jpayne@68: package tax;
jpayne@68: 
jpayne@68: import java.io.File;
jpayne@68: import java.io.PrintStream;
jpayne@68: 
jpayne@68: import dna.Data;
jpayne@68: import fileIO.ByteFile;
jpayne@68: import fileIO.ByteFile1;
jpayne@68: import fileIO.ByteFile2;
jpayne@68: import fileIO.ByteStreamWriter;
jpayne@68: import fileIO.FileFormat;
jpayne@68: import fileIO.ReadWrite;
jpayne@68: import shared.Parse;
jpayne@68: import shared.Parser;
jpayne@68: import shared.PreParser;
jpayne@68: import shared.ReadStats;
jpayne@68: import shared.Shared;
jpayne@68: import shared.Timer;
jpayne@68: import shared.Tools;
jpayne@68: import stream.FastaReadInputStream;
jpayne@68: import structures.ByteBuilder;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date April 4, 2017
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class ShrinkAccession {
jpayne@68: 	
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		ShrinkAccession x=new ShrinkAccession(args);
jpayne@68: 		x.process(t);
jpayne@68: 		
jpayne@68: 		//Close the print stream if it was redirected
jpayne@68: 		Shared.closeStream(x.outstream);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public ShrinkAccession(String[] args){
jpayne@68: 		
jpayne@68: 		{//Preparse block for help, config files, and outstream
jpayne@68: 			PreParser pp=new PreParser(args, getClass(), false);
jpayne@68: 			args=pp.args;
jpayne@68: 			outstream=pp.outstream;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		if(Data.PIGZ()){
jpayne@68: 			ReadWrite.ZIPLEVEL=Tools.max(ReadWrite.ZIPLEVEL, 6);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		Parser parser=new Parser();
jpayne@68: 		for(int i=0; i<args.length; i++){
jpayne@68: 			String arg=args[i];
jpayne@68: 			String[] split=arg.split("=");
jpayne@68: 			String a=split[0].toLowerCase();
jpayne@68: 			String b=split.length>1 ? split[1] : null;
jpayne@68: 
jpayne@68: 			if(parser.parse(arg, a, b)){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("verbose")){
jpayne@68: 				verbose=Parse.parseBoolean(b);
jpayne@68: 				ByteFile1.verbose=verbose;
jpayne@68: 				ByteFile2.verbose=verbose;
jpayne@68: 				ReadWrite.verbose=verbose;
jpayne@68: 			}else if(a.equals("gi")){
jpayne@68: 				KEEP_GI_NUMBERS=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("outgi") || a.equals("giout") || a.equals("gi")){
jpayne@68: 				giOut=b;
jpayne@68: 			}else if(parser.in1==null && i==0 && !arg.contains("=") && (arg.toLowerCase().startsWith("stdin") || new File(arg).exists())){
jpayne@68: 				parser.in1=arg;
jpayne@68: 			}else if(parser.out1==null && i==1 && !arg.contains("=")){
jpayne@68: 				parser.out1=arg;
jpayne@68: 			}else{
jpayne@68: 				outstream.println("Unknown parameter "+args[i]);
jpayne@68: 				assert(false) : "Unknown parameter "+args[i];
jpayne@68: 				//				throw new RuntimeException("Unknown parameter "+args[i]);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		{//Process parser fields
jpayne@68: 			Parser.processQuality();
jpayne@68: 			
jpayne@68: 			overwrite=ReadStats.overwrite=parser.overwrite;
jpayne@68: 			append=ReadStats.append=parser.append;
jpayne@68: 			
jpayne@68: 			in=parser.in1;
jpayne@68: 
jpayne@68: 			out=parser.out1;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		assert(FastaReadInputStream.settingsOK());
jpayne@68: 		
jpayne@68: 		if(in==null){throw new RuntimeException("Error - at least one input file is required.");}
jpayne@68: 		if(!ByteFile.FORCE_MODE_BF1 && !ByteFile.FORCE_MODE_BF2){
jpayne@68: 			ByteFile.FORCE_MODE_BF2=false;
jpayne@68: 			ByteFile.FORCE_MODE_BF1=true;
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		if(out!=null && out.equalsIgnoreCase("null")){out=null;}
jpayne@68: 		
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, append, false, out)){
jpayne@68: 			outstream.println((out==null)+", "+out);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+out+"\n");
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		ffout=FileFormat.testOutput(out, FileFormat.TXT, null, true, overwrite, append, false);
jpayne@68: 		ffoutGi=FileFormat.testOutput(giOut, FileFormat.TXT, null, true, overwrite, append, false);
