diff options
author | Kohei Yoshida <kohei.yoshida@collabora.com> | 2013-11-22 20:57:40 -0500 |
---|---|---|
committer | Michael Meeks <michael.meeks@collabora.com> | 2013-12-02 12:32:02 +0000 |
commit | 934941a4cf7c9ee7df69f03e6f0be246499d587f (patch) | |
tree | fc37e7a46e181ddee1d42a6fd0aedc8cd1290fc7 | |
parent | 1ab139fdbe9a7ed3b781730dfecb83a98e0b671b (diff) |
Allow worker threads to use their own FastParser instances.
To prevent deadlock during threaded sheet stream parsing. It now
deadlocks at a different place.
Conflicts:
oox/source/core/xmlfilterbase.cxx
sc/source/filter/oox/workbookfragment.cxx
Change-Id: I0ba0f2c9a257e71b0a340ab14e369b06d5fd8829
-rw-r--r-- | include/oox/core/xmlfilterbase.hxx | 8 | ||||
-rw-r--r-- | oox/source/core/xmlfilterbase.cxx | 170 | ||||
-rw-r--r-- | sc/source/filter/inc/workbookhelper.hxx | 6 | ||||
-rw-r--r-- | sc/source/filter/oox/workbookfragment.cxx | 195 | ||||
-rw-r--r-- | sc/source/filter/oox/workbookhelper.cxx | 7 |
5 files changed, 292 insertions, 94 deletions
diff --git a/include/oox/core/xmlfilterbase.hxx b/include/oox/core/xmlfilterbase.hxx index 87234fb22328..76eb091b3f2e 100644 --- a/include/oox/core/xmlfilterbase.hxx +++ b/include/oox/core/xmlfilterbase.hxx @@ -56,8 +56,7 @@ namespace oox { namespace core { class FragmentHandler; - -// ============================================================================ +class FastParser; struct TextField { com::sun::star::uno::Reference< com::sun::star::text::XText > xText; @@ -107,7 +106,8 @@ public: @return True, if the fragment could be imported. */ - bool importFragment( const ::rtl::Reference< FragmentHandler >& rxHandler ); + bool importFragment( const rtl::Reference<FragmentHandler>& rxHandler ); + bool importFragment( const rtl::Reference<FragmentHandler>& rxHandler, FastParser& rParser ); /** Imports a fragment into an xml::dom::XDocument. @@ -231,6 +231,8 @@ public: void importDocumentProperties(); + FastParser* createParser() const; + protected: virtual ::com::sun::star::uno::Reference< ::com::sun::star::io::XInputStream > implGetInputStream( utl::MediaDescriptor& rMediaDesc ) const; diff --git a/oox/source/core/xmlfilterbase.cxx b/oox/source/core/xmlfilterbase.cxx index 98c88863a21b..d1ae6a539c22 100644 --- a/oox/source/core/xmlfilterbase.cxx +++ b/oox/source/core/xmlfilterbase.cxx @@ -74,12 +74,6 @@ using utl::MediaDescriptor; using ::sax_fastparser::FSHelperPtr; using ::sax_fastparser::FastSerializerHelper; - - - - -// ============================================================================ - namespace { bool lclHasSuffix( const OUString& rFragmentPath, const OUString& rSuffix ) @@ -88,9 +82,77 @@ bool lclHasSuffix( const OUString& rFragmentPath, const OUString& rSuffix ) return (nSuffixPos >= 0) && rFragmentPath.match( rSuffix, nSuffixPos ); } -} // namespace +struct NamespaceIds: public rtl::StaticWithInit< + Sequence< beans::Pair< OUString, sal_Int32 > >, + NamespaceIds> +{ + Sequence< beans::Pair< OUString, sal_Int32 > > operator()() + { + static const char* const