xref: /aoo41x/main/sc/source/ui/unoobj/scdetect.cxx (revision b3f79822)
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_sc.hxx"
26 
27 #include "scdetect.hxx"
28 
29 #include <framework/interaction.hxx>
30 #include <com/sun/star/lang/XMultiServiceFactory.hpp>
31 #include <com/sun/star/beans/PropertyValue.hpp>
32 #include <com/sun/star/frame/XFrame.hpp>
33 #include <com/sun/star/frame/XModel.hpp>
34 #include <com/sun/star/awt/XWindow.hpp>
35 #include <com/sun/star/lang/XUnoTunnel.hpp>
36 #ifndef _UNOTOOLS_PROCESSFACTORY_HXX
37 #include <comphelper/processfactory.hxx>
38 #endif
39 #include <com/sun/star/beans/PropertyValue.hpp>
40 #include <com/sun/star/container/XNameAccess.hpp>
41 #include <com/sun/star/io/XInputStream.hpp>
42 #include <com/sun/star/task/XInteractionHandler.hpp>
43 #include <com/sun/star/lang/WrappedTargetRuntimeException.hpp>
44 #include <com/sun/star/ucb/CommandAbortedException.hpp>
45 #include <com/sun/star/ucb/InteractiveAppException.hpp>
46 #include <com/sun/star/ucb/XContent.hpp>
47 #include <com/sun/star/packages/zip/ZipIOException.hpp>
48 
49 
50 #include <framework/interaction.hxx>
51 
52 #ifndef _TOOLKIT_UNOHLP_HXX
53 #include <toolkit/helper/vclunohelper.hxx>
54 #endif
55 #include <ucbhelper/simpleinteractionrequest.hxx>
56 
57 #include <svtools/parhtml.hxx>
58 #include <rtl/ustring.h>
59 #include <rtl/logfile.hxx>
60 #include <svl/itemset.hxx>
61 #include <vcl/window.hxx>
62 #include <svl/eitem.hxx>
63 #include <svl/stritem.hxx>
64 #include <tools/urlobj.hxx>
65 #include <vos/mutex.hxx>
66 #include <svtools/sfxecode.hxx>
67 #include <svtools/ehdl.hxx>
68 #include <sot/storinfo.hxx>
69 #include <vcl/svapp.hxx>
70 #include <sfx2/sfxsids.hrc>
71 #include <sfx2/request.hxx>
72 #include <sfx2/docfile.hxx>
73 #include <sfx2/docfilt.hxx>
74 #include <sfx2/fcontnr.hxx>
75 #include <sfx2/app.hxx>
76 #include <sfx2/brokenpackageint.hxx>
77 #include <sot/storage.hxx>
78 
79 using namespace ::com::sun::star;
80 using namespace ::com::sun::star::uno;
81 using namespace ::com::sun::star::io;
82 using namespace ::com::sun::star::frame;
83 using namespace ::com::sun::star::task;
84 using namespace ::com::sun::star::beans;
85 using namespace ::com::sun::star::lang;
86 using namespace ::com::sun::star::ucb;
87 using ::rtl::OUString;
88 
ScFilterDetect(const REFERENCE<::com::sun::star::lang::XMultiServiceFactory> &)89 ScFilterDetect::ScFilterDetect( const REFERENCE < ::com::sun::star::lang::XMultiServiceFactory >& /* xFactory */ )
90 {
91 }
92 
~ScFilterDetect()93 ScFilterDetect::~ScFilterDetect()
94 {
95 }
96 
97 static const sal_Char __FAR_DATA pFilterSc50[]		= "StarCalc 5.0";
98 static const sal_Char __FAR_DATA pFilterSc50Temp[]	= "StarCalc 5.0 Vorlage/Template";
99 static const sal_Char __FAR_DATA pFilterSc40[]		= "StarCalc 4.0";
100 static const sal_Char __FAR_DATA pFilterSc40Temp[]	= "StarCalc 4.0 Vorlage/Template";
101 static const sal_Char __FAR_DATA pFilterSc30[]		= "StarCalc 3.0";
102 static const sal_Char __FAR_DATA pFilterSc30Temp[]	= "StarCalc 3.0 Vorlage/Template";
103 static const sal_Char __FAR_DATA pFilterSc10[]		= "StarCalc 1.0";
104 static const sal_Char __FAR_DATA pFilterXML[]		= "StarOffice XML (Calc)";
105 static const sal_Char __FAR_DATA pFilterAscii[]		= "Text - txt - csv (StarCalc)";
106 static const sal_Char __FAR_DATA pFilterLotus[]		= "Lotus";
107 static const sal_Char __FAR_DATA pFilterQPro6[]		= "Quattro Pro 6.0";
108 static const sal_Char __FAR_DATA pFilterExcel4[]	= "MS Excel 4.0";
109 static const sal_Char __FAR_DATA pFilterEx4Temp[]	= "MS Excel 4.0 Vorlage/Template";
110 static const sal_Char __FAR_DATA pFilterExcel5[]	= "MS Excel 5.0/95";
111 static const sal_Char __FAR_DATA pFilterEx5Temp[]	= "MS Excel 5.0/95 Vorlage/Template";
112 static const sal_Char __FAR_DATA pFilterExcel95[]	= "MS Excel 95";
113 static const sal_Char __FAR_DATA pFilterEx95Temp[]	= "MS Excel 95 Vorlage/Template";
114 static const sal_Char __FAR_DATA pFilterExcel97[]	= "MS Excel 97";
115 static const sal_Char __FAR_DATA pFilterEx97Temp[]	= "MS Excel 97 Vorlage/Template";
116 static const sal_Char __FAR_DATA pFilterExcelXML[]  = "MS Excel 2003 XML";
117 static const sal_Char __FAR_DATA pFilterDBase[]		= "dBase";
118 static const sal_Char __FAR_DATA pFilterDif[]		= "DIF";
119 static const sal_Char __FAR_DATA pFilterSylk[]		= "SYLK";
120 static const sal_Char __FAR_DATA pFilterHtml[]		= "HTML (StarCalc)";
121 static const sal_Char __FAR_DATA pFilterHtmlWeb[]	= "calc_HTML_WebQuery";
122 static const sal_Char __FAR_DATA pFilterRtf[]		= "Rich Text Format (StarCalc)";
123 
124 
lcl_MayBeAscii(SvStream & rStream)125 static sal_Bool lcl_MayBeAscii( SvStream& rStream )
126 {
127     // ASCII/CSV is considered possible if there are no null bytes, or a Byte
128     // Order Mark is present, or if, for Unicode UCS2/UTF-16, all null bytes
129     // are on either even or uneven byte positions.
