seq_id.cpp

ncbi源码

    case CSeq_id::e_Gibbmt:
        if ( (the_id =NStr::StringToNumeric (acc)) >= 0 ) {
            SetGibbmt(the_id);
        } else {
             s_InitThrow("Unexpected non-numeric accession.",
                         string(s_TextId[the_type]), acc_in, name_in,
                         NStr::IntToString(version), release_in);
        }
        break;
    case CSeq_id::e_Giim:
        {
            CGiimport_id &  giim = SetGiim();
            if ( (the_id =NStr::StringToNumeric (acc)) >= 0 ) {
                giim.SetId(the_id);
            } else {
                s_InitThrow("Unexpected non-numeric accession.",
                            string(s_TextId[the_type]), acc_in, name_in,
                            NStr::IntToString(version), release_in);
            }
            break;
        }
    case CSeq_id::e_Genbank:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release);
            SetGenbank(*text );
            break;
        }
    case CSeq_id::e_Embl:
        {
            CTextseq_id* text
               = new CTextseq_id(acc, name, version, release);
            SetEmbl(*text);
            break;
        }
    case CSeq_id::e_Pir:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release, false);
            SetPir(*text);
            break;
        }
    case CSeq_id::e_Swissprot:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release, false);
            SetSwissprot(*text);
            break;
        }
    case CSeq_id::e_Tpg:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release);
            SetTpg(*text);
            break;
        }
    case CSeq_id::e_Tpe:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release);
            SetTpe(*text);
            break;
        }
    case CSeq_id::e_Tpd:
        {
            CTextseq_id* text
                = new CTextseq_id(acc, name, version, release);
            SetTpd(*text);
            break;
        }
    case CSeq_id::e_Patent:
        {
            CPatent_seq_id&  pat    = SetPatent();
            CId_pat&         id_pat = pat.SetCit();
            CId_pat::C_Id&   id_pat_id = id_pat.SetId();
            id_pat.SetCountry(acc);

            const char      app_str[] = "App=";
            const SIZE_TYPE app_str_len = sizeof(app_str) - 1;
            if (name.substr(0, app_str_len) == app_str) {
                id_pat_id.SetApp_number(name.substr(app_str_len));
            } else {
                id_pat_id.SetNumber(name);
            }
            pat.SetSeqid(version);
            break;
        }
    case CSeq_id::e_Other: // RefSeq, allow dot version
        {
            CTextseq_id* text
                = new CTextseq_id(acc,name,version,release);
            SetOther(*text);
            break;
        }
    case CSeq_id::e_General:
        {
            CDbtag& dbt = SetGeneral();
            dbt.SetDb(acc);
            CObject_id& oid = dbt.SetTag();
            the_id = NStr::StringToNumeric(name);
            if (the_id >= 0  &&  (name.size() == 1 || name[0] != '0')) {
                oid.SetId(the_id);
            }else{
                oid.SetStr(name);
            }
            break;
        }
    case CSeq_id::e_Gi:
        the_id = NStr::StringToNumeric(acc);
        if (the_id >= 0 ) {
            SetGi(the_id);
        } else {
             s_InitThrow("Unexpected non-numeric accession.",
                         string(s_TextId[the_type]), acc_in, name_in,
                         NStr::IntToString(version), release_in);
        }
        break;
    case CSeq_id::e_Ddbj:
        {
            CTextseq_id* text
                = new CTextseq_id(acc,name,version,release);
            SetDdbj(*text);
            break;
        }
    case CSeq_id::e_Prf:
        {
            CTextseq_id* text
                = new CTextseq_id(acc,name,version,release,false);
            SetPrf(*text);
            break;
        }
    case CSeq_id::e_Pdb:
        {
            CPDB_seq_id& pdb     = SetPdb();
            CPDB_mol_id& pdb_mol = pdb.SetMol();
            pdb_mol.Set(acc);

            if (name.empty()) {
                pdb.SetChain(' ');
            } else if (name.size() == 1) {
                pdb.SetChain(static_cast<unsigned char> (name[0]));
            } else if ( name.compare("VB") == 0) {
                pdb.SetChain('|');
            } else if (name.size() == 2  &&  name[0] ==  name[1]) {
                pdb.SetChain( Locase(static_cast<unsigned char> (name[0])) );
            } else {
                s_InitThrow("Unexpected PDB chain id.",
                            string(s_TextId[the_type]), acc_in, name_in,
                            NStr::IntToString(version), release_in);
            }
            break;
        }
    default:
        THROW1_TRACE(invalid_argument, "Specified Seq-id type not supported");
    }
}


bool CSeq_id::Equals(const CSerialObject& object, ESerialRecursionMode how) const
{
    if ( typeid(object) != typeid(*this) ) {
        ERR_POST(Fatal <<
            "CSeq_id::Assign() -- Assignment of incompatible types: " <<
            typeid(*this).name() << " = " << typeid(object).name());
    }
    return CSerialObject::Equals(object, how);
}


