utilfun.cpp File Reference

#include <ncbi_pch.hpp>
#include <corelib/ncbitime.hpp>
#include "ftacpp.hpp"
#include <corelib/ncbistr.hpp>
#include <objmgr/scope.hpp>
#include <objmgr/object_manager.hpp>
#include <objects/seq/MolInfo.hpp>
#include <objects/seqloc/PDB_seq_id.hpp>
#include <corelib/tempstr.hpp>
#include "index.h"
#include "ftaerr.hpp"
#include "indx_def.h"
#include "utilfun.h"

Include dependency graph for utilfun.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Macros
#define	THIS_FILE   "utilfun.cpp"

Functions
	USING_SCOPE (objects)
CScope &	GetScope ()
static string	FTAitoa (Int4 m)
void	UnwrapAccessionRange (const CGB_block::TExtra_accessions &extra_accs, CGB_block::TExtra_accessions &hist)
static bool	sIsPrefixChar (char c)
bool	ParseAccessionRange (list< string > &tokens, unsigned skip)
bool	IsLeadPrefixChar (char c)
bool	IsDigit (char c)
bool	ParseAccessionRange (TokenStatBlkPtr tsbp, unsigned skip)
unique_ptr< TokenStatBlk >	TokenString (const char *str, Char delimiter)
Int2	fta_StringMatch (const Char **array, string_view text)
Int2	StringMatchIcase (const Char **array, string_view text)
Int2	MatchArrayString (const char **array, const char *text)
Int2	MatchArrayIString (const Char **array, const Char *text)
Int2	MatchArraySubString (const Char **array, string_view text)
Char *	StringIStr (const Char *where, const Char *what)
Int2	MatchArrayISubString (const Char **array, string_view text)
string	GetBlkDataReplaceNewLine (string_view instr, Uint2 indent)
static size_t	SeekLastAlphaChar (const Char *str, size_t len)
void	CleanTailNoneAlphaCharInString (string &str)
void	CleanTailNoneAlphaChar (char *str)
char *	PointToNextToken (char *ptr)
char *	GetTheCurrentToken (char **ptr)
char *	SrchTheChar (char *bptr, char *eptr, Char letter)
char *	SrchTheStr (char *bptr, char *eptr, const char *leadstr)
void	CpSeqId (InfoBioseqPtr ibp, const CSeq_id &id)
CRef< CDate_std >	get_full_date (const char *s, bool is_ref, Parser::ESource source)
int	SrchKeyword (const CTempString &ptr, const vector< string > &keywordList)
bool	CheckLineType (char *ptr, Int4 line, const vector< string > &keywordList, bool after_origin)
char *	SrchNodeType (DataBlkPtr entry, Int4 type, size_t *len)
char *	xSrchNodeType (const DataBlk &entry, Int4 type, size_t *len)
string	xGetNodeData (const DataBlk &entry, int nodeType)
DataBlkPtr	TrackNodeType (const DataBlk &entry, Int2 type)
const Section *	xTrackNodeType (const Entry &entry, int type)
bool	fta_tpa_keywords_check (const TKeywordList &kwds)
bool	fta_tsa_keywords_check (const TKeywordList &kwds, Parser::ESource source)
bool	fta_tls_keywords_check (const TKeywordList &kwds, Parser::ESource source)
bool	fta_is_tpa_keyword (const char *str)
bool	fta_is_tsa_keyword (const char *str)
bool	fta_is_tls_keyword (const char *str)
void	fta_keywords_check (const char *str, bool *estk, bool *stsk, bool *gssk, bool *htck, bool *flik, bool *wgsk, bool *tpak, bool *envk, bool *mgak, bool *tsak, bool *tlsk)
void	fta_remove_keywords (CMolInfo::TTech tech, TKeywordList &kwds)
void	fta_remove_tpa_keywords (TKeywordList &kwds)
void	fta_remove_tsa_keywords (TKeywordList &kwds, Parser::ESource source)
void	fta_remove_tls_keywords (TKeywordList &kwds, Parser::ESource source)
void	fta_remove_env_keywords (TKeywordList &kwds)
void	fta_remove_mag_keywords (TKeywordList &kwds)
void	xCheckEstStsGssTpaKeywords (const list< string > keywordList, bool tpa_check, IndexblkPtr entry)
void	check_est_sts_gss_tpa_kwds (ValNodePtr kwds, size_t len, IndexblkPtr entry, bool tpa_check, bool &specialist_db, bool &inferential, bool &experimental, bool &assembly)
ValNodePtr	ConstructValNode (CSeq_id::E_Choice choice, const char *data)
bool	fta_check_mga_keywords (CMolInfo &mol_info, const TKeywordList &kwds)
void	fta_StringCpy (char *dst, const char *src)
bool	SetTextId (Uint1 seqtype, CSeq_id &seqId, CTextseq_id &textId)
bool	IsCancelled (const TKeywordList &keywords)
bool	HasHtg (const TKeywordList &keywords)
void	RemoveHtgPhase (TKeywordList &keywords)
bool	HasHtc (const TKeywordList &keywords)

