NCBI C++ ToolKit
|
#include <ncbi_pch.hpp>
#include <set>
#include "ftacpp.hpp"
#include <objects/biblio/Id_pat.hpp>
#include <objects/biblio/Id_pat_.hpp>
#include <objects/seqloc/Textseq_id.hpp>
#include <objects/seqloc/PDB_seq_id.hpp>
#include <objects/general/Object_id.hpp>
#include <objects/seq/Bioseq.hpp>
#include <objects/seqset/Bioseq_set.hpp>
#include <objects/seq/Seq_inst.hpp>
#include <objects/general/Dbtag.hpp>
#include <objects/seq/Seq_data.hpp>
#include <objects/seq/Seq_descr.hpp>
#include <objects/seq/Seq_ext.hpp>
#include <objects/seqset/Seq_entry.hpp>
#include <objects/seq/Seg_ext.hpp>
#include <objects/general/Int_fuzz.hpp>
#include <objects/seqloc/Seq_loc.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/general/Date.hpp>
#include <objects/seq/Pubdesc.hpp>
#include <objects/seq/Numbering.hpp>
#include <objects/pub/Pub_equiv.hpp>
#include <objects/pub/Pub.hpp>
#include <objects/biblio/Cit_gen.hpp>
#include <objects/general/User_object.hpp>
#include <objects/seqfeat/OrgName.hpp>
#include <objects/seq/Seq_annot.hpp>
#include <objects/pub/Pub_set.hpp>
#include <objects/seqfeat/Imp_feat.hpp>
#include <objects/seq/seqport_util.hpp>
#include <util/sequtil/sequtil_convert.hpp>
#include <objects/seq/Delta_ext.hpp>
#include <objects/seq/Seq_literal.hpp>
#include <serial/iterator.hpp>
#include <objects/seqfeat/BioSource.hpp>
#include <objects/seqblock/EMBL_block.hpp>
#include <objects/seq/MolInfo.hpp>
#include "index.h"
#include "genbank.h"
#include "embl.h"
#include "sprot.h"
#include <objtools/flatfile/flatdefn.h>
#include "ftaerr.hpp"
#include "indx_blk.h"
#include "asci_blk.h"
#include "utilfun.h"
#include "fta_xml.h"
#include "add.h"
Go to the source code of this file.
Go to the SVN repository for this file.
Macros | |
#define | THIS_FILE "asci_blk.cpp" |
#define | Seq_descr_pub_same 50 |
Variables | |
const char * | magic_phrases [] |
vector< string > | genbankKeywords |
vector< string > | emblKeywords |
vector< string > | swissProtKeywords |
#define Seq_descr_pub_same 50 |
Definition at line 99 of file asci_blk.cpp.
#define THIS_FILE "asci_blk.cpp" |
Definition at line 97 of file asci_blk.cpp.
void AddNIDSeqId | ( | CBioseq & | bioseq, |
const DataBlk & | entry, | ||
Int2 | type, | ||
Int2 | coldata, | ||
Parser::ESource | source | ||
) |
Definition at line 2722 of file asci_blk.cpp.
References Parser::DDBJ, Parser::EMBL, CRef< C, Locker >::Empty(), DataBlk::mOffset, offset, CBioseq_Base::SetId(), rapidjson::source, StrToSeqId(), and TrackNodeType().
Referenced by EmblAscii(), and GenBankAsciiOrig().
void BuildBioSegHeader | ( | ParserPtr | pp, |
TEntryList & | entries, | ||
const CSeq_loc & | seqloc | ||
) |
Definition at line 2494 of file asci_blk.cpp.
References Parser::curindx, Indexblk::drop, CBioseq_set_Base::eClass_parts, CBioseq_set_Base::eClass_segset, entries, Parser::entrylist, GetBioseq(), GetBioseqSetDescr(), CBioseq_set_Base::SetClass(), CBioseq_set_Base::SetDescr(), CSeq_entry_Base::SetSeq(), CBioseq_set_Base::SetSeq_set(), and CSeq_entry_Base::SetSet().
Referenced by EmblAscii(), GenBankAsciiOrig(), and XMLAscii().
|
static |
Definition at line 377 of file asci_blk.cpp.
References InsertDatablkVal(), DataBlk::len, DataBlk::mOffset, DataBlk::mpData, ParFlat_COL_FEATKEY, ParFlat_FEATBLOCK, SrchTheChar(), and StringEquN().
Referenced by GetEmblSubBlock(), and GetGenBankSubBlock().
void BuildSubBlock | ( | DataBlkPtr | dbp, |
Int2 | subtype, | ||
const char * | subkw | ||
) |
Definition at line 812 of file asci_blk.cpp.
References GetSubNodeType(), InsertDatablkVal(), DataBlk::len, DataBlk::mOffset, and DataBlk::mpData.
Referenced by GetEmblSubBlock(), GetGenBankSubBlock(), and GetSprotSubBlock().
bool check_div | ( | bool | pat_acc, |
bool | pat_ref, | ||
bool | est_kwd, | ||
bool | sts_kwd, | ||
bool | gss_kwd, | ||
bool | if_cds, | ||
string & | div, | ||
CMolInfo::TTech * | tech, | ||
size_t | bases, | ||
Parser::ESource | source, | ||
bool & | drop | ||
) |
Definition at line 2565 of file asci_blk.cpp.
