gnRAWSource.cpp Source File

00001 
00002 // File:            gnRAWSource.h
00003 // Purpose:         Implements gnBaseSource for raw data files
00004 // Description:     
00005 // Changes:        
00006 // Version:         libGenome 0.1.0 
00007 // Author:          Aaron Darling 
00008 // Last Edited:     April 15, 2001, 11:13:00pm 
00009 // Modified by:     
00010 // Copyright:       (c) Aaron Darling 
00011 // Licenses:        Proprietary 
00013 #include "gn/gnFilter.h"
00014 #include "gn/gnRAWSource.h"
00015 #include "gn/gnGenomeSpec.h"
00016 #include "gn/gnFragmentSpec.h"
00017 #include "gn/gnSourceSpec.h"
00018 #include "gn/gnStringTools.h"
00019 #include "gn/gnDebug.h"
00020 
00021 gnRAWSource::gnRAWSource()
00022 {
00023         m_openString = "";
00024         m_contig = NULL;
00025 }
00026 
00027 gnRAWSource::gnRAWSource( const gnRAWSource& s ) : gnFileSource(s)
00028 {
00029         if(m_contig != NULL)
00030                 m_contig = s.m_contig->Clone();
00031 }
00032 
00033 gnRAWSource::~gnRAWSource()
00034 {
00035         m_ifstream.close();
00036         delete m_contig;
00037 }
00038 
00039 boolean gnRAWSource::HasContig( const string& name ) const
00040 {
00041         if( name.length() == 0 )
00042                 return true;
00043         return false;
00044 }
00045 
00046 uint32 gnRAWSource::GetContigID( const string& name ) const
00047 {
00048         return ALL_CONTIGS;
00049 }
00050 
00051 string gnRAWSource::GetContigName( const uint32 i ) const
00052 {
00053         return "";
00054 }
00055 
00056 gnSeqI gnRAWSource::GetContigSeqLength( const uint32 i ) const
00057 {
00058         if( m_contig && (i == 0 || i == ALL_CONTIGS))
00059                 return m_contig->GetSeqLength();
00060         return GNSEQI_ERROR;
00061 }
00062 
00063 boolean gnRAWSource::SeqRead( const gnSeqI start, char* buf, uint32& bufLen, const uint32 contigI ){
00064         return Read( start, buf, bufLen );
00065 }
00066 
00067 gnGenomeSpec *gnRAWSource::GetSpec() const{
00068         return m_spec->Clone();
00069 }
00070 
00071 boolean gnRAWSource::Write(gnSequence& seq, const string& filename){
00072         ofstream m_ofstream(filename.c_str(), ios::out | ios::binary);
00073         if(!m_ofstream.is_open())
00074                 return false;
00075 
00076         gnSeqC buf[BUFFER_SIZE + 1];
00077         buf[BUFFER_SIZE] = 0;
00078         gnSeqI readOffset = 0;
00079         gnSeqI readLength = seq.length();
00080 
00081         while(readLength > 0){  //buffer the read/writes
00082                 gnSeqI writeLen = readLength < BUFFER_SIZE ? readLength : BUFFER_SIZE;
00083                 
00084                 if(!seq.ToArray(buf, writeLen, readOffset))
00085                         return false;
00086                 m_ofstream << buf;
00087                 readLength -= writeLen;
00088                 readOffset += writeLen;
00089         }
00090         return true;
00091 }
00092 
00093 gnFileContig* gnRAWSource::GetFileContig( const uint32 contigI ) const{
00094         if(contigI > 0)
00095                 return NULL;
00096         return m_contig;
00097 }
00098 
00099 //File parsing access routine
00100 boolean gnRAWSource::ParseStream( istream& fin )
00101 {
00102         // init variables
00103         uint64 streamPos = 0;
00104         uint64 bufReadLen = 0;
00105         char* buf = new char[BUFFER_SIZE];
00106         gnSeqI seqLength = 0;
00107         
00108         m_contig->SetRepeatSeqGap(true);
00109         m_contig->SetSectStart(gnContigSequence, 0);
00110 
00111         while( !fin.eof() )
00112         {
00113                   // read chars
00114                 fin.read( buf , BUFFER_SIZE );
00115                 bufReadLen = fin.gcount();
00116                 
00117                 for( uint32 i=0 ; i < bufReadLen ; i++ )
00118                 {
00119                         if(m_pFilter->IsValid(buf[i]))
00120                                 seqLength++;
00121                         else
00122                                 m_contig->SetRepeatSeqGap(false);
00123                         break;
00124                 }
00125                 streamPos += bufReadLen;
00126         }
00127         m_contig->SetSectEnd(gnContigSequence, streamPos);
00128         m_contig->SetSeqLength(seqLength);
00129         m_spec = new gnGenomeSpec();
00130         gnFragmentSpec* fragspec = new gnFragmentSpec();
00131         gnSourceSpec* sspec = new gnSourceSpec(this);
00132         sspec->SetSourceName(m_openString);
00133         m_spec->AddSpec(fragspec);
00134         fragspec->AddSpec(sspec);
00135 
00136         m_ifstream.clear();
00137         delete[] buf;
00138         return true;
00139 }