Variables
static const char *	ParFlat_EST_kw_array []
static const char *	ParFlat_GSS_kw_array []
static const char *	ParFlat_STS_kw_array []
static const char *	ParFlat_HTC_kw_array []
static const char *	ParFlat_FLI_kw_array []
static const char *	ParFlat_WGS_kw_array []
static const char *	ParFlat_MGA_kw_array []
static const char *	ParFlat_MGA_more_kw_array []
static const char *	ParFlat_TSA_kw_array []
static const char *	ParFlat_TLS_kw_array []
static const char *	ParFlat_TPA_kw_array []
static const char *	ParFlat_TPA_kw_array_to_remove []
static const char *	ParFlat_ENV_kw_array []
static const char *	ParFlat_MAG_kw_array []

Macro Definition Documentation

THIS_FILE

#define THIS_FILE   "utilfun.cpp"

Definition at line 56 of file utilfun.cpp.

Function Documentation

check_est_sts_gss_tpa_kwds()

void check_est_sts_gss_tpa_kwds	(	ValNodePtr	kwds,
		size_t	len,
		IndexblkPtr	entry,
		bool	tpa_check,
		bool &	specialist_db,
		bool &	inferential,
		bool &	experimental,
		bool &	assembly
	)

Definition at line 1371 of file utilfun.cpp.

References ValNode::data, NStr::EqualNocase(), Indexblk::EST, fta_keywords_check(), Indexblk::GSS, Indexblk::HTC, Indexblk::is_tpa, len, MemFree(), ValNode::next, StringCat(), StringChr, StringNew(), and Indexblk::STS.

Referenced by GenBankIndex(), and XMLKeywordsCheck().

CheckLineType()

bool CheckLineType	(	char *	ptr,
		Int4	line,
		const vector< string > &	keywordList,
		bool	after_origin
	)

Definition at line 910 of file utilfun.cpp.

References ERR_ENTRY_InvalidLineType, ErrPostEx, i, msg(), SEV_ERROR, StringChr, StringEquN(), and StringNCpy().

Referenced by EmblIndex(), and GenBankIndex().

CleanTailNoneAlphaChar()

void CleanTailNoneAlphaChar

(

char *

str

)

Definition at line 697 of file utilfun.cpp.

References last(), SeekLastAlphaChar(), and str().

Referenced by GetOneGeneRef(), and GetTheCurrentToken().

CleanTailNoneAlphaCharInString()

void CleanTailNoneAlphaCharInString

(

string &

str

)

Definition at line 683 of file utilfun.cpp.

References SeekLastAlphaChar(), and str().

Referenced by GetBioseqSetDescrTitle(), GetEmblBlockXref(), and ParseRLDataSP().

ConstructValNode()

ValNodePtr ConstructValNode	(	CSeq_id::E_Choice	choice,
		const char *	data
	)

Definition at line 1437 of file utilfun.cpp.

References ValNode::choice, data, and ValNodeNew().

Referenced by AddToList(), GenBankIndex(), GetDRlineDataSP(), and XMLKeywordsCheck().

CpSeqId()

void CpSeqId	(	InfoBioseqPtr	ibp,
		const CSeq_id &	id
	)

Definition at line 791 of file utilfun.cpp.

References CSeq_id::Assign(), CTextseq_id_Base::GetAccession(), CTextseq_id_Base::GetName(), CTextseq_id_Base::GetVersion(), InfoBioseq::ids, CTextseq_id_Base::IsSetAccession(), CTextseq_id_Base::IsSetName(), CTextseq_id_Base::IsSetVersion(), InfoBioseq::mAccNum, InfoBioseq::mLocus, Ref(), and SetTextId().

Referenced by FindCd().

fta_check_mga_keywords()

bool fta_check_mga_keywords	(	CMolInfo &	mol_info,
		const TKeywordList &	kwds
	)

Definition at line 1447 of file utilfun.cpp.