References Parser::EMBL, ERR_DIVISION_ESTHasCDSFeature, ERR_DIVISION_ESTHasSTSKeywords, ERR_DIVISION_GSSHasCDSFeature, ERR_DIVISION_LongESTSequence, ERR_DIVISION_LongGSSSequence, ERR_DIVISION_LongSTSSequence, ERR_DIVISION_MappedtoEST, ERR_DIVISION_MappedtoGSS, ERR_DIVISION_MappedtoPAT, ERR_DIVISION_MappedtoSTS, ERR_DIVISION_MissingESTKeywords, ERR_DIVISION_MissingGSSKeywords, ERR_DIVISION_MissingPatentRef, ERR_DIVISION_MissingSTSKeywords, ERR_DIVISION_NotMappedtoEST, ERR_DIVISION_NotMappedtoGSS, ERR_DIVISION_NotMappedtoSTS, ERR_DIVISION_PATHasCDSFeature, ERR_DIVISION_PATHasESTKeywords, ERR_DIVISION_PATHasGSSKeywords, ERR_DIVISION_PATHasSTSKeywords, ERR_DIVISION_ShouldBePAT, ERR_DIVISION_STSHasCDSFeature, ErrPostEx, ErrPostStr, CMolInfo_Base::eTech_est, CMolInfo_Base::eTech_sts, CMolInfo_Base::eTech_survey, CMolInfo_Base::eTech_tsa, CMolInfo_Base::eTech_unknown, SEV_INFO, SEV_REJECT, SEV_WARNING, rapidjson::source, and StringEqu().
Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().
|
static |
Definition at line 2741 of file asci_blk.cpp.
References Begin(), NStr::CompareNocase(), Parser::curindx, Indexblk::division, Parser::entrylist, ERR_DIVISION_GBBlockDivision, ErrPostStr, CMolInfo_Base::eTech_tsa, CMolInfo_Base::eTech_unknown, CMolInfo_Base::GetTech(), CGB_block_Base::IsSetDiv(), CGB_block_Base::SetDiv(), and SEV_WARNING.
Referenced by EntryCheckDivCode().
|
static |
Definition at line 3024 of file asci_blk.cpp.
References buf, CleanVisString(), CleanVisStringList(), NStr::CompareNocase(), CGB_block_Base::GetDiv(), GetEmblDiv(), CGB_block_Base::GetExtra_accessions(), CGB_block_Base::GetKeywords(), CGB_block_Base::IsSetDate(), CGB_block_Base::IsSetDiv(), CGB_block_Base::IsSetEntry_date(), CGB_block_Base::IsSetExtra_accessions(), CGB_block_Base::IsSetKeywords(), CGB_block_Base::IsSetOrigin(), CGB_block_Base::IsSetSource(), CGB_block_Base::ResetDate(), CGB_block_Base::ResetDiv(), CGB_block_Base::ResetExtra_accessions(), CGB_block_Base::ResetKeywords(), CGB_block_Base::ResetOrigin(), CGB_block_Base::ResetSource(), CGB_block_Base::SetDate(), CGB_block_Base::SetDiv(), CGB_block_Base::SetExtra_accessions(), CGB_block_Base::SetKeywords(), CGB_block_Base::SetOrigin(), and CGB_block_Base::SetSource().
Referenced by fta_EntryCheckGBBlock().
void CheckHTGDivision | ( | const char * | div, |
CMolInfo::TTech | tech | ||
) |
Definition at line 2942 of file asci_blk.cpp.
References ERR_DIVISION_ShouldBeHTG, ERR_DIVISION_ShouldNotBeHTG, ErrPostStr, CMolInfo_Base::eTech_htgs_0, CMolInfo_Base::eTech_htgs_1, CMolInfo_Base::eTech_htgs_2, CMolInfo_Base::eTech_htgs_3, SEV_ERROR, SEV_WARNING, and StringEqu().
Referenced by GetDescrEmblBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().
|
static |
Definition at line 2079 of file asci_blk.cpp.
References CSerialObject::Assign(), CDate::Compare(), CSeqdesc_Base::e_Modif, CDate_Base::e_not_set, CSeqdesc_Base::e_Org, CDate::eCompare_same, entries, CSeq_descr_Base::Get(), GetDescrByChoice(), val, and CDate_Base::Which().
Referenced by SrchSegDescr().
|
static |
Definition at line 1934 of file asci_blk.cpp.
References entries, ERR_SEGMENT_PubMatch, ErrPostStr, CPub_equiv_Base::Get(), CPub_Base::GetGen(), CPubdesc_Base::GetPub(), CCit_gen_Base::GetSerial_number(), CPub_Base::IsGen(), CPub_equiv_Base::IsSet(), CPubdesc_Base::IsSetPub(), CCit_gen_Base::IsSetSerial_number(), not_found, SameCitation_PubEquivMatch_Logic(), CSeq_descr_Base::Set(), and SEV_WARNING.
Referenced by GetSegPub().
|
static |
Definition at line 2007 of file asci_blk.cpp.
References entries, and RemoveDescrByChoice().
Referenced by SrchSegDescr().
|
static |
Definition at line 2963 of file asci_blk.cpp.
References str().
Referenced by CheckGBBlock(), and CleanVisStringList().
