indx_blk.cpp File Reference

#include <ncbi_pch.hpp>
#include "ftacpp.hpp"
#include "index.h"
#include <objtools/flatfile/flatfile_parse_info.hpp>
#include "ftaerr.hpp"
#include "indx_blk.h"
#include "indx_def.h"
#include "utilfun.h"
#include <map>

Include dependency graph for indx_blk.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Macros
#define	THIS_FILE   "indx_blk.cpp"

Functions
	USING_SCOPE (objects)
bool	EmblIndex (ParserPtr pp, void(*fun)(IndexblkPtr entry, char *offset, Int4 len))
bool	GenBankIndex (ParserPtr pp)
bool	SprotIndex (ParserPtr pp, void(*fun)(IndexblkPtr entry, char *offset, Int4 len))
bool	XMLIndex (ParserPtr pp)
static const char *	GetResidue (TokenStatBlkPtr stoken)
static bool	XReadFile (FILE *fp, FinfoBlk &finfo)
static Int2	FileGetsBuf (char *res, Int4 size, FileBuf &fbuf)
bool	XReadFileBuf (FileBuf &fbuf, FinfoBlk &finfo)
NCBI_UNUSED bool	SkipTitle (FILE *fp, FinfoBlk &finfo, const char *str, size_t len)
NCBI_UNUSED bool	SkipTitle (FILE *fp, FinfoBlk &finfo, const CTempString &keyword)
bool	SkipTitleBuf (FileBuf &fbuf, FinfoBlk &finfo, const CTempString &keyword)
static bool	CheckLocus (const char *locus, Parser::ESource source)
static bool	CheckLocusSP (const char *locus)
static bool	CkDateFormat (const char *date)
int	CheckSTRAND (const string &str)
Int2	XMLCheckSTRAND (const char *str)
Int2	XMLCheckTPG (const char *str)
int	CheckTPG (const string &str)
Int2	CheckNADDBJ (const char *str)
Int2	CheckNA (const char *str)
Int2	CheckDIV (const char *str)
bool	CkLocusLinePos (char *offset, Parser::ESource source, LocusContPtr lcp, bool is_mga)
CRef< CDate_std >	GetUpdateDate (const char *ptr, Parser::ESource source)
static bool	fta_check_embl_moltype (char *str)
static bool	isSpace (char c)
static CTempString::const_iterator	sFindNextSpace (const CTempString &tempString, CTempString::const_iterator current_it)
static CTempString::const_iterator	sFindNextNonSpace (const CTempString &tempString, CTempString::const_iterator current_it)
static void	sSetLocusLineOffsets (const CTempString &locusLine, LocusCont &offsets)
IndexblkPtr	InitialEntry (ParserPtr pp, FinfoBlk &finfo)
void	DelNoneDigitTail (char *str)
void	DelNonDigitTail (string &str)
static bool	sIsUpperAlpha (char c)
Int4	IsNewAccessFormat (const Char *acnum)
static bool	IsValidAccessPrefix (const char *acc, const char **accpref)
static bool	fta_if_master_wgs_accession (const char *acnum, Int4 accformat)
static bool	s_IsVDBWGSScaffold (string_view accession)
static int	s_RefineWGSType (string_view accession, int initialType)
int	fta_if_wgs_acc (string_view accession)
bool	IsSPROTAccession (const char *acc)
static bool	sNotAllDigits (const char *first, const char *last)
static bool	CheckAccession (TokenStatBlkPtr stoken, Parser::ESource source, Parser::EMode mode, char *priacc, unsigned skip)
static bool	IsPatentedAccPrefix (const Parser &parseInfo, const char *acc)
static bool	IsTPAAccPrefix (const Parser &parseInfo, const char *acc)
static bool	IsWGSAccPrefix (const Parser &parseInfo, const char *acc)
static void	IsTSAAccPrefix (const Parser &parseInfo, const char *acc, IndexblkPtr ibp)
static void	IsTLSAccPrefix (const Parser &parseInfo, const char *acc, IndexblkPtr ibp)
bool	GetAccession (ParserPtr pp, const char *str, IndexblkPtr entry, unsigned skip)
void	ResetParserStruct (ParserPtr pp)
void	CloseFiles (ParserPtr pp)
void	MsgSkipTitleFail (const char *flatfile, FinfoBlk &finfo)
bool	FindNextEntryBuf (bool end_of_file, FileBuf &fbuf, FinfoBlk &finfo, const CTempString &keyword)
bool	FlatFileIndex (ParserPtr pp, void(*fun)(IndexblkPtr entry, char *offset, Int4 len))
const char **	GetAccArray (Parser::ESource source)
bool	isSupportedAccession (CSeq_id::E_Choice type)
CSeq_id::E_Choice	GetNucAccOwner (const CTempString &acc)
CSeq_id::E_Choice	GetProtAccOwner (const CTempString &acc)

Variables
static const char *	XML_STRAND_array []
static const char *	XML_TPG_array []
static const char *	ParFlat_NA_array_DDBJ []
static const char *	ParFlat_AA_array_DDBJ []
static const char *	ParFlat_NA_array []
static const char *	ParFlat_DIV_array []
static const char *	embl_accpref []
static const char *	lanl_accpref []
static const char *	sprot_accpref []
static const char *	ddbj_accpref []
static const char *	ncbi_accpref []
static const char *	refseq_accpref []
static const char *	acc_tsa_allowed []
static const char *	ncbi_tpa_accpref []
static const char *	ddbj_tpa_accpref []
static const char *	ncbi_wgs_accpref []
static const char *	ddbj_wgs_accpref []
static const set< string_view >	k_WgsScaffoldPrefix
static const map< Parser::ESource, string >	sourceNames
static const char *	month_name []
static const char *	ParFlat_RESIDUE_STR []
static const char *	ValidMolTypes []

Macro Definition Documentation

THIS_FILE

#define THIS_FILE   "indx_blk.cpp"

Definition at line 51 of file indx_blk.cpp.

