jpayne@68: package prok;
jpayne@68: 
jpayne@68: import java.io.File;
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.ArrayList;
jpayne@68: 
jpayne@68: import dna.Data;
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: 
jpayne@68: /**
jpayne@68:  * Static helpers for manipulating pgm files.
jpayne@68:  * main() merges pgm files.
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date Sep 24, 2018
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class PGMTools extends ProkObject {
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------             Main             ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Combines multiple pgm files into a single file */
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		
jpayne@68: 		//Start a timer immediately upon code entrance.
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		
jpayne@68: 		{//Preparse block for help, config files, and outstream
jpayne@68: 			PreParser pp=new PreParser(args, new Object() { }.getClass().getEnclosingClass(), false);
jpayne@68: 			args=pp.args;
jpayne@68: 			outstream=pp.outstream;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ProkObject.call18S=true;
jpayne@68: 		boolean overwrite=true;
jpayne@68: 		boolean allowDupes=false;
jpayne@68: 		String out=null;
jpayne@68: 		ArrayList<String> in=new ArrayList<String>(); 
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: 				if(b!=null && b.equalsIgnoreCase("null")){b=null;}
jpayne@68: 				
jpayne@68: 				if(a.equals("in")){
jpayne@68: 					assert(b!=null);
jpayne@68: 					Tools.addFiles(b, in);
jpayne@68: 				}else if(parseStatic(arg, a, b)){
jpayne@68: 					//do nothing
jpayne@68: 				}else if(a.equals("allowdupes") || a.equals("allowduplicates") || a.equals("dupes")){
jpayne@68: 					allowDupes=Parse.parseBoolean(b);
jpayne@68: 				}else if(a.equals("verbose")){
jpayne@68: 					verbose=Parse.parseBoolean(b);
jpayne@68: 					ReadWrite.verbose=verbose;
jpayne@68: 				}
jpayne@68: 
jpayne@68: 				else if(parser.parse(arg, a, b)){
jpayne@68: 					//do nothing
jpayne@68: 				}else if(b==null && new File(arg.split("@")[0]).exists()){
jpayne@68: 					in.add(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: 			overwrite=parser.overwrite;
jpayne@68: 			out=parser.out1;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		checkFileExistence(in, out, overwrite, allowDupes);
jpayne@68: 		
jpayne@68: 		ArrayList<GeneModel> models=loadModels(in);
jpayne@68: 		GeneModel gm=mergeModels(models);
jpayne@68: 		boolean errorState=writeModel(gm, out, overwrite);
jpayne@68: 		
jpayne@68: 		//Close the print stream if it was redirected
jpayne@68: 		Shared.closeStream(outstream);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static boolean parseStatic(String arg, String a, String b){
jpayne@68: 		if(a.equals("kinnercds")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			GeneModel.setInnerK(k);
jpayne@68: 		}else if(a.equals("kstartcds")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStartK(k);
jpayne@68: 		}else if(a.equals("kstopcds")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStopK(k);
jpayne@68: 		}else if(a.equals("kinnerrna")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			kInnerRNA=k;
jpayne@68: 		}else if(a.equals("kstartrna")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			kStartRNA=k;
jpayne@68: 		}else if(a.equals("kstoprna")){
jpayne@68: 			int k=Integer.parseInt(b);
jpayne@68: 			kStopRNA=k;
jpayne@68: 		}else if(a.equals("startleftoffset")){
jpayne@68: 			int x=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStartLeftOffset(x);