130 
131 	rStream.Seek(STREAM_SEEK_TO_BEGIN);
132 
133     const size_t nBufSize = 2048;
134 	sal_uInt16 aBuffer[ nBufSize ];
135     sal_uInt8* pByte = reinterpret_cast<sal_uInt8*>(aBuffer);
136 	sal_uLong nBytesRead = rStream.Read( pByte, nBufSize*2);
137 
138 	if ( nBytesRead >= 2 && (aBuffer[0] == 0xfffe || aBuffer[0] == 0xfeff) )
139 	{
140         // Unicode BOM file may contain null bytes.
141 		return sal_True;
142 	}
143 
144 	const sal_uInt16* p = aBuffer;
145     sal_uInt16 nMask = 0xffff;
146     nBytesRead /= 2;
147 	while( nBytesRead-- && nMask )
148     {
149         sal_uInt16 nVal = *p++ & nMask;
150         if (!(nVal & 0x00ff))
151             nMask &= 0xff00;
152         if (!(nVal & 0xff00))
153             nMask &= 0x00ff;
154     }
155 
156 	return nMask != 0;
157 }
158 
lcl_DetectExcelXML(SvStream & rStream,SfxFilterMatcher & rMatcher)159 static const SfxFilter* lcl_DetectExcelXML( SvStream& rStream, SfxFilterMatcher& rMatcher )
160 {
161     const SfxFilter* pFound = NULL;
162     rStream.Seek(STREAM_SEEK_TO_BEGIN);
163 
164     const size_t nBufSize = 4000;
165     sal_uInt8 aBuffer[ nBufSize ];
166     sal_uLong nBytesRead = rStream.Read( aBuffer, nBufSize );
167     sal_uLong nXMLStart = 0;
168 
169     // Skip UTF-8 BOM if present.
170     // No need to handle UTF-16 etc (also rejected in XMLFilterDetect).
171     if ( nBytesRead >= 3 && aBuffer[0] == 0xEF && aBuffer[1] == 0xBB && aBuffer[2] == 0xBF )
172         nXMLStart = 3;
173 
174     if ( nBytesRead >= nXMLStart + 5 && rtl_compareMemory( aBuffer+nXMLStart, "<?xml", 5 ) == 0 )
175     {
176         // Be consistent with XMLFilterDetect service: Check for presence of "Workbook" in XML file.
177 
178         rtl::OString aTryStr( "Workbook" );
179         rtl::OString aFileString(reinterpret_cast<const sal_Char*>(aBuffer), nBytesRead);
180 
181         if (aFileString.indexOf(aTryStr) >= 0)
182             pFound = rMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcelXML) );
183     }
184 
185     return pFound;
186 }
187 
lcl_MayBeDBase(SvStream & rStream)188 static sal_Bool lcl_MayBeDBase( SvStream& rStream )
189 {
190     // Look for dbf marker, see connectivity/source/inc/dbase/DTable.hxx
191     // DBFType for values.
192     const sal_uInt8 nValidMarks[] = {
193         0x03, 0x04, 0x05, 0x30, 0x43, 0xB3, 0x83, 0x8b, 0x8e, 0xf5 };
194     sal_uInt8 nMark;
195     rStream.Seek(STREAM_SEEK_TO_BEGIN);
196     rStream >> nMark;
197     bool bValidMark = false;
198     for (size_t i=0; i < sizeof(nValidMarks)/sizeof(nValidMarks[0]) && !bValidMark; ++i)
199     {
200         if (nValidMarks[i] == nMark)
201             bValidMark = true;
202     }
203     if ( !bValidMark )
204         return sal_False;
205 
206     const size_t nHeaderBlockSize = 32;
207     // Empty dbf is >= 32*2+1 bytes in size.
208     const size_t nEmptyDbf = nHeaderBlockSize * 2 + 1;
209 
210 	rStream.Seek(STREAM_SEEK_TO_END);
211 	sal_uLong nSize = rStream.Tell();
212 	if ( nSize < nEmptyDbf )
213 		return sal_False;
214 
215 	// length of header starts at 8
216 	rStream.Seek(8);
217 	sal_uInt16 nHeaderLen;
218 	rStream >> nHeaderLen;
219 
220 	if ( nHeaderLen < nEmptyDbf || nSize < nHeaderLen )
221 		return sal_False;
222 
223     // Last byte of header must be 0x0d, this is how it's specified.