Function Documentation

CheckAccession()

static bool CheckAccession ( TokenStatBlkPtr  stoken, Parser::ESource  source, Parser::EMode  mode, char *  priacc, unsigned  skip  )

static

Definition at line 1508 of file indx_blk.cpp.

References TokenBlk::c_str(), ERR_ACCESSION_BadAccessNum, ERR_ACCESSION_WGSProjectAccIsPri, ErrPostEx, fta_if_master_wgs_accession(), IsNewAccessFormat(), IsSPROTAccession(), len, TokenStatBlk::list, TokenBlk::next, Parser::Relaxed, SEV_ERROR, SEV_REJECT, sIsUpperAlpha(), sNotAllDigits(), rapidjson::source, Parser::SPROT, StringCpy(), and StringLen().

Referenced by GetAccession().

CheckDIV()

Int2 CheckDIV

(

const char *

str

)

Definition at line 532 of file indx_blk.cpp.

References fta_StringMatch(), ParFlat_DIV_array, and str().

Referenced by CkLocusLinePos(), GetGBBlock(), and XMLGetGBBlock().

CheckLocus()

static bool CheckLocus ( const char *  locus, Parser::ESource  source  )

static

Definition at line 379 of file indx_blk.cpp.

References Parser::DDBJ, ERR_LOCUS_BadLocusName, ErrPostEx, Parser::Flybase, Parser::NCBI, Parser::Refseq, SEV_ERROR, rapidjson::source, and StringEquN().

Referenced by InitialEntry().

CheckLocusSP()

static bool CheckLocusSP ( const char *  locus )

static

Definition at line 419 of file indx_blk.cpp.

References ERR_LOCUS_BadLocusName, ErrPostEx, and SEV_ERROR.

Referenced by InitialEntry().

CheckNA()

Int2 CheckNA

(

const char *

str

)

Definition at line 526 of file indx_blk.cpp.

References fta_StringMatch(), ParFlat_NA_array, and str().

Referenced by GetFlatBiomol().

CheckNADDBJ()

Int2 CheckNADDBJ

(

const char *

str

)

Definition at line 520 of file indx_blk.cpp.

References fta_StringMatch(), ParFlat_NA_array_DDBJ, and str().

Referenced by GetFlatBiomol().

CheckSTRAND()

int CheckSTRAND

(

const string &

str

)

Definition at line 467 of file indx_blk.cpp.

References i, NStr::StartsWith(), str(), and NStr::ToLower().

Referenced by CkLocusLinePos(), GetEmblInst(), GetEmblMolInfo(), GetGenBankInst(), and Entry::xInitSeqInst().

CheckTPG()

int CheckTPG

(

const string &

str

)

Definition at line 502 of file indx_blk.cpp.

References i, NStr::StartsWith(), str(), and NStr::ToLower().

Referenced by CkLocusLinePos(), GetGenBankInst(), and Entry::xInitSeqInst().

CkDateFormat()

static bool CkDateFormat ( const char *  date )

static

Definition at line 454 of file indx_blk.cpp.

References isdigit(), MatchArraySubString(), and month_name.

Referenced by CkLocusLinePos(), and GetUpdateDate().

CkLocusLinePos()

bool CkLocusLinePos	(	char *	offset,
		Parser::ESource	source,
		LocusContPtr	lcp,
		bool	is_mga
	)

Definition at line 538 of file indx_blk.cpp.

References LocusCont::bp, CheckDIV(), CheckSTRAND(), CheckTPG(), CkDateFormat(), LocusCont::date, Parser::DDBJ, LocusCont::div, ERR_FORMAT_IllegalCAGEMoltype, ERR_FORMAT_LocusLinePosition, ErrPostEx, i, Parser::LANL, MemCpy(), LocusCont::molecule, offset, ParFlat_AA_array_DDBJ, ParFlat_NA_array, ParFlat_NA_array_DDBJ, SEV_REJECT, SEV_WARNING, rapidjson::source, LocusCont::strand, StringChr, StringEquN(), StringEquNI(), StringMatchIcase(), and LocusCont::topology.

Referenced by GenBankIndex().

CloseFiles()

void CloseFiles

(

ParserPtr

pp

)

Definition at line 2163 of file indx_blk.cpp.

References Parser::qsfd.

Referenced by CloseAll().

DelNonDigitTail()

void DelNonDigitTail

(

string &

str

)

Definition at line 959 of file indx_blk.cpp.

References str().

Referenced by GetExtraAccession().

DelNoneDigitTail()

void DelNoneDigitTail

(

char *

str

)

Definition at line 945 of file indx_blk.cpp.

References str().

Referenced by GetAccession().

EmblIndex()

bool EmblIndex	(	ParserPtr	pp,
		void(*)(IndexblkPtr entry, char *offset, Int4 len)	fun
	)

Definition at line 192 of file em_index.cpp.