jpayne@68: 		}else if(a.equals("startrightoffset")){
jpayne@68: 			int x=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStartRightOffset(x);
jpayne@68: 		}else if(a.equals("stopleftoffset")){
jpayne@68: 			int x=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStopLeftOffset(x);
jpayne@68: 		}else if(a.equals("stoprightoffset")){
jpayne@68: 			int x=Integer.parseInt(b);
jpayne@68: 			GeneModel.setStopRightOffset(x);
jpayne@68: 		}else if(a.equalsIgnoreCase("callcdsonly") || a.equalsIgnoreCase("cdsonly")){
jpayne@68: 			callCDS=Parse.parseBoolean(b);
jpayne@68: 			calltRNA=call16S=call23S=call5S=call18S=!callCDS;
jpayne@68: 		}else if(a.equalsIgnoreCase("call16sonly") || a.equalsIgnoreCase("16sonly")){
jpayne@68: 			call16S=Parse.parseBoolean(b);
jpayne@68: 			calltRNA=call23S=call5S=call18S=callCDS=!call16S;
jpayne@68: 		}else if(a.equalsIgnoreCase("call23sonly") || a.equalsIgnoreCase("23sonly")){
jpayne@68: 			call23S=Parse.parseBoolean(b);
jpayne@68: 			calltRNA=call16S=call5S=call18S=callCDS=!call23S;
jpayne@68: 		}else if(a.equalsIgnoreCase("call5sonly") || a.equalsIgnoreCase("5sonly")){
jpayne@68: 			call5S=Parse.parseBoolean(b);
jpayne@68: 			calltRNA=call16S=call23S=call18S=callCDS=!call5S;
jpayne@68: 		}else if(a.equalsIgnoreCase("calltrnaonly") || a.equalsIgnoreCase("trnaonly")){
jpayne@68: 			calltRNA=Parse.parseBoolean(b);
jpayne@68: 			call16S=call23S=call5S=call18S=callCDS=!calltRNA;
jpayne@68: 		}else if(a.equalsIgnoreCase("call18sonly") || a.equalsIgnoreCase("18sonly")){
jpayne@68: 			call18S=Parse.parseBoolean(b);
jpayne@68: 			calltRNA=call16S=call23S=call5S=callCDS=!call18S;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		else if(a.equalsIgnoreCase("callcds") || a.equalsIgnoreCase("cds")){
jpayne@68: 			callCDS=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("calltrna") || a.equalsIgnoreCase("trna")){
jpayne@68: 			calltRNA=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("call16s") || a.equalsIgnoreCase("16s")){
jpayne@68: 			call16S=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("call18s") || a.equalsIgnoreCase("18s")){
jpayne@68: 			call18S=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("call23s") || a.equalsIgnoreCase("23s")){
jpayne@68: 			call23S=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("call5s") || a.equalsIgnoreCase("5s")){
jpayne@68: 			call5S=Parse.parseBoolean(b);
jpayne@68: 		}else if(a.equalsIgnoreCase("callrna") || a.equalsIgnoreCase("rna")){
jpayne@68: 			calltRNA=call16S=call18S=call5S=call23S=Parse.parseBoolean(b);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		else if(a.equalsIgnoreCase("normalize")){
jpayne@68: 			normalize=Parse.parseBoolean(b);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		else{
jpayne@68: 			return false;
jpayne@68: 		}
jpayne@68: 		return true;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static ArrayList<GeneModel> loadModels(ArrayList<String> fnames){
jpayne@68: 		ArrayList<GeneModel> models=new ArrayList<GeneModel>(fnames.size());
jpayne@68: 		ArrayList<Double> mults=new ArrayList<Double>(fnames.size());
jpayne@68: 		for(String s : fnames){
jpayne@68: 			double mult=1;
jpayne@68: 			String fname=s;
jpayne@68: 			if(s.indexOf('@')>=0){
jpayne@68: 				String[] split=s.split("@");
jpayne@68: 				fname=split[0];
jpayne@68: 				mult=Double.parseDouble(split[1]);
jpayne@68: 			}
jpayne@68: 			GeneModel pgm=GeneModelParser.loadModel(fname);
jpayne@68: 			mults.add(mult);
jpayne@68: 			models.add(pgm);
jpayne@68: 		}
jpayne@68: 		if(normalize){
jpayne@68: 			long max=0;
jpayne@68: 			for(GeneModel gm : models){