224     // #i9581#,#i26407# but some applications don't follow the specification
225     // and pad the header with one byte 0x00 to reach an
226     // even boundary. Some (#i88577# ) even pad more or pad using a 0x1a ^Z
227     // control character (#i8857#). This results in:
228     // Last byte of header must be 0x0d on 32 bytes boundary.
229 	sal_uInt16 nBlocks = (nHeaderLen - 1) / nHeaderBlockSize;
230 	sal_uInt8 nEndFlag = 0;
231 	while ( nBlocks > 1 && nEndFlag != 0x0d ) {
232 		rStream.Seek( nBlocks-- * nHeaderBlockSize );
233 		rStream >> nEndFlag;
234 	}
235 
236 	return ( 0x0d == nEndFlag );
237 }
238 
239 #if 0
240 static sal_Bool lcl_IsAnyXMLFilter( const SfxFilter* pFilter )
241 {
242 	if ( !pFilter )
243 		return sal_False;
244 
245 	//	sal_True for XML file or template
246 	//	(template filter has no internal name -> allow configuration key names)
247 
248 	String aName(pFilter->GetFilterName());
249 	return aName.EqualsAscii(pFilterXML) ||
250 		   aName.EqualsAscii("calc_StarOffice_XML_Calc") ||
251 		   aName.EqualsAscii("calc_StarOffice_XML_Calc_Template");
252 }
253 #endif
254 
detect(::com::sun::star::uno::Sequence<::com::sun::star::beans::PropertyValue> & lDescriptor)255 ::rtl::OUString SAL_CALL ScFilterDetect::detect( ::com::sun::star::uno::Sequence< ::com::sun::star::beans::PropertyValue >& lDescriptor ) throw( ::com::sun::star::uno::RuntimeException )
256 {
257     REFERENCE< XInputStream > xStream;
258     REFERENCE< XContent > xContent;
259     REFERENCE< XInteractionHandler > xInteraction;
260     String aURL;
261 	::rtl::OUString sTemp;
262     String aTypeName;            // a name describing the type (from MediaDescriptor, usually from flat detection)
263     String aPreselectedFilterName;      // a name describing the filter to use (from MediaDescriptor, usually from UI action)
264 
265 	::rtl::OUString aDocumentTitle; // interesting only if set in this method
266 
267 	// opening as template is done when a parameter tells to do so and a template filter can be detected
268     // (otherwise no valid filter would be found) or if the detected filter is a template filter and
269 	// there is no parameter that forbids to open as template
270 	sal_Bool bOpenAsTemplate = sal_False;
271     sal_Bool bWasReadOnly = sal_False, bReadOnly = sal_False;
272 
273 	sal_Bool bRepairPackage = sal_False;
274 	sal_Bool bRepairAllowed = sal_False;
275 
276 	// now some parameters that can already be in the array, but may be overwritten or new inserted here
277 	// remember their indices in the case new values must be added to the array
278 	sal_Int32 nPropertyCount = lDescriptor.getLength();
279     sal_Int32 nIndexOfFilterName = -1;
280     sal_Int32 nIndexOfInputStream = -1;
281     sal_Int32 nIndexOfContent = -1;
282     sal_Int32 nIndexOfReadOnlyFlag = -1;
283     sal_Int32 nIndexOfTemplateFlag = -1;
284     sal_Int32 nIndexOfDocumentTitle = -1;
285     bool bFakeXLS = false;
286 
287     for( sal_Int32 nProperty=0; nProperty<nPropertyCount; ++nProperty )
288 	{
289         // extract properties
290         if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("URL")) )
291 		{
292 			lDescriptor[nProperty].Value >>= sTemp;
293 			aURL = sTemp;
294 		}
295         else if( !aURL.Len() && lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("FileName")) )
296 		{
297 			lDescriptor[nProperty].Value >>= sTemp;
298 			aURL = sTemp;
299 		}
300         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("TypeName")) )
301 		{
302 			lDescriptor[nProperty].Value >>= sTemp;
303             aTypeName = sTemp;
304 		}
305         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("FilterName")) )
306 		{
307 			lDescriptor[nProperty].Value >>= sTemp;
308             aPreselectedFilterName = sTemp;
309 
310             // if the preselected filter name is not correct, it must be erased after detection
311             // remember index of property to get access to it later
312             nIndexOfFilterName = nProperty;
313 		}
314         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("InputStream")) )
315             nIndexOfInputStream = nProperty;
316         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("ReadOnly")) )
317             nIndexOfReadOnlyFlag = nProperty;
318         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("UCBContent")) )
319             nIndexOfContent = nProperty;
320         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("AsTemplate")) )
321 		{
322 			lDescriptor[nProperty].Value >>= bOpenAsTemplate;
323             nIndexOfTemplateFlag = nProperty;
324 		}
325         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("InteractionHandler")) )
326             lDescriptor[nProperty].Value >>= xInteraction;
327         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("RepairPackage")) )
328             lDescriptor[nProperty].Value >>= bRepairPackage;
329         else if( lDescriptor[nProperty].Name == OUString(RTL_CONSTASCII_USTRINGPARAM("DocumentTitle")) )
330             nIndexOfDocumentTitle = nProperty;
331 	}
332 
333     // can't check the type for external filters, so set the "dont" flag accordingly
334     ::vos::OGuard aGuard( Application::GetSolarMutex() );
335     //SfxFilterFlags nMust = SFX_FILTER_IMPORT, nDont = SFX_FILTER_NOTINSTALLED;
336 
337     SfxAllItemSet *pSet = new SfxAllItemSet( SFX_APP()->GetPool() );
338     TransformParameters( SID_OPENDOC, lDescriptor, *pSet );
339     SFX_ITEMSET_ARG( pSet, pItem, SfxBoolItem, SID_DOC_READONLY, sal_False );
340 
341     bWasReadOnly = pItem && pItem->GetValue();
342 
343 	const SfxFilter* pFilter = 0;
344 	String aPrefix = String::CreateFromAscii( "private:factory/" );
345 	if( aURL.Match( aPrefix ) == aPrefix.Len() )
346 	{
347 		String aPattern( aPrefix );
348 		aPattern += String::CreateFromAscii("scalc");
349 		if ( aURL.Match( aPattern ) >= aPattern.Len() )
350 			pFilter = SfxFilter::GetDefaultFilterFromFactory( aURL );
351 	}
352 	else
353 	{
354 		// container for Calc filters
355 		SfxFilterMatcher aMatcher( String::CreateFromAscii("scalc") );
