jpayne@68: package sketch;
jpayne@68: 
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.Arrays;
jpayne@68: 
jpayne@68: import fileIO.ByteFile;
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 structures.ByteBuilder;
jpayne@68: import tax.TaxTree;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date May 9, 2016
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class AddSSU {
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Initialization        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/**
jpayne@68: 	 * Code entrance from the command line.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		//Start a timer immediately upon code entrance.
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		
jpayne@68: 		//Create an instance of this class
jpayne@68: 		AddSSU x=new AddSSU(args);
jpayne@68: 		
jpayne@68: 		//Run the object
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: 	/**
jpayne@68: 	 * Constructor.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public AddSSU(String[] args){
jpayne@68: 		
jpayne@68: 		{//Preparse block for help, config files, and outstream
jpayne@68: 			PreParser pp=new PreParser(args, /*getClass()*/null, false);
jpayne@68: 			args=pp.args;
jpayne@68: 			outstream=pp.outstream;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Set shared static variables prior to parsing
jpayne@68: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		
jpayne@68: 		{//Parse the arguments
jpayne@68: 			final Parser parser=parse(args);
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: 		fixExtensions(); //Add or remove .gz or .bz2 as needed
jpayne@68: 		checkFileExistence(); //Ensure files can be read and written
jpayne@68: 		checkStatics(); //Adjust file-related static fields as needed for this program
jpayne@68: 
jpayne@68: 		ffout1=FileFormat.testOutput(out1, FileFormat.SKETCH, null, true, overwrite, append, false);
jpayne@68: 		ffin1=FileFormat.testInput(in1, FileFormat.SKETCH, null, true, false);
jpayne@68: 		
jpayne@68: 		if(verbose){
jpayne@68: 			System.err.println("Set r16SFile="+r16SFile);
jpayne@68: 			System.err.println("Set r18SFile="+r18SFile);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		tree=(treeFile!=null && (preferSSUMapEuks || preferSSUMapProks || clear16SEuks || clear18SEuks || 
jpayne@68: 				clear16SProks || clear18SProks || useSSUMapOnlyEuks || useSSUMapOnlyProks) ? TaxTree.loadTaxTree(treeFile, outstream, false, false) : null);
jpayne@68: 		
jpayne@68: 		if(preferSSUMapEuks || preferSSUMapProks || clear16SEuks || clear18SEuks || clear16SProks || clear18SProks || useSSUMapOnlyEuks || useSSUMapOnlyProks){
jpayne@68: 			assert(tree!=null) : "preferSSUMapForEuks, clear16SEuks, and clear18SEuks require a TaxTree.";
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------    Initialization Helpers    ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Parse arguments from the command line */
jpayne@68: 	private Parser parse(String[] args){
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: 			if(b!=null && b.equalsIgnoreCase("null")){b=null;}
jpayne@68: 
jpayne@68: 			if(a.equalsIgnoreCase("16S") || a.equalsIgnoreCase("16Sfile")){
jpayne@68: 				r16SFile=b;
jpayne@68: 			}else if(a.equalsIgnoreCase("18S") || a.equalsIgnoreCase("18Sfile")){
jpayne@68: 				r18SFile=b;
jpayne@68: 			}else if(a.equalsIgnoreCase("tree") || a.equalsIgnoreCase("treefile")){
jpayne@68: 				treeFile=b;
jpayne@68: 			}else 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: //				ReadWrite.verbose=verbose;
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			else if(a.equalsIgnoreCase("preferSSUMap")){
jpayne@68: 				preferSSUMap=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("preferSSUMapForEuks") || a.equalsIgnoreCase("preferSSUMapEuks")){
jpayne@68: 				preferSSUMapEuks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("useSSUMapOnly")){