References Parser::accver, CKeywordParser::AddDataLine(), TokenBlk::c_str(), checkedEmblKeywords, CheckLineType(), Parser::curindx, data, Indexblk::date, Parser::DDBJ, Indexblk::drop, em_err_field(), Parser::EMBL, Indexblk::embl_new_ID, EmblGetNewIDVersion(), emblKeywords, EmblSegment(), Parser::entrylist, ERR_ACCESSION_NoAccessNum, ERR_ENTRY_InvalidLineType, ERR_FORMAT_MissingEnd, ERR_FORMAT_Multiple_NI, ERR_FORMAT_Multiple_SV, ERR_ORGANISM_Multiple, ErrPostEx, ErrPostStr, Parser::ffbuf, FindNextEntryBuf(), FreeTokenstatblk(), FtaDeletePrefix(), FileBuf::get_offs(), GetAccession(), GetUpdateDate(), i, IndBlkNode::ibp, Parser::indx, InitialEntry(), Indexblk::is_contig, Indexblk::is_tpa, Indexblk::is_tpa_wgs_con, Indexblk::is_wgs, isalpha(), CKeywordParser::KeywordList(), Parser::KeywordParser(), Indexblk::len, FinfoBlk::line, TokenStatBlk::list, LoadEntry(), Indexblk::locusname, MemFree(), MsgSkipTitleFail(), TokenBlk::next, IndBlkNode::next, TokenStatBlk::num, Parser::num_drop, Indexblk::offset, Indexblk::origin, ParFlat_AC, ParFlat_AH, ParFlat_CO, ParFlat_COL_DATA_EMBL, ParFlat_DT, ParFlat_ID, ParFlat_KW, ParFlat_NI, ParFlat_OS, ParFlat_RN, ParFlat_SQ, ParFlat_SV, ParFlatEM_END, ParseEmblVersion(), PREFIX_ACCESSION, PREFIX_LOCUS, Parser::qsfd, QSIndex(), Parser::segment, SEV_ERROR, SEV_FATAL, SEV_INFO, SkipTitleBuf(), Parser::source, FinfoBlk::str, StringEquN(), StringNCpy(), StringNew(), TokenString(), Indexblk::tsa_allowed, xCheckEstStsGssTpaKeywords(), and XReadFileBuf().

Referenced by FlatFileIndex().

FileGetsBuf()

static Int2 FileGetsBuf ( char *  res, Int4  size, FileBuf &  fbuf  )

static

Definition at line 289 of file indx_blk.cpp.

References FileBuf::current, i, and ncbi::grid::netcache::search::fields::size.

Referenced by XReadFileBuf().

FindNextEntryBuf()

bool FindNextEntryBuf	(	bool	end_of_file,
		FileBuf &	fbuf,
		FinfoBlk &	finfo,
		const CTempString &	keyword
	)

Definition at line 2186 of file indx_blk.cpp.

References CTempString::data(), done, len, CTempString::size(), FinfoBlk::str, StringEquN(), and XReadFileBuf().

Referenced by EmblIndex(), GenBankIndex(), and SprotIndex().

FlatFileIndex()

bool FlatFileIndex	(	ParserPtr	pp,
		void(*)(IndexblkPtr entry, char *offset, Int4 len)	fun
	)

Definition at line 2205 of file indx_blk.cpp.

References Parser::EMBL, EmblIndex(), Parser::format, Parser::GenBank, GenBankIndex(), Parser::SPROT, SprotIndex(), Parser::XML, and XMLIndex().

Referenced by fta_parse_buf(), and sParseFlatfile().

fta_check_embl_moltype()

static bool fta_check_embl_moltype ( char *  str )

static

Definition at line 633 of file indx_blk.cpp.

References b, ERR_FORMAT_InvalidIDlineMolType, ErrPostEx, SEV_REJECT, str(), StringChr, StringEqu(), and ValidMolTypes.

Referenced by InitialEntry().

fta_if_master_wgs_accession()

static bool fta_if_master_wgs_accession ( const char *  acnum, Int4  accformat  )

static

Definition at line 1064 of file indx_blk.cpp.

Referenced by CheckAccession().

fta_if_wgs_acc()

int fta_if_wgs_acc

(

string_view

accession

)

Definition at line 1191 of file indx_blk.cpp.

References i, isalpha(), NStr::IsBlank(), isdigit(), k_WgsScaffoldPrefix, s_IsVDBWGSScaffold(), and s_RefineWGSType().

Referenced by BOOST_AUTO_TEST_CASE(), fta_add_hist(), fta_check_con_for_wgs(), fta_create_wgs_seqid(), fta_fix_secondaries(), fta_ranges_to_hist(), GetDescrEmblBlock(), GetExtraAccession(), IsTLSAccPrefix(), IsTSAAccPrefix(), and XMLGetEMBLBlock().

GenBankIndex()

bool GenBankIndex

(

ParserPtr

pp

)

Definition at line 343 of file gb_index.cpp.

References Parser::accver, Indexblk::acnum, Indexblk::assembly, check_est_sts_gss_tpa_kwds(), CheckLineType(), CkLocusLinePos(), ConstructValNode(), Parser::curindx, Parser::DDBJ, Indexblk::drop, e_not_set, Parser::EMBL, Parser::entrylist, ERR_ACCESSION_NoAccessNum, ERR_ENTRY_InvalidLineType, ERR_FORMAT_ContigInSegset, ERR_FORMAT_IncorrectMGALine, ERR_FORMAT_LineTypeOrder, ErrPostEx, Indexblk::experimental, Parser::ffbuf, FindNextEntryBuf(), Parser::Flybase, fta_check_mga_line(), FtaDeletePrefix(), gb_err_field(), genbankKeywords, FileBuf::get_offs(), GetAccession(), GetSegment(), Parser::HTGSCON, i, IndBlkNode::ibp, if(), Parser::ign_toks, Parser::indx, Indexblk::inferential, InitialEntry(), Indexblk::is_contig, Indexblk::is_mga, Indexblk::is_tpa, Indexblk::is_tpa_wgs_con, Indexblk::is_wgs, NStr::IsBlank(), Indexblk::lc, Indexblk::len, FinfoBlk::line, Indexblk::linenum, MemFree(), Parser::mode, MsgSkipTitleFail(), IndBlkNode::next, Parser::num_drop, Indexblk::offset, Indexblk::origin, ParFlat_ACCESSION, ParFlat_COL_DATA, ParFlat_COMMENT, ParFlat_CONTIG, ParFlat_DBLINK, ParFlat_DEFINITION, ParFlat_END, ParFlat_FEATURES, ParFlat_KEYWORDS, ParFlat_LOCUS, ParFlat_MGA, ParFlat_NCBI_GI, ParFlat_ORIGIN, ParFlat_PRIMARY, ParFlat_REFERENCE, ParFlat_SEGMENT, ParFlat_SOURCE, ParFlat_UNKW, ParFlat_USER, ParFlat_VERSION, ParseGenBankVersion(), PREFIX_ACCESSION, PREFIX_LOCUS, Parser::qsfd, QSIndex(), Parser::Refseq, Parser::Relaxed, Indexblk::segnum, SEV_ERROR, SEV_REJECT, SkipTitleBuf(), Parser::source, Indexblk::specialist_db, SrchKeyword(), FinfoBlk::str, StringEquN(), StringLen(), StringNCpy(), StringNew(), StringSave, StringStr, Indexblk::tsa_allowed, ValNodeFreeData(), ValNodeNew(), Indexblk::vernum, Indexblk::wgs_and_gi, and XReadFileBuf().

