jpayne@68: package sketch;
jpayne@68: 
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.HashMap;
jpayne@68: 
jpayne@68: import fileIO.ByteStreamWriter;
jpayne@68: import fileIO.ReadWrite;
jpayne@68: import shared.Parse;
jpayne@68: import shared.Tools;
jpayne@68: import structures.FloatList;
jpayne@68: import tax.TaxNode;
jpayne@68: import tax.TaxTree;
jpayne@68: 
jpayne@68: class ResultLineParser {
jpayne@68: 
jpayne@68: 	ResultLineParser(int mode_, TaxTree tree_, ByteStreamWriter bswBad_, ArrayList<RecordSet> recordSets_, boolean keepText_){
jpayne@68: 		mode=mode_;
jpayne@68: 		tree=tree_;
jpayne@68: 		bswBad=bswBad_;
jpayne@68: 		recordSets=recordSets_;
jpayne@68: 		keepText=keepText_ || bswBad!=null;
jpayne@68: 		for(int i=0; i<AnalyzeSketchResults.taxLevels; i++){
jpayne@68: 			aniLists[i]=new FloatList();
jpayne@68: 			ssuLists[i]=new FloatList();
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	void parse(byte[] line){
jpayne@68: 		if(keepText){text=line;}
jpayne@68: 		if(line[0]!='#'){
jpayne@68: 			if(mode==AnalyzeSketchResults.BBSKETCH_MODE){
jpayne@68: 				parseData(line);
jpayne@68: 			}else if(mode==AnalyzeSketchResults.MASH_MODE){
jpayne@68: 				parseDataMash(line);
jpayne@68: 			}else{
jpayne@68: 				assert(false) : "Bad mode: "+mode;
jpayne@68: 			}
jpayne@68: 		}else{
jpayne@68: 			parseHeader(line);
jpayne@68: 			if(bswBad!=null){bswBad.println(line);}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	private synchronized void parseHeader(byte[] line){
jpayne@68: 		ArrayList<byte[]> split=Tools.split(line, 0, (byte)'\t');
jpayne@68: 		for(int col=0; col<split.size(); col++){
jpayne@68: 			byte[] array=split.get(col);
jpayne@68: 			if(Tools.equals(array, "ANI") || Tools.equals(array, "AAI")){
jpayne@68: 				aniColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "QTaxID")){
jpayne@68: 				qTaxIDColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "RTaxID")){
jpayne@68: 				rTaxIDColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "SSU")){
jpayne@68: 				ssuColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "CALevel")){
jpayne@68: 				caLevelColumn=col;
jpayne@68: 			}
jpayne@68: 
jpayne@68: 			else if(Tools.equals(array, "QSize")){
jpayne@68: 				qSizeColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "RefSize") || Tools.equals(array, "RSize")){
jpayne@68: 				rSizeColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "QBases")){
jpayne@68: 				qBasesColumn=col;
jpayne@68: 			}else if(Tools.equals(array, "RBases")){
jpayne@68: 				rBasesColumn=col;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	private void parseData(byte[] line){
jpayne@68: 		ArrayList<byte[]> split=Tools.split(line, 0, (byte)'\t');
jpayne@68: 		qTaxID=Parse.parseInt(split.get(qTaxIDColumn), 0);
jpayne@68: 		rTaxID=Parse.parseInt(split.get(rTaxIDColumn), 0);
jpayne@68: 		qBases=Parse.parseLong(split.get(qBasesColumn), 0);
jpayne@68: 		rBases=Parse.parseLong(split.get(rBasesColumn), 0);
jpayne@68: 		qSize=Parse.parseLong(split.get(qSizeColumn), 0);
jpayne@68: 		rSize=Parse.parseLong(split.get(rSizeColumn), 0);
jpayne@68: 		ani=Parse.parseDouble(split.get(aniColumn), 0);
jpayne@68: 		byte[] ssuArray=split.get(ssuColumn);
jpayne@68: 		ssu=ssuArray.length==1 && ssuArray[0]=='.' ? -1 : Parse.parseDouble(ssuArray, 0);
jpayne@68: 		taxLevelExtended=TaxTree.stringToLevelExtended(new String(split.get(caLevelColumn)));
jpayne@68: 		if(taxLevelExtended<0) {
jpayne@68: 			System.err.println(new String(split.get(caLevelColumn)));
jpayne@68: 			taxLevelExtended=0;
jpayne@68: 		}
jpayne@68: 		processed=false;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private TaxNode getTaxNode(String fname){
jpayne@68: 		String name=ReadWrite.stripToCore(fname);
jpayne@68: 		if(name.startsWith("tid_")){
jpayne@68: 			int idx2=fname.indexOf('_', 4);
jpayne@68: 			int x=Parse.parseInt(fname, 4, idx2);
jpayne@68: 			return x>0 ? tree.getNode(x) : null;
jpayne@68: 			//name=name.substring(idx2+1); //This would allow fall-through to name parsing
jpayne@68: 		}
jpayne@68: 		try {
jpayne@68: 			return tree.getNodeByName(name);
jpayne@68: 		} catch (Throwable e) {
jpayne@68: 			return null;
