CCleanupHugeAsnReader Class Reference

Search Toolkit Book for CCleanupHugeAsnReader

#include "huge_file_cleanup.hpp"
(Private to src/app/asn_cleanup.)

Inheritance diagram for CCleanupHugeAsnReader:

Collaboration diagram for CCleanupHugeAsnReader:

Classes
struct	TFeatIdInfo

Public Types
enum	EOptions { eExtendedCleanup = 1 , eNoNcbiUserObjects = 1<<1 , eEnableSmallGenomeSets = 1<<2 }
using	TOptions = int
using	TParent = edit::CHugeAsnReader
using	TFeatId = CFeat_id::TLocal::TId

Public Member Functions
	CCleanupHugeAsnReader (TOptions options)
virtual	~CCleanupHugeAsnReader ()=default
void	FlattenGenbankSet () override
CRef< CSeq_entry >	LoadSeqEntry (const TBioseqSetInfo &info, eAddTopEntry add_top_entry=eAddTopEntry::yes) const override
const CCleanupChangeCore &	GetChanges () const

Private Types
using	TIdToFluLabel = map< CConstRef< CSeq_id >, string, CRefLess >

Private Member Functions
bool	x_IsExtendedCleanup () const
void	x_SetHooks (CObjectIStream &objStream, TContext &context) override
void	x_SetBioseqHooks (CObjectIStream &objStream, TContext &context) override
void	x_SetBioseqSetHooks (CObjectIStream &objStream, TContext &context) override
void	x_SetSeqFeatHooks (CObjectIStream &objStream, TContext &context)
void	x_RecordFeatureId (const CFeat_id &featId)
void	x_CreateSmallGenomeSets ()
void	x_PruneIfFeatsIncomplete ()
void	x_PruneAndReorderTopIds ()
void	x_PruneIfSegsMissing (const string &fluLabel, const set< size_t > &segsFound)
void	x_CleanupTopLevelDescriptors ()
bool	x_LooksLikeNucProtSet () const
void	x_AddTopLevelDescriptors (CSeq_entry &entry) const
TIdToFluLabel::iterator	x_GetFluLabel (const CConstRef< CSeq_id > &pId)

Private Attributes
list< CRef< CSeqdesc > >	m_TopLevelBiosources
CRef< CSeqdesc >	m_pTopLevelMolInfo
const TOptions	m_CleanupOptions
CCleanupChangeCore	m_Changes
TIdToFluLabel	m_IdToFluLabel
map< string, list< TBioseqSetInfo > >	m_FluLabelToSetInfo
map< TFileSize, string >	m_SetPosToFluLabel
set< CConstRef< CSeq_id >, CRefLess >	m_HasIncompleteFeats
TFeatIdInfo	m_FeatIdInfo

Detailed Description

Definition at line 49 of file huge_file_cleanup.hpp.

Member Typedef Documentation

TFeatId

using CCleanupHugeAsnReader::TFeatId = CFeat_id::TLocal::TId

Definition at line 69 of file huge_file_cleanup.hpp.

TIdToFluLabel

using CCleanupHugeAsnReader::TIdToFluLabel = map<CConstRef<CSeq_id>,string,CRefLess>

private

Definition at line 91 of file huge_file_cleanup.hpp.

TOptions

using CCleanupHugeAsnReader::TOptions = int

Definition at line 58 of file huge_file_cleanup.hpp.

TParent

using CCleanupHugeAsnReader::TParent = edit::CHugeAsnReader

Definition at line 62 of file huge_file_cleanup.hpp.

Member Enumeration Documentation

EOptions

enum CCleanupHugeAsnReader::EOptions

Enumerator
eExtendedCleanup
eNoNcbiUserObjects
eEnableSmallGenomeSets

Definition at line 52 of file huge_file_cleanup.hpp.

Constructor & Destructor Documentation

CCleanupHugeAsnReader()

CCleanupHugeAsnReader::CCleanupHugeAsnReader

(

TOptions

options

)

Definition at line 54 of file huge_file_cleanup.cpp.

~CCleanupHugeAsnReader()

virtual CCleanupHugeAsnReader::~CCleanupHugeAsnReader ( )

virtualdefault

Member Function Documentation

FlattenGenbankSet()

void CCleanupHugeAsnReader::FlattenGenbankSet ( )

override

Definition at line 64 of file huge_file_cleanup.cpp.