Referenced by FlatFileIndex().

GetAccArray()

const char** GetAccArray

(

Parser::ESource

source

)

Definition at line 2231 of file indx_blk.cpp.

References Parser::DDBJ, ddbj_accpref, Parser::EMBL, embl_accpref, Parser::LANL, lanl_accpref, Parser::NCBI, ncbi_accpref, Parser::Refseq, refseq_accpref, rapidjson::source, Parser::SPROT, and sprot_accpref.

Referenced by FillAccsBySource().

GetAccession()

bool GetAccession	(	ParserPtr	pp,
		const char *	str,
		IndexblkPtr	entry,
		unsigned	skip
	)

Definition at line 1937 of file indx_blk.cpp.

References Parser::accpref, Parser::accver, Indexblk::acnum, Parser::acprefix, TokenBlk::c_str(), CheckAccession(), Parser::DDBJ, DelNoneDigitTail(), Indexblk::drop, ERR_ACCESSION_BadAccessNum, ERR_ACCESSION_NoAccessNum, ErrPostEx, Parser::Flybase, Parser::format, FreeTokenstatblk(), FtaInstallPrefix(), i, Indexblk::is_mga, Indexblk::is_pat, Indexblk::is_tpa, Indexblk::is_wgs, IsNewAccessFormat(), IsPatentedAccPrefix(), IsTLSAccPrefix(), IsTPAAccPrefix(), IsTSAAccPrefix(), IsValidAccessPrefix(), IsWGSAccPrefix(), Indexblk::linenum, TokenStatBlk::list, Indexblk::locusname, MemFree(), Parser::mode, TokenBlk::next, TokenStatBlk::num, ParseAccessionRange(), PREFIX_ACCESSION, Parser::Relaxed, Indexblk::secaccs, SEV_ERROR, Parser::source, sourceNames, str(), StringChr, StringCpy(), StringEqu(), StringLen(), StringSave, TokenString(), Parser::USPTO, Indexblk::vernum, Indexblk::wgs_and_gi, and Parser::XML.

GetNucAccOwner()

CSeq_id::E_Choice GetNucAccOwner

(

const CTempString &

acc

)

Definition at line 2268 of file indx_blk.cpp.

References CSeq_id_Base::e_not_set, CSeq_id::fAcc_prot, CSeq_id::GetAccType(), CSeq_id::IdentifyAccession(), info, isSupportedAccession(), and type.

Referenced by fta_add_hist(), fta_check_compare_qual(), fta_check_non_tpa_tsa_tls_locations(), fta_parse_tpa_tsa_block(), GetDRlineDataSP(), GetExtraAccession(), and GetProtRefSeqId().

GetProtAccOwner()

CSeq_id::E_Choice GetProtAccOwner

(

const CTempString &

acc

)

Definition at line 2285 of file indx_blk.cpp.

References CSeq_id_Base::e_not_set, CSeq_id::fAcc_prot, CSeq_id::GetAccType(), CSeq_id::IdentifyAccession(), info, isSupportedAccession(), and type.

Referenced by fta_check_embl_drxref_dups(), GetDRlineDataSP(), and GetProtRefSeqId().

GetResidue()

static const char* GetResidue ( TokenStatBlkPtr  stoken )

static

Definition at line 234 of file indx_blk.cpp.

References b, TokenBlk::c_str(), NStr::CompareNocase(), i, TokenStatBlk::list, TokenBlk::next, TokenStatBlk::num, and ParFlat_RESIDUE_STR.

Referenced by InitialEntry(), and TestUtil::PrintSequence().

GetUpdateDate()

CRef<CDate_std> GetUpdateDate	(	const char *	ptr,
		Parser::ESource	source
	)

Definition at line 611 of file indx_blk.cpp.

References CkDateFormat(), CTime::eCurrent, get_full_date(), MemCpy(), rapidjson::source, Parser::SPROT, and StringEquN().

Referenced by EmblIndex(), GetEmblDate(), GetGenBankDescr(), GetSPDate(), InitialEntry(), SprotIndex(), XMLGetDescr(), XMLGetEMBLBlock(), and XMLInitialEntry().

InitialEntry()

IndexblkPtr InitialEntry	(	ParserPtr	pp,
		FinfoBlk &	finfo
	)

Definition at line 788 of file indx_blk.cpp.