jpayne@68: 		ffin=FileFormat.testInput(in, FileFormat.TXT, null, true, true);
jpayne@68: 		
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void process(Timer t){
jpayne@68: 		
jpayne@68: 		ByteFile bf=ByteFile.makeByteFile(ffin);
jpayne@68: 		ByteStreamWriter bsw=new ByteStreamWriter(ffout);
jpayne@68: 		bsw.start();
jpayne@68: 
jpayne@68: 		long linesProcessed=0;
jpayne@68: 		long charsProcessed=0;
jpayne@68: 		long badLines=0;
jpayne@68: 		
jpayne@68: 		byte[] line=bf.nextLine();
jpayne@68: 		ByteBuilder bb=new ByteBuilder(10000);
jpayne@68: 		int columns=4;
jpayne@68: 		while(line!=null){
jpayne@68: 			if(Tools.startsWith(line, "accession\t")){
jpayne@68: 				bb.append(line);
jpayne@68: 				bb.nl();
jpayne@68: 			}else if(Tools.startsWith(line, "accession.version\ttaxid")){
jpayne@68: 				columns=2;
jpayne@68: 				bb.append("accession\t\ttaxid\t");//dummy header
jpayne@68: 				bb.nl();
jpayne@68: 			}else{
jpayne@68: 				charsProcessed+=line.length+1;
jpayne@68: 				linesProcessed++;
jpayne@68: 				
jpayne@68: 				final int tid=(columns==4 ? AccessionToTaxid.parseLineToTaxid(line, (byte)'\t') : 
jpayne@68: 					AccessionToTaxid.parseLineToTaxid_2col(line, (byte)'\t'));
jpayne@68: 				if(tid<1){
jpayne@68: 					badLines++;
jpayne@68: 				}else{
jpayne@68: 					int i=0;
jpayne@68: 					
jpayne@68: 					while(i<line.length){//Accession
jpayne@68: 						byte b=line[i];
jpayne@68: 						bb.append(b);
jpayne@68: 						i++;
jpayne@68: 						if(b=='\t'){break;}
jpayne@68: 					}
jpayne@68: 					
jpayne@68: 					if(columns==4){
jpayne@68: 						while(i<line.length){//Accession with decimal
jpayne@68: 							byte b=line[i];
jpayne@68: 							//						bb.append(b);
jpayne@68: 							i++;
jpayne@68: 							if(b=='\t'){break;}
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 					bb.append('\t');
jpayne@68: 					
jpayne@68: 					while(i<line.length){//Taxid
jpayne@68: 						byte b=line[i];
jpayne@68: 						bb.append(b);
jpayne@68: 						i++;
jpayne@68: 						if(b=='\t'){break;}
jpayne@68: 					}
jpayne@68: 					
jpayne@68: 					if(KEEP_GI_NUMBERS){
jpayne@68: 						if(line.length>i && Tools.isDigit(line[i])){//GI number or "na"
jpayne@68: 							while(i<line.length){
jpayne@68: 								byte b=line[i];
jpayne@68: 								bb.append(b);
jpayne@68: 								i++;
jpayne@68: //								if(b=='\t'){break;}
jpayne@68: 							}
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 					bb.nl();
jpayne@68: 				}
jpayne@68: 				
jpayne@68: //				String[] split=new String(line).split("\t");
jpayne@68: //				bb.append(split[0]);
jpayne@68: //				bb.tab();
jpayne@68: //				bb.tab();
jpayne@68: //				bb.append(split[2]);
jpayne@68: //				bb.tab();
jpayne@68: //				bb.nl();
jpayne@68: 			}
jpayne@68: 			if(bb.length()>8000){
jpayne@68: 				bsw.print(bb);
jpayne@68: 				bb.clear();
jpayne@68: 			}
jpayne@68: 			line=bf.nextLine();
jpayne@68: 		}
jpayne@68: 		if(bb.length()>0){
jpayne@68: 			bsw.print(bb);
jpayne@68: 			bb.clear();
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		errorState|=bf.close();
jpayne@68: 		if(bsw!=null){errorState|=bsw.poisonAndWait();}
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 		outstream.println("Discarded "+badLines+" lines.\n");
jpayne@68: 		outstream.println(Tools.timeLinesBytesProcessed(t, linesProcessed, charsProcessed, 8));
jpayne@68: 		
jpayne@68: 		if(errorState){
jpayne@68: 			throw new RuntimeException(getClass().getName()+" terminated in an error state; the output may be corrupt.");
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private String in=null;
jpayne@68: 	private String out=null;
jpayne@68: 	private String giOut=null;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private final FileFormat ffin;
jpayne@68: 	private final FileFormat ffout;
jpayne@68: 	private final FileFormat ffoutGi;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private PrintStream outstream=System.err;
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	public static boolean KEEP_GI_NUMBERS=true;
jpayne@68: 	public boolean errorState=false;
jpayne@68: 	private boolean overwrite=false;
jpayne@68: 	private boolean append=false;
jpayne@68: 	
jpayne@68: }