Definition at line 3011 of file asci_blk.cpp.
References CleanVisString().
Referenced by CheckGBBlock().
Definition at line 1074 of file asci_blk.cpp.
References Parser::accver, Indexblk::acnum, Parser::curindx, CSeq_id_Base::e_Local, CSeq_id_Base::e_not_set, CSeq_id_Base::e_Tpe, Parser::EMBL, Indexblk::embl_new_ID, Parser::entrylist, Indexblk::is_tpa, NStr::IsBlank(), Indexblk::locusname, MakeUsptoPatSeqId(), Parser::mode, Ref(), Parser::Relaxed, Parser::seqtype, SetEmptyId(), CBioseq_Base::SetId(), SetTextId(), Parser::source, StringEqu(), Parser::USPTO, ValidSeqType(), and Indexblk::vernum.
Referenced by EmblAscii(), GenBankAscii(), GenBankAsciiOrig(), SpPrepareEntry(), and XMLAscii().
void DefVsHTGKeywords | ( | CMolInfo::TTech | tech, |
const DataBlk & | entry, | ||
Int2 | what, | ||
Int2 | ori, | ||
bool | cancelled | ||
) |
Definition at line 2817 of file asci_blk.cpp.
References b, count, ERR_DEFINITION_HTGNotInProgress, ERR_DEFINITION_HTGShouldBeComplete, ERR_SEQUENCE_UnknownBaseHTG3, ErrPostStr, CMolInfo_Base::eTech_htgs_0, CMolInfo_Base::eTech_htgs_1, CMolInfo_Base::eTech_htgs_2, CMolInfo_Base::eTech_htgs_3, fta_StringCpy(), DataBlk::len, magic_phrases, MemFree(), DataBlk::mOffset, r(), SEV_ERROR, SEV_WARNING, StringEquN(), StringNCpy(), StringSave, StringStr, tmp, and TrackNodeType().
Referenced by GetDescrEmblBlock(), and GetGBBlock().
Definition at line 3490 of file asci_blk.cpp.
References delta(), CSeq_inst_Base::GetExt(), CBioseq_Base::GetInst(), CSeq_ext_Base::IsDelta(), CSeq_inst_Base::IsSetExt(), CSeqportUtil::Pack(), and CBioseq_Base::SetInst().
Referenced by PackEntries().
Definition at line 3154 of file asci_blk.cpp.
References fta_if_pubs_sorted(), CPubdesc_Base::GetPub(), CSeqdesc_Base::GetPub(), CObject_id_Base::GetStr(), CUser_object_Base::GetType(), CSeqdesc_Base::GetUser(), CSeqdesc_Base::IsPub(), CUser_object_Base::IsSetType(), CObject_id_Base::IsStr(), CSeqdesc_Base::IsUser(), util::strcmp(), and CSeqdesc_Base::Which().
Referenced by fta_sort_descr().
void EntryCheckDivCode | ( | TEntryList & | seq_entries, |
ParserPtr | pp | ||
) |
Definition at line 2806 of file asci_blk.cpp.
References CheckDivCode(), and SeqEntryCheckTaxonDiv().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
|
static |
Definition at line 410 of file asci_blk.cpp.
References ERR_REFERENCE_MultipleIdentifiers, ErrPostStr, DataBlk::len, DataBlk::mOffset, SEV_ERROR, StringChr, StringEquN(), and StringLen().
Referenced by GetEmblSubBlock(), and GetGenBankSubBlock().
bool fta_EntryCheckGBBlock | ( | TEntryList & | seq_entries | ) |
Definition at line 3109 of file asci_blk.cpp.
References Begin(), and CheckGBBlock().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
void fta_fix_orgref_div | ( | const CBioseq::TAnnot & | annots, |
COrg_ref * | org_ref, | ||
CGB_block & | gbb | ||
) |
Definition at line 3266 of file asci_blk.cpp.
References count, fta_orgref_has_taxid(), COrg_ref_Base::GetDb(), CGB_block_Base::GetDiv(), CBioSource_Base::GetOrg(), COrg_ref_Base::GetOrgname(), CGB_block_Base::IsSetDiv(), COrgName_Base::IsSetDiv(), CBioSource_Base::IsSetOrg(), COrg_ref_Base::IsSetOrgname(), CGB_block_Base::ResetDiv(), and COrg_ref_Base::SetOrgname().
Referenced by GetEmblDescr(), GetGenBankDescr(), and XMLGetDescr().
|
static |
Definition at line 1268 of file asci_blk.cpp.
References fta_if_wgs_acc(), and next().
Referenced by GetExtraAccession().
|
static |
Definition at line 1478 of file asci_blk.cpp.
References buf, NStr::CompareNocase(), fta_is_tpa_keyword(), ncbi::grid::netcache::search::fields::key, StringEquNI(), and swap().
Referenced by GetSequenceOfKeywords().
|
static |
Definition at line 3145 of file asci_blk.cpp.
References GetSerialNumFromPubEquiv().
Referenced by descr_cmp().
bool fta_orgref_has_taxid | ( | const COrg_ref::TDb & | dbtags | ) |
Definition at line 3254 of file asci_blk.cpp.
References tag.
Referenced by fta_fix_orgref_div(), and GetEmblDescr().
void fta_set_strandedness | ( | TEntryList & | seq_entries | ) |
Definition at line 3337 of file asci_blk.cpp.