jpayne@68: 				useSSUMapOnly=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("useSSUMapOnlyEuks") || a.equalsIgnoreCase("SSUMapOnlyEuks")){
jpayne@68: 				useSSUMapOnlyEuks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("useSSUMapOnlyProks") || a.equalsIgnoreCase("SSUMapOnlyProks")){
jpayne@68: 				useSSUMapOnlyProks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("preferSSUMapForProks") || a.equalsIgnoreCase("preferSSUMapProks")){
jpayne@68: 				preferSSUMapProks=Parse.parseBoolean(b);
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			else if(a.equalsIgnoreCase("clearAll")){
jpayne@68: 				clear16S=clear18S=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear16S")){
jpayne@68: 				clear16S=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear18S")){
jpayne@68: 				clear18S=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear16SEuks")){
jpayne@68: 				clear16SEuks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear18SEuks")){
jpayne@68: 				clear18SEuks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear16SProks")){
jpayne@68: 				clear16SProks=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("clear18SProks")){
jpayne@68: 				clear18SProks=Parse.parseBoolean(b);
jpayne@68: 			}
jpayne@68: 			
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: 		if("auto".equalsIgnoreCase(r16SFile)){r16SFile=TaxTree.default16SFile();}
jpayne@68: 		if("auto".equalsIgnoreCase(r18SFile)){r18SFile=TaxTree.default18SFile();}
jpayne@68: 		SSUMap.r16SFile=r16SFile;
jpayne@68: 		SSUMap.r18SFile=r18SFile;
jpayne@68: 		
jpayne@68: 		return parser;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Add or remove .gz or .bz2 as needed */
jpayne@68: 	private void fixExtensions(){
jpayne@68: 		in1=Tools.fixExtension(in1);
jpayne@68: 		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Ensure files can be read and written */
jpayne@68: 	private void checkFileExistence(){
jpayne@68: 		//Ensure output files can be written
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 file "+out1+"\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		if(!Tools.testInputFiles(false, true, in1, r16SFile, r18SFile)){
jpayne@68: 			throw new RuntimeException("\nCan't read some input files.\n");  
jpayne@68: 		}
jpayne@68: 		assert(in1!=null) : "Input sketch file is required";
jpayne@68: 		assert(r16SFile!=null || r18SFile!=null) : "Input SSU file is required";
jpayne@68: 		
jpayne@68: 		//Ensure that no file was specified multiple times
jpayne@68: 		if(!Tools.testForDuplicateFiles(true, in1, out1, r16SFile, r18SFile)){
jpayne@68: 			throw new RuntimeException("\nSome file names were specified multiple times.\n");
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Adjust file-related static fields as needed for this program */
jpayne@68: 	private static void checkStatics(){
jpayne@68: 		//Adjust the number of threads for input file reading
jpayne@68: 		if(!ByteFile.FORCE_MODE_BF1 && !ByteFile.FORCE_MODE_BF2 && Shared.threads()>2){
jpayne@68: 			ByteFile.FORCE_MODE_BF2=true;
jpayne@68: 		}
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: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Outer Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	void process(Timer t){
jpayne@68: 		
jpayne@68: 		ByteFile bf=ByteFile.makeByteFile(ffin1);
jpayne@68: 		ByteStreamWriter bsw=makeBSW(ffout1);
jpayne@68: 		
jpayne@68: 		processInner(bf, bsw);
jpayne@68: 		
jpayne@68: 		errorState|=bf.close();
jpayne@68: 		if(bsw!=null){errorState|=bsw.poisonAndWait();}
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 
jpayne@68: 		outstream.println(Tools.timeLinesBytesProcessed(t, linesProcessed, bytesProcessed, 8));
jpayne@68: 		outstream.println(Tools.linesBytesOut(linesProcessed, bytesProcessed, linesOut, bytesOut, 8, true));
jpayne@68: 		
jpayne@68: 		outstream.println();
jpayne@68: 		outstream.println(Tools.number("Sketches:", sketchCount, 8));
jpayne@68: 		outstream.println(Tools.number("16S In:", r16Sin, 8));