References LocusCont::bases, Indexblk::bases, Indexblk::blocusname, LocusCont::bp, TokenBlk::c_str(), CheckLocus(), CheckLocusSP(), TokenBlk::data(), LocusCont::date, Indexblk::date, Parser::DDBJ, LocusCont::div, Parser::EMBL, Indexblk::embl_new_ID, TokenBlk::empty(), ERR_ENTRY_Skipped, ERR_FORMAT_BadlyFormattedIDLine, ERR_LOCUS_NoLocusName, ErrPostEx, ErrPostStr, Indexblk::EST, Parser::format, FreeTokenstatblk(), fta_check_embl_moltype(), FtaInstallPrefix(), Parser::GenBank, GetResidue(), GetUpdateDate(), Indexblk::GSS, Indexblk::HTC, i, Indexblk::is_pat, Indexblk::is_tls, Indexblk::is_tsa, Indexblk::lc, FinfoBlk::line, Indexblk::linenum, TokenStatBlk::list, Indexblk::locusname, Parser::mode, LocusCont::molecule, TokenBlk::next, TokenStatBlk::num, Indexblk::offset, ParFlat_COL_BASES, ParFlat_COL_BASES_NEW, ParFlat_COL_BP, ParFlat_COL_BP_NEW, ParFlat_COL_DATE, ParFlat_COL_DATE_NEW, ParFlat_COL_DIV, ParFlat_COL_DIV_NEW, ParFlat_COL_MOLECULE, ParFlat_COL_MOLECULE_NEW, ParFlat_COL_STRAND, ParFlat_COL_STRAND_NEW, ParFlat_COL_TOPOLOGY, ParFlat_COL_TOPOLOGY_NEW, FinfoBlk::pos, Indexblk::ppp, PREFIX_ACCESSION, PREFIX_LOCUS, Parser::Relaxed, SEV_ERROR, SEV_REJECT, Parser::source, Parser::SPROT, sSetLocusLineOffsets(), FinfoBlk::str, LocusCont::strand, StringChr, StringCpy(), StringEqu(), StringEquNI(), StringLen(), Indexblk::STS, TokenString(), and LocusCont::topology.

Referenced by EmblIndex(), GenBankIndex(), and SprotIndex().

IsNewAccessFormat()

Int4 IsNewAccessFormat

(

const Char *

acnum

)

Definition at line 993 of file indx_blk.cpp.

References isdigit(), and sIsUpperAlpha().

Referenced by CheckAccession(), fta_fix_seq_id(), fta_fix_seq_loc_id(), get_sub(), GetAccession(), GetGBBlock(), IsValidAccessPrefix(), and XMLGetGBBlock().

IsPatentedAccPrefix()

static bool IsPatentedAccPrefix ( const Parser &  parseInfo, const char *  acc  )

static

Definition at line 1606 of file indx_blk.cpp.

References Parser::all, Parser::DDBJ, Parser::EMBL, Parser::NCBI, Parser::source, StringEqu(), and StringEquN().

Referenced by GetAccession().

isSpace()

static bool isSpace ( char  c )

static

Definition at line 675 of file indx_blk.cpp.

References isspace().

Referenced by sFindNextNonSpace(), and sFindNextSpace().

IsSPROTAccession()

bool IsSPROTAccession

(

const char *

acc

)

Definition at line 1255 of file indx_blk.cpp.

References b, len, sprot_accpref, StringEquN(), and StringLen().

Referenced by CheckAccession(), and GetSprotDescr().

isSupportedAccession()

bool isSupportedAccession

(

CSeq_id::E_Choice

type

)

Definition at line 2248 of file indx_blk.cpp.

References CSeq_id_Base::e_Ddbj, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_Other, CSeq_id_Base::e_Tpd, CSeq_id_Base::e_Tpe, and CSeq_id_Base::e_Tpg.

Referenced by fta_fix_seq_id(), GetNucAccOwner(), and GetProtAccOwner().

IsTLSAccPrefix()

static void IsTLSAccPrefix ( const Parser &  parseInfo, const char *  acc, IndexblkPtr  ibp  )

static

Definition at line 1772 of file indx_blk.cpp.

References Indexblk::acnum, Parser::all, Parser::DDBJ, fta_if_wgs_acc(), Indexblk::is_tls, Parser::NCBI, Parser::source, and StringLen().

Referenced by GetAccession().

IsTPAAccPrefix()

static bool IsTPAAccPrefix ( const Parser &  parseInfo, const char *  acc  )

static

Definition at line 1673 of file indx_blk.cpp.

References Parser::all, Parser::DDBJ, ddbj_tpa_accpref, fta_StringMatch(), i, Parser::NCBI, ncbi_tpa_accpref, Parser::source, and StringLen().

Referenced by GetAccession().

IsTSAAccPrefix()

static void IsTSAAccPrefix ( const Parser &  parseInfo, const char *  acc, IndexblkPtr  ibp  )

static

Definition at line 1717 of file indx_blk.cpp.

References acc_tsa_allowed, Indexblk::acnum, Parser::all, Parser::DDBJ, Parser::EMBL, fta_if_wgs_acc(), fta_StringMatch(), Indexblk::is_tsa, Parser::NCBI, Parser::source, StringEqu(), StringEquN(), StringLen(), and Indexblk::tsa_allowed.

Referenced by GetAccession().

IsValidAccessPrefix()

static bool IsValidAccessPrefix ( const char *  acc, const char **  accpref  )

static

Definition at line 1045 of file indx_blk.cpp.

References b, i, IsNewAccessFormat(), StringEquN(), and StringLen().

Referenced by GetAccession().

IsWGSAccPrefix()

static bool IsWGSAccPrefix ( const Parser &  parseInfo, const char *  acc  )

static

Definition at line 1702 of file indx_blk.cpp.

References Parser::all, Parser::DDBJ, ddbj_wgs_accpref, fta_StringMatch(), Parser::NCBI, ncbi_wgs_accpref, Parser::source, and StringLen().

Referenced by GetAccession().

MsgSkipTitleFail()

void MsgSkipTitleFail	(	const char *	flatfile,
		FinfoBlk &	finfo
	)

Definition at line 2178 of file indx_blk.cpp.

References ERR_ENTRY_Begin, ErrPostEx, and SEV_ERROR.

Referenced by EmblIndex(), GenBankIndex(), and SprotIndex().

ResetParserStruct()

void ResetParserStruct

(

ParserPtr

pp

)

Definition at line 2111 of file indx_blk.cpp.

References Parser::curindx, Parser::entrylist, FreeIndexblk(), ProtBlk::ibp, Parser::indx, and Parser::pbp.

Referenced by fta_parse_buf(), and Parser::~Parser().

s_IsVDBWGSScaffold()

static bool s_IsVDBWGSScaffold ( string_view  accession )

static

Definition at line 1088 of file indx_blk.cpp.

References isalpha(), and isdigit().