References CCleanupChangeCore::eChangeBioseqSetClass, CBioseq_set_Base::eClass_genbank, CBioseq_set_Base::eClass_not_set, CBioseq_set_Base::eClass_nuc_prot, eEnableSmallGenomeSets, m_Changes, m_CleanupOptions, next(), CCleanupChangeCore::SetChanged(), x_CleanupTopLevelDescriptors(), x_CreateSmallGenomeSets(), x_LooksLikeNucProtSet(), and x_PruneAndReorderTopIds().

GetChanges()

const CCleanupChangeCore & CCleanupHugeAsnReader::GetChanges

(

)

const

Definition at line 58 of file huge_file_cleanup.cpp.

References m_Changes.

LoadSeqEntry()

CRef< CSeq_entry > CCleanupHugeAsnReader::LoadSeqEntry ( const TBioseqSetInfo &  info, eAddTopEntry  add_top_entry = eAddTopEntry::yes  ) const

override

Definition at line 344 of file huge_file_cleanup.cpp.

References CBioseq_set_Base::eClass_small_genome_set, eEnableSmallGenomeSets, map_checker< Container >::end(), map_checker< Container >::find(), info, m_CleanupOptions, m_FeatIdInfo, m_FluLabelToSetInfo, m_SetPosToFluLabel, CCleanupHugeAsnReader::TFeatIdInfo::PosToIdMap, Ref(), s_UpdateFeatureIds(), x_AddTopLevelDescriptors(), x_IsExtendedCleanup(), and yes.

x_AddTopLevelDescriptors()

void CCleanupHugeAsnReader::x_AddTopLevelDescriptors ( CSeq_entry &  entry ) const

private

Definition at line 209 of file huge_file_cleanup.cpp.

References CCleanupChangeCore::eAddDescriptor, CRef< C, Locker >::Empty(), CSeq_descr_Base::Get(), CSeq_entry::GetDescr(), CSeq_descr_Base::IsSet(), CSeq_entry::IsSetDescr(), m_Changes, m_pTopLevelMolInfo, m_TopLevelBiosources, CCleanupChangeCore::SetChanged(), CSeq_entry::SetDescr(), and x_IsExtendedCleanup().

Referenced by LoadSeqEntry().

x_CleanupTopLevelDescriptors()

void CCleanupHugeAsnReader::x_CleanupTopLevelDescriptors ( )

private

Definition at line 151 of file huge_file_cleanup.cpp.

References CCleanup::AddNcbiCleanupObject(), cleanup(), CCleanupChangeCore::eAddDescriptor, CCleanupChangeCore::eAddNcbiCleanupObject, CCleanupChangeCore::eMoveDescriptor, eNoNcbiUserObjects, CCleanupChangeCore::eRemoveDescriptor, m_Changes, m_CleanupOptions, m_pTopLevelMolInfo, m_TopLevelBiosources, CCleanup::NormalizeDescriptorOrder(), CRef< C, Locker >::Reset(), CCleanupChangeCore::SetChanged(), and x_IsExtendedCleanup().

Referenced by FlattenGenbankSet().

x_CreateSmallGenomeSets()

void CCleanupHugeAsnReader::x_CreateSmallGenomeSets ( )

private

Definition at line 473 of file huge_file_cleanup.cpp.

References CBioseq_set_Base::eClass_nuc_prot, NStr::IsBlank(), CSeq_inst::IsNa(), m_FluLabelToSetInfo, m_IdToFluLabel, m_SetPosToFluLabel, s_CheckForSegments(), s_GetInfluenzaLabel(), x_PruneIfFeatsIncomplete(), and x_PruneIfSegsMissing().

Referenced by FlattenGenbankSet().

x_GetFluLabel()

CCleanupHugeAsnReader::TIdToFluLabel::iterator CCleanupHugeAsnReader::x_GetFluLabel ( const CConstRef< CSeq_id > &  pId )

private

Definition at line 564 of file huge_file_cleanup.cpp.

References map_checker< Container >::end(), map_checker< Container >::lower_bound(), and m_IdToFluLabel.

Referenced by x_PruneAndReorderTopIds().

x_IsExtendedCleanup()

bool CCleanupHugeAsnReader::x_IsExtendedCleanup ( ) const

private

Definition at line 145 of file huge_file_cleanup.cpp.

References eExtendedCleanup, and m_CleanupOptions.

Referenced by LoadSeqEntry(), x_AddTopLevelDescriptors(), x_CleanupTopLevelDescriptors(), x_SetBioseqHooks(), x_SetBioseqSetHooks(), and x_SetSeqFeatHooks().

x_LooksLikeNucProtSet()

bool CCleanupHugeAsnReader::x_LooksLikeNucProtSet ( ) const

private

Definition at line 112 of file huge_file_cleanup.cpp.