356 		if ( aPreselectedFilterName.Len() )
357 			pFilter = SfxFilter::GetFilterByName( aPreselectedFilterName );
358 		else if( aTypeName.Len() )
359 			pFilter = aMatcher.GetFilter4EA( aTypeName );
360 
361 	    // ctor of SfxMedium uses owner transition of ItemSet
362 	    SfxMedium aMedium( aURL, bWasReadOnly ? STREAM_STD_READ : STREAM_STD_READWRITE, sal_False, NULL, pSet );
363 	    aMedium.UseInteractionHandler( sal_True );
364 
365 	    sal_Bool bIsStorage = aMedium.IsStorage();
366 	    if ( aMedium.GetErrorCode() == ERRCODE_NONE )
367 	    {
368 	        // remember input stream and content and put them into the descriptor later
369 			// should be done here since later the medium can switch to a version
370 	        xStream.set(aMedium.GetInputStream());
371 			xContent.set(aMedium.GetContent());
372         	bReadOnly = aMedium.IsReadOnly();
373 
374 	        // maybe that IsStorage() already created an error!
375 	        if ( bIsStorage )
376 			{
377                 uno::Reference < embed::XStorage > xStorage(aMedium.GetStorage( sal_False ));
378 				if ( aMedium.GetLastStorageCreationState() != ERRCODE_NONE )
379 				{
380 					// error during storage creation means _here_ that the medium
381 					// is broken, but we can not handle it in medium since unpossibility
382 					// to create a storage does not _always_ means that the medium is broken
383 					aMedium.SetError( aMedium.GetLastStorageCreationState(), ::rtl::OUString( RTL_CONSTASCII_USTRINGPARAM( OSL_LOG_PREFIX ) ) );
384 					if ( xInteraction.is() )
385 					{
386 						OUString empty;
387 						try
388 						{
389 							InteractiveAppException xException( empty,
390 															REFERENCE< XInterface >(),
391 															InteractionClassification_ERROR,
392 															aMedium.GetError() );
393 
394                             REFERENCE< XInteractionRequest > xRequest(
395 								new ucbhelper::SimpleInteractionRequest( makeAny( xException ),
396 																 	 ucbhelper::CONTINUATION_APPROVE ) );
397 							xInteraction->handle( xRequest );
398 						}
399 						catch ( Exception & ) {};
400 					}
401 				}
402                 else if ( xStorage.is() )
403 				{
404 					try
405 					{
406                         String aFilterName;
407                         if ( pFilter )
408                             aFilterName = pFilter->GetName();
409                         aTypeName = SfxFilter::GetTypeFromStorage( xStorage, pFilter ? pFilter->IsOwnTemplateFormat() : sal_False, &aFilterName );
410 					}
411 					catch( lang::WrappedTargetException& aWrap )
412 					{
413 						packages::zip::ZipIOException aZipException;
414 
415 						// repairing is done only if this type is requested from outside
416 						if ( ( aWrap.TargetException >>= aZipException ) && aTypeName.Len() )
417 						{
418 							if ( xInteraction.is() )
419 							{
420 								// the package is broken one
421        							aDocumentTitle = aMedium.GetURLObject().getName(
422 															INetURLObject::LAST_SEGMENT,
423 															true,
424 															INetURLObject::DECODE_WITH_CHARSET );
425 
426 								if ( !bRepairPackage )
427 								{
428                                     // ask the user whether he wants to try to repair
429                                     RequestPackageReparation aRequest( aDocumentTitle );
430                                     xInteraction->handle( aRequest.GetRequest() );
431                                     bRepairAllowed = aRequest.isApproved();
432 								}
433 
434 								if ( !bRepairAllowed )
435 								{
436 									// repair either not allowed or not successful
437                                     NotifyBrokenPackage aNotifyRequest( aDocumentTitle );
438                                     xInteraction->handle( aNotifyRequest.GetRequest() );
439 								}
440 							}
441 
442 							if ( !bRepairAllowed )
443 								aTypeName.Erase();
444 						}
445 					}
446 					catch( uno::RuntimeException& )
447 					{
448 						throw;
449 					}
450 					catch( uno::Exception& )
451 					{
452 						aTypeName.Erase();
453 					}
454 
455                    	if ( aTypeName.Len() )
456                        	pFilter = SfxFilterMatcher( String::CreateFromAscii("scalc") ).GetFilter4EA( aTypeName );
457 
458 				}
459 			}
460 			else
461 			{
462                 bool bIsXLS = false;
463                 SvStream* pStream = aMedium.GetInStream();
464                 const SfxFilter* pPreselectedFilter = pFilter;
465                 if ( pPreselectedFilter && pPreselectedFilter->GetName().SearchAscii("Excel") != STRING_NOTFOUND )
466                     bIsXLS = true;
467                 pFilter = 0;
468                 if ( pStream )
469                 {
470                     SotStorageRef aStorage = new SotStorage ( pStream, sal_False );
471                     if ( !aStorage->GetError() )
472                     {
473                         // Excel-5: detect through contained streams
474                         // there are some "excel" formats from 3rd party vendors that need to be distinguished
475                         String aStreamName(RTL_CONSTASCII_STRINGPARAM("Workbook"));
476                         sal_Bool bExcel97Stream = ( aStorage->IsStream( aStreamName ) );
477 
478                         aStreamName = String(RTL_CONSTASCII_STRINGPARAM("Book"));
479                         sal_Bool bExcel5Stream = ( aStorage->IsStream( aStreamName ) );
480                         if ( bExcel97Stream || bExcel5Stream )
481                         {
482                             if ( bExcel97Stream )
483                             {
484                                 String aOldName;
485                                 sal_Bool bIsCalcFilter = sal_True;
486                                 if ( pPreselectedFilter )
487                                 {
488                                     // cross filter; now this should be a type detection only, not a filter detection
489                                     // we can simulate it by preserving the preselected filter if the type matches
