ftanet.cpp File Reference

#include <ncbi_pch.hpp>
#include "ftacpp.hpp"
#include <objtools/data_loaders/genbank/gbloader.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/taxon1/taxon1.hpp>
#include <objects/seqset/Bioseq_set.hpp>
#include <objects/seq/Seq_descr.hpp>
#include <objects/seq/Pubdesc.hpp>
#include <objects/pub/Pub_equiv.hpp>
#include <objects/pub/Pub.hpp>
#include <objects/biblio/Cit_gen.hpp>
#include <objects/biblio/Cit_art.hpp>
#include <objects/biblio/Cit_sub.hpp>
#include <objects/biblio/Cit_book.hpp>
#include <objects/biblio/Cit_let.hpp>
#include <objects/biblio/Cit_pat.hpp>
#include <objects/biblio/Cit_jour.hpp>
#include <objects/biblio/Auth_list.hpp>
#include <objects/biblio/Affil.hpp>
#include <objects/biblio/Author.hpp>
#include <objects/biblio/Imprint.hpp>
#include <objects/seq/Seq_annot.hpp>
#include <objects/pub/Pub_set.hpp>
#include <objects/biblio/ArticleIdSet.hpp>
#include <objects/biblio/ArticleId.hpp>
#include <objmgr/util/sequence.hpp>
#include <objtools/edit/pub_fix.hpp>
#include <dbapi/driver/drivers.hpp>
#include "index.h"
#include <objtools/flatfile/flatdefn.h>
#include <objtools/flatfile/flatfile_parser.hpp>
#include <corelib/ncbi_message.hpp>
#include "ftaerr.hpp"
#include "asci_blk.h"
#include "ftamed.h"
#include "utilfun.h"
#include "ref.h"
#include "flatfile_message_reporter.hpp"

Include dependency graph for ftanet.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Classes
struct	KwordBlk
class	CFindPub

Macros
#define	THIS_FILE   "ftanet.cpp"
#define	HEALTHY_ACC   "U12345"
#define	TAX_SERVER_TIMEOUT   3

Functions
	USING_SCOPE (objects)
void	fta_strip_pub_comment (string &comment, const KwordBlk *kbp)
static void	fta_fix_last_initials (CName_std &namestd, bool initials)
static void	fta_fix_affil (TPubList &pub_list, Parser::ESource source)
static void	fta_fix_imprint_language (TPubList &pub_list)
static void	fta_strip_er_remarks (CPubdesc &pub_descr)
static Uint1	fta_init_med_server (bool normalize)
static Uint1	fta_init_tax_server (void)
void	fta_init_servers (ParserPtr pp)
void	fta_fini_servers (ParserPtr pp)
void	fta_fill_find_pub_option (ParserPtr pp, bool htag, bool rtag)
static void	fta_check_pub_ids (TPubList &pub_list)
void	fta_find_pub_explore (ParserPtr pp, TEntryList &seq_entries)
static void	fix_synonyms (CTaxon1 &taxon, COrg_ref &org_ref)
static CRef< COrg_ref >	fta_get_orgref_byid (ParserPtr pp, bool *drop, TTaxId taxid, bool isoh)
CRef< COrg_ref >	fta_fix_orgref_byid (ParserPtr pp, TTaxId taxid, bool *drop, bool isoh)
static CRef< COrg_ref >	fta_replace_org (ParserPtr pp, bool *drop, COrg_ref &org_ref, const Char *pn, int merge, Int4 attempt)
void	fta_fix_orgref (ParserPtr pp, COrg_ref &org_ref, bool *drop, char *organelle)
static TGi	fta_get_gi_for_seq_id (const CSeq_id &id)
Int4	fta_is_con_div (ParserPtr pp, const CSeq_id &id, const Char *acc)
void	fta_init_gbdataloader ()

Variables
static const KwordBlk	PubStatus []
static const STimeout	s_timeout = { 3 , 0 }

Macro Definition Documentation

HEALTHY_ACC

#define HEALTHY_ACC   "U12345"

Definition at line 86 of file ftanet.cpp.

TAX_SERVER_TIMEOUT

#define TAX_SERVER_TIMEOUT   3

Definition at line 793 of file ftanet.cpp.