Referenced by fta_if_wgs_acc().

s_RefineWGSType()

static int s_RefineWGSType ( string_view  accession, int  initialType  )

static

Definition at line 1130 of file indx_blk.cpp.

Referenced by fta_if_wgs_acc().

sFindNextNonSpace()

static CTempString::const_iterator sFindNextNonSpace ( const CTempString &  tempString, CTempString::const_iterator  current_it  )

static

Definition at line 689 of file indx_blk.cpp.

References CTempString::end(), and isSpace().

Referenced by sSetLocusLineOffsets().

sFindNextSpace()

static CTempString::const_iterator sFindNextSpace ( const CTempString &  tempString, CTempString::const_iterator  current_it  )

static

Definition at line 681 of file indx_blk.cpp.

References CTempString::end(), and isSpace().

Referenced by sSetLocusLineOffsets().

sIsUpperAlpha()

static bool sIsUpperAlpha ( char  c )

static

Definition at line 988 of file indx_blk.cpp.

Referenced by CheckAccession(), and IsNewAccessFormat().

SkipTitle() [1/2]

NCBI_UNUSED bool SkipTitle	(	FILE *	fp,
		FinfoBlk &	finfo,
		const char *	str,
		size_t	len
	)

Definition at line 342 of file indx_blk.cpp.

References fp, len, str(), FinfoBlk::str, StringEquN(), and XReadFile().

Referenced by SkipTitle().

SkipTitle() [2/2]

NCBI_UNUSED bool SkipTitle	(	FILE *	fp,
		FinfoBlk &	finfo,
		const CTempString &	keyword
	)

Definition at line 352 of file indx_blk.cpp.

References CTempString::data(), fp, CTempString::size(), and SkipTitle().

SkipTitleBuf()

bool SkipTitleBuf	(	FileBuf &	fbuf,
		FinfoBlk &	finfo,
		const CTempString &	keyword
	)

Definition at line 358 of file indx_blk.cpp.

References CTempString::data(), len, CTempString::size(), FinfoBlk::str, StringEquN(), and XReadFileBuf().

Referenced by EmblIndex(), GenBankIndex(), and SprotIndex().

sNotAllDigits()

static bool sNotAllDigits ( const char *  first, const char *  last  )

static

Definition at line 1491 of file indx_blk.cpp.

References first(), isdigit(), and last().

Referenced by CheckAccession().

SprotIndex()

bool SprotIndex	(	ParserPtr	pp,
		void(*)(IndexblkPtr entry, char *offset, Int4 len)	fun
	)

Definition at line 109 of file sp_index.cpp.

References TokenBlk::c_str(), Parser::curindx, data, Indexblk::date, Indexblk::drop, Parser::entrylist, ERR_ACCESSION_NoAccessNum, ERR_ENTRY_InvalidLineType, ERR_FORMAT_MissingEnd, ErrPostEx, ErrPostStr, Parser::ffbuf, FindNextEntryBuf(), FreeTokenstatblk(), FileBuf::get_offs(), GetAccession(), GetUpdateDate(), i, IndBlkNode::ibp, Parser::indx, InitialEntry(), isalpha(), Indexblk::len, TokenStatBlk::list, LoadEntry(), MsgSkipTitleFail(), TokenBlk::next, IndBlkNode::next, TokenStatBlk::num, Parser::num_drop, Indexblk::offset, ParFlat_COL_DATA_SP, ParFlatSP_AC, ParFlatSP_DT, ParFlatSP_END, ParFlatSP_ID, ParFlatSP_OS, ParFlatSP_RN, ParFlatSP_SQ, PointToNextToken(), SEV_ERROR, SkipTitleBuf(), Parser::source, Parser::sp_dt_seq_ver, sp_err_field(), SPGetVerNum(), FinfoBlk::str, StringEquN(), StringEquNI(), swissProtKeywords, TokenString(), Indexblk::vernum, and XReadFileBuf().

Referenced by FlatFileIndex().

sSetLocusLineOffsets()

static void sSetLocusLineOffsets ( const CTempString &  locusLine, LocusCont &  offsets  )

static

Definition at line 696 of file indx_blk.cpp.

References LocusCont::bases, CTempString::begin(), LocusCont::bp, LocusCont::date, LocusCont::div, CTempString::end(), LocusCont::molecule, sFindNextNonSpace(), sFindNextSpace(), LocusCont::strand, NStr::StringToNonNegativeInt(), CTempString::substr(), and LocusCont::topology.

Referenced by InitialEntry().

USING_SCOPE()

USING_SCOPE

(

objects

)

XMLCheckSTRAND()

Int2 XMLCheckSTRAND

(

const char *

str

)

Definition at line 485 of file indx_blk.cpp.

References str(), StringMatchIcase(), and XML_STRAND_array.

Referenced by XMLGetInst().

XMLCheckTPG()

Int2 XMLCheckTPG

(

const char *

str

)

Definition at line 491 of file indx_blk.cpp.

References i, str(), StringMatchIcase(), and XML_TPG_array.

Referenced by XMLGetInst().

XMLIndex()

bool XMLIndex

(

ParserPtr

pp

)

Definition at line 1420 of file xm_index.cpp.

References Parser::accver, Indexblk::bases, Parser::curindx, Indexblk::drop, Parser::entrylist, ERR_FORMAT_MissingEnd, ERR_FORMAT_XMLFormatError, ERR_INPUT_CannotReadEntry, ErrPostEx, Parser::indx, Indexblk::len, Indexblk::linenum, MemFree(), Parser::num_drop, SEV_ERROR, SEV_FATAL, Parser::source, Indexblk::xip, xmkwl, XMLAccessionsCheck(), XMLCheckRequiredTags(), XMLGetSegment(), XMLIndexFeatures(), XMLIndexReferences(), XMLInitialEntry(), XMLKeywordsCheck(), XMLLoadEntry(), XMLPerformIndex(), and XMLTagCheck().

Referenced by FlatFileIndex().