END_objects_SCOPE // namespace ncbi::objects::
END_NCBI_SCOPE

/*
 * ===========================================================================
 *
 * $Log: Seq_id.cpp,v $
 * Revision 1000.4  2004/06/01 19:34:28  gouriano
 * PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R6.81
 *
 * Revision 6.81  2004/06/01 15:26:07  johnson
 * Make CompareOrdered a true model of strict weak ordering
 *
 * Revision 6.80  2004/05/28 20:09:44  johnson
 * Added Compare for seq-id type General (CDbtag)
 *
 * Revision 6.79  2004/05/21 14:41:46  dicuccio
 * Moved database tag for general IDs to content part of label
 *
 * Revision 6.78  2004/05/19 17:26:25  gorelenk
 * Added include of PCH - ncbi_pch.hpp
 *
 * Revision 6.77  2004/05/16 16:57:44  dicuccio
 * Removed insertion of db type in general seq-id labels of type content - led to
 * duplicates of type
 *
 * Revision 6.76  2004/05/14 14:34:02  dicuccio
 * Include database name in label content if the db-tag is just an integer
 *
 * Revision 6.75  2004/03/25 15:59:06  gouriano
 * Added possibility to copy and compare serial object non-recursively
 *
 * Revision 6.74  2004/03/22 16:24:19  ucko
 * CN is now specifically assigned to GenBank ESTs.
 *
 * Revision 6.73  2004/01/22 21:03:58  dicuccio
 * Separated functionality of enums in GetLabel() into discrete mode and flags
 *
 * Revision 6.72  2004/01/22 18:45:46  dicuccio
 * Added new API: CSeq_id::GetLabel().  Rewired GetSeqIdString() to feed into
 * GetLabel().  Rewired GetStringDescr() to feed into GetLabel() directly instead
 * of feeding through GetSeqIdString().
 *
 * Revision 6.71  2004/01/21 22:55:47  ucko
 * GetSeqIdString: drop the database name from general IDs for
 * compatibility with code that can't handle its presence.
 *
 * Revision 6.70  2004/01/21 18:04:20  dicuccio
 * Added ctor to create a seq-id from a given dbtag, performing conversion to
 * specific seq-id types where possible
 *
 * Revision 6.69  2004/01/20 16:59:38  ucko
 * CSeq_id::IdentifyAccession: identify IDs consisting solely of digits as GIs.
 *
 * Revision 6.68  2004/01/16 17:39:17  vasilche
 * Fixed parsing 'gnl|xxx|999' format - integer tag detection.
 *
 * Revision 6.67  2004/01/16 15:58:19  ucko
 * CM now specifically assigned to eAcc_gb_con.
 *
 * Revision 6.66  2003/12/18 18:55:59  ucko
 * CSeq_id::IdentifyAccession: CQ is EMBL patents, CR is EMBL genomes,
 * CS-CU are reserved for future EMBL nucleotide use.
 *
 * Revision 6.65  2003/12/16 16:00:16  ucko
 * CSeq_id::IdentifyAccession: CL is GenBank GSS, CM-CP are GenBank
 * nucleotides to be assigned to specific projects as needed.
 *
 * Revision 6.64  2003/11/10 15:05:42  ucko
 * +CK to eAcc_gb_est
 *
 * Revision 6.63  2003/10/31 20:16:07  ucko
 * CSeq_id::IdentifyAccession: CI and CJ are both DDBJ EST.
 *
 * Revision 6.62  2003/10/24 14:57:03  ucko
 * IdentifyAccession: CH -> eAcc_gb_con.
 * GetSeqIdString: include the database name for IDs of type general.
 *
 * Revision 6.61  2003/08/25 21:15:41  ucko
 * Tweak slightly for efficiency.
 *
 * Revision 6.60  2003/08/22 15:16:48  dondosha
 * Correction in CSeq_id constructor, to allow id strings starting with a gi id
