--- ncbi-blast-2.6.0+-src/c++/src/algo/blast/blastinput/blast_fasta_input.cpp 2016-10-27 12:13:57.000000000 -0700 +++ ncbi-blast-2.6.0+-src/c++/src/algo/blast/blastinput/blast_fasta_input.cpp 2017-02-07 11:56:27.299726089 -0800 @@ -292,6 +292,12 @@ CBlastFastaInputSource::x_InitInputReade CFastaReader::fAllSeqIds : (CFastaReader::fNoParseID | CFastaReader::fDLOptional); + + // Allow CFastaReader fSkipCheck flag to be set based + // on new CBlastInputSourceConfig property - GetSkipSeqCheck() -RMH- + flags += ( m_Config.GetSkipSeqCheck() ? + CFastaReader::fSkipCheck : 0 ); + flags += (m_ReadProteins ? CFastaReader::fAssumeProt : CFastaReader::fAssumeNuc); --- ncbi-blast-2.6.0+-src/c++/src/algo/blast/blastinput/blast_input.cpp 2016-06-20 08:45:40.000000000 -0700 +++ ncbi-blast-2.6.0+-src/c++/src/algo/blast/blastinput/blast_input.cpp 2017-02-07 11:56:27.300726093 -0800 @@ -53,12 +53,14 @@ CBlastInputSourceConfig::CBlastInputSour TSeqRange range /* = TSeqRange() */, bool retrieve_seq_data /* = true */, int local_id_counter /* = 1 */, - unsigned int seqlen_thresh2guess /* = numeric_limits::max()*/ ) + unsigned int seqlen_thresh2guess /* = numeric_limits::max()*/, + bool skip_seq_check /* = false -RMH- */ ) : m_Strand(strand), m_LowerCaseMask(lowercase), m_BelieveDeflines(believe_defline), m_Range(range), m_DLConfig(dlconfig), m_RetrieveSeqData(retrieve_seq_data), m_LocalIdCounter(local_id_counter), m_SeqLenThreshold2Guess(seqlen_thresh2guess), + m_SkipSeqCheck(skip_seq_check), /* -RMH- */ m_GapsToNs(false) { // Set an appropriate default for the strand --- ncbi-blast-2.6.0+-src/c++/include/algo/blast/blastinput/blast_input.hpp 2016-06-20 08:45:40.000000000 -0700 +++ ncbi-blast-2.6.0+-src/c++/include/algo/blast/blastinput/blast_input.hpp 2017-02-07 11:56:27.394726411 -0800 @@ -93,6 +93,8 @@ public: /// type guessing (see @ref kSeqLenThreshold2Guess) [in] /// @param local_id_counter counter used to create the CSeqidGenerator to /// create local identifiers for sequences read [in] + /// @param skip_seq_check When set this will avoid the sequence + /// validation step when using the CFastaReader. -RMH- CBlastInputSourceConfig(const SDataLoaderConfig& dlconfig, objects::ENa_strand strand = objects::eNa_strand_other, bool lowercase = false, @@ -101,7 +103,8 @@ public: bool retrieve_seq_data = true, int local_id_counter = 1, unsigned int seqlen_thresh2guess = - numeric_limits::max()); + numeric_limits::max(), + bool skip_seq_check = false /* -RMH- */ ); /// Destructor /// @@ -136,6 +139,18 @@ public: /// bool GetBelieveDeflines() const { return m_BelieveDeflines; } + /// Retrieve status of sequence alphabet validation + /// @return boolean to toggle validation of seq data + /// -RMH- + /// + bool GetSkipSeqCheck() const { return m_SkipSeqCheck; } + + /// Turn validation of sequence on/off + /// @param skip boolean to toggle validation of sequence + /// -RMH- + /// + void SetSkipSeqCheck(bool skip) { m_SkipSeqCheck = skip; } + /// Set range for all sequences /// @param r range to use [in] void SetRange(const TSeqRange& r) { m_Range = r; } @@ -201,6 +216,8 @@ private: bool m_LowerCaseMask; /// Whether to parse sequence IDs bool m_BelieveDeflines; + /// Whether to validate sequence data -RMH- + bool m_SkipSeqCheck; /// Sequence range TSeqRange m_Range; /// Configuration object for data loaders, used by CBlastInputReader diff -rupN ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.in ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.in --- ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.in 1969-12-31 16:00:00.000000000 -0800 +++ ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.in 2017-02-07 11:56:27.408726459 -0800 @@ -0,0 +1,16 @@ +# $Id: Makefile.in 371962 2012-08-14 09:45:56Z coulouri $ + +# Meta-makefile("APP" project) +################################# + +REQUIRES = objects algo + +APP_PROJ = rmblastn + +srcdir = @srcdir@ +include @builddir@/Makefile.meta + +.PHONY: all $(APP_PROJ) + +rmblastn: + ${MAKE} ${MFLAGS} -f Makefile.rmblastn_app diff -rupN ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.rmblastn.app ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.rmblastn.app --- ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.rmblastn.app 1969-12-31 16:00:00.000000000 -0800 +++ ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/Makefile.rmblastn.app 2017-02-07 11:56:27.426726519 -0800 @@ -0,0 +1,16 @@ +WATCHERS = camacho madden maning + +APP = rmblastn +SRC = rmblastn_app +LIB_ = $(BLAST_INPUT_LIBS) $(BLAST_LIBS) $(OBJMGR_LIBS) +LIB = blast_app_util $(LIB_:%=%$(STATIC)) + +# De-universalize Mac builds to work around a PPC toolchain limitation +CFLAGS = $(FAST_CFLAGS:ppc=i386) +CXXFLAGS = $(FAST_CXXFLAGS:ppc=i386) +LDFLAGS = $(FAST_LDFLAGS:ppc=i386) + +CPPFLAGS = $(ORIG_CPPFLAGS) +LIBS = $(CMPRS_LIBS) $(DL_LIBS) $(PCRE_LIBS) $(NETWORK_LIBS) $(ORIG_LIBS) + +REQUIRES = objects -Cygwin diff -rupN ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/rmblastn_app.cpp ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/rmblastn_app.cpp --- ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/rmblastn_app.cpp 1969-12-31 16:00:00.000000000 -0800 +++ ncbi-blast-2.6.0+-src/c++/src/app/rmblastn/rmblastn_app.cpp 2017-02-07 11:56:27.434726546 -0800 @@ -0,0 +1,183 @@ +/* $Id: rmblastn_app.cpp 371962 2012-08-14 09:45:56Z coulouri $ + * =========================================================================== + * + * PUBLIC DOMAIN NOTICE + * + * =========================================================================== + * + * Authors: Robert M. Hubley + * Christiam Camacho ( original blastn_app.cpp ) + * + */ + +/** @file rmblastn_app.cpp + * RMBLASTN command line application + */ + +#ifndef SKIP_DOXYGEN_PROCESSING +static char const rcsid[] = + "$Id: rmblastn_app.cpp 371962 2012-08-14 09:45:56Z coulouri $"; +#endif /* SKIP_DOXYGEN_PROCESSING */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include "../blast/blast_app_util.hpp" + +#ifndef SKIP_DOXYGEN_PROCESSING +USING_NCBI_SCOPE; +USING_SCOPE(blast); +USING_SCOPE(objects); +#endif + +class CRMBlastnApp : public CNcbiApplication +{ +public: + /** @inheritDoc */ + CRMBlastnApp() { + CRef version(new CVersion()); + version->SetVersionInfo(new CBlastVersion()); + SetFullVersion(version); + } +private: + /** @inheritDoc */ + virtual void Init(); + /** @inheritDoc */ + virtual int Run(); + + /// This application's command line args + CRef m_CmdLineArgs; +}; + + + +void CRMBlastnApp::Init() +{ + // formulate command line arguments + m_CmdLineArgs.Reset(new CRMBlastnAppArgs()); + + // read the command line + + HideStdArgs(fHideLogfile | fHideConffile | fHideFullVersion | fHideXmlHelp | fHideDryRun); + SetupArgDescriptions(m_CmdLineArgs->SetCommandLine()); +} + +int CRMBlastnApp::Run(void) +{ + int status = BLAST_EXIT_SUCCESS; + + try { + + // Allow the fasta reader to complain on invalid sequence input + SetDiagPostLevel(eDiag_Warning); + + /*** Get the BLAST options ***/ + const CArgs& args = GetArgs(); + CRef opts_hndl; + if(RecoverSearchStrategy(args, m_CmdLineArgs)) { + opts_hndl.Reset(&*m_CmdLineArgs->SetOptionsForSavedStrategy(args)); + } + else { + opts_hndl.Reset(&*m_CmdLineArgs->SetOptions(args)); + } + const CBlastOptions& opt = opts_hndl->GetOptions(); + + /*** Get the query sequence(s) ***/ + CRef query_opts = + m_CmdLineArgs->GetQueryOptionsArgs(); + SDataLoaderConfig dlconfig(query_opts->QueryIsProtein()); + dlconfig.OptimizeForWholeLargeSequenceRetrieval(); + CBlastInputSourceConfig