References _ASSERT, CBioseq_set_Base::eClass_parts, CBioseq_set_Base::eClass_segset, CSeq_inst::IsAa(), CSeq_inst::IsNa(), and next().

Referenced by FlattenGenbankSet().

x_PruneAndReorderTopIds()

void CCleanupHugeAsnReader::x_PruneAndReorderTopIds ( )

private

Definition at line 88 of file huge_file_cleanup.cpp.

References map_checker< Container >::begin(), map_checker< Container >::end(), map_checker< Container >::erase(), map_checker< Container >::find(), m_IdToFluLabel, and x_GetFluLabel().

Referenced by FlattenGenbankSet().

x_PruneIfFeatsIncomplete()

void CCleanupHugeAsnReader::x_PruneIfFeatsIncomplete ( )

private

Definition at line 536 of file huge_file_cleanup.cpp.

References map_checker< Container >::begin(), map_checker< Container >::end(), map_checker< Container >::erase(), map_checker< Container >::find(), set< Key, Compare >::insert(), m_FluLabelToSetInfo, m_HasIncompleteFeats, m_IdToFluLabel, m_SetPosToFluLabel, s_IdInSet(), and s_RemoveEntriesWithVal().

Referenced by x_CreateSmallGenomeSets().

x_PruneIfSegsMissing()

void CCleanupHugeAsnReader::x_PruneIfSegsMissing ( const string &  fluLabel, const set< size_t > &  segsFound  )

private

Definition at line 522 of file huge_file_cleanup.cpp.

References map_checker< Container >::end(), map_checker< Container >::erase(), map_checker< Container >::find(), CInfluenzaSet::GetInfluenzaType(), CInfluenzaSet::GetNumRequired(), m_FluLabelToSetInfo, m_IdToFluLabel, m_SetPosToFluLabel, s_RemoveEntriesWithVal(), and set< Key, Compare >::size().

Referenced by x_CreateSmallGenomeSets().

x_RecordFeatureId()

void CCleanupHugeAsnReader::x_RecordFeatureId ( const CFeat_id &  featId )

private

Definition at line 594 of file huge_file_cleanup.cpp.

References map_checker< Container >::end(), set< Key, Compare >::end(), CCleanupHugeAsnReader::TFeatIdInfo::ExistingIds, map_checker< Container >::find(), set< Key, Compare >::find(), CObject_id_Base::GetId(), CFeat_id_Base::GetLocal(), CCleanupHugeAsnReader::TFeatIdInfo::IdOffset, set< Key, Compare >::insert(), CObject_id_Base::IsId(), CFeat_id_Base::IsLocal(), m_FeatIdInfo, CCleanupHugeAsnReader::TFeatIdInfo::NewExistingIds, CCleanupHugeAsnReader::TFeatIdInfo::NewIds, CCleanupHugeAsnReader::TFeatIdInfo::RemappedIds, and s_FindNextOffset().

Referenced by x_SetSeqFeatHooks().

x_SetBioseqHooks()

void CCleanupHugeAsnReader::x_SetBioseqHooks ( CObjectIStream &  objStream, TContext &  context  )

overrideprivate

Definition at line 620 of file huge_file_cleanup.cpp.

References set< Key, Compare >::begin(), map_checker< Container >::clear(), set< Key, Compare >::clear(), context, CBioseq_set_Base::eClass_genbank, map_checker< Container >::empty(), set< Key, Compare >::end(), CCleanupHugeAsnReader::TFeatIdInfo::ExistingIds, in(), set< Key, Compare >::insert(), m_FeatIdInfo, CCleanupHugeAsnReader::TFeatIdInfo::NewExistingIds, CCleanupHugeAsnReader::TFeatIdInfo::NewIds, CCleanupHugeAsnReader::TFeatIdInfo::PosToIdMap, CCleanupHugeAsnReader::TFeatIdInfo::RemappedIds, SetLocalSkipHook(), and x_IsExtendedCleanup().

x_SetBioseqSetHooks()

void CCleanupHugeAsnReader::x_SetBioseqSetHooks ( CObjectIStream &  objStream, TContext &  context  )

overrideprivate

Definition at line 656 of file huge_file_cleanup.cpp.