namespaceURIs[] = { + "http://www.w3.org/XML/1998/namespace", + "http://schemas.openxmlformats.org/package/2006/relationships", + "http://schemas.openxmlformats.org/officeDocument/2006/relationships", + "http://schemas.openxmlformats.org/drawingml/2006/main", + "http://schemas.openxmlformats.org/drawingml/2006/diagram", + "http://schemas.openxmlformats.org/drawingml/2006/chart", + "http://schemas.openxmlformats.org/drawingml/2006/chartDrawing", + "urn:schemas-microsoft-com:vml", + "urn:schemas-microsoft-com:office:office", + "urn:schemas-microsoft-com:office:word", + "urn:schemas-microsoft-com:office:excel", + "urn:schemas-microsoft-com:office:powerpoint", + "http://schemas.microsoft.com/office/2006/activeX", + "http://schemas.openxmlformats.org/spreadsheetml/2006/main", + "http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing", + "http://schemas.microsoft.com/office/excel/2006/main", + "http://schemas.openxmlformats.org/presentationml/2006/main", + "http://schemas.openxmlformats.org/markup-compatibility/2006", + "http://schemas.openxmlformats.org/spreadsheetml/2006/main/v2", + "http://schemas.microsoft.com/office/drawing/2008/diagram", + "http://schemas.microsoft.com/office/spreadsheetml/2009/9/main" + }; + + static const sal_Int32 namespaceIds[] = { + NMSP_xml, + NMSP_packageRel, + NMSP_officeRel, + NMSP_dml, + NMSP_dmlDiagram, + NMSP_dmlChart, + NMSP_dmlChartDr, + NMSP_dmlSpreadDr, + NMSP_vml, + NMSP_vmlOffice, + NMSP_vmlWord, + NMSP_vmlExcel, + NMSP_vmlPowerpoint, + NMSP_xls, + NMSP_ppt, + NMSP_ax, + NMSP_xm, + NMSP_mce, + NMSP_mceTest, + NMSP_dsp, + NMSP_xlsExtLst + }; + + Sequence< beans::Pair< OUString, sal_Int32 > > aRet(STATIC_ARRAY_SIZE(namespaceIds)); + for( sal_Int32 i=0; i<aRet.getLength(); ++i ) + aRet[i] = make_Pair( + OUString::createFromAscii(namespaceURIs[i]), + namespaceIds[i]); + return aRet; + } +}; -// ============================================================================ +void registerNamespaces( FastParser& rParser ) +{ + const Sequence< beans::Pair<OUString, sal_Int32> > ids = NamespaceIds::get(); + for (sal_Int32 i = 0; i < ids.getLength(); ++i) + rParser.registerNamespace(ids[i].Second); +} + +} // namespace struct XmlFilterBaseImpl { @@ -105,75 +167,6 @@ struct XmlFilterBaseImpl explicit XmlFilterBaseImpl( const Reference< XComponentContext >& rxContext ) throw( RuntimeException ); }; -// ---------------------------------------------------------------------------- - -namespace -{ - struct NamespaceIds: public rtl::StaticWithInit< - Sequence< beans::Pair< OUString, sal_Int32 > >, - NamespaceIds> - { - Sequence< beans::Pair< OUString, sal_Int32 > > operator()() - { - static const char* const namespaceURIs[] = { - "http://www.w3.org/XML/1998/namespace", - "http://schemas.openxmlformats.org/package/2006/relationships", - "http://schemas.openxmlformats.org/officeDocument/2006/relationships", - "http://schemas.openxmlformats.org/drawingml/2006/main", - "http://schemas.openxmlformats.org/drawingml/2006/diagram", - "http://schemas.openxmlformats.org/drawingml/2006/chart", - "http://schemas.openxmlformats.org/drawingml/2006/chartDrawing", - "urn:schemas-microsoft-com:vml", - "urn:schemas-microsoft-com:office:office", - "urn:schemas-microsoft-com:office:word", - "urn:schemas-microsoft-com:office:excel", - "urn:schemas-microsoft-com:office:powerpoint", - "http://schemas.microsoft.com/office/2006/activeX", - "http://schemas.openxmlformats.org/spreadsheetml/2006/main", - "http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing", - "http://schemas.microsoft.com/office/excel/2006/main", - "http://schemas.openxmlformats.org/presentationml/2006/main", - "http://schemas.openxmlformats.org/markup-compatibility/2006", - "http://schemas.openxmlformats.org/spreadsheetml/2006/main/v2", - "http://schemas.microsoft.com/office/drawing/2008/diagram", - "http://schemas.microsoft.com/office/spreadsheetml/2009/9/main" - }; - - static const