References Begin(), CSeq_inst_Base::eMol_aa, CSeq_inst_Base::eMol_dna, CSeq_inst_Base::eMol_rna, CSeq_inst_Base::eStrand_ds, and CSeq_inst_Base::eStrand_ss.
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
void fta_sort_descr | ( | TEntryList & | seq_entries | ) |
Definition at line 3181 of file asci_blk.cpp.
References Begin(), and descr_cmp().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
void fta_sort_seqfeat_cit | ( | TEntryList & | seq_entries | ) |
Definition at line 3238 of file asci_blk.cpp.
References Begin(), and sort_feat_cit().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
|
static |
Definition at line 2409 of file asci_blk.cpp.
References Indexblk::acnum, CSeq_loc::Assign(), CSeq_loc::begin(), Indexblk::blocusname, Parser::curindx, CInt_fuzz_Base::eLim_gt, CSeq_loc::end(), entries, Parser::entrylist, CSeq_inst_Base::eRepr_seg, GetBioseqSetDescrTitle(), CBioseq_Base::GetDescr(), CSeq_entry_Base::GetSeq(), Indexblk::is_tpa, MakeSegSetSeqId(), Parser::seg_acc, Parser::seqtype, CBioseq_Base::SetDescr(), CSeq_inst_Base::SetExt(), CSeq_inst_Base::SetFuzz(), CBioseq_Base::SetId(), CBioseq_Base::SetInst(), CSeq_inst_Base::SetLength(), CSeq_inst_Base::SetMol(), CSeq_loc::SetNull(), CSeq_inst_Base::SetRepr(), CSeqdesc_Base::SetTitle(), SrchSegLength(), and SrchSegSeqMol().
Referenced by BuildBioSegHeader().
|
static |
Definition at line 2330 of file asci_blk.cpp.
References entries, GetSegSetDblink(), and SrchSegDescr().
Referenced by BuildBioSegHeader().
|
static |
Definition at line 2137 of file asci_blk.cpp.
References CleanTailNoneAlphaCharInString(), and CSeq_descr_Base::Get().
Referenced by GetBioseq(), and SrchSegDescr().
|
static |
Definition at line 1835 of file asci_blk.cpp.
References CSeq_descr_Base::Get().
Referenced by CheckSegDescrChoice().
Definition at line 1159 of file asci_blk.cpp.
References fta_StringCpy(), len, MemCpy(), MemFree(), offset, r(), ncbi::grid::netcache::search::fields::size, SrchTheChar(), SrchTheStr(), str(), StringChr, StringEquN(), StringNew(), and StringStr.
Referenced by GetEmblDescr(), and GetGenBankDescr().
const CSeq_descr& GetDescrPointer | ( | const CSeq_entry & | entry | ) |
Definition at line 2954 of file asci_blk.cpp.
References CBioseq_Base::GetDescr(), CBioseq_set_Base::GetDescr(), CSeq_entry_Base::GetSeq(), CSeq_entry_Base::GetSet(), and CSeq_entry_Base::IsSeq().
Referenced by GetGcode(), and GetTopBiosource().
unique_ptr<unsigned char[]> GetDNAConv | ( | void | ) |
Definition at line 1786 of file asci_blk.cpp.
References eSeq_code_type_iupacna, CSeqportUtil::GetCode(), CSeqportUtil::GetCodeIndexFromTo(), i, int, MemSet(), compile_time_bits::range(), and tolower().
Referenced by EmblAscii(), GenBankAscii(), GenBankAsciiOrig(), get_bioseq(), and XMLAscii().
char* GetEmblBlock | ( | DataBlkPtr * | chain, |
char * | ptr, | ||
short * | retkw, | ||
Parser::EFormat | format, | ||
char * | eptr | ||
) |
Definition at line 545 of file asci_blk.cpp.
References emblKeywords, format, InsertDatablkVal(), len, offset, ParFlat_END, ParFlat_UNKW, Parser::SPROT, SrchKeyword(), StringEquN(), and swissProtKeywords.
Referenced by CheckOutsideEntry(), EmblAscii(), get_bioseq(), and SpPrepareEntry().
|
static |
Definition at line 684 of file asci_blk.cpp.
References Parser::EMBL, GetSubNodeType(), DataBlk::len, DataBlk::mOffset, DataBlk::mType, Parser::NCBI, ParFlat_REF_BTW, ParFlat_REF_END, ParFlat_REF_NO_TARGET, ParFlat_REF_SITES, rapidjson::source, SrchTheStr(), and str().
Referenced by GetEmblSubBlock().
void GetEmblSubBlock | ( | size_t | bases, |
Parser::ESource | source, | ||
const DataBlk & | entry | ||
) |
Definition at line 740 of file asci_blk.cpp.
References BuildFeatureBlock(), BuildSubBlock(), EntryBlk::chain, ERR_FEATURE_NoFeatData, ErrPostStr, fta_check_mult_ids(), GetEmblRefType(), GetLenSubNode(), DataBlk::mpData, DataBlk::mpNext, DataBlk::mType, ParFlat_FH, ParFlat_OC, ParFlat_OG, ParFlat_OS, ParFlat_RA, ParFlat_RC, ParFlat_RG, ParFlat_RL, ParFlat_RN, ParFlat_RP, ParFlat_RT, ParFlat_RX, SEV_WARNING, rapidjson::source, TrackNodeType(), and TrimEmblFeatBlk().