THIS_FILE

#define THIS_FILE   "ftanet.cpp"

Definition at line 84 of file ftanet.cpp.

Function Documentation

fix_synonyms()

static void fix_synonyms ( CTaxon1 &  taxon, COrg_ref &  org_ref  )

static

Definition at line 796 of file ftanet.cpp.

References COrg_ref_Base::SetSyn(), and CTaxon1::SetSynonyms().

Referenced by fta_get_orgref_byid().

fta_check_pub_ids()

static void fta_check_pub_ids ( TPubList &  pub_list )

static

Definition at line 521 of file ftanet.cpp.

References ERR_REFERENCE_ArticleIdDiscarded, ErrPostEx, and SEV_ERROR.

Referenced by CFindPub::fix_pub_equiv().

fta_fill_find_pub_option()

void fta_fill_find_pub_option	(	ParserPtr	pp,
		bool	htag,
		bool	rtag
	)

Definition at line 479 of file ftanet.cpp.

References Parser::SFindPubOptions::always_look, Parser::fpo, Parser::SFindPubOptions::merge_ids, and Parser::SFindPubOptions::replace_cit.

Referenced by Flat2AsnCheck().

fta_find_pub_explore()

void fta_find_pub_explore	(	ParserPtr	pp,
		TEntryList &	seq_entries
	)

Definition at line 753 of file ftanet.cpp.

References CFindPub::Apply(), fta_init_med_server(), Parser::medserver, and Parser::normalize.

Referenced by GenBankAsciiOrig(), OutputEmblAsn(), SpPrepareEntry(), and XMLAscii().

fta_fini_servers()

void fta_fini_servers

(

ParserPtr

pp

)

Definition at line 408 of file ftanet.cpp.

References GetPubmedClient(), and Parser::medserver.

fta_fix_affil()

static void fta_fix_affil ( TPubList &  pub_list, Parser::ESource  source  )

static

Definition at line 225 of file ftanet.cpp.

References CAuth_list_Base::CanGetAffil(), CCit_art_Base::CanGetAuthors(), CCit_book_Base::CanGetAuthors(), CCit_pat_Base::CanGetAuthors(), CCit_sub_Base::CanGetAuthors(), CCit_let_Base::CanGetCit(), CAuth_list_Base::CanGetNames(), CAuth_list_Base::C_Names::e_Std, CAffil_Base::e_Str, fta_fix_last_initials(), gen, CAuth_list_Base::GetAffil(), CAuth_list_Base::GetNames(), CAuth_list_Base::IsSetAffil(), CCit_art_Base::IsSetAuthors(), CCit_book_Base::IsSetAuthors(), CCit_pat_Base::IsSetAuthors(), CCit_sub_Base::IsSetAuthors(), CCit_let_Base::IsSetCit(), CAuth_list_Base::IsSetNames(), CName_std_Base::IsSetSuffix(), names, CAuth_list_Base::SetAffil(), CCit_art_Base::SetAuthors(), CCit_book_Base::SetAuthors(), CCit_pat_Base::SetAuthors(), CCit_sub_Base::SetAuthors(), CCit_let_Base::SetCit(), CAuth_list_Base::SetNames(), CAffil_Base::SetStr(), ShrinkSpaces(), CAffil_Base::Which(), and CAuth_list_Base::C_Names::Which().

Referenced by CFindPub::fix_pub_annot().

fta_fix_imprint_language()

static void fta_fix_imprint_language ( TPubList &  pub_list )

static

Definition at line 310 of file ftanet.cpp.

References CCit_art_Base::GetFrom(), CCit_art_Base::C_From::IsJournal(), CCit_art_Base::IsSetFrom(), journal(), and CCit_art_Base::SetFrom().

Referenced by CFindPub::fix_pub_annot().

fta_fix_last_initials()

static void fta_fix_last_initials ( CName_std &  namestd, bool  initials  )

static

Definition at line 147 of file ftanet.cpp.

References i, CName_std_Base::IsSetInitials(), CName_std_Base::IsSetLast(), CName_std_Base::SetInitials(), CName_std_Base::SetLast(), CName_std_Base::SetSuffix(), str(), and util::strcmp().