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	private void parseDataMash(byte[] line){
jpayne@68: 		///dev/shm/tid_123_Zymomonas_mobilis.fna.gz	/dev/shm/tid_456_bacterium_endosymbiont_of_Bathymodiolus_sp._5_South.fna.gz	0.43859	0.00515848	1/20000
jpayne@68: 
jpayne@68: 		String[] split=new String(line).split("\t");
jpayne@68: 
jpayne@68: 		String fraction=split[split.length-1];
jpayne@68: 		int numerator=Integer.parseInt(fraction.split("/")[0]);
jpayne@68: 		if(numerator<MIN_HITS){return;}
jpayne@68: 		int denominator=Integer.parseInt(fraction.split("/")[1]);
jpayne@68: 
jpayne@68: 		//The default ordering is reversed since mash output is ordered first by ref, then query
jpayne@68: 		//The normal ordering (as below) requires a linux sort
jpayne@68: 		{
jpayne@68: 			TaxNode qNode=getTaxNode(split[0]);
jpayne@68: 			TaxNode rNode=getTaxNode(split[1]);
jpayne@68: 
jpayne@68: 			if(qNode==null || rNode==null){return;}
jpayne@68: 			qTaxID=qNode.id;
jpayne@68: 			rTaxID=rNode.id;
jpayne@68: 			TaxNode ancestor=tree.commonAncestor(qNode, rNode);
jpayne@68: 			taxLevelExtended=ancestor.levelExtended;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ani=numerator/(float)denominator;
jpayne@68: 		ssu=-1;
jpayne@68: 		if(taxLevelExtended<0){taxLevelExtended=0;}
jpayne@68: 		processed=false;
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	//Returns a complete set when a new set is started
jpayne@68: 	RecordSet processData(HashMap<Long, Float> map, boolean saveRecord){
jpayne@68: 		RecordSet old=null;
jpayne@68: 		if(processed){return null;}
jpayne@68: 		levelAniSums[taxLevelExtended]+=ani;
jpayne@68: 		levelCounts[taxLevelExtended]++;
jpayne@68: 		aniLists[taxLevelExtended].add((float)ani);
jpayne@68: 
jpayne@68: 		if(ssu>0){
jpayne@68: 			levelSSUSums[taxLevelExtended]+=ssu;
jpayne@68: 			levelCountsSSU[taxLevelExtended]++;
jpayne@68: 			ssuLists[taxLevelExtended].add((float)ssu);
jpayne@68: 		}
jpayne@68: 		if(map!=null){
jpayne@68: 			long key=(((long)qTaxID)<<32)|rTaxID;
jpayne@68: 			map.put(key, (float)ani);
jpayne@68: 		}
jpayne@68: 		if(saveRecord){
jpayne@68: 			if(currentSet==null || currentSet.qID!=qTaxID){
jpayne@68: 				old=currentSet;
jpayne@68: 				currentSet=new RecordSet(qTaxID);
jpayne@68: 				if(recordSets!=null){
jpayne@68: 					recordSets.add(currentSet);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			currentSet.records.add(new Record(this));
jpayne@68: 		}
jpayne@68: 		processed=true;
jpayne@68: 		return old;
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	//		final static int taxLevels=TaxTree.numTaxaNamesExtended;
jpayne@68: 	final long[] levelCounts=new long[AnalyzeSketchResults.taxLevels];
jpayne@68: 	final long[] levelCountsSSU=new long[AnalyzeSketchResults.taxLevels];
jpayne@68: 
jpayne@68: 	final double[] levelAniSums=new double[AnalyzeSketchResults.taxLevels];
jpayne@68: 	final double[] levelSSUSums=new double[AnalyzeSketchResults.taxLevels];
jpayne@68: 
jpayne@68: 	final FloatList[] aniLists=new FloatList[AnalyzeSketchResults.taxLevels];
jpayne@68: 	final FloatList[] ssuLists=new FloatList[AnalyzeSketchResults.taxLevels];
jpayne@68: 
jpayne@68: 	final ArrayList<RecordSet> recordSets;
jpayne@68: 
jpayne@68: 	final int mode;
jpayne@68: 	final TaxTree tree;
jpayne@68: 	final ByteStreamWriter bswBad;
jpayne@68: 
jpayne@68: 	int qTaxID=-1;
jpayne@68: 	int rTaxID=-1;
jpayne@68: 	long qBases;
jpayne@68: 	long rBases;
jpayne@68: 	long qSize;
jpayne@68: 	long rSize;
jpayne@68: 	double ani=-1;
jpayne@68: 	double ssu=-1;
jpayne@68: 	int taxLevelExtended=-1;
jpayne@68: 	boolean processed=true;
jpayne@68: 	RecordSet currentSet=null;
jpayne@68: 	final boolean keepText;
jpayne@68: 
jpayne@68: 	byte[] text=null;
jpayne@68: 
jpayne@68: 	private static int qTaxIDColumn=7;
jpayne@68: 	private static int rTaxIDColumn=8;
jpayne@68: 	private static int qSizeColumn=3;
jpayne@68: 	private static int rSizeColumn=4;
jpayne@68: 	private static int qBasesColumn=5;
jpayne@68: 	private static int rBasesColumn=6;
jpayne@68: 	private static int aniColumn=2;
jpayne@68: 	private static int ssuColumn=11;
jpayne@68: 	private static int caLevelColumn=12;
jpayne@68: 
jpayne@68: 	static int MIN_HITS=3;
jpayne@68: 	
jpayne@68: }