Referenced by EmblAscii().
void GetExtraAccession | ( | IndexblkPtr | ibp, |
bool | allow_uwsec, | ||
Parser::ESource | source, | ||
TAccessionList & | accessions | ||
) |
Definition at line 1317 of file asci_blk.cpp.
References a, Indexblk::acnum, NStr::CommonSuffixSize(), Parser::DDBJ, DelNonDigitTail(), Indexblk::drop, CSeq_id_Base::e_Embl, Parser::EMBL, ERR_ACCESSION_ScfldHasWGSContigSec, ERR_ACCESSION_UnusualWGS_Secondary, ERR_ACCESSION_WGSMasterAsSecondary, ERR_ACCESSION_WGSWithNonWGS_Sec, ErrPostEx, ErrPostStr, fta_fix_secondaries(), fta_if_wgs_acc(), GetNucAccOwner(), i, Indexblk::is_contig, MemFree(), Parser::NCBI, Parser::Refseq, Indexblk::secaccs, SEV_REJECT, SEV_WARNING, rapidjson::source, StringEquN(), StringLen(), StringSave, text(), and Indexblk::wgssec.
Referenced by GetDescrEmblBlock(), GetDescrSPBlock(), GetEmblGBBlock(), GetGBBlock(), XMLGetEMBLBlock(), and XMLGetGBBlock().
|
static |
Definition at line 1858 of file asci_blk.cpp.
References CSeq_descr_Base::Set(), and CBioseq_Base::SetDescr().
Referenced by SrchSegDescr().
char* GetGenBankBlock | ( | DataBlkPtr * | chain, |
char * | ptr, | ||
Int2 * | retkw, | ||
char * | eptr | ||
) |
Definition at line 284 of file asci_blk.cpp.
References genbankKeywords, InsertDatablkVal(), len, offset, ParFlat_UNKW, SrchKeyword(), and StringEquN().
Referenced by GenBankAsciiOrig(), and get_bioseq().
|
static |
Definition at line 337 of file asci_blk.cpp.
References NStr::Find(), DataBlk::len, DataBlk::mOffset, DataBlk::mType, NPOS, ParFlat_REF_BTW, ParFlat_REF_END, ParFlat_REF_NO_TARGET, ParFlat_REF_SITES, and str().
Referenced by GetGenBankSubBlock().
Definition at line 454 of file asci_blk.cpp.
References BuildFeatureBlock(), BuildSubBlock(), fta_check_mult_ids(), GetGenBankRefType(), GetLenSubNode(), DataBlk::mpNext, DataBlk::mType, ParFlat_AUTHORS, ParFlat_CONSRTM, ParFlat_FEATURES, ParFlat_JOURNAL, ParFlat_MEDLINE, ParFlat_ORGANISM, ParFlat_PUBMED, ParFlat_REFERENCE, ParFlat_REMARK, ParFlat_SOURCE, ParFlat_STANDARD, ParFlat_TITLE, and TrackNodeType().
Referenced by GenBankAsciiOrig().
void GetLenSubNode | ( | DataBlkPtr | dbp | ) |
Definition at line 835 of file asci_blk.cpp.
References done, ERR_FORMAT_LineTypeOrder, ErrPostEx, isdigit(), l(), len, DataBlk::len, DataBlk::mOffset, DataBlk::mpData, DataBlk::mpNext, n, offset, and SEV_WARNING.
Referenced by GetEmblSubBlock(), GetGenBankSubBlock(), and GetSprotSubBlock().
Definition at line 2343 of file asci_blk.cpp.
Referenced by SrchSegSeqMol().
unique_ptr<unsigned char[]> GetProteinConv | ( | void | ) |
Definition at line 1814 of file asci_blk.cpp.
References eSeq_code_type_iupacaa, CSeqportUtil::GetCode(), CSeqportUtil::GetCodeIndexFromTo(), i, int, MemSet(), and compile_time_bits::range().
Referenced by GenBankAscii(), GenBankAsciiOrig(), get_bioseq(), SprotAscii(), and XMLAscii().
|
static |
Definition at line 2050 of file asci_blk.cpp.
References CheckSegPub(), CleanUpSeqDescrPub(), entries, CSeq_descr_Base::Set(), CBioseq_Base::SetDescr(), and CPubdesc_Base::SetPub().
Referenced by SrchSegDescr().
|
static |
Definition at line 2210 of file asci_blk.cpp.
References CUser_object_Base::CanGetType(), CRef< C, Locker >::Empty(), entries, CSerialObject::Equals(), ERR_SEGMENT_DBLinkMissingOrNonUnique, ERR_SEGMENT_GPIDMissingOrNonUnique, ErrPostStr, CSeq_descr_Base::Get(), CObject_id_Base::GetStr(), CUser_object_Base::GetType(), CRef< C, Locker >::NotEmpty(), CRef< C, Locker >::Reset(), CSeq_descr_Base::Set(), and SEV_REJECT.
