CMultiReader Class Reference

Search Toolkit Book for CMultiReader

#include "multireader.hpp"
(Private to src/app/table2asn.)

Collaboration diagram for CMultiReader:

Public Types
using	TAnnots = list< CRef< objects::CSeq_annot > >

Public Member Functions
	CMultiReader (CTable2AsnContext &context)
	~CMultiReader ()
void	LoadTemplate (const string &ifname)
void	LoadGFF3Fasta (istream &in, TAnnots &annots)
void	AddAnnots (IIndexedFeatureReader *reader, CBioseq &bioseq) const
void	LoadIndexedAnnot (std::unique_ptr< IIndexedFeatureReader > &reader, const string &filename)
void	GetIndexedAnnot (std::unique_ptr< IIndexedFeatureReader > &reader, TAnnots &annots)
void	LoadDescriptors (const string &ifname, CRef< objects::CSeq_descr > &out_desc) const
void	ApplyDescriptors (objects::CSeq_entry &obj, const objects::CSeq_descr &source) const
void	WriteObject (const CSerialObject &, ostream &)
CRef< objects::CSeq_entry >	ReadAlignment (CNcbiIstream &instream, const CArgs &args)
CRef< CSerialObject >	ReadNextEntry ()
CFormatGuess::EFormat	OpenFile (const string &filename, CRef< CSerialObject > &input_sequence, TAnnots &annots)
CRef< CSerialObject >	FetchEntry (const CFormatGuess::EFormat &format, const string &objectType, unique_ptr< CNcbiIstream > &pIstr, TAnnots &annots)

Static Public Member Functions
static void	GetSeqEntry (CRef< objects::CSeq_entry > &entry, CRef< objects::CSeq_submit > &submit, CRef< CSerialObject > obj)

Static Public Attributes
static const set< TTypeInfo >	kSupportedTypes

Private Member Functions
void	MergeDescriptors (objects::CSeq_descr &dest, const objects::CSeq_descr &source) const
void	MergeDescriptors (objects::CSeq_descr &dest, const objects::CSeqdesc &source) const
bool	AtSeqenceData () const
CRef< objects::CSeq_entry >	xReadFasta (CNcbiIstream &instream)
CRef< CSerialObject >	xApplyTemplate (CRef< CSerialObject > obj, bool merge_template_descriptors) const
CRef< CSerialObject >	xReadASN1Text (CObjectIStream &pObjIstrm) const
CRef< CSerialObject >	xReadASN1Binary (CObjectIStream &pObjIstrm, const string &content_type) const
TAnnots	xReadGFF3 (CNcbiIstream &instream, bool post_process)
TAnnots	xReadGTF (CNcbiIstream &instream) const
CRef< objects::CSeq_entry >	xReadFlatfile (CFormatGuess::EFormat format, const string &filename, CNcbiIstream &instream)
void	x_PostProcessAnnots (TAnnots &annots) const
unique_ptr< CObjectIStream >	xCreateASNStream (const string &filename) const
unique_ptr< CObjectIStream >	xCreateASNStream (CFormatGuess::EFormat format, unique_ptr< istream > &instream) const
CFormatGuess::EFormat	xInputGetFormat (CNcbiIstream &, CFileContentInfo *=nullptr) const
void	xAnnotGetFormat (objects::edit::CHugeFile &file) const

Private Attributes
int	m_iFlags
string	m_AnnotName
string	m_AnnotTitle
CTable2AsnContext &	m_context
unique_ptr< CObjectIStream >	m_obj_stream
shared_ptr< objects::CGff3LocationMerger >	m_gff3_merger
bool	mAtSequenceData
shared_mutex	m_Mutex

Detailed Description

Definition at line 41 of file multireader.hpp.

Member Typedef Documentation

TAnnots

using CMultiReader::TAnnots = list<CRef<objects::CSeq_annot> >

Definition at line 48 of file multireader.hpp.

Constructor & Destructor Documentation

CMultiReader()

CMultiReader::CMultiReader

(

CTable2AsnContext &

context

)

Definition at line 463 of file multireader.cpp.

~CMultiReader()

CMultiReader::~CMultiReader

(

)

Definition at line 1017 of file multireader.cpp.