iconfig(dlconfig, query_opts->GetStrand(), + query_opts->UseLowercaseMasks(), + query_opts->GetParseDeflines(), + query_opts->GetRange()); + iconfig.SetSkipSeqCheck(true); + CBlastFastaInputSource fasta(m_CmdLineArgs->GetInputStream(), iconfig); + CBlastInput input(&fasta, m_CmdLineArgs->GetQueryBatchSize()); + + /*** Initialize the database/subject ***/ + CRef db_args(m_CmdLineArgs->GetBlastDatabaseArgs()); + CRef db_adapter; + CRef scope; + InitializeSubject(db_args, opts_hndl, m_CmdLineArgs->ExecuteRemotely(), + db_adapter, scope); + _ASSERT(db_adapter && scope); + + // Initialize the megablast database index now so we can know whether an indexed search will be run. + // This is only important for the reference in the report, but would be done anyway. + if (opt.GetUseIndex() && !m_CmdLineArgs->ExecuteRemotely()) { + CRef my_options(&(opts_hndl->SetOptions())); + CSetupFactory::InitializeMegablastDbIndex(my_options); + } + + /*** Get the formatting options ***/ + CRef fmt_args(m_CmdLineArgs->GetFormattingArgs()); + CBlastFormat formatter(opt, *db_adapter, + fmt_args->GetFormattedOutputChoice(), + query_opts->GetParseDeflines(), + m_CmdLineArgs->GetOutputStream(), + fmt_args->GetNumDescriptions(), + fmt_args->GetNumAlignments(), + *scope, + opt.GetMatrixName(), + fmt_args->ShowGis(), + fmt_args->DisplayHtmlOutput(), + opt.GetQueryGeneticCode(), + opt.GetDbGeneticCode(), + opt.GetSumStatisticsMode(), + m_CmdLineArgs->ExecuteRemotely(), + db_adapter->GetFilteringAlgorithm(), + fmt_args->GetCustomOutputFormatSpec(), + m_CmdLineArgs->GetTask() == "megablast", + opt.GetMBIndexLoaded()); + + + formatter.PrintProlog(); + + /*** Process the input ***/ + for (; !input.End(); formatter.ResetScopeHistory()) { + + CRef query_batch(input.GetNextSeqBatch(*scope)); + CRef queries(new CObjMgr_QueryFactory(*query_batch)); + + SaveSearchStrategy(args, m_CmdLineArgs, queries, opts_hndl); + + CRef results; + + if (m_CmdLineArgs->ExecuteRemotely()) { + CRef rmt_blast = + InitializeRemoteBlast(queries, db_args, opts_hndl, + m_CmdLineArgs->ProduceDebugRemoteOutput()); + results = rmt_blast->GetResultSet(); + } else { + CLocalBlast lcl_blast(queries, opts_hndl, db_adapter); + lcl_blast.SetNumberOfThreads(m_CmdLineArgs->GetNumThreads()); + results = lcl_blast.Run(); + } + + ITERATE(CSearchResultSet, result, *results) { + formatter.PrintOneResultSet(**result, query_batch); + } + } + + formatter.PrintEpilog(opt); + + if (m_CmdLineArgs->ProduceDebugOutput()) { + opts_hndl->GetOptions().DebugDumpText(NcbiCerr, "BLAST options", 1); + } + + } CATCH_ALL(status) + return status; +} + +#ifndef SKIP_DOXYGEN_PROCESSING +int main(int argc, const char* argv[] /*, const char* envp[]*/) +{ + return CRMBlastnApp().AppMain(argc, argv, 0, eDS_Default, 0); +} +#endif /* SKIP_DOXYGEN_PROCESSING */ --- ncbi-blast-2.6.0+-src/c++/src/objtools/blast/seqdb_reader/seqdbimpl.cpp 2016-10-12 07:27:34.000000000 -0700 +++ ncbi-blast-2.6.0+-src/c++/src/objtools/blast/seqdb_reader/seqdbimpl.cpp 2017-03-29 11:14:01.852864789 -0700 @@ -1813,6 +1813,9 @@ void CSeqDBImpl::HashToOids(unsigned has CSeqDBIdSet CSeqDBImpl::GetIdSet() { + // RMH: Added to eliminate race condition for vector garbage collection + CFastMutexGuard guard(m_MutexForIdSet); + if (m_IdSet.Blank()) { if (! m_UserGiList.Empty()) { // Note: this returns a 'blank' IdSet list for positive --- ncbi-blast-2.6.0+-src/c++/src/objtools/blast/seqdb_reader/seqdbimpl.hpp 2016-02-03 07:19:38.000000000 -0800 +++ ncbi-blast-2.6.0+-src/c++/src/objtools/blast/seqdb_reader/seqdbimpl.hpp 2017-03-29 11:12:33.521480161 -0700 @@ -1133,6 +1133,9 @@ public: private: CLASS_MARKER_FIELD("IMPL") + // RMH + CFastMutex m_MutexForIdSet; + /// Adjust string length to offset of first embedded NUL byte. /// /// This is a work-around for bad data in the database; probably