Referenced by GetBioseqSetDescr().
bool GetSeqData | ( | ParserPtr | pp, |
const DataBlk & | entry, | ||
CBioseq & | bioseq, | ||
Int4 | nodetype, | ||
unsigned char * | seqconv, | ||
Uint1 | seq_data_type | ||
) |
Definition at line 1674 of file asci_blk.cpp.
References Indexblk::bases, buf, Parser::curindx, Parser::DDBJ, Indexblk::drop, CSeq_data_Base::e_Iupacaa, Parser::EMBL, Parser::entrylist, ERR_SEQUENCE_AllNs, ERR_SEQUENCE_SeqLenNotEq, ERR_SEQUENCE_TooShort, ERR_SEQUENCE_TooShortIsPatent, ErrPostEx, ErrPostStr, Parser::format, Parser::GenBank, CBioseq::GetLength(), INSDSEQ_SEQUENCE, Indexblk::is_contig, Indexblk::is_mga, Indexblk::is_pat, Indexblk::is_prot, isalpha(), len, MemFree(), DataBlk::mOffset, ScanSequence(), CBioseq_Base::SetInst(), SEV_INFO, SEV_REJECT, SEV_WARNING, Parser::source, str(), StringLen(), StringSave, Parser::USPTO, Indexblk::xip, Parser::XML, XMLFindTagValue(), and xSrchNodeType().
Referenced by get_bioseq(), GetEmblInst(), GetGenBankInst(), GetSPInst(), s_DoNWalign(), s_GetCdsSequence(), COligoSpecificityCheck::x_NW_alignment(), and XMLGetInst().
Definition at line 2469 of file asci_blk.cpp.
References Parser::accver, Indexblk::acnum, CSeq_loc::Add(), Parser::curindx, Parser::entrylist, MakeAccSeqId(), Parser::seqtype, CSeq_loc::SetWhole(), and Indexblk::vernum.
Referenced by EmblAscii(), GenBankAsciiOrig(), and XMLAscii().
void GetSequenceOfKeywords | ( | const DataBlk & | entry, |
int | type, | ||
Uint2 | col_data, | ||
TKeywordList & | keywords | ||
) |
Definition at line 1547 of file asci_blk.cpp.
References fta_fix_tpa_keywords(), GetBlkDataReplaceNewLine(), last(), ParFlatSP_KW, NStr::Split(), StripECO(), NStr::TrimSuffixInPlace(), NStr::TruncateSpacesInPlace(), xFixEMBLKeywords(), and xGetNodeData().
Referenced by CollectGapFeats(), GetDescrSPBlock(), and GetGBBlock().
|
static |
Definition at line 3129 of file asci_blk.cpp.
References CPub_equiv_Base::Get().
Referenced by fta_if_pubs_sorted().
Definition at line 648 of file asci_blk.cpp.
References SrchTheChar(), StringEquN(), and StringLen().
Referenced by BuildSubBlock(), and GetEmblRefType().
|
static |
Definition at line 2781 of file asci_blk.cpp.
References GetDescrPointer().
Referenced by SeqEntryCheckTaxonDiv().
|
static |
Definition at line 231 of file asci_blk.cpp.
Referenced by BuildFeatureBlock(), BuildSubBlock(), GetEmblBlock(), and GetGenBankBlock().
Definition at line 3373 of file asci_blk.cpp.
References CCit_gen_Base::IsSetAuthors(), CCit_gen_Base::IsSetCit(), CCit_gen_Base::IsSetDate(), CCit_gen_Base::IsSetIssue(), CCit_gen_Base::IsSetJournal(), CCit_gen_Base::IsSetMuid(), CCit_gen_Base::IsSetPages(), CCit_gen_Base::IsSetPmid(), CCit_gen_Base::IsSetSerial_number(), CCit_gen_Base::IsSetTitle(), and CCit_gen_Base::IsSetVolume().
Referenced by RemoveSerials().
Definition at line 2536 of file asci_blk.cpp.
References CSeq_id_Base::e_Patent, CTextseq_id_Base::GetName(), CTextseq_id_Base::IsSetAccession(), CTextseq_id_Base::IsSetName(), and StringEquN().
Referenced by FindCd(), FindFeatSeqLoc(), and FindGene().
Definition at line 960 of file asci_blk.cpp.
References CSeq_id_Base::e_not_set, SetTextId(), and ValidSeqType().
Referenced by GetDRlineDataSP(), GetSeqExt(), GetSeqLocIntSP(), GetSPSeqLoc(), and LoadFeat().
CRef<CSeq_id> MakeLocusSeqId | ( | const char * | locus, |
CSeq_id::E_Choice | seqtype | ||
) |
Definition at line 990 of file asci_blk.cpp.
References CRef< C, Locker >::Reset(), and SetTextId().
Referenced by GetDRlineDataSP().
|
static |
Definition at line 1008 of file asci_blk.cpp.
References CSeq_id_Base::e_not_set, CRef< C, Locker >::Reset(), SetTextId(), and ValidSeqType().
Referenced by GetBioseq().
CRef<CPatent_seq_id> MakeUsptoPatSeqId | ( | const char * | acc | ) |
Definition at line 884 of file asci_blk.cpp.
References StringChr.
Referenced by CreateEntryBioseq(), fta_fix_seq_id(), GetProtRefSeqId(), and LoadFeat().
void PackEntries | ( | TEntryList & | seq_entries | ) |
Definition at line 3502 of file asci_blk.cpp.