XReadFile()

static bool XReadFile ( FILE *  fp, FinfoBlk &  finfo  )

static

Definition at line 263 of file indx_blk.cpp.

References fp, FinfoBlk::line, n, FinfoBlk::pos, FinfoBlk::str, StringCpy(), and StringEquN().

Referenced by SkipTitle().

XReadFileBuf()

bool XReadFileBuf	(	FileBuf &	fbuf,
		FinfoBlk &	finfo
	)

Definition at line 314 of file indx_blk.cpp.

References FileGetsBuf(), FileBuf::get_offs(), FinfoBlk::line, FinfoBlk::pos, FinfoBlk::str, StringCpy(), and StringEquN().

Referenced by EmblIndex(), FindNextEntryBuf(), GenBankIndex(), SkipTitleBuf(), and SprotIndex().

Variable Documentation

acc_tsa_allowed

const char* acc_tsa_allowed[]

static

Initial value:

= {
    "AF", "AY", "DQ", "EF", "EU", "FJ", "GQ", "HQ", "JF", "JN", "JQ", "JX",
    "KC", "KF", "KJ", "KM", "KP", "KR", "KT", "KU", "KX", "KY", "MF", "MG",
    "MH", "MK", "MN", "MT", nullptr
}

Definition at line 145 of file indx_blk.cpp.

Referenced by IsTSAAccPrefix().

ddbj_accpref

const char* ddbj_accpref[]

static

Initial value:

= {
    "AB", "AG", "AK", "AP", "AT", "AU", "AV", "BA", "BB", "BD", "BJ", "BP",
    "BR", "BS", "BW", "BY", "CI", "CJ", "DA", "DB", "DC", "DD", "DE", "DF",
    "DG", "DH", "DI", "DJ", "DK", "DL", "DM", "FS", "FT", "FU", "FV", "FW",
    "FX", "FY", "FZ", "GA", "GB", "HT", "HU", "HV", "HW", "HX", "HY", "HZ",
    "LA", "LB", "LC", "LD", "LE", "LF", "LG", "LH", "LI", "LJ", "LU", "LV",
    "LX", "LY", "LZ", "MA", "MB", "MC", "MD", "ME", "OF", "OG", "OH", "OI",
    "OJ", "PA", "PE", "PF", "PG", "PH", "PI", "PJ", "PK", nullptr
}

Definition at line 104 of file indx_blk.cpp.

Referenced by GetAccArray().

ddbj_tpa_accpref

const char* ddbj_tpa_accpref[]

static

Initial value:

= {
    "BR", "HT", "HU", nullptr
}

Definition at line 155 of file indx_blk.cpp.

Referenced by IsTPAAccPrefix().

ddbj_wgs_accpref

const char* ddbj_wgs_accpref[]

static

Initial value:

= {
    "HT", "HU", nullptr
}

Definition at line 163 of file indx_blk.cpp.

Referenced by IsWGSAccPrefix().

embl_accpref

const char* embl_accpref[]

static

Initial value:

= {
    "AJ", "AL", "AM", "AN", "AX", "BN", "BX", "CQ", "CR", "CS", "CT", "CU",
    "FB", "FM", "FN", "FO", "FP", "FQ", "FR", "GM", "GN", "HA", "HB", "HC",
    "HD", "HE", "HF", "HG", "HH", "HI", "JA", "JB", "JC", "JD", "JE", "LK",
    "LL", "LM", "LN", "LO", "LP", "LQ", "LR", "LS", "LT", "MP", "MQ", "MR",
    "MS", "OA", "OB", "OC", "OD", "OE", "OU", "OV", "OW", "OX", "OY", "OZ",
    nullptr
}

Definition at line 87 of file indx_blk.cpp.

Referenced by GetAccArray().

k_WgsScaffoldPrefix

const set<string_view> k_WgsScaffoldPrefix

static

Initial value:

= {
    "CH", "CT", "CU", "DF", "DG", "DS",
    "EM", "EN", "EP", "EQ", "FA", "FM",
    "GG", "GJ", "GK", "GL", "HT", "HU",
    "JH", "KB", "KD", "KE", "KI", "KK",
    "KL", "KN", "KQ", "KV", "KZ", "LD",
    "ML", "MU", "PS"
}

Definition at line 167 of file indx_blk.cpp.

Referenced by fta_if_wgs_acc().

lanl_accpref

const char* lanl_accpref[]

static

Initial value:

= {
    "AD", nullptr
}

Definition at line 96 of file indx_blk.cpp.

Referenced by GetAccArray().

month_name

const char* month_name[]

static

Initial value:

= {
    "Ill", "JAN", "FEB", "MAR", "APR", "MAY", "JUN", "JUL", "AUG", "SEP", "OCT", "NOV", "DEC", nullptr
}

Definition at line 187 of file indx_blk.cpp.

Referenced by CkDateFormat(), and get_month_code_by_name().

ncbi_accpref

const char* ncbi_accpref[]

static

Initial value:

