test_pin.cpp

来自「ncbi源码」· C++ 代码 · 共 1,309 行 · 第 1/3 页

/*
 * ===========================================================================
 * PRODUCTION $Log: test_pin.cpp,v $
 * PRODUCTION Revision 1000.1  2004/06/01 19:47:01  gouriano
 * PRODUCTION PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.19
 * PRODUCTION
 * ===========================================================================
 */


#include <ncbi_pch.hpp>
#include <objtools/readers/seqdb/seqdb.hpp>

#include <serial/serial.hpp>
#include <serial/objostr.hpp>
#include <iostream>
#include <string>
#include <corelib/ncbimtx.hpp>
#include <objmgr/util/sequence.hpp>
#include <objects/seqloc/Seq_id.hpp>

#include <sys/types.h>
#include <sys/wait.h>
#include <assert.h>

USING_NCBI_SCOPE;


#include <sys/time.h>
#include <unistd.h>

inline double dbl_time(void)
{
    struct timeval tv;
    gettimeofday(& tv, 0);
    
    return tv.tv_sec + double(tv.tv_usec) / 1000000.0;
}


struct charbox {
    char xyz[10234];
};

//#include "thr_test.cpp"

int hang10()
{
    if (long(& dbl_time) == 1102000L) {
        cout << "inconceivable!" << endl;
    }
    return 0;
}

class CAccelThread {
public:
    CAccelThread(CSeqDB & db)
        : _db(db)
    {
    }
    
    void RunInline(void)
    {
//         _pid = fork();
        
//         if (! _pid) {
            _Iterate();
            exit(0);
//         }
    }
    
    void Run(void)
    {
        _pid = fork();
        
        if (! _pid) {
            _Iterate();
            exit(0);
        }
    }
    
    void Wait(void)
    {
        if (_pid) {
            waitpid(_pid, 0, 0);
            _pid = 0;
        }
    }
    
private:
    void _Iterate(void)
    {
        Uint4 oidx = 0;
        
        try {
            Uint4 oid = 0;
            const char * buf = 0;
        
            Uint4 hashesque = 0;
        
            Uint4 part = _db.GetNumSeqs() / 40;
            Uint4 parts = part;
        
            cout << "ITER:" << flush;
        
            while(_db.CheckOrFindOID(oid)) {
                if (oid > parts) {
                    parts += part;
                    cout << "+" << flush;
                }
                
                /*Uint4 oid_len =*/ _db.GetSequence(oid, & buf);
                
//                 for(Uint4 ii = 0; ii < oid_len; ii += 511) {
//                     hashesque += buf[ii];
//                 }
            
                _db.RetSequence(& buf);
            
                oid ++;
                oidx = oid;
                
                if (oid > 2763324) {
                    x_PossiblyStopMaybe();
                }
            }
            _hash = hashesque;
            cout << "!" << endl;
        }
        catch(...) {
            cout << "***" << oidx << "-" << _db.GetNumSeqs() << endl;
        }
    }
    
    void x_PossiblyStopMaybe(void)
    {
        //cout << " It is not widely known that I once leveled Rome. " << endl;
    }
    
    CSeqDB & _db;
    Uint4    _hash;
    Uint4    _pid;
};

int test1(int argc, char ** argv)
{
    string dbpath = "/net/fridge/vol/export/blast/db/blast";
    
    list<string> args;
    
    while(argc > 1) {
        args.push_front(string(argv[--argc]));
    }
    
    bool use_mm        = true;
    bool deletions     = true;
    Int8 num_display   = -1;
    Int4 num_itera     = 1;
    bool look_seq      = false;
    bool show_bioseq   = false;
    bool show_fasta    = false;
    bool get_bioseq    = false;
    bool show_progress = true;
    bool approx        = true;
    bool accel_thread  = false;
    bool accel_inline  = false;
    
    string dbname("nr");
    char seqtype = kSeqTypeProt;
    
    bool failed      = false;
    
    while(! args.empty()) {
        string desc;
        
        string s = args.front();
        args.pop_front();
        
        if (s == "-accel1") {
            s = "-accel2";
            accel_thread = true;
        } else desc += " [-accel1]";
        
        if (s == "-accel3") {
            s = "-accel2";
            accel_thread = true;
            accel_inline = true;
        } else desc += " [-accel1]";
        
        if (s == "-accel2") {
            CSeqDB nt("nr", 'p');
            
            CAccelThread accel(nt);
            
            vector<Uint4> V;
            
            Uint4 numseq = nt.GetNumSeqs();
            
            for(Uint4 i = 0; i < numseq; i+=2) {
                // Pick a random location in the array; push the
                // value at that location onto the end.
                
                // Put i into the old location for that value.
                
                V.push_back(0);
                
                Uint4 loc = (rand() + (rand() << 16)) % V.size();
                
                V[i/2] = V[loc];
                V[loc] = i/2;
            }
            
            double t_s(dbl_time());
            
            if (accel_thread) {
                if (accel_inline) {
                    accel.RunInline();
                } else {
                    accel.Run();
                }
            }
            
            Uint4 part = V.size() / 40;
            Uint4 parts = part;
            
            cout << "iter:" << flush;
            
            for(Uint4 i = 0; i < V.size(); i++) {
                if (i > parts) {
                    parts += part;
                    cout << "-" << flush;
                }
                
                Uint4 oid = V[i];
                
                const char * buf = 0;
                nt.GetSequence(oid, & buf);
                nt.RetSequence(& buf);
            }
            
            double t_e(dbl_time());
            
            cout << "#" << endl;
            cout << "time elapsed: " << (t_e - t_s) << endl;
            