References set< Key, Compare >::begin(), map_checker< Container >::clear(), set< Key, Compare >::clear(), context, CBioseq_set_Base::eClass_genbank, map_checker< Container >::empty(), set< Key, Compare >::end(), CCleanupHugeAsnReader::TFeatIdInfo::ExistingIds, CObjectInfoMI::GetMember(), CObjectInfo::GetObjectPtr(), in(), set< Key, Compare >::insert(), last(), m_FeatIdInfo, CCleanupHugeAsnReader::TFeatIdInfo::NewExistingIds, CCleanupHugeAsnReader::TFeatIdInfo::NewIds, CCleanupHugeAsnReader::TFeatIdInfo::PosToIdMap, prev(), CCleanupHugeAsnReader::TFeatIdInfo::RemappedIds, CTypeConverter< T >::SafeCast(), SetLocalSkipHook(), and x_IsExtendedCleanup().

x_SetHooks()

void CCleanupHugeAsnReader::x_SetHooks ( CObjectIStream &  objStream, TContext &  context  )

overrideprivate

Definition at line 790 of file huge_file_cleanup.cpp.

References context, and x_SetSeqFeatHooks().

x_SetSeqFeatHooks()

void CCleanupHugeAsnReader::x_SetSeqFeatHooks ( CObjectIStream &  objStream, TContext &  context  )

private

Definition at line 716 of file huge_file_cleanup.cpp.

References eEnableSmallGenomeSets, eExtreme_Biological, in(), set< Key, Compare >::insert(), m_CleanupOptions, m_HasIncompleteFeats, Ref(), CTypeConverter< T >::SafeCast(), SetLocalReadHook(), SetLocalSkipHook(), x_IsExtendedCleanup(), and x_RecordFeatureId().

Referenced by x_SetHooks().

Member Data Documentation

m_Changes

CCleanupChangeCore CCleanupHugeAsnReader::m_Changes

mutableprivate

Definition at line 90 of file huge_file_cleanup.hpp.

Referenced by FlattenGenbankSet(), GetChanges(), x_AddTopLevelDescriptors(), and x_CleanupTopLevelDescriptors().

m_CleanupOptions

const TOptions CCleanupHugeAsnReader::m_CleanupOptions

private

Definition at line 89 of file huge_file_cleanup.hpp.

Referenced by FlattenGenbankSet(), LoadSeqEntry(), x_CleanupTopLevelDescriptors(), x_IsExtendedCleanup(), and x_SetSeqFeatHooks().

m_FeatIdInfo

TFeatIdInfo CCleanupHugeAsnReader::m_FeatIdInfo

private

Definition at line 111 of file huge_file_cleanup.hpp.

Referenced by LoadSeqEntry(), x_RecordFeatureId(), x_SetBioseqHooks(), and x_SetBioseqSetHooks().

m_FluLabelToSetInfo

map<string, list<TBioseqSetInfo> > CCleanupHugeAsnReader::m_FluLabelToSetInfo

private

Definition at line 94 of file huge_file_cleanup.hpp.

Referenced by LoadSeqEntry(), x_CreateSmallGenomeSets(), x_PruneIfFeatsIncomplete(), and x_PruneIfSegsMissing().

m_HasIncompleteFeats

set<CConstRef<CSeq_id>, CRefLess> CCleanupHugeAsnReader::m_HasIncompleteFeats

private

Definition at line 96 of file huge_file_cleanup.hpp.

Referenced by x_PruneIfFeatsIncomplete(), and x_SetSeqFeatHooks().

m_IdToFluLabel

TIdToFluLabel CCleanupHugeAsnReader::m_IdToFluLabel

private

Definition at line 92 of file huge_file_cleanup.hpp.

Referenced by x_CreateSmallGenomeSets(), x_GetFluLabel(), x_PruneAndReorderTopIds(), x_PruneIfFeatsIncomplete(), and x_PruneIfSegsMissing().

m_pTopLevelMolInfo

CRef<CSeqdesc> CCleanupHugeAsnReader::m_pTopLevelMolInfo

private

Definition at line 88 of file huge_file_cleanup.hpp.

Referenced by x_AddTopLevelDescriptors(), and x_CleanupTopLevelDescriptors().

m_SetPosToFluLabel

map<TFileSize, string> CCleanupHugeAsnReader::m_SetPosToFluLabel

private

Definition at line 95 of file huge_file_cleanup.hpp.

Referenced by LoadSeqEntry(), x_CreateSmallGenomeSets(), x_PruneIfFeatsIncomplete(), and x_PruneIfSegsMissing().

m_TopLevelBiosources

list<CRef<CSeqdesc> > CCleanupHugeAsnReader::m_TopLevelBiosources

private

Definition at line 87 of file huge_file_cleanup.hpp.

Referenced by x_AddTopLevelDescriptors(), and x_CleanupTopLevelDescriptors().

---

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

• src/app/asn_cleanup/huge_file_cleanup.hpp
• src/app/asn_cleanup/huge_file_cleanup.cpp