References NStr::EqualNocase(), ERR_KEYWORD_ConflictingMGAKeywords, ERR_KEYWORD_MissingMGAKeywords, ErrPostStr, CMolInfo_Base::GetTechexp(), CMolInfo_Base::IsSetTechexp(), ncbi::grid::netcache::search::fields::key, MatchArrayIString(), ParFlat_MGA_more_kw_array, CMolInfo_Base::SetTechexp(), and SEV_REJECT.

Referenced by GetGBBlock(), and XMLGetGBBlock().

fta_is_tls_keyword()

bool fta_is_tls_keyword

(

const char *

str

)

Definition at line 1187 of file utilfun.cpp.

References MatchArrayIString(), ParFlat_TLS_kw_array, and str().

fta_is_tpa_keyword()

bool fta_is_tpa_keyword

(

const char *

str

)

Definition at line 1170 of file utilfun.cpp.

References MatchArrayIString(), ParFlat_TPA_kw_array, and str().

Referenced by fta_fix_tpa_keywords().

fta_is_tsa_keyword()

bool fta_is_tsa_keyword

(

const char *

str

)

Definition at line 1179 of file utilfun.cpp.

References MatchArrayIString(), ParFlat_TSA_kw_array, and str().

fta_keywords_check()

void fta_keywords_check	(	const char *	str,
		bool *	estk,
		bool *	stsk,
		bool *	gssk,
		bool *	htck,
		bool *	flik,
		bool *	wgsk,
		bool *	tpak,
		bool *	envk,
		bool *	mgak,
		bool *	tsak,
		bool *	tlsk
	)

Definition at line 1195 of file utilfun.cpp.

References MatchArrayString(), ParFlat_ENV_kw_array, ParFlat_EST_kw_array, ParFlat_FLI_kw_array, ParFlat_GSS_kw_array, ParFlat_HTC_kw_array, ParFlat_MGA_kw_array, ParFlat_STS_kw_array, ParFlat_TLS_kw_array, ParFlat_TPA_kw_array, ParFlat_TSA_kw_array, ParFlat_WGS_kw_array, and str().

Referenced by check_est_sts_gss_tpa_kwds(), GetDescrEmblBlock(), GetGBBlock(), xCheckEstStsGssTpaKeywords(), XMLGetEMBLBlock(), and XMLGetGBBlock().

fta_remove_env_keywords()

void fta_remove_env_keywords

(

TKeywordList &

kwds

)

Definition at line 1307 of file utilfun.cpp.

References ncbi::grid::netcache::search::fields::key, MatchArrayIString(), and ParFlat_ENV_kw_array.

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

fta_remove_keywords()

void fta_remove_keywords	(	CMolInfo::TTech	tech,
		TKeywordList &	kwds
	)

Definition at line 1232 of file utilfun.cpp.

References b, CMolInfo_Base::eTech_est, CMolInfo_Base::eTech_fli_cdna, CMolInfo_Base::eTech_htc, CMolInfo_Base::eTech_sts, CMolInfo_Base::eTech_survey, CMolInfo_Base::eTech_wgs, ncbi::grid::netcache::search::fields::key, MatchArrayString(), ParFlat_EST_kw_array, ParFlat_FLI_kw_array, ParFlat_GSS_kw_array, ParFlat_HTC_kw_array, ParFlat_STS_kw_array, and ParFlat_WGS_kw_array.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

fta_remove_mag_keywords()

void fta_remove_mag_keywords

(

TKeywordList &

kwds

)

Definition at line 1321 of file utilfun.cpp.

References ncbi::grid::netcache::search::fields::key, MatchArrayIString(), and ParFlat_MAG_kw_array.

Referenced by GetGBBlock().

fta_remove_tls_keywords()

void fta_remove_tls_keywords	(	TKeywordList &	kwds,
		Parser::ESource	source
	)

Definition at line 1292 of file utilfun.cpp.

References Parser::EMBL, NStr::EqualNocase(), ncbi::grid::netcache::search::fields::key, MatchArrayIString(), ParFlat_TLS_kw_array, and rapidjson::source.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

fta_remove_tpa_keywords()

void fta_remove_tpa_keywords

(

TKeywordList &

kwds

)

Definition at line 1263 of file utilfun.cpp.

References ncbi::grid::netcache::search::fields::key, MatchArrayIString(), and ParFlat_TPA_kw_array_to_remove.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

fta_remove_tsa_keywords()

void fta_remove_tsa_keywords	(	TKeywordList &	kwds,
		Parser::ESource	source
	)

Definition at line 1277 of file utilfun.cpp.

