jpayne@68: package tax;
jpayne@68: 
jpayne@68: import java.io.File;
jpayne@68: import java.io.PrintStream;
jpayne@68: 
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.Shared;
jpayne@68: import shared.Timer;
jpayne@68: import shared.Tools;
jpayne@68: import stream.ConcurrentGenericReadInputStream;
jpayne@68: import stream.FastaReadInputStream;
jpayne@68: import structures.ByteBuilder;
jpayne@68: import structures.IntHashSet;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date May 9, 2016
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class RenameIMG {
jpayne@68: 	
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		RenameIMG x=new RenameIMG(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 RenameIMG(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: 		
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(a.equals("lines")){
jpayne@68: 				maxLines=Long.parseLong(b);
jpayne@68: 				if(maxLines<0){maxLines=Long.MAX_VALUE;}
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: 				stream.FastaReadInputStream.verbose=verbose;
jpayne@68: 				ConcurrentGenericReadInputStream.verbose=verbose;
jpayne@68: 				stream.FastqReadInputStream.verbose=verbose;
jpayne@68: 				ReadWrite.verbose=verbose;
jpayne@68: 			}else if(a.equals("img")){
jpayne@68: 				imgFile=b;
jpayne@68: 			}else if(parser.parse(arg, a, b)){
jpayne@68: 				//do nothing
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: 			overwrite=parser.overwrite;
jpayne@68: 			append=parser.append;
jpayne@68: 			
jpayne@68: 			in1=parser.in1;
jpayne@68: 
jpayne@68: 			out1=parser.out1;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		assert(FastaReadInputStream.settingsOK());
jpayne@68: 		
jpayne@68: 		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
jpayne@68: 		if("auto".equalsIgnoreCase(imgFile)){imgFile=TaxTree.defaultImgFile();}//TODO: why are these set to the same default?
jpayne@68: 		if("auto".equalsIgnoreCase(in1)){in1=TaxTree.defaultImgFile();}
jpayne@68: 		
jpayne@68: 		if(!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(out1!=null && out1.equalsIgnoreCase("null")){out1=null;}
jpayne@68: 		
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, append, false, out1)){
jpayne@68: 			outstream.println((out1==null)+", "+out1);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+out1+"\n");
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		ffout1=FileFormat.testOutput(out1, FileFormat.FA, null, true, overwrite, append, false);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void copyFiles(ImgRecord[] array){
jpayne@68: 		if(useSet){set=new IntHashSet(10000);}
jpayne@68: 		ByteStreamWriter bsw=new ByteStreamWriter(ffout1);
jpayne@68: 		bsw.start();
jpayne@68: 		for(ImgRecord ir : array){
jpayne@68: 			if(ir.taxID>0){set.add(ir.taxID);}
jpayne@68: 			else{unknownTaxid++;}
jpayne@68: 			FileFormat ffin=FileFormat.testInput(ir.path(), FileFormat.FA, null, true, true);
jpayne@68: 			process_inner(ffin, bsw, ir.imgID);
jpayne@68: 		}
jpayne@68: 		knownTaxid=set.size();
jpayne@68: 		set=null;
jpayne@68: 		if(bsw!=null){errorState|=bsw.poisonAndWait();}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void process(Timer t){
jpayne@68: 		ImgRecord[] array=ImgRecord.toArray(in1, TaxTree.IMG_HQ);
jpayne@68: 		if(imgFile==null){
jpayne@68: 			TaxTree.loadIMG(array);
jpayne@68: 		}else{
jpayne@68: 			ImgRecord[] array2=ImgRecord.toArray(imgFile, TaxTree.IMG_HQ);
jpayne@68: 			TaxTree.loadIMG(array2);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		copyFiles(array);
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 
jpayne@68: 		final int spaces=8;
jpayne@68: 		String fpstring=""+filesProcessed;
jpayne@68: 		String cpstring=Tools.padKM(sequencesProcessed, spaces);
jpayne@68: 		String bapstring=Tools.padKM(basesProcessed, spaces);