490                                     // example: Excel filters for Writer
491                                     aOldName = pPreselectedFilter->GetFilterName();
492                                     bIsCalcFilter = pPreselectedFilter->GetServiceName().EqualsAscii("com.sun.star.sheet.SpreadsheetDocument");
493                                 }
494 
495                                 if ( aOldName.EqualsAscii(pFilterEx97Temp) || !bIsCalcFilter )
496                                 {
497                                     //  Excel 97 template selected -> keep selection
498                                 }
499                                 else if ( bExcel5Stream &&
500                                             ( aOldName.EqualsAscii(pFilterExcel5) || aOldName.EqualsAscii(pFilterEx5Temp) ||
501                                             aOldName.EqualsAscii(pFilterExcel95) || aOldName.EqualsAscii(pFilterEx95Temp) ) )
502                                 {
503                                     //  dual format file and Excel 5 selected -> keep selection
504                                 }
505                                 else
506                                 {
507                                     //  else use Excel 97 filter
508                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcel97) );
509                                 }
510                             }
511                             else if ( bExcel5Stream )
512                             {
513                                 String aOldName;
514                                 sal_Bool bIsCalcFilter = sal_True;
515                                 if ( pPreselectedFilter )
516                                 {
517                                     // cross filter; now this should be a type detection only, not a filter detection
518                                     // we can simulate it by preserving the preselected filter if the type matches
519                                     // example: Excel filters for Writer
520                                     aOldName = pPreselectedFilter->GetFilterName();
521                                     bIsCalcFilter = pPreselectedFilter->GetServiceName().EqualsAscii("com.sun.star.sheet.SpreadsheetDocument");
522                                 }
523 
524                                 if ( aOldName.EqualsAscii(pFilterExcel95) || aOldName.EqualsAscii(pFilterEx95Temp) ||
525                                         aOldName.EqualsAscii(pFilterEx5Temp) || !bIsCalcFilter )
526                                 {
527                                     //  Excel 95 oder Vorlage (5 oder 95) eingestellt -> auch gut
528                                 }
529                                 else if ( aOldName.EqualsAscii(pFilterEx97Temp) )
530                                 {
531                                     // #101923# auto detection has found template -> return Excel5 template
532                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterEx5Temp) );
533                                 }
534                                 else
535                                 {
536                                     //  sonst wird als Excel 5-Datei erkannt
537                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterExcel5) );
538                                 }
539                             }
540                         }
541                     }
542                     else
543                     {
544                         SvStream &rStr = *pStream;
545 
546                         // Tabelle mit Suchmustern
547                         // Bedeutung der Sequenzen
548                         // 0x00??: genau Byte 0x?? muss an dieser Stelle stehen
549                         // 0x0100: ein Byte ueberlesen (don't care)
550                         // 0x02nn: ein Byte aus 0xnn Alternativen folgt
551                         // 0x8000: Erkennung abgeschlossen
552                         //
553 
554         #define M_DC        0x0100
555         #define M_ALT(ANZ)  (0x0200+(ANZ))
556         #define M_ENDE      0x8000
557 
558                         static const sal_uInt16 pLotus[] =      // Lotus 1/1A/2
559                             { 0x0000, 0x0000, 0x0002, 0x0000,
560                             M_ALT(2), 0x0004, 0x0006,
561                             0x0004, M_ENDE };
562 
563                         static const sal_uInt16 pLotusNew[] =   // Lotus >= 9.7
564                             { 0x0000, 0x0000, M_DC, 0x0000,     // Rec# + Len (0x1a)
565                               M_ALT(3), 0x0003, 0x0004, 0x0005, // File Revision Code 97->ME
566                               0x0010, 0x0004, 0x0000, 0x0000,
567                               M_ENDE };
568 
569                         static const sal_uInt16 pExcel1[] =     // Excel BIFF2, BIFF3, BIFF4
570                             {   0x09,                                   // lobyte of BOF rec ID (0x0009, 0x0209, 0x0409)
571                                 M_ALT(3), 0x00, 0x02, 0x04,             // hibyte of BOF rec ID (0x0009, 0x0209, 0x0409)
572                                 M_ALT(3), 4, 6, 8,                      // lobyte of BOF rec size (4, 6, 8, 16)
573                                 0x00,                                   // hibyte of BOF rec size (4, 6, 8, 16)
574                                 M_DC, M_DC,                             // any version
575                                 M_ALT(3), 0x10, 0x20, 0x40,             // lobyte of data type (0x0010, 0x0020, 0x0040)
576                                 0x00,                                   // hibyte of data type (0x0010, 0x0020, 0x0040)
577                                 M_ENDE };
578 
579                         static const sal_uInt16 pExcel2[] =     // Excel BIFF4 Workspace
580                             {   0x09,                                   // lobyte of BOF rec ID (0x0409)
581                                 0x04,                                   // hibyte of BOF rec ID (0x0409)