jpayne@68: 		outstream.println(Tools.number("18S In:", r18Sin, 8));
jpayne@68: 		outstream.println(Tools.number("16S Added:", r16SfromMap, 8));
jpayne@68: 		outstream.println(Tools.number("18S Added:", r18SfromMap, 8));
jpayne@68: 		outstream.println(Tools.numberPercent("16S Out:", r16Sout, r16Sout*100.0/sketchCount, 2, 8));
jpayne@68: 		outstream.println(Tools.numberPercent("18S Out:", r18Sout, r18Sout*100.0/sketchCount, 2, 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: 	/*----------------         Inner Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private static ByteStreamWriter makeBSW(FileFormat ff){
jpayne@68: 		if(ff==null){return null;}
jpayne@68: 		ByteStreamWriter bsw=new ByteStreamWriter(ff);
jpayne@68: 		bsw.start();
jpayne@68: 		return bsw;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: //	private void processInner_old(ByteFile bf, ByteStreamWriter bsw){
jpayne@68: //		SSUMap.load(outstream);
jpayne@68: //		
jpayne@68: //		if(verbose){
jpayne@68: //			System.err.println("Loaded SSUMap; |16S|="+SSUMap.r16SCount()+", |18S|="+SSUMap.r18SCount());
jpayne@68: //		}
jpayne@68: //		
jpayne@68: //		byte[] line=bf.nextLine();
jpayne@68: ////		ByteBuilder bb=new ByteBuilder();
jpayne@68: //
jpayne@68: //		final byte[] ssuBytes="SSU:".getBytes();
jpayne@68: //		final byte[] r16SBytes="16S:".getBytes();
jpayne@68: //		final byte[] r18SBytes="18S:".getBytes();
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+1);
jpayne@68: //
jpayne@68: //				final boolean header=(line[0]=='#');
jpayne@68: //
jpayne@68: //				linesOut++;
jpayne@68: //				bytesOut+=(line.length+1);
jpayne@68: //
jpayne@68: //				if(header){
jpayne@68: //					if(Tools.startsWith(line, "#SZ:")){
jpayne@68: //						sketchCount++;
jpayne@68: //
jpayne@68: //						bsw.print(line);
jpayne@68: //
jpayne@68: //						final int tid=parseTaxID(line);
jpayne@68: //						final boolean has16S=Tools.contains(line, ssuBytes, 0) || Tools.contains(line, r16SBytes, 0);
jpayne@68: //						final boolean has18S=Tools.contains(line, r18SBytes, 0);
jpayne@68: //
jpayne@68: //						if(verbose){
jpayne@68: //							System.err.println("For line "+new String(line)+":");
jpayne@68: //							System.err.println("tid="+tid+", has16S="+has16S+", has18S="+has18S);
jpayne@68: //						}
jpayne@68: //
jpayne@68: //						if(tid>0){
jpayne@68: //							final byte[] r16S=has16S ? null : SSUMap.r16SMap.get(tid);
jpayne@68: //							final byte[] r18S=has18S ? null : SSUMap.r18SMap.get(tid);
jpayne@68: //							if(r16S!=null){bsw.print("\t16S:").print(r16S.length); ssuOut++;}
jpayne@68: //							if(r18S!=null){bsw.print("\t18S:").print(r18S.length); ssuOut++;}
jpayne@68: //							if(r16S!=null){bsw.print("\n#16S:").print(r16S);}
jpayne@68: //							if(r18S!=null){bsw.print("\n#18S:").print(r18S);}
jpayne@68: //
jpayne@68: //							if(verbose){System.err.println("Found 16S: "+(r16S!=null)+"; found 18S: "+(r18S!=null));}
jpayne@68: //						}
jpayne@68: //						bsw.println();
jpayne@68: //					}else if(Tools.startsWith(line, "#16S:") || Tools.startsWith(line, "#18S:") || Tools.startsWith(line, "#SSU:")){
jpayne@68: //						bsw.println(line);
jpayne@68: //						ssuIn++;
jpayne@68: //						ssuOut++;
jpayne@68: //					}else{
jpayne@68: //						assert(Tools.startsWith(line, "##")) : new String(line);
jpayne@68: //						bsw.println(line);
jpayne@68: //					}
jpayne@68: //				}else{
jpayne@68: //					bsw.println(line);
jpayne@68: //				}
jpayne@68: //			}
jpayne@68: //			line=bf.nextLine();
jpayne@68: //		}
jpayne@68: //	}
jpayne@68: 	
jpayne@68: 	private void processInner(ByteFile bf, ByteStreamWriter bsw){
jpayne@68: 		SSUMap.load(outstream);
jpayne@68: 		
jpayne@68: 		if(verbose){
jpayne@68: 			System.err.println("Loaded SSUMap; |16S|="+SSUMap.r16SCount()+", |18S|="+SSUMap.r18SCount());
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		byte[] line=bf.nextLine();
jpayne@68: //		ByteBuilder bb=new ByteBuilder();
jpayne@68: 
jpayne@68: //		final byte[] ssuBytes="SSU:".getBytes();