References Begin(), DeltaBioseqPack(), CSeq_inst_Base::eRepr_const, CSeq_inst_Base::eRepr_delta, CSeq_inst_Base::eRepr_raw, and RawBioseqPack().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), and XMLAscii().
|
static |
Definition at line 3435 of file asci_blk.cpp.
References CSeqConvert::Convert(), CSeq_data_Base::e_Iupacaa, CSeqUtil::e_Iupacaa, CSeq_data_Base::e_Ncbi8aa, CSeqUtil::e_Ncbi8aa, CSeqUtil::e_Ncbieaa, CSeq_data_Base::e_Ncbistdaa, CSeqUtil::e_Ncbistdaa, CSeqUtil::e_not_set, CAliasBase< TPrim >::Get(), CSeq_data_Base::GetIupacaa(), CSeq_data_Base::GetNcbi8aa(), CSeq_data_Base::GetNcbistdaa(), CAliasBase< TPrim >::Set(), and CSeq_data_Base::SetNcbieaa().
Referenced by RawBioseqPack().
Definition at line 3197 of file asci_blk.cpp.
References CCit_gen_Base::GetCit(), CPub_Base::GetGen(), CPub_Base::GetMuid(), CPub_Base::IsGen(), CPub_Base::IsMuid(), CCit_gen_Base::IsSetCit(), and CPub_Base::Which().
Referenced by sort_feat_cit().
Definition at line 3478 of file asci_blk.cpp.
References CBioseq_Base::GetInst(), CSeq_inst_Base::GetSeq_data(), CSeq_data_Base::IsGap(), CSeq_inst::IsNa(), CSeq_inst_Base::IsSetMol(), CSeq_inst_Base::IsSetSeq_data(), CSeqportUtil::Pack(), PackSeqData(), CBioseq_Base::SetInst(), and CSeq_data_Base::Which().
Referenced by PackEntries().
|
static |
Definition at line 1988 of file asci_blk.cpp.
References CSeq_descr_Base::Set().
Referenced by CleanUpSeqDescrChoice().
Definition at line 3385 of file asci_blk.cpp.
References IsCitEmpty().
Referenced by StripSerialNumbers().
|
static |
Definition at line 1879 of file asci_blk.cpp.
References a, b, CPub_Base::e_Gen, CCit_gen_Base::GetSerial_number(), CCit_gen_Base::IsSetAuthors(), CCit_gen_Base::IsSetCit(), CCit_gen_Base::IsSetDate(), CCit_gen_Base::IsSetIssue(), CCit_gen_Base::IsSetJournal(), CCit_gen_Base::IsSetMuid(), CCit_gen_Base::IsSetPages(), CCit_gen_Base::IsSetSerial_number(), CCit_gen_Base::IsSetTitle(), and CCit_gen_Base::IsSetVolume().
Referenced by CheckSegPub().
Int4 ScanSequence | ( | bool | warn, |
char ** | seqptr, | ||
std::vector< char > & | bsp, | ||
unsigned char * | conv, | ||
Char | replacechar, | ||
int * | numns | ||
) |
Definition at line 1618 of file asci_blk.cpp.
References buf, conv, copy(), count, ERR_SEQUENCE_BadResidue, ErrPostEx, int, isalpha(), and SEV_ERROR.
Referenced by GetSeqData().
|
static |
Definition at line 2793 of file asci_blk.cpp.
References CBioSource_Base::GetOrg(), COrg_ref_Base::GetOrgname(), GetTopBiosource(), COrgName_Base::IsSetDiv(), CBioSource_Base::IsSetOrg(), and COrg_ref_Base::IsSetOrgname().
Referenced by EntryCheckDivCode().
Definition at line 1062 of file asci_blk.cpp.
References CBioseq_Base::SetId(), CObject_id::SetId8(), and CSeq_id_Base::SetLocal().
Referenced by CreateEntryBioseq().
void ShrinkSpaces | ( | char * | line | ) |
Definition at line 118 of file asci_blk.cpp.
References fta_StringCpy().
Referenced by DescrRefs(), fta_fix_affil(), fta_get_trna_from_comment(), fta_get_trna_from_product(), fta_remark_is_er(), fta_strip_pub_comment(), fta_vnp_structured_comment(), GetSPDescrTitle(), GetViralHostsFrom_OH(), ParseFeatureBlock(), RemoveSourceFeatSpaces(), SPFeatProtRef(), StripCDSComment(), and XMLParseFeatureBlock().
void ShrinkSpaces | ( | string & | line | ) |
Definition at line 166 of file asci_blk.cpp.
References i.
|
static |
Definition at line 3215 of file asci_blk.cpp.
References pub_cmp(), and swap().
Referenced by fta_sort_seqfeat_cit().
Definition at line 1039 of file asci_blk.cpp.
References len, DataBlk::len, DataBlk::mOffset, DataBlk::mpData, DataBlk::mpNext, DataBlk::mType, and TrackNodeType().
Referenced by FakeGenBankBioSources().
|
static |
Definition at line 2181 of file asci_blk.cpp.