582                                 M_ALT(3), 4, 6, 8,                      // lobyte of BOF rec size (4, 6, 8, 16)
583                                 0x00,                                   // hibyte of BOF rec size (4, 6, 8, 16)
584                                 M_DC, M_DC,                             // any version
585                                 0x00,                                   // lobyte of data type (0x0100)
586                                 0x01,                                   // hibyte of data type (0x0100)
587                                 M_ENDE };
588 
589                         static const sal_uInt16 pExcel3[] =     // #i23425# Excel BIFF5, BIFF7, BIFF8 (simple book stream)
590                             {   0x09,                                   // lobyte of BOF rec ID (0x0809)
591                                 0x08,                                   // hibyte of BOF rec ID (0x0809)
592                                 M_ALT(4), 4, 6, 8, 16,                  // lobyte of BOF rec size
593                                 0x00,                                   // hibyte of BOF rec size
594                                 M_DC, M_DC,                             // any version
595                                 M_ALT(5), 0x05, 0x06, 0x10, 0x20, 0x40, // lobyte of data type
596                                 0x00,                                   // hibyte of data type
597                                 M_ENDE };
598 
599                         static const sal_uInt16 pSc10[] =       // StarCalc 1.0 Dokumente
600                             { 'B', 'l', 'a', 'i', 's', 'e', '-', 'T', 'a', 'b', 'e', 'l', 'l',
601                             'e', 0x000A, 0x000D, 0x0000,    // Sc10CopyRight[16]
602                             M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC, M_DC,
603                             M_DC, M_DC,                   // Sc10CopyRight[29]
604                             M_ALT(2), 0x0065, 0x0066,     // Versionsnummer 101 oder 102
605                             0x0000,
606                             M_ENDE };
607 
608                         static const sal_uInt16 pLotus2[] =     // Lotus >3
609                             { 0x0000, 0x0000, 0x001A, 0x0000,   // Rec# + Len (26)
610                             M_ALT(2), 0x0000, 0x0002,         // File Revision Code
611                             0x0010,
612                             0x0004, 0x0000,                   // File Revision Subcode
613                             M_ENDE };
614 
615                         static const sal_uInt16 pQPro[] =
616                                { 0x0000, 0x0000, 0x0002, 0x0000,
617                                  M_ALT(4), 0x0001, 0x0002, // WB1, WB2
618                                  0x0006, 0x0007,           // QPro 6/7 (?)
619                                  0x0010,
620                                  M_ENDE };
621 
622                         static const sal_uInt16 pDIF1[] =       // DIF mit CR-LF
623                             {
624                             'T', 'A', 'B', 'L', 'E',
625                             M_DC, M_DC,
626                             '0', ',', '1',
627                             M_DC, M_DC,
628                             '\"',
629                             M_ENDE };
630 
631                         static const sal_uInt16 pDIF2[] =       // DIF mit CR oder LF
632                             {
633                             'T', 'A', 'B', 'L', 'E',
634                             M_DC,
635                             '0', ',', '1',
636                             M_DC,
637                             '\"',
638                             M_ENDE };
639 
640                         static const sal_uInt16 pSylk[] =       // Sylk
641                             {
642                             'I', 'D', ';',
643                             M_ALT(3), 'P', 'N', 'E',        // 'P' plus undocumented Excel extensions 'N' and 'E'
644                             M_ENDE };
645 
646                         static const sal_uInt16 *ppFilterPatterns[] =      // Arrays mit Suchmustern
647                             {
648                             pLotus,
649                             pExcel1,
650                             pExcel2,
651                             pExcel3,
652                             pSc10,
653                             pDIF1,
654                             pDIF2,
655                             pSylk,
656                             pLotusNew,
657                             pLotus2,
658                             pQPro
659                             };
660                         const sal_uInt16 nFilterCount = sizeof(ppFilterPatterns) / sizeof(ppFilterPatterns[0]);
661 
662                         static const sal_Char* const pFilterName[] =     // zugehoerige Filter
663                             {
664                             pFilterLotus,
665                             pFilterExcel4,
666                             pFilterExcel4,
667                             pFilterExcel4,
668                             pFilterSc10,
669                             pFilterDif,
670                             pFilterDif,
671                             pFilterSylk,
672                             pFilterLotus,
673                             pFilterLotus,
674                             pFilterQPro6
675                             };
676 
677                         // const sal_uInt16 nByteMask = 0xFF;
678 
679                         // suchen Sie jetzt!
680                         // ... realisiert ueber 'Mustererkennung'
681 
682                         sal_uInt8            nAkt;
683                         sal_Bool            bSync;          // Datei und Muster stimmen ueberein
684                         sal_uInt16          nFilter;        // Zaehler ueber alle Filter
685                         const sal_uInt16    *pSearch;       // aktuelles Musterwort
686 
687                         for ( nFilter = 0 ; nFilter < nFilterCount ; nFilter++ )
688                         {
689                             rStr.Seek( 0 ); // am Anfang war alles Uebel...