jpayne@68: //		final byte[] r16SBytes="16S:".getBytes();
jpayne@68: //		final byte[] r18SBytes="18S:".getBytes();
jpayne@68: 
jpayne@68: 		SketchHeader header=null;
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+1);
jpayne@68: 
jpayne@68: 				final boolean isHeader=(line[0]=='#');
jpayne@68: 
jpayne@68: 				if(isHeader){
jpayne@68: 					if(Tools.startsWith(line, "#SZ:")){
jpayne@68: 						assert(header==null) : "\nReplacing this:\n"+header.toBytes()+"\nWith this:\n"+new String(line)+"\n";
jpayne@68: 						header=new SketchHeader(line);
jpayne@68: 						sketchCount++;
jpayne@68: 					}else if(Tools.startsWith(line, "##")){
jpayne@68: 						bsw.println(line);
jpayne@68: 
jpayne@68: 						linesOut++;
jpayne@68: 						bytesOut+=(line.length+1);
jpayne@68: 					}else{
jpayne@68: 						header.addLine(line);
jpayne@68: 					}
jpayne@68: 				}else{
jpayne@68: 					if(header!=null){
jpayne@68: 						try {
jpayne@68: 							processHeader(header);
jpayne@68: 						} catch (Throwable e) {
jpayne@68: 							e.printStackTrace();
jpayne@68: 							assert(false) : header.toBytes();
jpayne@68: 						}
jpayne@68: 						r16Sout+=(header.r16S==null ? 0 : 1);
jpayne@68: 						r18Sout+=(header.r18S==null ? 0 : 1);
jpayne@68: 						linesOut+=1+(header.r16S==null ? 0 : 1)+(header.r18S==null ? 0 : 1);
jpayne@68: 						ByteBuilder bb=header.toBytes();
jpayne@68: 						bytesOut+=(bb.length+1);
jpayne@68: 						bsw.println(bb);
jpayne@68: 						header=null;
jpayne@68: 					}
jpayne@68: 					bsw.println(line);
jpayne@68: 
jpayne@68: 					linesOut++;
jpayne@68: 					bytesOut+=(line.length+1);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			line=bf.nextLine();
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void processHeader(SketchHeader header){
jpayne@68: 		
jpayne@68: 		if(verbose){System.err.println("Processing tid "+header.tid+":\n"+header.toBytes()+"\n");}
jpayne@68: 
jpayne@68: 		final boolean euk=(tree!=null && header.tid>0 && header.tid<SketchObject.minFakeID) ? tree.isEukaryote(header.tid) : false;
jpayne@68: 		final boolean prok=(tree!=null && header.tid>0 && header.tid<SketchObject.minFakeID) ? tree.isProkaryote(header.tid) : false;
jpayne@68: 		if(useSSUMapOnly || (useSSUMapOnlyEuks && euk) || (useSSUMapOnlyProks && prok)){header.r16S=header.r18S=null;}
jpayne@68: 		if(header.tid>0){
jpayne@68: 			final boolean preferMap=(preferSSUMap || (preferSSUMapEuks && euk) || (preferSSUMapProks && prok));
jpayne@68: 			byte[] r16S=(SSUMap.r16SMap==null ? null : SSUMap.r16SMap.get(header.tid));
jpayne@68: 			byte[] r18S=(SSUMap.r18SMap==null ? null : SSUMap.r18SMap.get(header.tid));
jpayne@68: 			if(r16S!=null && (preferMap || header.r16S==null)){
jpayne@68: 				header.r16S=r16S;
jpayne@68: 				r16SfromMap++;
jpayne@68: 			}
jpayne@68: 			if(r18S!=null && (preferMap || header.r18S==null)){
jpayne@68: 				header.r18S=r18S;
jpayne@68: 				r18SfromMap++;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		if(clear16S || (clear16SEuks && euk) || (clear16SProks && prok)){header.r16S=null;}
jpayne@68: 		if(clear18S || (clear18SEuks && euk) || (clear18SProks && prok)){header.r18S=null;}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	int parseTaxID(byte[] line){
jpayne@68: 		String[] split=Tools.tabPattern.split(new String(line));
jpayne@68: 		for(String s : split){
jpayne@68: 			if(s.startsWith("ID:") || s.startsWith("TAXID:")){
jpayne@68: 				final int colon=s.indexOf(':');
jpayne@68: 				final String sub=s.substring(colon+1);
jpayne@68: 				return Integer.parseInt(sub);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		return -1;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	//A very limited parser
jpayne@68: 	private class SketchHeader {
jpayne@68: 		
jpayne@68: 		SketchHeader(byte[] line){
jpayne@68: 			this(new String(line, 1, line.length-1));
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		SketchHeader(String line){
