ftamain.cpp File Reference

#include <ncbi_pch.hpp>
#include "ftacpp.hpp"
#include <objects/seqset/Bioseq_set.hpp>
#include <objtools/data_loaders/genbank/gbloader.hpp>
#include <objects/submit/Seq_submit.hpp>
#include <objects/submit/Submit_block.hpp>
#include <objects/biblio/Cit_sub.hpp>
#include <objects/biblio/Auth_list.hpp>
#include "index.h"
#include "sprot.h"
#include "embl.h"
#include "genbank.h"
#include <objtools/flatfile/ff2asn.h>
#include "ftanet.h"
#include <objtools/flatfile/flatfile_parser.hpp>
#include <objtools/flatfile/flatdefn.h>
#include "keyword_parse.hpp"
#include "flatfile_message_reporter.hpp"
#include "ftaerr.hpp"
#include <objtools/logging/listener.hpp>
#include "indx_blk.h"
#include "asci_blk.h"
#include "add.h"
#include "loadfeat.h"
#include "gb_ascii.h"
#include "sp_ascii.h"
#include "em_ascii.h"
#include "utilfeat.h"
#include "buf_data_loader.h"
#include "utilfun.h"
#include "entry.h"

Include dependency graph for ftamain.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Macros
#define	THIS_FILE   "ftamain.cpp"

Functions
	USING_SCOPE (objects)
bool	XMLAscii (ParserPtr pp)
void	fta_init_gbdataloader ()
static void	CkSegmentSet (ParserPtr pp)
static bool	CompareAccs (const Indexblk *p1, const Indexblk *p2)
static bool	CompareAccsV (const Indexblk *p1, const Indexblk *p2)
static bool	CompareData (const Indexblk *p1, const Indexblk *p2)
static bool	CompareDataV (const Indexblk *p1, const Indexblk *p2)
static void	CheckDupEntries (ParserPtr pp)
static CRef< CSerialObject >	MakeBioseqSet (ParserPtr pp)
static CRef< CSerialObject >	MakeSeqSubmit (ParserPtr pp)
static void	SetReleaseStr (ParserPtr pp)
static void	GetAuthorsStr (ParserPtr pp)
static CRef< CSerialObject >	CloseAll (ParserPtr pp)
static bool	sParseFlatfile (CRef< CSerialObject > &ret, ParserPtr pp, bool already=false)
Int2	fta_main (ParserPtr pp, bool already)
static bool	FillAccsBySource (Parser &pp, const string &source, bool all)
void	Flat2AsnCheck (char *ffentry, char *source, char *format, bool accver, Parser::EMode mode, Int4 limit)
static void	s_ReportFatalError (const string &msg, IObjtoolsListener *pListener)
TEntryList &	fta_parse_buf (Parser &pp, const char *buf)
bool	fta_set_format_source (Parser &pp, const string &format, const string &source)
void	fta_init_pp (Parser &pp)

Macro Definition Documentation

THIS_FILE

#define THIS_FILE   "ftamain.cpp"

Definition at line 76 of file ftamain.cpp.

Function Documentation

CheckDupEntries()

static void CheckDupEntries ( ParserPtr  pp )

static

Definition at line 265 of file ftamain.cpp.

References Parser::accver, Indexblk::acnum, CompareAccs(), CompareAccsV(), Indexblk::date, Indexblk::drop, CDate::eCompare_before, CDate::eCompare_same, Parser::entrylist, ERR_ENTRY_Repeated, ErrPostEx, first(), i, Parser::indx, Indexblk::locusname, Indexblk::offset, SEV_WARNING, ct::sort(), StringCmp(), and Indexblk::vernum.

Referenced by fta_parse_buf(), and sParseFlatfile().

CkSegmentSet()

static void CkSegmentSet ( ParserPtr  pp )

static

Definition at line 102 of file ftamain.cpp.

References Parser::curindx, Parser::debug, Parser::entrylist, ERR_SEGMENT_MissSegEntry, ErrPostEx, i, if(), Parser::indx, Parser::num_drop, SEV_ERROR, and StringEqu().

Referenced by fta_parse_buf(), and sParseFlatfile().

CloseAll()

static CRef<CSerialObject> CloseAll ( ParserPtr  pp )

static

Definition at line 416 of file ftamain.cpp.

References Parser::BioseqSet, CloseFiles(), Parser::entries, MakeBioseqSet(), MakeSeqSubmit(), Parser::output_format, and Parser::Seqsubmit.

Referenced by CTcpWorker::Execute(), and sParseFlatfile().

CompareAccs()

static bool CompareAccs ( const Indexblk *  p1, const Indexblk *  p2  )

static

Definition at line 172 of file ftamain.cpp.