References Parser::EMBL, NStr::EqualNocase(), ncbi::grid::netcache::search::fields::key, MatchArrayIString(), ParFlat_TSA_kw_array, and rapidjson::source.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

fta_StringCpy()

void fta_StringCpy	(	char *	dst,
		const char *	src
	)

Definition at line 1496 of file utilfun.cpp.

Referenced by CheckDelGbblockSourceFromDescrs(), DefVsHTGKeywords(), DelParenthesis(), ExtractErratum(), FakeEmblBioSources(), fta_get_part_sup(), fta_parse_rrna_feat(), fta_parse_structured_comment(), fta_strip_aa(), GetDescrComment(), GetSPDescrComment(), GetSPDescrTitle(), ind_subdbp(), LoadEntry(), LoadEntryGenbank(), ShrinkSpaces(), SPPostProcVarSeq(), SPValidateEcnum(), stripStr(), XMLGetDescr(), and XMLGetDescrComment().

fta_StringMatch()

Int2 fta_StringMatch	(	const Char **	array,
		string_view	text
	)

Definition at line 486 of file utilfun.cpp.

References NStr::EqualCase(), i, StringLen(), and text().

Referenced by CheckDIV(), CheckNA(), CheckNADDBJ(), CkSPComTopics(), EmblGetDivisionNewID(), GetDescrEmblBlock(), GetEmblBlockXref(), IsTPAAccPrefix(), IsTSAAccPrefix(), IsWGSAccPrefix(), ParseSpComment(), and XMLGetEMBLBlock().

fta_tls_keywords_check()

bool fta_tls_keywords_check	(	const TKeywordList &	kwds,
		Parser::ESource	source
	)

Definition at line 1136 of file utilfun.cpp.

References Parser::EMBL, NStr::EqualNocase(), ERR_KEYWORD_MissingTLSKeywords, ErrPostStr, i, ncbi::grid::netcache::search::fields::key, MatchArrayIString(), ParFlat_TLS_kw_array, SEV_REJECT, and rapidjson::source.

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

fta_tpa_keywords_check()

bool fta_tpa_keywords_check

(

const TKeywordList &

kwds

)

Definition at line 1020 of file utilfun.cpp.

References b, buf, NStr::EqualNocase(), ERR_KEYWORD_ConflictingTPATiers, ERR_KEYWORD_InvalidTPATier, ERR_KEYWORD_MissingTPAKeywords, ERR_KEYWORD_MissingTPATier, ERR_KEYWORD_UnexpectedTPA, ErrPostEx, ErrPostStr, i, ncbi::grid::netcache::search::fields::key, len, MatchArrayIString(), ParFlat_TPA_kw_array, SEV_ERROR, SEV_REJECT, and SEV_WARNING.

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

fta_tsa_keywords_check()

bool fta_tsa_keywords_check	(	const TKeywordList &	kwds,
		Parser::ESource	source
	)

Definition at line 1102 of file utilfun.cpp.

References Parser::EMBL, NStr::EqualNocase(), ERR_KEYWORD_MissingTSAKeywords, ErrPostStr, i, ncbi::grid::netcache::search::fields::key, MatchArrayIString(), ParFlat_TSA_kw_array, SEV_REJECT, and rapidjson::source.

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

FTAitoa()

static string FTAitoa ( Int4  m )

static

Definition at line 179 of file utilfun.cpp.

Referenced by UnwrapAccessionRange().

get_full_date()

CRef<CDate_std> get_full_date	(	const char *	s,
		bool	is_ref,
		Parser::ESource	source
	)

Definition at line 827 of file utilfun.cpp.

References ERR_DATE_IllegalDate, ERR_REFERENCE_IllegalDate, ErrPostEx, int, isdigit(), months, msg(), CRef< C, Locker >::Reset(), CDate_std_Base::SetDay(), CDate_std_Base::SetMonth(), CDate_std_Base::SetYear(), SEV_ERROR, SEV_WARNING, rapidjson::source, Parser::SPROT, StringNCpy(), and CTime::Year().

Referenced by get_pat(), get_sub(), get_sub_gsdb(), GetCitPatent(), GetUpdateDate(), and MakeDatePtr().

GetBlkDataReplaceNewLine()

string GetBlkDataReplaceNewLine	(	string_view	instr,
		Uint2	indent
	)

Definition at line 644 of file utilfun.cpp.