Member Function Documentation

AddAnnots()

void CMultiReader::AddAnnots	(	IIndexedFeatureReader *	reader,
		CBioseq &	bioseq
	)		const

Definition at line 1113 of file multireader.cpp.

References IIndexedFeatureReader::GetAndUseAnnot(), IIndexedFeatureReader::GetAnnotId(), CBioseq_Base::GetId(), l(), r(), and ct::sort().

ApplyDescriptors()

void CMultiReader::ApplyDescriptors	(	objects::CSeq_entry &	obj,
		const objects::CSeq_descr &	source
	)		const

Definition at line 701 of file multireader.cpp.

References MergeDescriptors(), CSeq_entry::SetDescr(), and rapidjson::source.

AtSeqenceData()

bool CMultiReader::AtSeqenceData ( ) const

inlineprivate

Definition at line 75 of file multireader.hpp.

References mAtSequenceData.

Referenced by LoadGFF3Fasta().

FetchEntry()

CRef< CSerialObject > CMultiReader::FetchEntry	(	const CFormatGuess::EFormat &	format,
		const string &	objectType,
		unique_ptr< CNcbiIstream > &	pIstr,
		TAnnots &	annots
	)

Definition at line 742 of file multireader.cpp.

References CFormatGuess::eBinaryASN, CFormatGuess::eFasta, CFormatGuess::eGff3, eSerial_AsnBinary, eSerial_AsnText, eTakeOwnership, CFormatGuess::eTextASN, CFastaReader::fNoUserObjs, format, LoadGFF3Fasta(), m_iFlags, m_obj_stream, NCBI_THROW2, CObjectIStream::Open(), xApplyTemplate(), xReadASN1Binary(), xReadASN1Text(), and xReadFasta().

GetIndexedAnnot()

void CMultiReader::GetIndexedAnnot	(	std::unique_ptr< IIndexedFeatureReader > &	reader,
		TAnnots &	annots
	)

Definition at line 1222 of file multireader.cpp.

References m_context, and CTable2AsnContext::m_genome_center_id.

GetSeqEntry()

void CMultiReader::GetSeqEntry ( CRef< objects::CSeq_entry > &  entry, CRef< objects::CSeq_submit > &  submit, CRef< CSerialObject >  obj  )

static

Definition at line 826 of file multireader.cpp.

References CRef< C, Locker >::GetPointer(), CSerialObject::GetThisTypeInfo(), CRef< C, Locker >::Reset(), and CSeq_submit_Base::SetData().

Referenced by LoadIndexedAnnot(), and xApplyTemplate().

LoadDescriptors()

void CMultiReader::LoadDescriptors	(	const string &	ifname,
		CRef< objects::CSeq_descr > &	out_desc
	)		const

Definition at line 469 of file multireader.cpp.

References CObjectIStream::eNoFileHeader, NStr::EqualNocase(), CSeq_descr_Base::Get(), CException::GetMsg(), ObjectInfo(), CRef< C, Locker >::Reset(), CSeq_descr_Base::Set(), CSeqdesc_Base::SetPub(), and xCreateASNStream().

LoadGFF3Fasta()

void CMultiReader::LoadGFF3Fasta	(	istream &	in,
		TAnnots &	annots
	)

Definition at line 725 of file multireader.cpp.

References AtSeqenceData(), eDiag_Info, CFormatGuess::eGff3, g_LogDiagMessage(), CFormatGuess::GetFormatName(), in(), m_context, CTable2AsnContext::m_logger, NCBI_THROW2, x_PostProcessAnnots(), and xReadGFF3().

Referenced by FetchEntry(), and OpenFile().

LoadIndexedAnnot()

void CMultiReader::LoadIndexedAnnot	(	std::unique_ptr< IIndexedFeatureReader > &	reader,
		const string &	filename
	)

Definition at line 1021 of file multireader.cpp.