jpayne@68: 			if(line.charAt(0)=='#'){line=line.substring(1);}
jpayne@68: 			assert(line.startsWith("SZ:"));
jpayne@68: 			String[] split=Tools.tabPattern.split(line);
jpayne@68: 			fields=new ArrayList<String>(line.length()+2);
jpayne@68: 			int tid_=-1;
jpayne@68: 			for(String s : split){
jpayne@68: 				if(s.startsWith("16S:") || s.startsWith("18S:") || s.startsWith("SSU:")){
jpayne@68: 					//do nothing
jpayne@68: 				}else{
jpayne@68: 					if(s.startsWith("ID:") || s.startsWith("TAXID:")){
jpayne@68: 						final int colon=s.indexOf(':');
jpayne@68: 						final String sub=s.substring(colon+1);
jpayne@68: 						tid_=Integer.parseInt(sub);
jpayne@68: 					}
jpayne@68: 					fields.add(s);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			tid=tid_;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		void addLine(byte[] line){
jpayne@68: 			assert(line[0]=='#');
jpayne@68: 			assert(line[1]=='1' || line[1]=='S') : new String(line);
jpayne@68: 			if(Tools.startsWith(line, "#16S:") || Tools.startsWith(line, "#SSU:")){
jpayne@68: 				assert(r16S==null);
jpayne@68: 				r16S=Arrays.copyOfRange(line, 5, line.length);
jpayne@68: 				r16Sin++;
jpayne@68: 			}else if(Tools.startsWith(line, "#18S:")){
jpayne@68: 				assert(r18S==null);
jpayne@68: 				r18S=Arrays.copyOfRange(line, 5, line.length);
jpayne@68: 				r18Sin++;
jpayne@68: 			}else{
jpayne@68: 				assert(false) : new String(line);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ByteBuilder toBytes(){
jpayne@68: 			ByteBuilder bb=new ByteBuilder(1000);
jpayne@68: 			bb.append('#');
jpayne@68: 			for(int i=0; i<fields.size(); i++){
jpayne@68: 				if(i>0){bb.tab();}
jpayne@68: 				bb.append(fields.get(i));
jpayne@68: 			}
jpayne@68: 			if(r16S!=null){bb.tab().append("16S:").append(r16S.length);}
jpayne@68: 			if(r18S!=null){bb.tab().append("18S:").append(r18S.length);}
jpayne@68: 			
jpayne@68: 			if(r16S!=null){bb.nl().append("#16S:").append(r16S);}
jpayne@68: 			if(r18S!=null){bb.nl().append("#18S:").append(r18S);}
jpayne@68: 			return bb;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		final int tid;
jpayne@68: 		ArrayList<String> fields;
jpayne@68: 		byte[] r16S;
jpayne@68: 		byte[] r18S;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------            Fields            ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private String in1=null;
jpayne@68: 	private String out1=null;
jpayne@68: 	private String r16SFile="auto";
jpayne@68: 	private String r18SFile="auto";
jpayne@68: 	private String treeFile="auto";
jpayne@68: 
jpayne@68: 	boolean preferSSUMap=false;
jpayne@68: 	boolean preferSSUMapEuks=false;
jpayne@68: 	boolean preferSSUMapProks=false;
jpayne@68: 	boolean useSSUMapOnly=false;
jpayne@68: 	boolean useSSUMapOnlyEuks=false;
jpayne@68: 	boolean useSSUMapOnlyProks=false;
jpayne@68: 	boolean clear16S=false;
jpayne@68: 	boolean clear18S=false;
jpayne@68: 	boolean clear16SEuks=false;
jpayne@68: 	boolean clear18SEuks=false;
jpayne@68: 	boolean clear16SProks=false;
jpayne@68: 	boolean clear18SProks=false;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private long linesProcessed=0;
jpayne@68: 	private long linesOut=0;
jpayne@68: 	private long bytesProcessed=0;
jpayne@68: 	private long bytesOut=0;
jpayne@68: 	
jpayne@68: 	private long sketchCount=0;
jpayne@68: 	
jpayne@68: 	private long r16Sin=0;
jpayne@68: 	private long r16Sout=0;
jpayne@68: 	private long r16SfromMap=0;
jpayne@68: 	private long r18Sin=0;
jpayne@68: 	private long r18Sout=0;
jpayne@68: 	private long r18SfromMap=0;
jpayne@68: 	
jpayne@68: 	private long maxLines=Long.MAX_VALUE;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Final Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private final FileFormat ffin1;
jpayne@68: 	private final FileFormat ffout1;
jpayne@68: 	
jpayne@68: 	private final TaxTree tree;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Common Fields         ----------------*/
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: }