References indent(), last(), NStr::Split(), NStr::StartsWith(), and NStr::TruncateSpacesInPlace().

Referenced by GetDescrEmblBlock(), GetEmblDescr(), GetGBBlock(), GetGenBankDescr(), GetGenBankLineage(), GetSequenceOfKeywords(), GetSPDescrTitle(), SPFeatGeneRef(), SPFeatProtRef(), and SprotRefString().

GetScope()

CScope& GetScope

(

void

)

Definition at line 62 of file utilfun.cpp.

References CScope::AddTopLevelSeqEntry(), CObjectManager::GetInstance(), MSerial_AsnText, CRef< C, Locker >::Reset(), and sc_TestEntry.

GetTheCurrentToken()

char* GetTheCurrentToken

(

char **

ptr

)

Definition at line 727 of file utilfun.cpp.

References CleanTailNoneAlphaChar(), str(), and StringSave.

Referenced by ParseRLDataSP().

HasHtc()

bool HasHtc

(

const TKeywordList &

keywords

)

Definition at line 1603 of file utilfun.cpp.

References NStr::EqualNocase(), and ncbi::grid::netcache::search::fields::key.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

HasHtg()

bool HasHtg

(

const TKeywordList &

keywords

)

Definition at line 1574 of file utilfun.cpp.

References ncbi::grid::netcache::search::fields::key.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

IsCancelled()

bool IsCancelled

(

const TKeywordList &

keywords

)

Definition at line 1563 of file utilfun.cpp.

References NStr::EqualNocase(), and ncbi::grid::netcache::search::fields::key.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

IsDigit()

bool IsDigit ( char  c )

inline

Definition at line 344 of file utilfun.cpp.

Referenced by ParseAccessionRange().

IsLeadPrefixChar()

bool IsLeadPrefixChar ( char  c )

inline

Definition at line 340 of file utilfun.cpp.

Referenced by ParseAccessionRange().

MatchArrayIString()

Int2 MatchArrayIString	(	const Char **	array,
		const Char *	text
	)

Definition at line 549 of file utilfun.cpp.

References NStr::EqualNocase(), i, and text().

Referenced by DbxrefQualToDbtag(), fta_check_mga_keywords(), fta_is_tls_keyword(), fta_is_tpa_keyword(), fta_is_tsa_keyword(), fta_remove_env_keywords(), fta_remove_mag_keywords(), fta_remove_tls_keywords(), fta_remove_tpa_keywords(), fta_remove_tsa_keywords(), fta_tls_keywords_check(), fta_tpa_keywords_check(), fta_tsa_keywords_check(), and SPGetPEValue().

MatchArrayISubString()

Int2 MatchArrayISubString	(	const Char **	array,
		string_view	text
	)

Definition at line 623 of file utilfun.cpp.

References NStr::FindNoCase(), i, NPOS, and text().

MatchArrayString()

Int2 MatchArrayString	(	const char **	array,
		const char *	text
	)

Definition at line 533 of file utilfun.cpp.

References NStr::Equal(), i, and text().

Referenced by fta_check_ncrna(), fta_check_pseudogene_qual(), fta_check_satellite(), fta_keywords_check(), fta_remove_keywords(), GetFeatNameAndLoc(), and GetRnaRef().

MatchArraySubString()

Int2 MatchArraySubString	(	const Char **	array,
		string_view	text
	)

Definition at line 578 of file utilfun.cpp.

References NStr::Find(), i, NPOS, and text().

Referenced by CkDateFormat().

ParseAccessionRange() [1/2]

bool ParseAccessionRange	(	list< string > &	tokens,
		unsigned	skip
	)

Definition at line 265 of file utilfun.cpp.

References NStr::EqualCase(), ERR_ACCESSION_2ndAccPrefixMismatch, ERR_ACCESSION_Invalid2ndAccRange, ErrPostEx, first(), last(), SEV_REJECT, sIsPrefixChar(), NStr::SplitInTwo(), and NStr::StringToInt().

Referenced by GetAccession().

ParseAccessionRange() [2/2]

bool ParseAccessionRange	(	TokenStatBlkPtr	tsbp,
		unsigned	skip
	)

Definition at line 349 of file utilfun.cpp.

References ERR_ACCESSION_2ndAccPrefixMismatch, ERR_ACCESSION_Invalid2ndAccRange, ErrPostEx, NStr::fConvErr_NoThrow, first(), IsDigit(), IsLeadPrefixChar(), last(), TokenStatBlk::list, TokenStatBlk::num, SEV_REJECT, sIsPrefixChar(), NStr::StringToInt(), and tmp.