References Indexblk::acnum, and StringCmp().

Referenced by CheckDupEntries().

CompareAccsV()

static bool CompareAccsV ( const Indexblk *  p1, const Indexblk *  p2  )

static

Definition at line 178 of file ftamain.cpp.

References Indexblk::acnum, i, StringCmp(), and Indexblk::vernum.

Referenced by CheckDupEntries().

CompareData()

static bool CompareData ( const Indexblk *  p1, const Indexblk *  p2  )

static

Definition at line 200 of file ftamain.cpp.

References Indexblk::acnum, Indexblk::blocusname, Indexblk::offset, Indexblk::segnum, Indexblk::segtotal, and StringCmp().

Referenced by fta_parse_buf(), and sParseFlatfile().

CompareDataV()

static bool CompareDataV ( const Indexblk *  p1, const Indexblk *  p2  )

static

Definition at line 229 of file ftamain.cpp.

References Indexblk::acnum, Indexblk::blocusname, Indexblk::offset, Indexblk::segnum, Indexblk::segtotal, StringCmp(), and Indexblk::vernum.

Referenced by fta_parse_buf(), and sParseFlatfile().

FillAccsBySource()

static bool FillAccsBySource ( Parser &  pp, const string &  source, bool  all  )

static

Definition at line 541 of file ftamain.cpp.

References Parser::accpref, Parser::accver, Parser::acprefix, Parser::all, all, Parser::citat, Parser::DDBJ, CSeq_id_Base::e_Ddbj, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_Other, CSeq_id_Base::e_Swissprot, Parser::EMBL, NStr::EqualNocase(), ErrPostEx, Parser::Flybase, Parser::format, Parser::GenBank, GetAccArray(), Parser::LANL, Parser::mode, Parser::NCBI, ParFlat_DDBJ_AC, ParFlat_EMBL_AC, ParFlat_LANL_AC, ParFlat_NCBI_AC, ParFlat_SPROT_AC, Parser::Refseq, Parser::Relaxed, Parser::seqtype, SEV_FATAL, rapidjson::source, Parser::source, Parser::SPROT, Parser::USPTO, and Parser::XML.

Referenced by Flat2AsnCheck(), and fta_set_format_source().

Flat2AsnCheck()

void Flat2AsnCheck	(	char *	ffentry,
		char *	source,
		char *	format,
		bool	accver,
		Parser::EMode	mode,
		Int4	limit
	)

Definition at line 631 of file ftamain.cpp.

References Parser::accver, Parser::allow_crossdb_featloc, Parser::cleanup, Parser::convert, Parser::EMBL, NStr::EqualNocase(), ErrPostEx, Parser::ffbuf, FillAccsBySource(), Parser::format, format, fta_fill_find_pub_option(), fta_main(), Parser::GenBank, Parser::genenull, Parser::limit, Parser::medserver, Parser::mode, Parser::qamode, Parser::qsfd, Parser::qsfile, FileBuf::set(), SEV_ERROR, Parser::sort, rapidjson::source, Parser::sp_dt_seq_ver, Parser::SPROT, Parser::taxserver, and Parser::XML.

fta_init_gbdataloader()

void fta_init_gbdataloader

(

)

Definition at line 1071 of file ftanet.cpp.

References CObjectManager::GetInstance(), and CGBDataLoader::RegisterInObjectManager().

Referenced by sParseFlatfile().

fta_init_pp()

void fta_init_pp

(

Parser &

pp

)

Definition at line 941 of file ftamain.cpp.

References Parser::accpref, Parser::accver, Parser::acprefix, Parser::all, Parser::allow_crossdb_featloc, Parser::citat, Parser::cleanup, Parser::convert, Parser::curindx, Parser::date, Parser::debug, CSeq_id_Base::e_not_set, Parser::entrez_fetch, Parser::entrylist, Parser::ff_get_entry, Parser::ff_get_entry_pp, Parser::ff_get_entry_v, Parser::ff_get_entry_v_pp, Parser::ff_get_qscore, Parser::ff_get_qscore_pp, Parser::ffdb, Parser::genenull, Parser::histacc, Parser::ign_bad_qs, Parser::ign_toks, Parser::indx, Parser::limit, Parser::medserver, Parser::mode, Parser::no_code, Parser::no_date, Parser::num_drop, Parser::pbp, Parser::qamode, Parser::qsfd, Parser::qsfile, Parser::Release, Parser::seg_acc, Parser::segment, Parser::seqtype, Parser::simple_genes, Parser::sort, Parser::sp_dt_seq_ver, Parser::taxserver, Parser::transl, and Parser::user_data.

fta_main()

Int2 fta_main	(	ParserPtr	pp,
		bool	already
	)

Definition at line 531 of file ftamain.cpp.