690                             rStr >> nAkt;
691                             pSearch = ppFilterPatterns[ nFilter ];
692                             bSync = sal_True;
693                             while( !rStr.IsEof() && bSync )
694                             {
695                                 register sal_uInt16 nMuster = *pSearch;
696 
697                                 if( nMuster < 0x0100 )
698                                 { //                                direkter Byte-Vergleich
699                                     if( ( sal_uInt8 ) nMuster != nAkt )
700                                         bSync = sal_False;
701                                 }
702                                 else if( nMuster & M_DC )
703                                 { //                                             don't care
704                                 }
705                                 else if( nMuster & M_ALT(0) )
706                                 { //                                      alternative Bytes
707                                     sal_uInt8 nAnzAlt = ( sal_uInt8 ) nMuster;
708                                     bSync = sal_False;          // zunaechst unsynchron
709                                     while( nAnzAlt > 0 )
710                                     {
711                                         pSearch++;
712                                         if( ( sal_uInt8 ) *pSearch == nAkt )
713                                             bSync = sal_True;   // jetzt erst Synchronisierung
714                                         nAnzAlt--;
715                                     }
716                                 }
717                                 else if( nMuster & M_ENDE )
718                                 { //                                        Format detected
719                                     if ( pFilterName[nFilter] == pFilterExcel4 && pPreselectedFilter &&
720                                         ( (pPreselectedFilter)->GetFilterName().EqualsAscii(pFilterEx4Temp) || pPreselectedFilter->GetTypeName().EqualsAscii("calc_MS_Excel_40") ) )
721                                     {
722                                         //  Excel 4 erkannt, Excel 4 Vorlage eingestellt -> auch gut
723                                         // oder Excel 4 Filter anderer Applikation (simulated type detection!)
724                                     }
725                                     else
726                                     {   // gefundenen Filter einstellen
727                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterName[ nFilter ]) );
728                                     }
729                                     bSync = sal_False;              // leave inner loop
730                                     nFilter = nFilterCount;     // leave outer loop
731                                 }
732                                 else
733                                 { //                                         Tabellenfehler
734                                     DBG_ERROR( "-ScApplication::DetectFilter(): Fehler in Mustertabelle");
735                                 }
736 
737                                 pSearch++;
738                                 rStr >> nAkt;
739                             }
740                         }
741 
742                         if ( pPreselectedFilter && !pFilter )
743                         {
744                             // further checks for filters only if they are preselected: ASCII, HTML, RTF, DBase
745                             // without the preselection other filters (Writer) take precedence
746                             // DBase can't be detected reliably, so it also needs preselection
747                             bool bMaybeText = lcl_MayBeAscii( rStr );
748                             if ( pPreselectedFilter->GetFilterName().EqualsAscii(pFilterAscii) && bMaybeText )
749                             {
750                                 // Text filter is accepted if preselected
751                                 pFilter = pPreselectedFilter;
752                             }
753                             else
754                             {
755                                 // get file header
756                                 rStr.Seek( 0 );
757                                 const int nTrySize = 80;
758                                 ByteString aHeader;
759                                 for ( int j = 0; j < nTrySize && !rStr.IsEof(); j++ )
760                                 {
761                                     sal_Char c;
762                                     rStr >> c;
763                                     aHeader += c;
764                                 }
765                                 aHeader += '\0';
766 
767                                 if ( HTMLParser::IsHTMLFormat( aHeader.GetBuffer() ) )
768                                 {
769                                     // test for HTML
770                                     if ( pPreselectedFilter->GetName().EqualsAscii(pFilterHtml) )
771                                     {
772                                         pFilter = pPreselectedFilter;
773                                     }
774                                     else
775                                     {
776                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterHtmlWeb) );
777                                         if ( bIsXLS )
778                                             bFakeXLS = true;
779                                     }
780                                 }
781                                 else if ( bIsXLS && bMaybeText )
782                                 {
783                                     // Detect Excel 2003 XML here only if XLS was preselected.
784                                     // The configured detection for Excel 2003 XML is still in XMLFilterDetect.