References eDiag_Info, CFormatGuess::eFiveColFeatureTable, CFormatGuess::eFlatFileEna, CFormatGuess::eFlatFileGenbank, CFormatGuess::eFlatFileUniProt, CFormatGuess::eGff3, CFormatGuess::eGffAugustus, CFormatGuess::eGtf, CFormatGuess::eTextASN, CFormatGuess::eUnknown, CFeature_table_reader::fAllIdsAsLocal, CFeature_table_reader::fCreateGenesFromCDSs, CFeature_table_reader::fLeaveProteinIds, CFeature_table_reader::fPreferGenbankId, g_LogDiagMessage(), CSeq_entry::GetAnnot(), CFormatGuess::GetFormatName(), GetSeqEntry(), in(), CSeq_entry::IsSetAnnot(), m_context, CTable2AsnContext::m_genome_center_id, CTable2AsnContext::m_logger, NCBI_THROW2, CDirEntry::SplitPath(), NStr::ToLower(), unused, xAnnotGetFormat(), xReadASN1Text(), xReadFlatfile(), xReadGFF3(), and xReadGTF().

LoadTemplate()

void CMultiReader::LoadTemplate

(

const string &

ifname

)

Definition at line 510 of file multireader.cpp.

References CSerialObject::Assign(), context, CSeqdesc_Base::e_Pub, CSeqdesc_Base::e_Source, CTime::eCurrent, CSeq_inst_Base::eMol_dna, CRef< C, Locker >::Empty(), CObjectIStream::eNoFileHeader, CDate::ePrecision_day, CSeq_inst_Base::eRepr_raw, CSeq_descr_Base::Get(), CSeqdesc_Base::GetUser(), CUser_object::IsDBLink(), CSeqdesc_Base::IsUser(), ITERATE, m_context, CTable2AsnContext::m_entry_template, CTable2AsnContext::m_submit_template, NCBI_USER_THROW_FMT, CRef< C, Locker >::NotEmpty(), ObjectInfo(), CRef< C, Locker >::Reset(), CUser_object_Base::SetData(), CBioseq_Base::SetId(), CBioseq_Base::SetInst(), CPubdesc_Base::SetPub(), CSeqdesc_Base::SetPub(), CSeq_entry_Base::SetSeq(), CPub_Base::SetSub(), CSeqdesc_Base::SetUser(), tmp, and xCreateASNStream().

MergeDescriptors() [1/2]

void CMultiReader::MergeDescriptors ( objects::CSeq_descr &  dest, const objects::CSeq_descr &  source  ) const

private

Referenced by ApplyDescriptors().

MergeDescriptors() [2/2]

void CMultiReader::MergeDescriptors ( objects::CSeq_descr &  dest, const objects::CSeqdesc &  source  ) const

private

OpenFile()

CFormatGuess::EFormat CMultiReader::OpenFile	(	const string &	filename,
		CRef< CSerialObject > &	input_sequence,
		TAnnots &	annots
	)

Definition at line 776 of file multireader.cpp.

References CFormatGuess::eBinaryASN, CFormatGuess::eFasta, CFormatGuess::eGff3, CRef< C, Locker >::Empty(), eSerial_AsnBinary, eSerial_AsnText, CFormatGuess::eTextASN, CFastaReader::fNoUserObjs, format, CBufferedStream< _Stream, _bufsize >::get(), in(), LoadGFF3Fasta(), m_iFlags, m_obj_stream, CFileContentInfo::mInfoGenbank, CFileContentInfoGenbank::mObjectType, NCBI_THROW2, CObjectIStream::Open(), xApplyTemplate(), xInputGetFormat(), xReadASN1Binary(), xReadASN1Text(), and xReadFasta().

ReadAlignment()

CRef< CSeq_entry > CMultiReader::ReadAlignment	(	CNcbiIstream &	instream,
		const CArgs &	args
	)

Definition at line 285 of file multireader.cpp.

References CAlnReader::eAlpha_Nucleotide, CAlnReader::eAlpha_Protein, CBioseq_set_Base::eClass_eco_set, CBioseq_set_Base::eClass_genbank, CBioseq_set_Base::eClass_mut_set, CBioseq_set_Base::eClass_phy_set, CBioseq_set_Base::eClass_pop_set, CBioseq_set_Base::eClass_small_genome_set, map_checker< Container >::end(), CAlnReader::fGenerateLocalIDs, map_checker< Container >::find(), CAlnReader::GetSeqEntry(), m_context, CTable2AsnContext::m_logger, CAlnReader::Read(), CAlnReader::SetAllGap(), CAlnReader::SetAlphabet(), and CAlnReader::SetMissing().