sal_Int32 namespaceIds[] = { - NMSP_xml, - NMSP_packageRel, - NMSP_officeRel, - NMSP_dml, - NMSP_dmlDiagram, - NMSP_dmlChart, - NMSP_dmlChartDr, - NMSP_dmlSpreadDr, - NMSP_vml, - NMSP_vmlOffice, - NMSP_vmlWord, - NMSP_vmlExcel, - NMSP_vmlPowerpoint, - NMSP_xls, - NMSP_ppt, - NMSP_ax, - NMSP_xm, - NMSP_mce, - NMSP_mceTest, - NMSP_dsp, - NMSP_xlsExtLst - }; - - Sequence< beans::Pair< OUString, sal_Int32 > > aRet(STATIC_ARRAY_SIZE(namespaceIds)); - for( sal_Int32 i=0; i<aRet.getLength(); ++i ) - aRet[i] = make_Pair( - OUString::createFromAscii(namespaceURIs[i]), - namespaceIds[i]); - return aRet; - } - }; -} - -// ---------------------------------------------------------------------------- XmlFilterBaseImpl::XmlFilterBaseImpl( const Reference< XComponentContext >& rxContext ) throw( RuntimeException ) : maFastParser( rxContext ), @@ -181,10 +174,7 @@ XmlFilterBaseImpl::XmlFilterBaseImpl( const Reference< XComponentContext >& rxCo maVmlSuffix( ".vml" ) { // register XML namespaces - const Sequence< beans::Pair< OUString, sal_Int32 > > ids= - NamespaceIds::get(); - for( sal_Int32 i=0; i<ids.getLength(); ++i ) - maFastParser.registerNamespace( ids[i].Second ); + registerNamespaces(maFastParser); } XmlFilterBase::XmlFilterBase( const Reference< XComponentContext >& rxContext ) throw( RuntimeException ) : @@ -220,13 +210,25 @@ void XmlFilterBase::importDocumentProperties() xImporter->importProperties( xDocumentStorage, xPropSupplier->getDocumentProperties() ); } +FastParser* XmlFilterBase::createParser() const +{ + FastParser* pParser = new FastParser(getComponentContext()); + registerNamespaces(*pParser); + return pParser; +} + OUString XmlFilterBase::getFragmentPathFromFirstType( const OUString& rType ) { // importRelations() caches the relations map for subsequence calls return importRelations( OUString() )->getFragmentPathFromFirstType( rType ); } -bool XmlFilterBase::importFragment( const ::rtl::Reference< FragmentHandler >& rxHandler ) +bool XmlFilterBase::importFragment( const rtl::Reference<FragmentHandler>& rxHandler ) +{ + return importFragment(rxHandler, mxImpl->maFastParser); +} + +bool XmlFilterBase::importFragment( const rtl::Reference<FragmentHandler>& rxHandler, FastParser& rParser ) { OSL_ENSURE( rxHandler.is(), "XmlFilterBase::importFragment - missing fragment handler" ); if( !rxHandler.is() ) @@ -280,8 +282,8 @@ bool XmlFilterBase::importFragment( const ::rtl::Reference< FragmentHandler >& r // own try/catch block for showing parser failure assertion with fragment path if( xInStrm.is() ) try { - mxImpl->maFastParser.setDocumentHandler( xDocHandler ); - mxImpl->maFastParser.parseStream( xInStrm, aFragmentPath ); + rParser.setDocumentHandler(xDocHandler); + rParser.parseStream(xInStrm, aFragmentPath); return true; } catch( Exception& ) diff --git a/sc/source/filter/inc/workbookhelper.hxx b/sc/source/filter/inc/workbookhelper.hxx index abafb20b7421..1f472c3860f6 100644 --- a/sc/source/filter/inc/workbookhelper.hxx +++ b/sc/source/filter/inc/workbookhelper.hxx @@ -53,6 +53,7 @@ namespace oox { namespace core { class FilterBase; class FragmentHandler; class XmlFilterBase; + class FastParser; } } class ScDocument; @@ -269,7 +270,10 @@ public: /** Imports a fragment using the passed fragment handler, which contains the full path to the fragment stream. */ - bool importOoxFragment( const ::rtl::Reference< ::oox::core::FragmentHandler >& rxHandler ); + bool importOoxFragment( const rtl::Reference<oox::core::FragmentHandler>& rxHandler ); + + bool importOoxFragment( const rtl::Reference<oox::core::FragmentHandler>& rxHandler, oox::core::FastParser& rParser ); + // BIFF2-BIFF8 specific (MUST NOT be called in OOXML/BIFF12 filter) ------- diff --git a/sc/source/filter/oox/workbookfragment.cxx b/sc/source/filter/oox/workbookfragment.cxx index 485642c9df45..e9884dacb63e 100644 --- a/sc/source/filter/oox/workbookfragment.cxx +++ b/sc/source/filter/oox/workbookfragment.cxx @@ -48,6 +48,8 @@ #include "calcconfig.hxx" #include <oox/core/fastparser.hxx> +#include <comphelper/processfactory.hxx> +#include <officecfg/Office/Calc.hxx> #include <salhelper/thread.hxx> #include <osl/conditn.hxx> @@ -197,6 +199,193 @@ const RecordInfo* WorkbookFragment::getRecordInfos() const return spRecInfos; } +namespace { + +class WorkerThread; + +typedef std::pair<WorksheetGlobalsRef, FragmentHandlerRef> SheetFragmentHandler; +typedef std::vector<SheetFragmentHandler> SheetFragmentVector; +typedef rtl::Reference<WorkerThread> WorkerThreadRef; + +struct WorkerThreadData +{ + osl::Mutex maMtx; + std::vector<WorkerThreadRef> maThreads; +}; + +struct IdleWorkerThreadData +{ + osl::Mutex maMtx; + osl::Condition maCondAdded; + std::queue<WorkerThread*> maThreads; +}; + +struct +{ + boost::scoped_ptr<WorkerThreadData> mpWorkerThreads; + boost::scoped_ptr<IdleWorkerThreadData> mpIdleThreads; + +} aThreadGlobals; + +enum WorkerAction +{ + None = 0, + TerminateThread, + Work +}; + +class WorkerThread : public salhelper::Thread +{ + WorkbookFragment& mrWorkbookHandler; + size_t mnID; + FragmentHandlerRef mxHandler; + boost::scoped_ptr<oox::core::FastParser> mxParser; + osl::Mutex maMtxAction; + osl::Condition maCondActionChanged; + WorkerAction meAction; +public: + WorkerThread( WorkbookFragment& rWorkbookHandler, size_t nID ) : + salhelper::Thread("sheet-import-worker-thread"), + mrWorkbookHandler(rWorkbookHandler), + mnID(nID), + mxParser(rWorkbookHandler.getOoxFilter().createParser()), + meAction(None) {} + + virtual void execute() + { + announceIdle(); + + // Keep looping until the terminate request is set. + for (maCondActionChanged.wait(); true; maCondActionChanged.wait()) + { + osl::MutexGuard aGuard(maMtxAction); + if (!maCondActionChanged.check()) + // Wait again. + continue; + + maCondActionChanged.reset(); + + if (meAction == TerminateThread) + // End the thread. + return; + + if (meAction != Work) + continue; + +#if 0 + // TODO : This still deadlocks in the fast parser code. + mrWorkbookHandler.importOoxFragment(mxHandler, *mxParser); +#else + double val = rand() / static_cast<double>(RAND_MAX); + val *= 1000000; // normalize to 1 second. + val *= 1.5; // inflate it a bit. + usleep(val); // pretend to be working while asleep. +#endif + announceIdle(); + } + } + + void announceIdle() + { + // Set itself idle to receive a new task from the main thread. + osl::MutexGuard aGuard(aThreadGlobals.mpIdleThreads->maMtx); + aThreadGlobals.mpIdleThreads->maThreads.push(this); + aThreadGlobals.mpIdleThreads->maCondAdded.set(); + } + + void terminate() + { + osl::MutexGuard aGuard(maMtxAction); + meAction = TerminateThread; + maCondActionChanged.set(); + } + + void assign( const FragmentHandlerRef& rHandler ) + { + osl::MutexGuard aGuard(maMtxAction); + mxHandler = rHandler; + meAction = Work; + maCondActionChanged.set(); + } +}; + +void importSheetFragments( WorkbookFragment& rWorkbookHandler, SheetFragmentVector& rSheets ) +{ +#if 0 // threaded version + size_t nThreadCount = 3; + if (nThreadCount > rSheets.size()) + nThreadCount = rSheets.size(); + + // Create new thread globals. + aThreadGlobals.mpWorkerThreads.reset(new WorkerThreadData); + aThreadGlobals.mpIdleThreads.reset(new IdleWorkerThreadData); + + SheetFragmentVector::iterator it = rSheets.begin(), itEnd = rSheets.end(); + + { + // Initialize worker threads. + osl::MutexGuard aGuard(aThreadGlobals.mpWorkerThreads->maMtx); + for (size_t i = 0; i < nThreadCount; ++i) + { + WorkerThreadRef pThread(new WorkerThread(rWorkbookHandler, i)); + aThreadGlobals.mpWorkerThreads->maThreads.push_back(pThread); + pThread->launch(); + } + } + + for (aThreadGlobals.mpIdleThreads->maCondAdded.wait(); true; aThreadGlobals.mpIdleThreads->maCondAdded.wait()) + { + osl::MutexGuard aGuard(aThreadGlobals.mpIdleThreads->maMtx); + if (!aThreadGlobals.mpIdleThreads->maCondAdded.check()) + // Wait again. + continue; + + aThreadGlobals.mpIdleThreads->maCondAdded.reset(); + + // Assign work to all idle threads. + while (!aThreadGlobals.mpIdleThreads->maThreads.empty()) + { + if (it == itEnd) + break; + + WorkerThread* p = aThreadGlobals.mpIdleThreads->maThreads.front(); + aThreadGlobals.mpIdleThreads->maThreads.pop(); + p->assign(it->second); + ++it; + } + + if (it == itEnd) + // Finished! Exit the loop. + break; + } + + { + // Terminate all worker threads. + osl::MutexGuard aGuard(aThreadGlobals.mpWorkerThreads->maMtx); + for (size_t i = 0, n = aThreadGlobals.mpWorkerThreads->maThreads.size(); i < n; ++i) + { + WorkerThreadRef pWorker = aThreadGlobals.mpWorkerThreads->maThreads[i]; + pWorker->terminate(); + if (pWorker.is()) + pWorker->join(); + } + } + + // Delete all thread globals. + aThreadGlobals.mpWorkerThreads.reset(); + aThreadGlobals.mpIdleThreads.reset(); + +#else // non-threaded version + for( SheetFragmentVector::iterator it = rSheets.begin(), itEnd = rSheets.end(); it != itEnd; ++it) + { + // import the sheet fragment + rWorkbookHandler.importOoxFragment(it->second); + } +#endif +} + +} + void WorkbookFragment::finalizeImport() { ISegmentProgressBarRef xGlobalSegment = getProgressBar().createSegment( PROGRESS_LENGTH_GLOBALS ); @@ -318,11 +507,7 @@ void WorkbookFragment::finalizeImport() } // load all worksheets - for( SheetFragmentVector::iterator aIt = aSheetFragments.begin(), aEnd = aSheetFragments.end(); aIt != aEnd; ++aIt ) - { - // import the sheet fragment - importOoxFragment( aIt->second ); - } + importSheetFragments(*this, aSheetFragments); for( std::vector<WorksheetHelper*>::iterator aIt = maHelpers.begin(), aEnd = maHelpers.end(); aIt != aEnd; ++aIt ) { diff --git a/sc/source/filter/oox/workbookhelper.cxx b/sc/source/filter/oox/workbookhelper.cxx index 0fca1513c6cf..32c8bd62cc19 100644 --- a/sc/source/filter/oox/workbookhelper.cxx +++ b/sc/source/filter/oox/workbookhelper.cxx @@ -1012,11 +1012,16 @@ XmlFilterBase& WorkbookHelper::getOoxFilter() const return mrBookGlob.getOoxFilter(); } -bool WorkbookHelper::importOoxFragment( const ::rtl::Reference< FragmentHandler >& rxHandler ) +bool WorkbookHelper::importOoxFragment( const rtl::Reference<FragmentHandler>& rxHandler ) { return getOoxFilter().importFragment( rxHandler ); } +bool WorkbookHelper::importOoxFragment( const rtl::Reference<FragmentHandler>& rxHandler, oox::core::FastParser& rParser ) +{ + return getOoxFilter().importFragment(rxHandler, rParser); +} + // BIFF specific -------------------------------------------------------------- BiffType WorkbookHelper::getBiff() const |