 *
 * Revision 6.59  2003/08/11 14:37:20  ucko
 * IdentifyAccession: "CG" is GenBank GSS.
 *
 * Revision 6.58  2003/07/14 20:11:59  ucko
 * +CF (eAcc_gb_est)
 *
 * Revision 6.57  2003/07/02 13:46:14  ucko
 * +CE (eAcc_gb_gss)
 *
 * Revision 6.56  2003/06/24 16:33:48  ucko
 * CSeq_id::IdentifyAccession: always return unknown for accessions that
 * contain no digits, even if they happen to look like prefixes.
 *
 * Revision 6.55  2003/05/09 14:22:56  ucko
 * CSeq_id::x_Init: treat missing (chain) names as spaces (reported by
 * Michel Dumontier) and get rid of some unnecessary calls to c_str().
 *
 * Revision 6.54  2003/04/30 14:41:01  ucko
 * CSeq_id::IdentifyAccession: CDnnnnnn -> eAcc_gb_est
 *
 * Revision 6.53  2003/03/25 15:37:18  ucko
 * CSeq_id::IdentifyAccession("CC...") -> eAcc_gb_gss
 *
 * Revision 6.52  2003/03/11 15:55:44  kuznets
 * iterate -> ITERATE
 *
 * Revision 6.51  2003/02/06 22:23:29  vasilche
 * Added CSeq_id::Assign(), CSeq_loc::Assign().
 * Added int CSeq_id::Compare() (not safe).
 * Added caching of CSeq_loc::GetTotalRange().
 *
 * Revision 6.50  2003/02/04 15:15:12  grichenk
 * Overrided Assign() for CSeq_loc and CSeq_id
 *
 * Revision 6.49  2003/01/18 08:40:03  kimelman
 * addes seqid constructor for numeric types
 *
 * Revision 6.48  2003/01/15 18:27:13  ucko
 * +AK (accidentally skipped earlier -- sigh)
 *
 * Revision 6.47  2003/01/10 15:57:30  ucko
 * Identify CB as gb_est.
 *
 * Revision 6.46  2003/01/08 16:48:03  ucko
 * +AL (accidentally skipped earlier)
 *
 * Revision 6.45  2003/01/07 19:52:15  ucko
 * Add more refseq types (NR_, NS_, NW_).
 *
 * Revision 6.44  2002/12/30 23:44:42  vakatov
 * CSeq_id::GetStringDescr() -- un-freeze "strstream" to avoid a mem.leak
 *
 * Revision 6.43  2002/12/26 16:39:25  vasilche
 * Object manager class CSeqMap rewritten.
 *
 * Revision 6.42  2002/12/03 15:55:12  dicuccio
 * Corrected processing of text id accessions with no accession set (in
 * GetSeqIdString()) - use name instead.
 *
 * Revision 6.41  2002/11/26 15:13:32  dicuccio
 * Added CSeq_id::GetStringDescr() - provides text representations of seq-ids in a
 * number of formats.
 *
 * Revision 6.40  2002/10/23 18:23:07  ucko
 * Add self-classification (using known type information).
 *
 * Revision 6.39  2002/10/22 20:19:14  jianye
 * Added GetSeqIdString()
 *
 * Revision 6.38  2002/10/18 16:03:08  ucko
 * +CA (eAcc_gb_est)
 *
 * Revision 6.37  2002/10/03 18:55:04  clausen
 * Removed extra whitespace
 *
 * Revision 6.36  2002/10/03 17:17:11  clausen
 * Added BestRank() and WorstRank()
 *
 * Revision 6.35  2002/10/03 16:03:00  ucko
 * +BZ (eAcc_gb_gss)
 *
 * Revision 6.34  2002/09/23 16:43:46  ucko
 * Change check for absence of '|' to use string::find.
 *
 * Revision 6.33  2002/09/20 19:55:29  ucko
 * +BY (eAcc_ddbj_est)
 *
 * Revision 6.32  2002/08/28 14:07:56  ucko
 * IdentifyAccession: handle BX (EMBL genome)
 *
 * Revision 6.31  2002/08/26 20:44:06  ucko
 * CSeq_id::IdentifyAccession: +BW (DDBJ EST)
 *
 * Revision 6.30  2002/08/19 16:54:30  ucko
 * IdentifyAccession: add BV (eAcc_gb_sts).
 *
 * Revision 6.29  2002/08/19 15:42:14  ucko
 * IdentifyAccession: add BU (eAcc_gb_est).