References sParseFlatfile().

Referenced by Flat2AsnCheck().

fta_parse_buf()

TEntryList& fta_parse_buf	(	Parser &	pp,
		const char *	buf
	)

Definition at line 796 of file ftamain.cpp.

References Parser::accver, CScope::AddDefaults(), buf, CheckDupEntries(), CkSegmentSet(), CompareData(), CompareDataV(), CObjectManager::eDefault, Parser::EMBL, EmblAscii(), Parser::entrez_fetch, Parser::entries, Parser::entrylist, ERR_ACCESSION_InvalidAccessNum, ERR_ENTRY_ParsingSetup, ErrPostEx, Parser::ffbuf, FlatFileIndex(), Parser::format, fta_fini_servers(), fta_init_servers(), FtaDeletePrefix(), FtaInstallPrefix(), Parser::GenBank, GenBankAscii(), CObjectManager::GetInstance(), GetScope(), ProtBlk::ibp, Parser::indx, CObjectManager::kPriority_Default, Parser::medserver, Parser::num_drop, Parser::pbp, PREFIX_ACCESSION, PREFIX_LOCUS, CBuffer_DataLoader::RegisterInObjectManager(), ResetParserStruct(), FileBuf::set(), SEV_INFO, SEV_WARNING, Parser::sort, ct::sort(), Parser::SPROT, SprotAscii(), Parser::taxserver, Parser::XML, and XMLAscii().

fta_set_format_source()

bool fta_set_format_source	(	Parser &	pp,
		const string &	format,
		const string &	source
	)

Definition at line 923 of file ftamain.cpp.

References Parser::all, Parser::EMBL, ErrPostEx, FillAccsBySource(), Parser::format, format, Parser::GenBank, SEV_FATAL, rapidjson::source, Parser::SPROT, and Parser::XML.

GetAuthorsStr()

static void GetAuthorsStr ( ParserPtr  pp )

static

Definition at line 400 of file ftamain.cpp.

References Parser::authors_str, Parser::DDBJ, Parser::EMBL, Parser::LANL, Parser::NCBI, Parser::Refseq, Parser::source, and Parser::SPROT.

Referenced by sParseFlatfile().

MakeBioseqSet()

static CRef<CSerialObject> MakeBioseqSet ( ParserPtr  pp )

static

Definition at line 321 of file ftamain.cpp.

References CBioseq_set_Base::eClass_genbank, CTime::eCurrent, Parser::entries, CDate::ePrecision_day, Parser::qamode, Parser::release_str, CBioseq_set_Base::SetClass(), CBioseq_set_Base::SetDate(), CBioseq_set_Base::SetRelease(), and CBioseq_set_Base::SetSeq_set().

Referenced by CloseAll().

MakeSeqSubmit()

static CRef<CSerialObject> MakeSeqSubmit ( ParserPtr  pp )

static

Definition at line 339 of file ftamain.cpp.

References Parser::authors_str, Parser::entries, entries, CSubmit_block_Base::SetCit(), CSeq_submit_Base::SetData(), and CSeq_submit_Base::SetSub().

Referenced by CloseAll().

s_ReportFatalError()

static void s_ReportFatalError ( const string &  msg, IObjtoolsListener *  pListener  )

static

Definition at line 731 of file ftamain.cpp.

References eDiag_Fatal, eUnknown, NCBI_THROW, and IObjtoolsListener::PutMessage().

Referenced by CFlatFileParser::Parse().

SetReleaseStr()

static void SetReleaseStr ( ParserPtr  pp )

static

Definition at line 353 of file ftamain.cpp.

References Parser::DDBJ, Parser::EMBL, Parser::Flybase, Parser::format, Parser::GenBank, Parser::LANL, Parser::NCBI, Parser::Refseq, Parser::release_str, Parser::source, Parser::SPROT, Parser::USPTO, Parser::XML, and Parser::xml_comp.

Referenced by sParseFlatfile().

sParseFlatfile()

static bool sParseFlatfile ( CRef< CSerialObject > &  ret, ParserPtr  pp, bool  already = false  )

static

Definition at line 433 of file ftamain.cpp.