Referenced by fta_fix_affil().

fta_fix_orgref()

void fta_fix_orgref	(	ParserPtr	pp,
		COrg_ref &	org_ref,
		bool *	drop,
		char *	organelle
	)

Definition at line 936 of file ftanet.cpp.

References CSerialObject::Assign(), CRef< C, Locker >::Empty(), ERR_SERVER_TaxNameWasFound, ERR_SERVER_TaxServerDown, ErrPostEx, fta_init_tax_server(), fta_replace_org(), COrg_ref_Base::GetSyn(), COrg_ref_Base::GetTaxname(), COrg_ref_Base::IsSetSyn(), COrg_ref_Base::IsSetTaxname(), isspace(), CRef< C, Locker >::NotEmpty(), COrg_ref_Base::ResetSyn(), COrg_ref_Base::SetTaxname(), SEV_FATAL, SEV_INFO, and Parser::taxserver.

Referenced by CreateRawBioSources(), and GetSprotDescr().

fta_fix_orgref_byid()

CRef<COrg_ref> fta_fix_orgref_byid	(	ParserPtr	pp,
		TTaxId	taxid,
		bool *	drop,
		bool	isoh
	)

Definition at line 848 of file ftanet.cpp.

References ERR_SERVER_TaxNameWasFound, ERR_SERVER_TaxServerDown, ErrPostEx, fta_get_orgref_byid(), fta_init_tax_server(), CRef< C, Locker >::NotEmpty(), SEV_FATAL, SEV_INFO, TAX_ID_TO, and Parser::taxserver.

Referenced by fta_replace_org(), and GetSprotDescr().

fta_get_gi_for_seq_id()

static TGi fta_get_gi_for_seq_id ( const CSeq_id &  id )

static

Definition at line 1001 of file ftanet.cpp.

References GetGiForId(), GetScope(), GI_CONST, HEALTHY_ACC, i, CTextseq_id_Base::SetAccession(), CSeq_id_Base::SetGenbank(), SleepSec(), and ZERO_GI.

Referenced by fta_is_con_div().

fta_get_orgref_byid()

static CRef<COrg_ref> fta_get_orgref_byid ( ParserPtr  pp, bool *  drop, TTaxId  taxid, bool  isoh  )

static

Definition at line 806 of file ftanet.cpp.

References CSerialObject::Assign(), CConstRef< C, Locker >::Empty(), ERR_ORGANISM_TaxIdNotSpecLevel, ERR_ORGANISM_TaxNameNotFound, ERR_SERVER_TaxServerDown, ErrPostEx, fix_synonyms(), CTaxon1::GetById(), COrg_ref_Base::GetSyn(), i, CTaxon1::Init(), COrg_ref_Base::IsSetSyn(), CRef< C, Locker >::Reset(), COrg_ref_Base::ResetSyn(), s_timeout, SEV_ERROR, SEV_FATAL, SEV_WARNING, and TAX_ID_TO.

Referenced by fta_fix_orgref_byid().

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_med_server()

static Uint1 fta_init_med_server ( bool  normalize )

static

Definition at line 368 of file ftanet.cpp.

References GetPubmedClient(), Init(), and InitPubmedClient().

Referenced by fta_find_pub_explore(), and fta_init_servers().

fta_init_servers()

void fta_init_servers

(

ParserPtr

pp

)

Definition at line 386 of file ftanet.cpp.

References ERR_SERVER_Failed, ERR_SERVER_NoPubMedLookup, ERR_SERVER_NoTaxLookup, ErrPostStr, fta_init_med_server(), fta_init_tax_server(), Parser::medserver, Parser::normalize, SEV_ERROR, SEV_WARNING, and Parser::taxserver.

fta_init_tax_server()

static Uint1 fta_init_tax_server ( void  )

static

Definition at line 377 of file ftanet.cpp.

References CTaxon1::Init().

Referenced by fta_fix_orgref(), fta_fix_orgref_byid(), and fta_init_servers().

fta_is_con_div()

Int4 fta_is_con_div	(	ParserPtr	pp,
		const CSeq_id &	id,
		const Char *	acc
	)

Definition at line 1033 of file ftanet.cpp.

