
745 
746                         if ( pPreselectedFilter && !pFilter )
747                         {
748                             // further checks for filters only if they are preselected: ASCII, HTML, RTF, DBase
749                             // without the preselection other filters (Writer) take precedence
750                             // DBase can't be detected reliably, so it also needs preselection
751                             bool bMaybeText = lcl_MayBeAscii( rStr );
752                             if ( pPreselectedFilter->GetFilterName().EqualsAscii(pFilterAscii) && bMaybeText )
753                             {
754                                 // Text filter is accepted if preselected
755                                 pFilter = pPreselectedFilter;
756                             }
757                             else
758                             {
759                                 // get file header
760                                 rStr.Seek( 0 );
761                                 const int nTrySize = 80;
762                                 ByteString aHeader;
763                                 for ( int j = 0; j < nTrySize && !rStr.IsEof(); j++ )
764                                 {
765                                     sal_Char c;
766                                     rStr >> c;
767                                     aHeader += c;
768                                 }
769                                 aHeader += '\0';
770 
771                                 if ( HTMLParser::IsHTMLFormat( aHeader.GetBuffer() ) )
772                                 {
773                                     // test for HTML
774                                     if ( pPreselectedFilter->GetName().EqualsAscii(pFilterHtml) )
775                                     {
776                                         pFilter = pPreselectedFilter;
777                                     }
778                                     else
779                                     {
780                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterHtmlWeb) );
781                                         if ( bIsXLS )
782                                             bFakeXLS = true;
783                                     }
784                                 }
785                                 else if ( bIsXLS && bMaybeText )
786                                 {
787                                     // Detect Excel 2003 XML here only if XLS was preselected.
788                                     // The configured detection for Excel 2003 XML is still in XMLFilterDetect.
789                                     pFilter = lcl_DetectExcelXML( rStr, aMatcher );
790                                     if (!pFilter)
791                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterAscii) );
792                                     bFakeXLS = true;
793                                 }
794                                 else if ( aHeader.CompareTo( "{\\rtf", 5 ) == COMPARE_EQUAL )
795                                 {
796                                     // test for RTF
797                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterRtf) );
798                                 }
799                                 else if ( pPreselectedFilter->GetName().EqualsAscii(pFilterDBase) && lcl_MayBeDBase( rStr ) )
800                                     pFilter = pPreselectedFilter;
801                             }
802                         }
803                     }
804 				}
805 			}
806 		}
807 	}
808 
809     if ( nIndexOfInputStream == -1 && xStream.is() )
810     {
811         // if input stream wasn't part of the descriptor, now it should be, otherwise the content would be opend twice
812         lDescriptor.realloc( nPropertyCount + 1 );
813         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("InputStream");
814         lDescriptor[nPropertyCount].Value <<= xStream;
815         nPropertyCount++;
816     }
817 
818     if ( nIndexOfContent == -1 && xContent.is() )
819     {
820         // if input stream wasn't part of the descriptor, now it should be, otherwise the content would be opend twice
821         lDescriptor.realloc( nPropertyCount + 1 );
822         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("UCBContent");
823         lDescriptor[nPropertyCount].Value <<= xContent;
824         nPropertyCount++;
825     }
826 