References CheckSegDescrChoice(), CleanUpSeqDescrChoice(), CSeqdesc_Base::e_Modif, CSeqdesc_Base::e_Org, CSeqdesc_Base::e_Update_date, entries, GetBioseqSetDescrTitle(), CBioseq_Base::GetDescr(), GetFirstSegDescrChoice(), GetSegPub(), CSeq_descr_Base::Set(), CSeq_entry_Base::SetSeq(), and CSeqdesc_Base::SetTitle().
Referenced by GetBioseqSetDescr().
|
static |
Definition at line 2390 of file asci_blk.cpp.
References entries, and CBioseq::GetLength().
Referenced by GetBioseq().
|
static |
Definition at line 2365 of file asci_blk.cpp.
References CSeq_inst_Base::eMol_na, entries, ERR_SEGMENT_DiffMolType, ErrPostEx, CBioseq_Base::GetInst(), CSeq_inst_Base::GetMol(), GetMoleculeClassString(), and SEV_WARNING.
Referenced by GetBioseq().
void StripSerialNumbers | ( | TEntryList & | seq_entries | ) |
Definition at line 3402 of file asci_blk.cpp.
References Begin(), CImp_feat_Base::GetKey(), CImp_feat_Base::IsSetKey(), RemoveSerials(), and SwissProtIDPresent().
Referenced by GenBankAsciiOrig(), OutputEmblAsn(), SpPrepareEntry(), and XMLAscii().
Definition at line 2690 of file asci_blk.cpp.
References CDbtag_Base::SetDb(), CDbtag_Base::SetTag(), and tag.
Referenced by AddNIDSeqId(), and QualsToSeqID().
|
static |
Definition at line 3356 of file asci_blk.cpp.
References Begin().
Referenced by StripSerialNumbers().
|
static |
Definition at line 604 of file asci_blk.cpp.
References DataBlk::len, DataBlk::mOffset, and SrchTheChar().
Referenced by GetEmblSubBlock().
USING_SCOPE | ( | objects | ) |
Definition at line 918 of file asci_blk.cpp.
References CSeq_id_Base::e_Ddbj, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_not_set, 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, CSeq_id::GetAccType(), CSeq_id::IdentifyAccession(), and type.
Referenced by CreateEntryBioseq(), MakeAccSeqId(), and MakeSegSetSeqId().
void xFixEMBLKeywords | ( | string & | keywordData | ) |
Definition at line 1508 of file asci_blk.cpp.
References NStr::FindNoCase(), and i.
Referenced by GetSequenceOfKeywords().
void xGetGenBankBlocks | ( | Entry & | entry | ) |
Definition at line 256 of file asci_blk.cpp.
References genbankKeywords, Entry::mBaseData, Entry::mSections, ParFlat_LOCUS, ParFlat_UNKW, NStr::Split(), SrchKeyword(), and NStr::StartsWith().
Referenced by GenBankAscii().
void xGetGenBankSubBlocks | ( | Entry & | entry, |
size_t | bases | ||
) |
Definition at line 493 of file asci_blk.cpp.
References Entry::mSections, ParFlat_AUTHORS, ParFlat_CONSRTM, ParFlat_FEATURES, ParFlat_JOURNAL, ParFlat_MEDLINE, ParFlat_ORGANISM, ParFlat_PUBMED, ParFlat_REFERENCE, ParFlat_REMARK, ParFlat_SOURCE, ParFlat_STANDARD, and ParFlat_TITLE.
Referenced by GenBankAscii().
bool XMLCheckCDS | ( | const char * | entry, |
XmlIndexPtr | xip | ||
) |
Definition at line 3306 of file asci_blk.cpp.
References XmlIndex::end, INSDFEATURE_KEY, INSDSEQ_FEATURE_TABLE, XmlIndex::next, XmlIndex::start, StringEquN(), XmlIndex::subtags, and XmlIndex::tag.
Referenced by XMLGetEMBLBlock(), and XMLGetGBBlock().
void XMLDefVsHTGKeywords | ( | CMolInfo::TTech | tech, |
const char * | entry, | ||
XmlIndexPtr | xip, | ||
bool | cancelled | ||
) |
Definition at line 2885 of file asci_blk.cpp.
References b, count, ERR_DEFINITION_HTGNotInProgress, ERR_DEFINITION_HTGShouldBeComplete, ERR_SEQUENCE_UnknownBaseHTG3, ErrPostStr, CMolInfo_Base::eTech_htgs_0, CMolInfo_Base::eTech_htgs_1, CMolInfo_Base::eTech_htgs_2, CMolInfo_Base::eTech_htgs_3, INSDSEQ_DEFINITION, INSDSEQ_SEQUENCE, magic_phrases, MemFree(), r(), SEV_ERROR, SEV_WARNING, StringSave, StringStr, tmp, and XMLFindTagValue().
Referenced by XMLGetEMBLBlock(), and XMLGetGBBlock().
|
extern |
Definition at line 56 of file em_index.cpp.
Referenced by EmblIndex(), and GetEmblBlock().
|
extern |
Definition at line 56 of file gb_index.cpp.
Referenced by GenBankIndex(), GetGenBankBlock(), and xGetGenBankBlocks().
const char* magic_phrases[] |
Definition at line 104 of file asci_blk.cpp.
Referenced by DefVsHTGKeywords(), and XMLDefVsHTGKeywords().
|
extern |
Definition at line 54 of file sp_index.cpp.
Referenced by GetEmblBlock(), and SprotIndex().