jpayne@68: 				max=Tools.max(gm.basesProcessed, max);
jpayne@68: 			}
jpayne@68: 			for(GeneModel gm : models){
jpayne@68: 				if(max!=gm.basesProcessed){
jpayne@68: 					double mult=max/(double)(Tools.max(100, gm.basesProcessed));
jpayne@68: 					gm.multiplyBy(mult);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		for(int i=0; i<models.size(); i++){
jpayne@68: 			double mult=mults.get(i);
jpayne@68: 			GeneModel gm=models.get(i);
jpayne@68: 			if(mult!=1){gm.multiplyBy(mult);}
jpayne@68: 		}
jpayne@68: 		return models;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static GeneModel mergeModels(ArrayList<GeneModel> models){
jpayne@68: 		if(models.size()==1){return models.get(0);}
jpayne@68: 		GeneModel pgmSum=new GeneModel(true);
jpayne@68: 		for(GeneModel pgm : models){
jpayne@68: 			pgmSum.add(pgm);
jpayne@68: 		}
jpayne@68: 		return pgmSum;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static GeneModel loadAndMerge(ArrayList<String> in) {
jpayne@68: 		ArrayList<GeneModel> models=loadModels(in);
jpayne@68: 		return mergeModels(models);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static boolean writeModel(GeneModel pgm, String out, boolean overwrite){
jpayne@68: 		FileFormat ffout=FileFormat.testOutput(out, FileFormat.PGM, null, true, overwrite, false, false);
jpayne@68: 		return writeModel(pgm, ffout);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static boolean writeModel(GeneModel pgm, FileFormat ffout){
jpayne@68: 		ByteStreamWriter bsw=ByteStreamWriter.makeBSW(ffout);
jpayne@68: 
jpayne@68: 		ByteBuilder bb=new ByteBuilder();
jpayne@68: 		pgm.appendTo(bb);
jpayne@68: 
jpayne@68: 		boolean errorState=false;
jpayne@68: 		if(bsw!=null){
jpayne@68: 			bsw.addJob(bb);
jpayne@68: 			errorState|=bsw.poisonAndWait();
jpayne@68: 		}
jpayne@68: 		return errorState;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Ensure files can be read and written */
jpayne@68: 	private static void checkFileExistence(ArrayList<String> in, String out, boolean overwrite, boolean allowDupes){
jpayne@68: 		//Ensure output files can be written
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, false, false, out)){
jpayne@68: 			outstream.println((out==null)+", "+out);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output file "+out+"\n");
jpayne@68: 		}
jpayne@68: 		ArrayList<String> in2=new ArrayList<String>();
jpayne@68: 		for(String s : in){
jpayne@68: 			in2.add(s.split("@")[0]);
jpayne@68: 		}
jpayne@68: 		in=null;
jpayne@68: 		
jpayne@68: 		for(int i=0; i<in2.size(); i++){
jpayne@68: 			String s=in2.get(i);
jpayne@68: 			if(s.equalsIgnoreCase("auto") || s.equalsIgnoreCase("default")){
jpayne@68: 				in2.set(i, Data.findPath("?model.pgm"));
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		ArrayList<String> foo=new ArrayList<String>();
jpayne@68: 		foo.addAll(in2);
jpayne@68: 		if(!Tools.testInputFiles(allowDupes, true, foo.toArray(new String[0]))){
jpayne@68: 			throw new RuntimeException("\nCan't read some input files.\n");  
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure that no file was specified multiple times
jpayne@68: 		if(!allowDupes){
jpayne@68: 			foo.add(out);
jpayne@68: 			if(!Tools.testForDuplicateFiles(true, foo.toArray(new String[0]))){
jpayne@68: 				throw new RuntimeException("\nSome file names were specified multiple times.\n");
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private static PrintStream outstream=System.err;
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	/** Mix models equally */
jpayne@68: 	public static boolean normalize=false;
jpayne@68: 	
jpayne@68: }