 *
 * Revision 6.28  2002/08/16 19:27:01  ucko
 * Recognize new WGS RefSeq accessions.
 *
 * Revision 6.27  2002/08/14 15:52:27  ucko
 * Add BT and XR_.
 *
 * Revision 6.26  2002/08/06 18:22:19  ucko
 * Properly handle versioned PDB accessions.
 *
 * Revision 6.25  2002/08/01 20:33:10  ucko
 * s_IdentifyAccession -> IdentifyAccession; s_ is only for module-static names.
 *
 * Revision 6.24  2002/07/30 19:42:44  ucko
 * Add s_IdentifyAccession, and use it in the string-based constructor if
 * the input isn't FASTA-format.
 * Move CVS log to end.
 *
 * Revision 6.23  2002/06/06 20:31:33  clausen
 * Moved methods using object manager to objects/util
 *
 * Revision 6.22  2002/05/22 14:03:40  grichenk
 * CSerialUserOp -- added prefix UserOp_ to Assign() and Equals()
 *
 * Revision 6.21  2002/05/06 03:39:12  vakatov
 * OM/OM1 renaming
 *
 * Revision 6.20  2002/05/03 21:28:17  ucko
 * Introduce T(Signed)SeqPos.
 *
 * Revision 6.19  2002/01/16 18:56:32  grichenk
 * Removed CRef<> argument from choice variant setter, updated sources to
 * use references instead of CRef<>s
 *
 * Revision 6.18  2002/01/10 19:00:04  clausen
 * Added GetLength
 *
 * Revision 6.17  2002/01/09 15:59:30  grichenk
 * Fixed includes
 *
 * Revision 6.16  2001/10/15 23:00:00  vakatov
 * CSeq_id::x_Init() -- get rid of unreachable "break;"
 *
 * Revision 6.15  2001/08/31 20:05:44  ucko
 * Fix ICC build.
 *
 * Revision 6.14  2001/08/31 16:02:10  clausen
 * Added new constructors for Fasta and added new id types, tpd, tpe, tpg
 *
 * Revision 6.13  2001/07/16 16:22:48  grichenk
 * Added CSerialUserOp class to create Assign() and Equals() methods for
 * user-defind classes.
 * Added SerialAssign<>() and SerialEquals<>() functions.
 *
 * Revision 6.12  2001/05/24 20:24:27  grichenk
 * Renamed seq/objmgrstub.hpp -> obgmgr/objmgr_base.hpp
 * Added Genbank, Embl and Ddbj support in CSeq_id::Compare()
 * Fixed General output by CSeq_id::WriteAsFasta()
 *
 * Revision 6.11  2001/04/17 04:14:49  vakatov
 * CSeq_id::AsFastaString() --> CSeq_id::WriteAsFasta()
 *
 * Revision 6.10  2001/01/03 16:39:05  vasilche
 * Added CAbstractObjectManager - stub for object manager.
 * CRange extracted to separate file.
 *
 * Revision 6.9  2000/12/26 17:28:55  vasilche
 * Simplified and formatted code.
 *
 * Revision 6.8  2000/12/15 19:30:31  ostell
 * Used Upcase() in AsFastaString() and changed to PNocase().Equals() style
 *
 * Revision 6.7  2000/12/08 22:19:45  ostell
 * changed MakeFastString to AsFastaString and to use ostream instead of string
 *
 * Revision 6.6  2000/12/08 20:45:14  ostell
 * added MakeFastaString()
 *
 * Revision 6.5  2000/12/04 15:09:41  vasilche
 * Added missing include.
 *
 * Revision 6.4  2000/11/30 22:08:18  ostell
 * finished Match()
 *
 * Revision 6.3  2000/11/30 16:13:12  ostell
 * added support for Textseq_id to Seq_id.Match()
 *
 * Revision 6.2  2000/11/28 12:47:41  ostell
 * fixed first switch statement to break properly
 *
 * Revision 6.1  2000/11/21 18:58:29  vasilche
 * Added Match() methods for CSeq_id, CObject_id and CDbtag.
 *
 * ===========================================================================
 */