PointToNextToken()

char* PointToNextToken

(

char *

ptr

)

Definition at line 707 of file utilfun.cpp.

Referenced by GetDescrEmblBlock(), GetDescrSPBlock(), GetEmblBlockXref(), GetEmblInst(), GetEmblMolInfo(), GetReleaseInfo(), ParseRLDataSP(), and SprotIndex().

RemoveHtgPhase()

void RemoveHtgPhase

(

TKeywordList &

keywords

)

Definition at line 1588 of file utilfun.cpp.

References NStr::EqualNocase(), and ncbi::grid::netcache::search::fields::key.

Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().

SeekLastAlphaChar()

static size_t SeekLastAlphaChar ( const Char *  str, size_t  len  )

static

Definition at line 667 of file utilfun.cpp.

References len, and str().

Referenced by CleanTailNoneAlphaChar(), and CleanTailNoneAlphaCharInString().

SetTextId()

bool SetTextId	(	Uint1	seqtype,
		CSeq_id &	seqId,
		CTextseq_id &	textId
	)

Definition at line 1507 of file utilfun.cpp.

References CSeq_id_Base::e_Ddbj, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_Gpipe, CSeq_id_Base::e_Named_annot_track, CSeq_id_Base::e_Other, CSeq_id_Base::e_Pdb, CSeq_id_Base::e_Pir, CSeq_id_Base::e_Prf, CSeq_id_Base::e_Swissprot, CSeq_id_Base::e_Tpd, CSeq_id_Base::e_Tpe, CSeq_id_Base::e_Tpg, CPDB_seq_id_Base::SetChain_id(), CSeq_id_Base::SetDdbj(), CSeq_id_Base::SetEmbl(), CSeq_id_Base::SetGenbank(), CSeq_id_Base::SetGpipe(), CSeq_id_Base::SetNamed_annot_track(), CSeq_id_Base::SetOther(), CSeq_id_Base::SetPdb(), CSeq_id_Base::SetPir(), CSeq_id_Base::SetPrf(), CSeq_id_Base::SetSwissprot(), CSeq_id_Base::SetTpd(), CSeq_id_Base::SetTpe(), and CSeq_id_Base::SetTpg().

Referenced by CpSeqId(), CreateEntryBioseq(), fta_fix_seq_id(), fta_parse_tpa_tsa_block(), GetProtRefSeqId(), MakeAccSeqId(), MakeLocusSeqId(), and MakeSegSetSeqId().

sIsPrefixChar()

static bool sIsPrefixChar ( char  c )

static

Definition at line 260 of file utilfun.cpp.

Referenced by ParseAccessionRange().

SrchKeyword()

int SrchKeyword	(	const CTempString &	ptr,
		const vector< string > &	keywordList
	)

Definition at line 897 of file utilfun.cpp.

References i, ParFlat_UNKW, and NStr::StartsWith().

Referenced by GenBankIndex(), GetEmblBlock(), GetGenBankBlock(), and xGetGenBankBlocks().

SrchNodeType()

char* SrchNodeType	(	DataBlkPtr	entry,
		Int4	type,
		size_t *	len
	)

Definition at line 949 of file utilfun.cpp.

References len, DataBlk::len, DataBlk::mOffset, and TrackNodeType().

Referenced by GetDRlineDataSP(), GetSPDate(), GetSPDescrComment(), GetSprotDescr(), ParseSPFeat(), SpAddToIndexBlk(), SPFeatGeneRef(), SPFeatProtRef(), and SPGetPEValue().

SrchTheChar()

char* SrchTheChar	(	char *	bptr,
		char *	eptr,
		Char	letter
	)

Definition at line 759 of file utilfun.cpp.

References i, and letter().

Referenced by BuildFeatureBlock(), GetDescrComment(), GetEmblBlockXref(), GetEmblDate(), GetGBBlock(), GetReleaseInfo(), GetSubNodeType(), ParseSPFeat(), and TrimEmblFeatBlk().

SrchTheStr()

char* SrchTheStr	(	char *	bptr,
		char *	eptr,
		const char *	leadstr
	)

Definition at line 779 of file utilfun.cpp.

References i.

Referenced by check_cds(), GetDescrComment(), GetEmblRefType(), GetFlatBiomol(), and GetReleaseInfo().

StringIStr()

Char* StringIStr	(	const Char *	where,
		const Char *	what
	)

Definition at line 591 of file utilfun.cpp.