= {
    "AA", "AC", "AD", "AE", "AF", "AH", "AI", "AQ", "AR", "AS", "AW", "AY",
    "AZ", "BC", "BE", "BF", "BG", "BH", "BI", "BK", "BL", "BM", "BQ", "BT",
    "BU", "BV", "BZ", "CA", "CB", "CC", "CD", "CE", "CF", "CG", "CH", "CK",
    "CL", "CM", "CN", "CO", "CP", "CV", "CW", "CX", "CY", "CZ", "DN", "DP",
    "DQ", "DR", "DS", "DT", "DU", "DV", "DW", "DX", "DY", "DZ", "EA", "EB",
    "EC", "ED", "EE", "EF", "EG", "EH", "EI", "EJ", "EK", "EL", "EM", "EN",
    "EP", "EQ", "ER", "ES", "ET", "EU", "EV", "EW", "EX", "EY", "EZ", "FA",
    "FC", "FD", "FE", "FF", "FG", "FH", "FI", "FJ", "FK", "FL", "GC", "GD",
    "GE", "GF", "GG", "GH", "GJ", "GK", "GL", "GO", "GP", "GQ", "GR", "GS",
    "GT", "GU", "GV", "GW", "GX", "GY", "GZ", "HJ", "HK", "HL", "HM", "HN",
    "HO", "HP", "HQ", "HR", "HS", "JF", "JG", "JH", "JI", "JJ", "JK", "JL",
    "JM", "JN", "JO", "JP", "JQ", "JR", "JS", "JT", "JU", "JV", "JW", "JX",
    "JY", "JZ", "KA", "KB", "KC", "KD", "KE", "KF", "KG", "KH", "KI", "KJ",
    "KK", "KL", "KM", "KN", "KO", "KP", "KQ", "KR", "KS", "KT", "KU", "KV",
    "KX", "KY", "KZ", "MF", "MG", "MH", "MI", "MJ", "MK", "ML", "MM", "MN",
    "MO", "MT", "MU", "MV", "MW", "MX", "MY", "MZ", "OK", "OL", "OM", "ON",
    "OO", "OP", "OQ", "OR", "OS", "OT", "PP", "PR", "PS", "PT", nullptr
}

Definition at line 114 of file indx_blk.cpp.

Referenced by GetAccArray().

ncbi_tpa_accpref

const char* ncbi_tpa_accpref[]

static

Initial value:

= {
    "BK", "BL", "GJ", "GK", nullptr
}

Definition at line 151 of file indx_blk.cpp.

Referenced by IsTPAAccPrefix().

ncbi_wgs_accpref

const char* ncbi_wgs_accpref[]

static

Initial value:

= {
    "GJ", "GK", nullptr
}

Definition at line 159 of file indx_blk.cpp.

Referenced by IsWGSAccPrefix().

ParFlat_AA_array_DDBJ

const char* ParFlat_AA_array_DDBJ[]

static

Initial value:

= {
    "PRT", nullptr
}

Definition at line 70 of file indx_blk.cpp.

Referenced by CkLocusLinePos().

ParFlat_DIV_array

const char* ParFlat_DIV_array[]

static

Initial value:

= {
    "   ", "PRI", "ROD", "MAM", "VRT", "INV", "PLN", "BCT", "RNA",
    "VRL", "PHG", "SYN", "UNA", "EST", "PAT", "STS", "ORG", "GSS",
    "HUM", "HTG", "CON", "HTC", "ENV", "TSA", nullptr
}

Definition at line 81 of file indx_blk.cpp.

Referenced by CheckDIV().

ParFlat_NA_array

const char* ParFlat_NA_array[]

static

Initial value:

= {
    "    ", "NA", "DNA", "genomic DNA", "other DNA", "unassigned DNA", "RNA",
    "mRNA", "rRNA", "tRNA", "uRNA", "scRNA", "snRNA", "snoRNA", "pre-RNA",
    "pre-mRNA", "genomic RNA", "other RNA", "unassigned RNA", "cRNA",
    "viral cRNA", nullptr
}

Definition at line 74 of file indx_blk.cpp.

Referenced by CheckNA(), and CkLocusLinePos().

ParFlat_NA_array_DDBJ

const char* ParFlat_NA_array_DDBJ[]

static

Initial value:

= {
    "cDNA", nullptr
}

Definition at line 66 of file indx_blk.cpp.

Referenced by CheckNADDBJ(), and CkLocusLinePos().

ParFlat_RESIDUE_STR

const char* ParFlat_RESIDUE_STR[]

static

Initial value:

= {
    "bp", "bp.", "bp,", "AA", "AA.", "AA,", nullptr
}

Definition at line 191 of file indx_blk.cpp.

Referenced by GetResidue().

refseq_accpref

const char* refseq_accpref[]

static

Initial value:

= {
    "NC_", "NG_", "NM_", "NP_", "NR_", "NT_", "NW_", "XM_", "XP_", "XR_",
    "NZ_", nullptr
}

Definition at line 134 of file indx_blk.cpp.

Referenced by GetAccArray().

sourceNames

const map<Parser::ESource, string> sourceNames

static

Initial value:

= {
    { Parser::ESource::unknown, "unknown" },
    { Parser::ESource::EMBL, "EMBL" },
    { Parser::ESource::GenBank, "GENBANK" },
    { Parser::ESource::SPROT, "Swiss-Prot" },
    { Parser::ESource::NCBI, "NCBI" },
    { Parser::ESource::LANL, "GSDB" },
    { Parser::ESource::Flybase, "FlyBase" },
    { Parser::ESource::Refseq, "RefSeq" }
}

Definition at line 176 of file indx_blk.cpp.

Referenced by GetAccession().

sprot_accpref

const char* sprot_accpref[]

static

Initial value:

= {
    "DD", nullptr
}

Definition at line 100 of file indx_blk.cpp.

Referenced by GetAccArray(), and IsSPROTAccession().

ValidMolTypes

const char* ValidMolTypes[]

static

Initial value:

= {
    "genomic DNA",
    "genomic RNA",
    "mRNA",
    "tRNA",
    "rRNA",
    "snoRNA",
    "snRNA",
    "scRNA",
    "pre-RNA",
    "pre-mRNA",
    "other RNA",
    "other DNA",
    "transcribed RNA",
    "unassigned RNA",
    "unassigned DNA",
    "viral cRNA",
    nullptr
}

Definition at line 195 of file indx_blk.cpp.

Referenced by fta_check_embl_moltype().

XML_STRAND_array

const char* XML_STRAND_array[]

static

Initial value:

= {
    "   ", "single", "double", "mixed", nullptr
}

Definition at line 58 of file indx_blk.cpp.

Referenced by XMLCheckSTRAND().

XML_TPG_array

const char* XML_TPG_array[]

static

Initial value:

= {
    "   ", "Linear", "Circular", "Tandem", nullptr
}

Definition at line 62 of file indx_blk.cpp.

Referenced by XMLCheckTPG().