Enter your search terms GNU-Darwin Web Submit search form

gnDNXSource.cpp Go to the documentation of this file. // File: gnDNXSource.h // Purpose: Implements gnBaseSource for .DNX files // Description: // Changes: // Version: libGenome 0.1.0 // Author: Aaron Darling // Last Edited: April 15, 2001, 11:13:00pm // Modified by: // Copyright: (c) Aaron Darling // Licenses: Proprietary #include "gn/gnDNXSource.h" #include "gn/gnSourceSpec.h" #include "gn/gnStringSpec.h" #include "gn/gnSourceFactory.h" #include "gn/gnFASSource.h" #include "gn/gnGBKSource.h" #include "gn/gnBaseHeader.h" #include "gn/gnFilter.h" #include "gn/gnDebug.h" #include "gn/gnStringTools.h" #include <string> gnDNXSource::gnDNXSource() { m_DNXSpec = new gnGenomeSpec(); m_pFilter = gnFilter::fullDNASeqFilter(); if(m_pFilter == NULL){ DebugMsg("Error using static sequence filters."); } } gnDNXSource::gnDNXSource( const gnDNXSource& s ) : gnFileSource(s) { if(s.m_DNXSpec != NULL) m_DNXSpec = s.m_DNXSpec->Clone(); } gnDNXSource::~gnDNXSource() { m_ifstream.close(); delete m_DNXSpec; } boolean gnDNXSource::HasContig( const string& name ) const { for(uint32 contigI = 0; contigI < m_DNXSpec->GetSpecListLength(); contigI++){ if(m_DNXSpec->GetSpec(contigI)->GetName() == name) return true; } return false; } uint32 gnDNXSource::GetContigID( const string& name ) const { for(uint32 contigI = 0; contigI < m_DNXSpec->GetSpecListLength(); contigI++){ if(m_DNXSpec->GetSpec(contigI)->GetName() == name) return contigI; } return ALL_CONTIGS; } string gnDNXSource::GetContigName( const uint32 i ) const { if(i < m_DNXSpec->GetSpecListLength()){ gnBaseSpec *gnbs = m_DNXSpec->GetSpec(i); return gnbs->GetName(); } return ""; } gnSeqI gnDNXSource::GetContigSeqLength( const uint32 i ) const { if( i == ALL_CONTIGS){ return m_DNXSpec->GetLength(); }else if(i < m_DNXSpec->GetSpecListLength()){ gnBaseSpec *gnbs = m_DNXSpec->GetSpec(i); return gnbs->GetLength(); } return 0; } //read raw data from the file void gnDNXSource::ValidateName(string& name){ if(name == ""){ //make a random one. name.resize(4); srand(time(NULL)); for(int i=0; i < 4; i++) name[i] = (rand() % 26) + 64; } } boolean gnDNXSource::Write(gnGenomeSpec* spec, const string& filename){ ofstream m_ofstream(filename.c_str(), ios::out | ios::binary); gnSourceFactory* m_sSourceFactory = gnSourceFactory::GetSourceFactory(); if(!m_ofstream.is_open()) return false; for(uint32 i=0; i < spec->GetSpecListLength(); i++){ //each of these will be dnx statements gnFragmentSpec* curStatementSpec = spec->GetSpec(i); string sourceName = spec->GetSourceName(); string statementName = spec->GetName(); if(!m_sSourceFactory->HasSource(sourceName)){ ValidateName(statementName); statementName += ".seq"; m_ofstream << statementName << "="; }else m_ofstream << sourceName << "="; for(uint32 j=0; j < curStatementSpec->GetSpecListLength(); j++){ //each of these will be the files //referred to by the dnx statement gnContigSpec* curSubSpec = curStatementSpec->GetSpec(i); sourceName = curStatementSpec->GetSourceName(); string contigName = curStatementSpec->GetName(); if(!m_sSourceFactory->HasSource(sourceName)){ ValidateName(contigName); string writename = contigName+".seq"; gnSequence gns = *curSubSpec; gnGBKSource::Write(gns, writename); m_ofstream << writename; }else m_ofstream << sourceName; if( j + 1 < curStatementSpec->GetSpecListLength()) m_ofstream << "+"; } m_ofstream << ";"; gnBaseHeader *gpbh = spec->GetHeader(0); string header = ""; if(gpbh != NULL){ header = gpbh->GetHeader(); //delete everything after the first newline. uint32 newlinepos = header.find_first_of('\n', 0); if(newlinepos != string::npos) header = header.substr(0, newlinepos - 1); } m_ofstream << header << "\r\n"; } m_ofstream.close(); return true; } gnFileContig* gnDNXSource::GetFileContig( const uint32 contigI ) const{ return NULL; //returning NULL } //reads an inputstream and creates fills the spec vector appropriately boolean gnDNXSource::ParseStream( istream& fin ) { // INIT temp varables uint32 readState = 0; //10 - currently inside a comment uint32 sectionStart = 0; gnFragmentSpec* currentFragSpec = 0; gnBaseSource *currentSource; string currentSourceName; uint32 currentContig = ALL_CONTIGS; uint32 currentSeqStart = 0; boolean currentRevComp = false; // INIT buffer uint64 bufReadLen = 0; uint64 remainingBuffer = 0; char* buf = new char[BUFFER_SIZE]; string curliteral; //Get the source factory and add the current dnx path to it. gnSourceFactory *sourceFactory = gnSourceFactory::GetSourceFactory(); sourceFactory->AddPath(getPathString(m_openString)); while( !fin.eof() ) { if(sectionStart > 0){ remainingBuffer = bufReadLen - sectionStart; if(readState == 5){ //add literal curliteral += string(buf, sectionStart, remainingBuffer); remainingBuffer = 0; sectionStart = bufReadLen; }else memmove(buf, buf+sectionStart, remainingBuffer); } // read chars fin.read( buf + remainingBuffer, BUFFER_SIZE - (bufReadLen - sectionStart)); sectionStart = 0; bufReadLen = fin.gcount() + remainingBuffer; for( uint32 i=0 ; i < bufReadLen ; i++ ) { char ch = buf[i]; switch( readState ) { case 0: // Get name of genome if(ch == '='){ //genome name is from sectionStart to i string contigName(buf+sectionStart, i - sectionStart); currentFragSpec = new gnFragmentSpec(); currentFragSpec->SetName(contigName); currentFragSpec->SetSourceName(m_openString); m_DNXSpec->AddSpec(currentFragSpec); sectionStart = i+1; readState = 1; } break; case 1: // Ignore whitespace before filename or literal if((ch == ' ')||(ch == ' ')) break; case 2: // Are we getting a new source file name or a literal? if(ch == '"'){ //getting a literal readState = 5; sectionStart = i+1; break; } readState = 3; sectionStart = i; case 3: // Get a new source file name //stop on >, (, +, and \n if(ch == '\n' && sectionStart == i -1){ if(buf[sectionStart]=='\r'){ sectionStart = i + 1; break; } } if((ch == '+')||(ch == '>')||(ch == '(')||(ch == '\n')||(ch == ';')){ //use the entire source file string seqfile(buf, sectionStart, i - sectionStart); currentSourceName = seqfile; currentSource = sourceFactory->AddSource(seqfile, true); if (currentSource==NULL) { delete[] buf; return false; } if((ch == '+')||(ch == '\n')||(ch == ';')){ gnSourceSpec* tmp_spec = new gnSourceSpec(currentSource); tmp_spec->SetSourceName(seqfile); currentFragSpec->AddSpec(tmp_spec); readState = 1; if(ch == '\n'){ //reached the end of the statement. parse another. readState = 0; }else if(ch == ';'){ //hit a comment. readState = 9; } }else if(ch == '>'){ //select a contig to use readState = 4; }else if(ch == '('){ // use a specified section of the entire file readState = 6; } sectionStart = i + 1; } break; case 4: // Get a specific contig to use //stop on (, +, ;, and \n if((ch == '+')||(ch == '\n')||(ch == ';')||(ch == '(')){ //use the entire contig string contigname(buf, sectionStart, i - sectionStart); currentContig = currentSource->GetContigID(contigname); if((ch == '+')||(ch == '\n')||(ch == ';')){ gnSourceSpec* tmp_spec = new gnSourceSpec(currentSource, currentContig); tmp_spec->SetSourceName(currentSourceName); currentFragSpec->AddSpec(tmp_spec); readState = 1; if(ch == '\n'){ //reached the end of the statement. parse another. readState = 0; }else if(ch == ';'){ //hit a comment. readState = 9; } }else if(ch == '('){ //use the specified section readState = 6; } sectionStart = i + 1; } break; case 5: // read in a literal // stop on " if(ch == '"'){ //now create a string spec from sectionStart to i-1 string literal(buf, sectionStart, i - sectionStart); if(curliteral.length() > 0){ literal += curliteral; curliteral = ""; } gnStringSpec *gpss = new gnStringSpec(literal, currentFragSpec->GetSpecListLength()); currentFragSpec->AddSpec(gpss); } case 6: // read in a specified section //stop on , or < or > if((ch == ',') || (ch == '<') || (ch == '>')){ string seqstartstring(buf, sectionStart, i - sectionStart); if(seqstartstring == "lend"){ currentSeqStart = 0; }else if (seqstartstring == "rend"){ currentSeqStart = GNSEQI_END; }else currentSeqStart = atoi(seqstartstring.c_str()) - 1; if(ch == '<') currentRevComp = true; sectionStart = i + 1; readState = 7; } break; case 7: // read in the second half of a specified section //stop on ) if(ch == ')'){ string seqendstring(buf, sectionStart, i - sectionStart); uint32 currentSeqEnd = GNSEQI_END; if(seqendstring == "lend"){ currentSeqEnd = 0; }else if (seqendstring == "rend"){ currentSeqEnd = GNSEQI_END; }else currentSeqEnd = atoi(seqendstring.c_str()) - 1; gnSourceSpec* tmp_spec = new gnSourceSpec(currentSource, currentContig, currentSeqStart, currentSeqEnd, currentRevComp); tmp_spec->SetSourceName(currentSourceName); currentFragSpec->AddSpec(tmp_spec); currentRevComp = false; //set it back to its default value. sectionStart = i + 1; readState = 8; //look for connective operator } break; case 8: //skip whitespace until a connective or terminating operator is reached. if(ch == '+'){ sectionStart = i + 1; readState = 1; //start over } if(ch == '\n'){ sectionStart = i + 1; readState = 0; } if(ch == ';'){ sectionStart = i + 1; readState = 9; } break; case 9: //skip comment until newline. if(ch == '\n'){ sectionStart = i + 1; readState = 0; } break; default: DebugMsg("ERROR in file\n"); return false; break; } }// for all buf }// while !eof // CLEAN UP delete[] buf; return true; }