Referenced by CheckForUnusualFullLengthOrgs(), fill_orgref(), fta_get_trna_from_product(), fta_if_special_org(), fta_parse_rrna_feat(), fta_remark_is_er(), fta_strip_pub_comment(), get_embl_pmid(), get_embl_str_pub_id(), get_muid(), get_plasmid(), GetCitBook(), GetCitBookOld(), GetFlatBiomol(), GetRnaRef(), GetSPDate(), GetSPDescrTitle(), GetTaxnameNameFromDescrs(), GetViralHostsFrom_OH(), QualsToSeqID(), SPFeatGeneRef(), and SPGetVerNum().

StringMatchIcase()

Int2 StringMatchIcase	(	const Char **	array,
		string_view	text
	)

Definition at line 507 of file utilfun.cpp.

References NStr::EqualNocase(), i, StringLen(), and text().

Referenced by CheckDelGbblockSourceFromDescrs(), CkLocusLinePos(), get_plasmid(), GetGenomeInfo(), GetSPGenome(), UpdateRawBioSource(), XMLCheckSTRAND(), and XMLCheckTPG().

TokenString()

unique_ptr<TokenStatBlk> TokenString	(	const char *	str,
		Char	delimiter
	)

Definition at line 445 of file utilfun.cpp.

References delimiter, TokenStatBlk::list, TokenStatBlk::num, and str().

Referenced by EmblIndex(), GetAccession(), GetSegment(), InitialEntry(), SprotIndex(), and XMLGetSegment().

TrackNodeType()

DataBlkPtr TrackNodeType	(	const DataBlk &	entry,
		Int2	type
	)

Definition at line 994 of file utilfun.cpp.

References EntryBlk::chain, DataBlk::mpData, DataBlk::mpNext, and DataBlk::mType.

Referenced by AddNIDSeqId(), check_cds(), DefVsHTGKeywords(), FakeEmblBioSources(), get_plasmid(), GetEmblDescr(), GetEmblInstContig(), GetEmblSubBlock(), GetFlatBiomol(), GetGenBankDescr(), GetGenBankInstContig(), GetGenBankSubBlock(), GetSprotDescr(), GetSprotSubBlock(), ImpFeatPub(), LoadFeat(), SeqFeatPub(), SrchNodeSubType(), SrchNodeType(), xGetNodeData(), and xSrchNodeType().

UnwrapAccessionRange()

void UnwrapAccessionRange	(	const CGB_block::TExtra_accessions &	extra_accs,
		CGB_block::TExtra_accessions &	hist
	)

Definition at line 197 of file utilfun.cpp.

References first(), FTAitoa(), i, and last().

Referenced by fta_add_hist().

USING_SCOPE()

USING_SCOPE

(

objects

)

xCheckEstStsGssTpaKeywords()

void xCheckEstStsGssTpaKeywords	(	const list< string >	keywordList,
		bool	tpa_check,
		IndexblkPtr	entry
	)

Definition at line 1335 of file utilfun.cpp.

References Indexblk::assembly, NStr::EqualNocase(), Indexblk::EST, Indexblk::experimental, fta_keywords_check(), Indexblk::GSS, Indexblk::HTC, Indexblk::inferential, Indexblk::is_tpa, Indexblk::specialist_db, and Indexblk::STS.

Referenced by EmblIndex().

xGetNodeData()

string xGetNodeData	(	const DataBlk &	entry,
		int	nodeType
	)

Definition at line 977 of file utilfun.cpp.

References string, tmp, and TrackNodeType().

Referenced by GetSequenceOfKeywords().

xSrchNodeType()

char* xSrchNodeType	(	const DataBlk &	entry,
		Int4	type,
		size_t *	len
	)

Definition at line 963 of file utilfun.cpp.

References len, DataBlk::len, DataBlk::mOffset, and TrackNodeType().

Referenced by fta_get_user_object(), GetDescrEmblBlock(), GetEmblBlockXref(), GetEmblDate(), GetEmblDescr(), GetFlatBiomol(), GetGBBlock(), GetGenBankDescr(), GetReleaseInfo(), and GetSeqData().

xTrackNodeType()

const Section* xTrackNodeType	(	const Entry &	entry,
		int	type
	)

Definition at line 1008 of file utilfun.cpp.

References Entry::mSections, and Section::mType.