ReadNextEntry()

CRef< CSerialObject > CMultiReader::ReadNextEntry

(

)

Definition at line 877 of file multireader.cpp.

References m_obj_stream, and xReadASN1Text().

WriteObject()

void CMultiReader::WriteObject	(	const CSerialObject &	object,
		ostream &	ostr
	)

Definition at line 453 of file multireader.cpp.

References eSerial_AsnBinary, eSerial_AsnText, CTable2AsnContext::m_binary_asn1_output, and m_context.

x_PostProcessAnnots()

void CMultiReader::x_PostProcessAnnots ( TAnnots &  annots ) const

private

Definition at line 939 of file multireader.cpp.

References data, m_context, CTable2AsnContext::m_eukaryote, m_gff3_merger, CTable2AsnContext::m_locus_tag_prefix, CTable2AsnContext::m_locus_tags_needed, CTable2AsnContext::m_logger, NCBI_THROW, s_RemoveGenBankDbxrefs(), and CSeq_annot_Base::SetData().

Referenced by LoadGFF3Fasta(), xReadGFF3(), and xReadGTF().

xAnnotGetFormat()

void CMultiReader::xAnnotGetFormat ( objects::edit::CHugeFile &  file ) const

private

Definition at line 417 of file multireader.cpp.

References CFormatGuess::CFormatHints::AddPreferredFormat(), CFormatGuess::CFormatHints::DisableAllNonpreferred(), CFormatGuess::eBinaryASN, CFormatGuess::eFiveColFeatureTable, CFormatGuess::eFlatFileEna, CFormatGuess::eFlatFileGenbank, CFormatGuess::eFlatFileUniProt, CFormatGuess::eGff3, CFormatGuess::eGffAugustus, CFormatGuess::eGtf, eSerial_AsnBinary, eSerial_AsnText, CFormatGuess::eTextASN, file, CFormatGuess::GetFormatHints(), CFormatGuess::GuessFormat(), and in().

Referenced by LoadIndexedAnnot().

xApplyTemplate()

CRef< CSerialObject > CMultiReader::xApplyTemplate ( CRef< CSerialObject >  obj, bool  merge_template_descriptors  ) const

private

Definition at line 836 of file multireader.cpp.

References CLineError::Create(), eDiag_Warning, CRef< C, Locker >::Empty(), ILineError::eProblem_Unset, CBioseq_set_Base::GetSeq_set(), GetSeqEntry(), CSeq_entry_Base::GetSet(), CSeq_entry_Base::IsSet(), m_context, CTable2AsnContext::m_logger, CTable2AsnContext::m_t, CTable2AsnContext::MergeWithTemplate(), msg(), CRef< C, Locker >::NotEmpty(), CSeq_entry::Parentize(), CSeq_entry::ResetParentEntry(), CBioseq_set_Base::SetSeq_set(), and CSeq_entry_Base::SetSet().

Referenced by FetchEntry(), and OpenFile().

xCreateASNStream() [1/2]

unique_ptr< CObjectIStream > CMultiReader::xCreateASNStream ( CFormatGuess::EFormat  format, unique_ptr< istream > &  instream  ) const

private

Definition at line 982 of file multireader.cpp.

References CFormatGuess::eBinaryASN, eSerial_AsnBinary, eSerial_AsnText, eSerial_None, eSerial_Xml, eTakeOwnership, CFormatGuess::eTextASN, CFormatGuess::eUnknown, CFormatGuess::eXml, format, CFormatGuess::GetFormatName(), NCBI_USER_THROW_FMT, CObjectIStream::Open(), and xInputGetFormat().

xCreateASNStream() [2/2]

unique_ptr< CObjectIStream > CMultiReader::xCreateASNStream ( const string &  filename ) const

private

Definition at line 976 of file multireader.cpp.

References CFormatGuess::eUnknown.