jpayne@68: 		String tpstring=""+knownTaxid;
jpayne@68: 		
jpayne@68: 		outstream.println("Time:                         \t"+t);
jpayne@68: 		outstream.println("Files Processed:    "+fpstring);
jpayne@68: 		outstream.println("Contigs Processed:  "+cpstring);
jpayne@68: 		outstream.println("Bases Processed:    "+bapstring);
jpayne@68: 		if(useSet){outstream.println("TaxIDs Processed:   "+tpstring+" \t"+"("+unknownTaxid+" unknown)");}
jpayne@68: 		outstream.println(Tools.linesBytesProcessed(t.elapsed, linesProcessed, bytesProcessed, spaces));
jpayne@68: 		
jpayne@68: 		outstream.println();
jpayne@68: 		outstream.println("Valid Files:       \t"+filesValid);
jpayne@68: 		outstream.println("Invalid Files:     \t"+(filesProcessed-filesValid));
jpayne@68: 		outstream.println("Valid Lines:       \t"+linesValid);
jpayne@68: 		outstream.println("Invalid Lines:     \t"+(linesProcessed-linesValid));
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: 	void process_inner(final FileFormat ffin, final ByteStreamWriter bsw, final long img){
jpayne@68: 		
jpayne@68: 		filesProcessed++;
jpayne@68: 		{
jpayne@68: 			File f=new File(ffin.name());
jpayne@68: 			if(!f.exists() || !f.canRead()){
jpayne@68: 				System.err.println("Can't find "+f);
jpayne@68: 				errorState=true;
jpayne@68: 				return;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		final int tid=TaxTree.imgToTaxid(img);
jpayne@68: 		ByteFile bf=ByteFile.makeByteFile(ffin);
jpayne@68: 		
jpayne@68: 		byte[] line=bf.nextLine();
jpayne@68: 		ByteBuilder bb=new ByteBuilder();
jpayne@68: 		
jpayne@68: 		while(line!=null){
jpayne@68: 			if(line.length>0){
jpayne@68: 				if(maxLines>0 && linesProcessed>=maxLines){break;}
jpayne@68: 				linesProcessed++;
jpayne@68: 				bytesProcessed+=line.length;
jpayne@68: 
jpayne@68: 				linesValid++;
jpayne@68: 				if(line[0]=='>'){
jpayne@68: 					sequencesProcessed++;
jpayne@68: 					bb.append('>');
jpayne@68: 					if(tid>=0){
jpayne@68: 						bb.append("tid|");
jpayne@68: 						bb.append(tid);
jpayne@68: 						bb.append('|');
jpayne@68: 					}
jpayne@68: 					bb.append("img|");
jpayne@68: 					bb.append(img);
jpayne@68: 					bb.append(' ');
jpayne@68: 					for(int i=1; i<line.length; i++){
jpayne@68: 						bb.append(line[i]);
jpayne@68: 					}
jpayne@68: 				}else{
jpayne@68: 					basesProcessed+=line.length;
jpayne@68: 					bb.append(line);
jpayne@68: 				}
jpayne@68: 				bb.nl();
jpayne@68: 				bsw.print(bb.toBytes());
jpayne@68: 				bb.clear();
jpayne@68: 			}
jpayne@68: 			line=bf.nextLine();
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		filesValid++;
jpayne@68: 		errorState|=bf.close();
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private String in1=null;
jpayne@68: 	private String out1=null;
jpayne@68: 	private String imgFile=null;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private IntHashSet set=null;
jpayne@68: 	private int knownTaxid=0;
jpayne@68: 	private int unknownTaxid=0;
jpayne@68: 	private boolean useSet=true;
jpayne@68: 	
jpayne@68: 	private long linesProcessed=0;
jpayne@68: 	private long linesValid=0;
jpayne@68: 	private long bytesProcessed=0;
jpayne@68: 
jpayne@68: 	private long basesProcessed=0;
jpayne@68: 	private long sequencesProcessed=0;
jpayne@68: 	private long filesProcessed=0;
jpayne@68: 	private long filesValid=0;
jpayne@68: 	
jpayne@68: 	private long maxLines=Long.MAX_VALUE;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private final FileFormat ffout1;
jpayne@68: 	
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private PrintStream outstream=System.err;
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	public boolean errorState=false;
jpayne@68: 	private boolean overwrite=false;
jpayne@68: 	private boolean append=false;
jpayne@68: 	
jpayne@68: }