Variable Documentation

ParFlat_ENV_kw_array

const char* ParFlat_ENV_kw_array[]

static

Initial value:

= {
    "ENV",
    nullptr
}

Definition at line 167 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_env_keywords().

ParFlat_EST_kw_array

const char* ParFlat_EST_kw_array[]

static

Initial value:

= {
    "EST",
    "EST PROTO((expressed sequence tag)",
    "expressed sequence tag",
    "EST (expressed sequence tag)",
    "EST (expressed sequence tags)",
    "EST(expressed sequence tag)",
    "transcribed sequence fragment",
    nullptr
}

Definition at line 69 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().

ParFlat_FLI_kw_array

const char* ParFlat_FLI_kw_array[]

static

Initial value:

= {
    "FLI_CDNA",
    nullptr
}

Definition at line 101 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().

ParFlat_GSS_kw_array

const char* ParFlat_GSS_kw_array[]

static

Initial value:

= {
    "GSS",
    "GSS (genome survey sequence)",
    "trapped exon",
    nullptr
}

Definition at line 80 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().

ParFlat_HTC_kw_array

const char* ParFlat_HTC_kw_array[]

static

Initial value:

= {
    "HTC",
    nullptr
}

Definition at line 96 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().

ParFlat_MAG_kw_array

const char* ParFlat_MAG_kw_array[]

static

Initial value:

= {
    "Metagenome Assembled Genome",
    "MAG",
    nullptr
}

Definition at line 172 of file utilfun.cpp.

Referenced by fta_remove_mag_keywords().

ParFlat_MGA_kw_array

const char* ParFlat_MGA_kw_array[]

static

Initial value:

= {
    "MGA",
    "CAGE (Cap Analysis Gene Expression)",
    "5'-SAGE",
    nullptr
}

Definition at line 111 of file utilfun.cpp.

Referenced by fta_keywords_check().

ParFlat_MGA_more_kw_array

const char* ParFlat_MGA_more_kw_array[]

static

Initial value:

= {
    "CAGE (Cap Analysis Gene Expression)",
    "5'-SAGE",
    "5'-end tag",
    "unspecified tag",
    "small RNA",
    nullptr
}

Definition at line 118 of file utilfun.cpp.

Referenced by fta_check_mga_keywords().

ParFlat_STS_kw_array

const char* ParFlat_STS_kw_array[]

static

Initial value:

= {
    "STS",
    "STS(sequence tagged site)",
    "STS (sequence tagged site)",
    "STS sequence",
    "sequence tagged site",
    nullptr
}

Definition at line 87 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().

ParFlat_TLS_kw_array

const char* ParFlat_TLS_kw_array[]

static

Initial value:

= {
    "TLS",
    "Targeted Locus Study",
    nullptr
}

Definition at line 139 of file utilfun.cpp.

Referenced by fta_is_tls_keyword(), fta_keywords_check(), fta_remove_tls_keywords(), and fta_tls_keywords_check().

ParFlat_TPA_kw_array

const char* ParFlat_TPA_kw_array[]

static

Initial value:

= {
    "TPA",
    "THIRD PARTY ANNOTATION",
    "THIRD PARTY DATA",
    "TPA:INFERENTIAL",
    "TPA:EXPERIMENTAL",
    "TPA:REASSEMBLY",
    "TPA:ASSEMBLY",
    "TPA:SPECIALIST_DB",
    nullptr
}

Definition at line 148 of file utilfun.cpp.

Referenced by fta_is_tpa_keyword(), fta_keywords_check(), and fta_tpa_keywords_check().

ParFlat_TPA_kw_array_to_remove

const char* ParFlat_TPA_kw_array_to_remove[]

static

Initial value:

= {
    "TPA",
    "THIRD PARTY ANNOTATION",
    "THIRD PARTY DATA",
    nullptr
}

Definition at line 160 of file utilfun.cpp.

Referenced by fta_remove_tpa_keywords().

ParFlat_TSA_kw_array

const char* ParFlat_TSA_kw_array[]

static

Initial value:

= {
    "TSA",
    "Transcriptome Shotgun Assembly",
    nullptr
}

Definition at line 130 of file utilfun.cpp.

Referenced by fta_is_tsa_keyword(), fta_keywords_check(), fta_remove_tsa_keywords(), and fta_tsa_keywords_check().

ParFlat_WGS_kw_array

const char* ParFlat_WGS_kw_array[]

static

Initial value:

= {
    "WGS",
    nullptr
}

Definition at line 106 of file utilfun.cpp.

Referenced by fta_keywords_check(), and fta_remove_keywords().