References Parser::accver, CScope::AddDefaults(), Parser::BioseqSet, CheckDupEntries(), CkSegmentSet(), CloseAll(), CompareData(), CompareDataV(), Parser::EMBL, EmblAscii(), Parser::entrylist, ERR_ACCESSION_InvalidAccessNum, ERR_ENTRY_ParsingSetup, ErrPostEx, FlatFileIndex(), Parser::format, fta_fini_servers(), fta_init_gbdataloader(), fta_init_servers(), FtaDeletePrefix(), FtaInstallPrefix(), Parser::GenBank, GenBankAsciiOrig(), GetAuthorsStr(), GetScope(), ProtBlk::ibp, Parser::indx, Parser::InitializeKeywordParser(), Parser::num_drop, Parser::output_format, Parser::pbp, PREFIX_ACCESSION, PREFIX_LOCUS, CScope::ResetHistory(), Parser::Seqsubmit, SetReleaseStr(), SEV_INFO, SEV_WARNING, Parser::sort, ct::sort(), Parser::SPROT, SprotAscii(), Parser::XML, and XMLAscii().

Referenced by fta_main(), and CFlatFileParser::Parse().

USING_SCOPE()

USING_SCOPE

(

objects

)

XMLAscii()

bool XMLAscii

(

ParserPtr

pp

)

Definition at line 1128 of file xm_ascii.cpp.

References Parser::accver, Indexblk::acnum, CScope::AddBioseq(), AssemblyGapsToDelta(), Indexblk::bases, BuildBioSegHeader(), CheckDupDates(), CheckFeatSeqLoc(), Parser::citat, Parser::cleanup, Parser::convert, CreateEntryBioseq(), Parser::curindx, Parser::DDBJ, DealWithGenes(), Parser::debug, Indexblk::division, Indexblk::drop, Parser::entries, EntryCheckDivCode(), Parser::entrylist, CSeq_inst_Base::eRepr_raw, ERR_DATE_IllegalDate, ERR_DIVISION_Mismatch, ERR_ENTRY_GBBlock_not_Empty, ERR_ENTRY_LongHTGSSequence, ERR_ENTRY_LongSequence, ERR_ENTRY_Parsed, ERR_ENTRY_ParsingComplete, ERR_ENTRY_Skipped, err_install(), ERR_QSCORE_FailedToParse, ERR_REFERENCE_No_references, ERR_SEGMENT_OnlyOneMember, ERR_SEGMENT_Rejected, ERR_SEQUENCE_BadData, ERR_TSA_UnexpectedPrimaryAccession, ErrPostEx, ErrPostStr, Parser::ff_get_qscore, Parser::ff_get_qscore_pp, FinalCleanup(), Parser::Flybase, Parser::format, fta_EntryCheckGBBlock(), fta_find_pub_explore(), fta_remove_cleanup_user_object(), fta_set_molinfo_completeness(), fta_set_strandedness(), fta_sort_descr(), fta_sort_seqfeat_cit(), fta_tsa_tls_comment_dblink_check(), FtaDeletePrefix(), Indexblk::gaps, GapsToDelta(), CSeq_descr_Base::Get(), CBioseq_Base::GetDescr(), GetDNAConv(), CBioseq_Base::GetInst(), GetProteinConv(), GetQSFromFile(), CSeq_inst_Base::GetRepr(), GetScope(), GetSeqExt(), Indexblk::htg, i, Parser::ign_bad_qs, Parser::indx, Indexblk::is_pat, Indexblk::is_prot, Indexblk::is_tls, Indexblk::is_tsa, Indexblk::is_wgs, CSeq_inst::IsNa(), CBioseq_Base::IsSetAnnot(), DataBlk::len, Parser::limit, LoadFeat(), Indexblk::locusname, MaybeCutGbblockSource(), MemFree(), DataBlk::mOffset, DataBlk::mpData, DataBlk::mpQscore, Parser::no_date, no_date(), no_reference(), CRef< C, Locker >::NotEmpty(), PackEntries(), PREFIX_ACCESSION, PREFIX_LOCUS, ProcessCitations(), ProcNucProt(), Indexblk::psip, Parser::qamode, QscoreToSeqAnnot(), Parser::qsfd, Indexblk::qslength, Parser::Refseq, CRef< C, Locker >::Reset(), CScope::ResetHistory(), Indexblk::segnum, Indexblk::segtotal, EntryBlk::seq_entry, SeqToDelta(), CBioseq_Base::SetId(), SEV_ERROR, SEV_INFO, SEV_REJECT, SEV_WARNING, Parser::source, StringEqu(), StringEquN(), StringLen(), StripSerialNumbers(), Indexblk::tsa_allowed, Parser::USPTO, GeneRefFeats::valid, Indexblk::vernum, Indexblk::xip, Parser::xml_comp, XMLCheckContigEverywhere(), XMLFakeBioSources(), XMLGetDescr(), XMLGetDivision(), XMLGetInst(), and XMLLoadEntry().

Referenced by fta_parse_buf(), and sParseFlatfile().