            
            double w_s(dbl_time());
            
            accel.Wait();
            
            double w_e(dbl_time());
            
            cout << "time elapsed: " << (w_e - w_s) << endl;
            
            return 0;
        } else desc += " [-accel2]";
        
        if (s == "-fromcpp") {
            const char * ge = getenv("BLASTDB");
            string pe("BLASTDB=/home/bealer/code/ut/internal/blast/unit_test/data:" + string(ge ? ge : ""));
            putenv((char*)pe.c_str());
            
            CSeqDB local1("seqp", 'p');
            CSeqDB local2("seqp", 'p', 0, 0, false);
            
            Int4 num1 = local1.GetNumSeqs();
            Int4 num2 = local1.GetNumSeqs();
        
            assert(num1 >= 1);
            assert(num1 == num2);
            
            cout << "Test worked." << endl;
            return 0;
        } else desc += " [-fromcpp]";
        
        if (s == "-alphabeta") {
            
            // Note: this test is NOT EXPECTED to work, unless the
            // user has databases named "alpha" and "beta" somewhere
            // in their path.
            
            CSeqDB ab("alpha beta", 'p');
            
            for(Uint4 i = 0; i < ab.GetNumSeqs(); i++) {
                cout << "-----seq " << i << " length " << ab.GetSeqLength(i) << " ------------" << endl;
            }
            
            return 0;
        } else desc += " [-alphabeta]";
        
        if (s == "-here") {
            CSeqDB nr("tenth", 'p');
            
            for(int i = 0; i<100; i++) {
                CRef<CBioseq> bs = nr.GetBioseq(i);
                
                cout << "-----seq " << i << "------------" << endl;
                
                auto_ptr<CObjectOStream>
                    outpstr(CObjectOStream::Open(eSerial_AsnText, cout));
                
                *outpstr << *bs;
            }
            
            return 0;
        } else desc += " [-here]";
        
        if (s == "-dyn") {
            CSeqDB db("nr", 'p');
            
            cout << "Num oids: " << db.GetNumSeqs() << endl;
            
            char * buf1 = 0;
            
            Int4 len = db.GetAmbigSeqAlloc(10,
                                           & buf1,
                                           kSeqDBNuclBlastNA8,
                                           eNew);
            
            cout << "Length (10): " << len << endl;
            
            delete[] buf1;
            
            return 0;
        } else desc += " [-dyn]";
        
        if (s == "-limit") {
            {
                CSeqDB db("/home/bealer/seqdb/tenth", 'p', 10, 20, true);
                
                cout << "Num oids: " << db.GetNumSeqs() << endl;
                
                CSeqDBIter i = db.Begin();
                
                while(i) {
                    CRef<CBioseq> bs = db.GetBioseq(i.GetOID());
                    
                    cout << "-----seq "
                         << i.GetOID() << " length "
                         << i.GetLength() << "-------" << endl;
                    
                    ++i;
                }
            }
            {
                CSeqDB db("swissprot", 'p', 135, 175, true);
                
                cout << "Num oids: " << db.GetNumSeqs() << endl;
                
                CSeqDBIter i = db.Begin();
                
                while(i) {
                    CRef<CBioseq> bs = db.GetBioseq(i.GetOID());
                    
                    cout << "-----seq "
                         << i.GetOID() << " length "
                         << i.GetLength() << "-------" << endl;
                    
                    ++i;
                }
            }
            
            return 0;
        } else desc += " [-local]";
        
        if (s == "-local") {
            CSeqDB nr("/home/bealer/seqdb/tenth", 'p');
            
            for(int i = 0; i<100; i++) {
                CRef<CBioseq> bs = nr.GetBioseq(i);
                
                cout << "-----seq " << i << "------------" << endl;
                
                auto_ptr<CObjectOStream>
                    outpstr(CObjectOStream::Open(eSerial_AsnText, cout));
                
                *outpstr << *bs;
            }
            
            return 0;
        } else desc += " [-local]";
        
        if (s == "-seqids") {
            CSeqDB nr(/*dbpath,*/ "nr", 'p');
            
            for(int i = 0; i<100; i++) {
                list< CRef<CSeq_id> > seqids =
                    nr.GetSeqIDs(i);
                
                cout << "-----seq " << i << "------------" << endl;
                
                for(list< CRef<CSeq_id> >::iterator j = seqids.begin();
                    j != seqids.end();
                    j++) {
                    
                    cout << "SEQID----*:" << endl;
                    
                    auto_ptr<CObjectOStream>
                        outpstr(CObjectOStream::Open(eSerial_AsnText, cout));
                    
                    CRef<CSeq_id> ident = *j;
                    
                    *outpstr << *ident;
                }
            }
            
            return 0;
        } else desc += " [-seqids]";
        
        if (s == "-memtest") {
            CSeqDB nt("nt", 'n', 0, 0, false);
            
            Uint4 oid = 0;
            
            for(int i = 0; i<100; i++) {
                const char * buf(0);
                
                if (! nt.CheckOrFindOID(oid))
                    break;
                
                cout << "-----------------" << endl;
                
                if (1) {
                    int length = nt.GetSequence(oid, & buf);
                    
                    cout << "NT OID = " << oid << ", length is " << length << endl;
                    
                    int y = (length > 16) ? 16 : length;
                    
                    for(int x = 0; x < y; x++) {
                        if ((x & 3) == 0)
                            cout << " ";
                        
                        if (unsigned(buf[x]) < 0x10)
                            cout << "0";
                        
                        cout << hex << (unsigned(buf[x]) & 0xFF) << " ";