827     if ( bReadOnly != bWasReadOnly )
828     {
829         if ( nIndexOfReadOnlyFlag == -1 )
830         {
831             lDescriptor.realloc( nPropertyCount + 1 );
832             lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("ReadOnly");
833             lDescriptor[nPropertyCount].Value <<= bReadOnly;
834             nPropertyCount++;
835         }
836         else
837             lDescriptor[nIndexOfReadOnlyFlag].Value <<= bReadOnly;
838     }
839 
840 	if ( !bRepairPackage && bRepairAllowed )
841 	{
842         lDescriptor.realloc( nPropertyCount + 1 );
843         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("RepairPackage");
844         lDescriptor[nPropertyCount].Value <<= bRepairAllowed;
845         nPropertyCount++;
846 
847 		bOpenAsTemplate = sal_True;
848 
849 		// TODO/LATER: set progress bar that should be used
850 	}
851 
852 	if ( bOpenAsTemplate )
853 	{
854 		if ( nIndexOfTemplateFlag == -1 )
855 		{
856         	lDescriptor.realloc( nPropertyCount + 1 );
857         	lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("AsTemplate");
858         	lDescriptor[nPropertyCount].Value <<= bOpenAsTemplate;
859         	nPropertyCount++;
860 		}
861 		else
862         	lDescriptor[nIndexOfTemplateFlag].Value <<= bOpenAsTemplate;
863 	}
864 
865 	if ( aDocumentTitle.getLength() )
866 	{
867 		// the title was set here
868 		if ( nIndexOfDocumentTitle == -1 )
869 		{
870         	lDescriptor.realloc( nPropertyCount + 1 );
871         	lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("DocumentTitle");
872         	lDescriptor[nPropertyCount].Value <<= aDocumentTitle;
873         	nPropertyCount++;
874 		}
875 		else
876         	lDescriptor[nIndexOfDocumentTitle].Value <<= aDocumentTitle;
877 	}
878 
879     if ( bFakeXLS )
880     {
881         if ( nIndexOfFilterName == -1 )
882         {
883             lDescriptor.realloc( nPropertyCount + 1 );
884             lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("FilterName");
885             lDescriptor[nPropertyCount].Value <<= rtl::OUString(pFilter->GetName());
886             nPropertyCount++;
887         }
888         else
889             lDescriptor[nIndexOfFilterName].Value <<= rtl::OUString(pFilter->GetName());
890     }
891 
892     if ( pFilter )
893         aTypeName = pFilter->GetTypeName();
894     else
895         aTypeName.Erase();
896     return aTypeName;
897 }
898 
899 SFX_IMPL_SINGLEFACTORY( ScFilterDetect )
900 
901 /* XServiceInfo */
902 UNOOUSTRING SAL_CALL ScFilterDetect::getImplementationName() throw( UNORUNTIMEEXCEPTION )
903 {
904     return impl_getStaticImplementationName();
905 }
906                                                                                                                                 \
907 /* XServiceInfo */
908 sal_Bool SAL_CALL ScFilterDetect::supportsService( const UNOOUSTRING& sServiceName ) throw( UNORUNTIMEEXCEPTION )
909 {
910     UNOSEQUENCE< UNOOUSTRING >  seqServiceNames(getSupportedServiceNames());
911     const UNOOUSTRING*          pArray          =   seqServiceNames.getConstArray();
912     for ( sal_Int32 nCounter=0; nCounter<seqServiceNames.getLength(); nCounter++ )
913     {
914         if ( pArray[nCounter] == sServiceName )
915         {
916             return sal_True ;
917         }
918     }
919     return sal_False ;
920 }
921 
922 /* XServiceInfo */
923 UNOSEQUENCE< UNOOUSTRING > SAL_CALL ScFilterDetect::getSupportedServiceNames() throw( UNORUNTIMEEXCEPTION )
924 {
925     return impl_getStaticSupportedServiceNames();
926 }
927 
928 /* Helper for XServiceInfo */
929 UNOSEQUENCE< UNOOUSTRING > ScFilterDetect::impl_getStaticSupportedServiceNames()
930 {
931     UNOMUTEXGUARD aGuard( UNOMUTEX::getGlobalMutex() );
932     UNOSEQUENCE< UNOOUSTRING > seqServiceNames( 1 );
933     seqServiceNames.getArray() [0] = UNOOUSTRING::createFromAscii( "com.sun.star.frame.ExtendedTypeDetection"  );
934     return seqServiceNames ;
935 }
936 
937 /* Helper for XServiceInfo */
938 UNOOUSTRING ScFilterDetect::impl_getStaticImplementationName()
939 {
940     return UNOOUSTRING::createFromAscii( "com.sun.star.comp.calc.FormatDetector" );
941 }
942 
943 /* Helper for registry */
944 UNOREFERENCE< UNOXINTERFACE > SAL_CALL ScFilterDetect::impl_createInstance( const UNOREFERENCE< UNOXMULTISERVICEFACTORY >& xServiceManager ) throw( UNOEXCEPTION )
945 {
946     return UNOREFERENCE< UNOXINTERFACE >( *new ScFilterDetect( xServiceManager ) );
947 }
948 
949