Referenced by LoadDescriptors(), and LoadTemplate().

xInputGetFormat()

CFormatGuess::EFormat CMultiReader::xInputGetFormat ( CNcbiIstream &  istr, CFileContentInfo *  content_info = nullptr  ) const

private

Definition at line 398 of file multireader.cpp.

References CFormatGuess::CFormatHints::AddPreferredFormat(), CFormatGuess::CFormatHints::DisableAllNonpreferred(), CFormatGuess::eBinaryASN, CFormatGuess::eGff3, CFormatGuess::eTextASN, CFormatGuessEx::GetFormatHints(), CFormatGuessEx::GuessFormat(), CFormatGuessEx::GuessFormatAndContent(), kSupportedTypes, and CFormatGuessEx::SetRecognizedGenbankTypes().

Referenced by OpenFile(), and xCreateASNStream().

xReadASN1Binary()

CRef< CSerialObject > CMultiReader::xReadASN1Binary ( CObjectIStream &  pObjIstrm, const string &  content_type  ) const

private

Definition at line 189 of file multireader.cpp.

References ObjectInfo(), CObjectIStream::Read(), and Ref().

Referenced by FetchEntry(), and OpenFile().

xReadASN1Text()

CRef< CSerialObject > CMultiReader::xReadASN1Text ( CObjectIStream &  pObjIstrm ) const

private

Definition at line 219 of file multireader.cpp.

References CGapsEditor::ConvertNs2Gaps(), CRef< C, Locker >::Empty(), CObjectIStream::eNoFileHeader, CSeq_submit_Base::GetData(), CSeq_submit_Base::C_Data::GetEntrys(), m_context, CTable2AsnContext::m_DefaultEvidence, CTable2AsnContext::m_gap_type, CTable2AsnContext::m_gap_Unknown_length, CTable2AsnContext::m_gapNmin, CTable2AsnContext::m_GapsizeToEvidence, ObjectInfo(), CObjectIStream::Read(), CObjectIStream::ReadFileHeader(), CRef< C, Locker >::Reset(), CBioseq_Base::SetAnnot(), CSeq_submit_Base::SetData(), CSeq_entry_Base::SetSeq(), CBioseq_set_Base::SetSeq_set(), and CSeq_entry_Base::SetSet().

Referenced by FetchEntry(), LoadIndexedAnnot(), OpenFile(), and ReadNextEntry().

xReadFasta()

CRef< CSeq_entry > CMultiReader::xReadFasta ( CNcbiIstream &  instream )

private

Definition at line 326 of file multireader.cpp.

References eDiag_Warning, CFastaReader::fAssumeNuc, CFastaReader::fDisableParseRange, CFastaReader::fForceType, CFastaReader::fHyphensIgnoreAndWarn, CFastaReader::fIgnoreMods, CFastaReader::fLetterGaps, CFastaReader::fNoSplit, CFastaReader::fParseGaps, CFastaReader::fParseRawID, CFastaReader::fValidate, g_LogGeneralParsingError(), kMax_Int, CTable2AsnContext::m_allow_accession, CTable2AsnContext::m_ClassValue, m_context, CTable2AsnContext::m_current_file, CTable2AsnContext::m_d_fasta, CTable2AsnContext::m_DefaultEvidence, CTable2AsnContext::m_di_fasta, CTable2AsnContext::m_gap_type, CTable2AsnContext::m_gap_Unknown_length, CTable2AsnContext::m_gapNmin, CTable2AsnContext::m_GapsizeToEvidence, CTable2AsnContext::m_HandleAsSet, m_iFlags, CTable2AsnContext::m_logger, CTable2AsnContext::MakeGenomeCenterId(), NCBI_THROW2, and result.

Referenced by FetchEntry(), and OpenFile().

xReadFlatfile()

CRef< CSeq_entry > CMultiReader::xReadFlatfile ( CFormatGuess::EFormat  format, const string &  filename, CNcbiIstream &  instream  )

private

Definition at line 1173 of file multireader.cpp.

References Parser::BioseqSet, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_Swissprot, CFormatGuess::eFlatFileEna, CFormatGuess::eFlatFileGenbank, CFormatGuess::eFlatFileUniProt, Parser::EMBL, format, Parser::GenBank, CBioseq_Base::IsSetAnnot(), m_context, CTable2AsnContext::m_logger, NCBI_THROW2, ParFlat_EMBL_AC, CFlatFileParser::Parse(), Ref(), CBioseq_Base::SetAnnot(), and Parser::SPROT.