References Parser::curindx, Parser::entrez_fetch, Parser::entrylist, NStr::EqualNocase(), ERR_ACCESSION_CannotGetDivForSecondary, ErrPostEx, fta_get_gi_for_seq_id(), SEV_ERROR, and ZERO_GI.

fta_replace_org()

static CRef<COrg_ref> fta_replace_org ( ParserPtr  pp, bool *  drop, COrg_ref &  org_ref, const Char *  pn, int  merge, Int4  attempt  )

static

Definition at line 873 of file ftanet.cpp.

References CSerialObject::Assign(), Parser::curindx, Parser::DDBJ, Indexblk::drop, Parser::EMBL, CConstRef< C, Locker >::Empty(), Parser::entrylist, ERR_ORGANISM_TaxIdNotSpecLevel, ERR_ORGANISM_TaxIdNotUnique, ERR_ORGANISM_TaxNameNotFound, ERR_SERVER_TaxServerDown, ErrPostEx, CTaxon1::Fini(), fta_fix_orgref_byid(), CTaxon1::GetTaxIdByOrgRef(), COrg_ref_Base::GetTaxname(), i, CTaxon1::Init(), Indexblk::is_pat, COrg_ref_Base::IsSetTaxname(), CTaxon1::Lookup(), CTaxon1::LookupMerge(), Indexblk::no_gc_warning, CRef< C, Locker >::NotEmpty(), Indexblk::organism, CRef< C, Locker >::Reset(), s_timeout, SEV_ERROR, SEV_FATAL, SEV_WARNING, Parser::source, Indexblk::taxid, and ZERO_TAX_ID.

Referenced by fta_fix_orgref().

fta_strip_er_remarks()

static void fta_strip_er_remarks ( CPubdesc &  pub_descr )

static

Definition at line 335 of file ftanet.cpp.

References ePubStatus_aheadofprint, ePubStatus_epublish, ePubStatus_ppublish, fta_strip_pub_comment(), CPub_equiv_Base::Get(), CPubdesc_Base::GetComment(), CCit_art_Base::GetFrom(), CCit_art_Base::C_From::GetJournal(), CPubdesc_Base::GetPub(), CCit_art_Base::C_From::IsJournal(), CPubdesc_Base::IsSetComment(), CCit_art_Base::IsSetFrom(), journal(), PubStatus, CPubdesc_Base::ResetComment(), and CPubdesc_Base::SetComment().

fta_strip_pub_comment()

void fta_strip_pub_comment	(	string &	comment,
		const KwordBlk *	kbp
	)

Definition at line 119 of file ftanet.cpp.

References ERR_REFERENCE_UnusualPubStatus, ErrPostEx, i, KwordBlk::len, nullptr, SEV_WARNING, ShrinkSpaces(), KwordBlk::str, StringEquNI(), and StringIStr().

Referenced by fta_strip_er_remarks().

USING_SCOPE()

USING_SCOPE

(

objects

)

Variable Documentation

PubStatus

const KwordBlk PubStatus[]

static

Initial value:

= {
    { "Publication Status: Available-Online prior to print", 51 },
    { "Publication Status : Available-Online prior to print", 52 },
    { "Publication_Status: Available-Online prior to print", 51 },
    { "Publication_Status : Available-Online prior to print", 52 },
    { "Publication-Status: Available-Online prior to print", 51 },
    { "Publication-Status : Available-Online prior to print", 52 },
    { "Publication Status: Online-Only", 31 },
    { "Publication Status : Online-Only", 32 },
    { "Publication_Status: Online-Only", 31 },
    { "Publication_Status : Online-Only", 32 },
    { "Publication-Status: Online-Only", 31 },
    { "Publication-Status : Online-Only", 32 },
    { "Publication Status: Available-Online", 36 },
    { "Publication Status : Available-Online", 37 },
    { "Publication_Status: Available-Online", 36 },
    { "Publication_Status : Available-Online", 37 },
    { "Publication-Status: Available-Online", 36 },
    { "Publication-Status : Available-Online", 37 },
    { nullptr, 0 }
}

Definition at line 96 of file ftanet.cpp.

Referenced by fta_strip_er_remarks().

s_timeout

const STimeout s_timeout = { 3 , 0 }

static

Definition at line 794 of file ftanet.cpp.

Referenced by fta_get_orgref_byid(), and fta_replace_org().