785                                     pFilter = lcl_DetectExcelXML( rStr, aMatcher );
786                                     if (!pFilter)
787                                         pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterAscii) );
788                                     bFakeXLS = true;
789                                 }
790                                 else if ( aHeader.CompareTo( "{\\rtf", 5 ) == COMPARE_EQUAL )
791                                 {
792                                     // test for RTF
793                                     pFilter = aMatcher.GetFilter4FilterName( String::CreateFromAscii(pFilterRtf) );
794                                 }
795                                 else if ( pPreselectedFilter->GetName().EqualsAscii(pFilterDBase) && lcl_MayBeDBase( rStr ) )
796                                     pFilter = pPreselectedFilter;
797                             }
798                         }
799                     }
800 				}
801 			}
802 		}
803 	}
804 
805     if ( nIndexOfInputStream == -1 && xStream.is() )
806     {
807         // if input stream wasn't part of the descriptor, now it should be, otherwise the content would be opend twice
808         lDescriptor.realloc( nPropertyCount + 1 );
809         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("InputStream");
810         lDescriptor[nPropertyCount].Value <<= xStream;
811         nPropertyCount++;
812     }
813 
814     if ( nIndexOfContent == -1 && xContent.is() )
815     {
816         // if input stream wasn't part of the descriptor, now it should be, otherwise the content would be opend twice
817         lDescriptor.realloc( nPropertyCount + 1 );
818         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("UCBContent");
819         lDescriptor[nPropertyCount].Value <<= xContent;
820         nPropertyCount++;
821     }
822 
823     if ( bReadOnly != bWasReadOnly )
824     {
825         if ( nIndexOfReadOnlyFlag == -1 )
826         {
827             lDescriptor.realloc( nPropertyCount + 1 );
828             lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("ReadOnly");
829             lDescriptor[nPropertyCount].Value <<= bReadOnly;
830             nPropertyCount++;
831         }
832         else
833             lDescriptor[nIndexOfReadOnlyFlag].Value <<= bReadOnly;
834     }
835 
836 	if ( !bRepairPackage && bRepairAllowed )
837 	{
838         lDescriptor.realloc( nPropertyCount + 1 );
839         lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("RepairPackage");
840         lDescriptor[nPropertyCount].Value <<= bRepairAllowed;
841         nPropertyCount++;
842 
843 		bOpenAsTemplate = sal_True;
844 
845 		// TODO/LATER: set progress bar that should be used
846 	}
847 
848 	if ( bOpenAsTemplate )
849 	{
850 		if ( nIndexOfTemplateFlag == -1 )
851 		{
852         	lDescriptor.realloc( nPropertyCount + 1 );
853         	lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("AsTemplate");
854         	lDescriptor[nPropertyCount].Value <<= bOpenAsTemplate;
855         	nPropertyCount++;
856 		}
857 		else
858         	lDescriptor[nIndexOfTemplateFlag].Value <<= bOpenAsTemplate;
859 	}
860 
861 	if ( aDocumentTitle.getLength() )
862 	{
863 		// the title was set here
864 		if ( nIndexOfDocumentTitle == -1 )
865 		{
866         	lDescriptor.realloc( nPropertyCount + 1 );
867         	lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("DocumentTitle");
868         	lDescriptor[nPropertyCount].Value <<= aDocumentTitle;
869         	nPropertyCount++;
870 		}
871 		else
872         	lDescriptor[nIndexOfDocumentTitle].Value <<= aDocumentTitle;
873 	}
874 
875     if ( bFakeXLS )
876     {
877         if ( nIndexOfFilterName == -1 )
878         {
879             lDescriptor.realloc( nPropertyCount + 1 );
880             lDescriptor[nPropertyCount].Name = ::rtl::OUString::createFromAscii("FilterName");
881             lDescriptor[nPropertyCount].Value <<= rtl::OUString(pFilter->GetName());
882             nPropertyCount++;
883         }
884         else
885             lDescriptor[nIndexOfFilterName].Value <<= rtl::OUString(pFilter->GetName());
886     }
887 
888     if ( pFilter )
889         aTypeName = pFilter->GetTypeName();
890     else
891         aTypeName.Erase();
892     return aTypeName;
893 }
894 
SFX_IMPL_SINGLEFACTORY(ScFilterDetect)895 SFX_IMPL_SINGLEFACTORY( ScFilterDetect )
896 
897 /* XServiceInfo */
898 UNOOUSTRING SAL_CALL ScFilterDetect::getImplementationName() throw( UNORUNTIMEEXCEPTION )
899 {
900     return impl_getStaticImplementationName();
901 }
902                                                                                                                                 \
903 /* XServiceInfo */
supportsService(const UNOOUSTRING & sServiceName)904 sal_Bool SAL_CALL ScFilterDetect::supportsService( const UNOOUSTRING& sServiceName ) throw( UNORUNTIMEEXCEPTION )
905 {
906     UNOSEQUENCE< UNOOUSTRING >  seqServiceNames(getSupportedServiceNames());
907     const UNOOUSTRING*          pArray          =   seqServiceNames.getConstArray();
908     for ( sal_Int32 nCounter=0; nCounter<seqServiceNames.getLength(); nCounter++ )
909     {
910         if ( pArray[nCounter] == sServiceName )
911         {
912             return sal_True ;
913         }
914     }
915     return sal_False ;
916 }
917 
918 /* XServiceInfo */
getSupportedServiceNames()919 UNOSEQUENCE< UNOOUSTRING > SAL_CALL ScFilterDetect::getSupportedServiceNames() throw( UNORUNTIMEEXCEPTION )
920 {
921     return impl_getStaticSupportedServiceNames();
922 }
923 
924 /* Helper for XServiceInfo */
impl_getStaticSupportedServiceNames()925 UNOSEQUENCE< UNOOUSTRING > ScFilterDetect::impl_getStaticSupportedServiceNames()
926 {
927     UNOMUTEXGUARD aGuard( UNOMUTEX::getGlobalMutex() );
928     UNOSEQUENCE< UNOOUSTRING > seqServiceNames( 1 );
929     seqServiceNames.getArray() [0] = UNOOUSTRING::createFromAscii( "com.sun.star.frame.ExtendedTypeDetection"  );
930     return seqServiceNames ;
931 }
932 
933 /* Helper for XServiceInfo */
impl_getStaticImplementationName()934 UNOOUSTRING ScFilterDetect::impl_getStaticImplementationName()
935 {
936     return UNOOUSTRING::createFromAscii( "com.sun.star.comp.calc.FormatDetector" );
937 }
938 
939 /* Helper for registry */
impl_createInstance(const UNOREFERENCE<UNOXMULTISERVICEFACTORY> & xServiceManager)940 UNOREFERENCE< UNOXINTERFACE > SAL_CALL ScFilterDetect::impl_createInstance( const UNOREFERENCE< UNOXMULTISERVICEFACTORY >& xServiceManager ) throw( UNOEXCEPTION )
941 {
942     return UNOREFERENCE< UNOXINTERFACE >( *new ScFilterDetect( xServiceManager ) );
943 }
944 
945