Referenced by LoadIndexedAnnot().

xReadGFF3()

CMultiReader::TAnnots CMultiReader::xReadGFF3 ( CNcbiIstream &  instream, bool  post_process  )

private

Definition at line 885 of file multireader.cpp.

References CReadUtil::AsSeqId(), CGff2Reader::AtSequenceData(), CReaderBase::fAllIdsAsLocal, CGff2Reader::fGenbankMode, CGff3Reader::fGeneXrefs, flags, CGff2Reader::fRetainLocusIds, CGff3Reader::GetLocationMerger(), lr(), m_AnnotName, m_AnnotTitle, m_context, m_gff3_merger, CTable2AsnContext::m_logger, mAtSequenceData, msg(), CGff2Reader::ReadSeqAnnots(), and x_PostProcessAnnots().

Referenced by LoadGFF3Fasta(), and LoadIndexedAnnot().

xReadGTF()

CMultiReader::TAnnots CMultiReader::xReadGTF ( CNcbiIstream &  instream ) const

private

Definition at line 1152 of file multireader.cpp.

References CReaderBase::fAllIdsAsLocal, CGff2Reader::fGenbankMode, CGtfReader::fGenerateChildXrefs, flags, lr(), m_AnnotName, m_AnnotTitle, m_context, CTable2AsnContext::m_logger, msg(), CGff2Reader::ReadSeqAnnots(), and x_PostProcessAnnots().

Referenced by LoadIndexedAnnot().

Member Data Documentation

kSupportedTypes

const set< TTypeInfo > CMultiReader::kSupportedTypes

static

Initial value:

= {
    CBioseq_set::GetTypeInfo(),
    CBioseq::GetTypeInfo(),
    CSeq_entry::GetTypeInfo(),
    CSeq_submit::GetTypeInfo(),
    CSeq_annot::GetTypeInfo(),
}

Definition at line 50 of file multireader.hpp.

Referenced by CTbl2AsnApp::ProcessOneFile(), and xInputGetFormat().

m_AnnotName

string CMultiReader::m_AnnotName

private

Definition at line 92 of file multireader.hpp.

Referenced by xReadGFF3(), and xReadGTF().

m_AnnotTitle

string CMultiReader::m_AnnotTitle

private

Definition at line 93 of file multireader.hpp.

Referenced by xReadGFF3(), and xReadGTF().

m_context

CTable2AsnContext& CMultiReader::m_context

private

Definition at line 94 of file multireader.hpp.

Referenced by GetIndexedAnnot(), LoadGFF3Fasta(), LoadIndexedAnnot(), LoadTemplate(), ReadAlignment(), WriteObject(), x_PostProcessAnnots(), xApplyTemplate(), xReadASN1Text(), xReadFasta(), xReadFlatfile(), xReadGFF3(), and xReadGTF().

m_gff3_merger

shared_ptr<objects::CGff3LocationMerger> CMultiReader::m_gff3_merger

private

Definition at line 96 of file multireader.hpp.

Referenced by x_PostProcessAnnots(), and xReadGFF3().

m_iFlags

int CMultiReader::m_iFlags

private

Definition at line 91 of file multireader.hpp.

Referenced by FetchEntry(), OpenFile(), and xReadFasta().

m_Mutex

shared_mutex CMultiReader::m_Mutex

mutableprivate

Definition at line 98 of file multireader.hpp.

m_obj_stream

unique_ptr<CObjectIStream> CMultiReader::m_obj_stream

private

Definition at line 95 of file multireader.hpp.

Referenced by FetchEntry(), OpenFile(), and ReadNextEntry().

mAtSequenceData

bool CMultiReader::mAtSequenceData

private

Definition at line 97 of file multireader.hpp.

Referenced by AtSeqenceData(), and xReadGFF3().

---

The documentation for this class was generated from the following files:

• src/app/table2asn/multireader.hpp
• src/app/table2asn/multireader.cpp