jpayne@68: package tax;
jpayne@68: 
jpayne@68: import java.io.File;
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.LinkedHashMap;
jpayne@68: import java.util.Locale;
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 fileIO.TextStreamWriter;
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.FastaReadInputStream;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Constructs a directory and file tree of sequences
jpayne@68:  * corresponding to a taxonomic tree.
jpayne@68:  * 
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date December 12, 2017
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class ExplodeTree {
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: 		Timer t=new Timer();
jpayne@68: 		ExplodeTree x=new ExplodeTree(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: 	/**
jpayne@68: 	 * Constructor.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public ExplodeTree(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: 		//Set shared static variables
jpayne@68: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		
jpayne@68: 		//Create a parser object
jpayne@68: 		Parser parser=new Parser();
jpayne@68: 		
jpayne@68: 		//Parse each argument
jpayne@68: 		for(int i=0; i<args.length; i++){
jpayne@68: 			String arg=args[i];
jpayne@68: 			
jpayne@68: 			//Break arguments into their constituent parts, in the form of "a=b"
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("verbose")){
jpayne@68: 				verbose=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("out") || a.equals("path") || a.equals("outpath")){
jpayne@68: 				outPath=b;
jpayne@68: 			}else if(a.equals("prefix")){
jpayne@68: 				prefix=b;
jpayne@68: 			}else if(a.equals("results") || a.equals("result")){
jpayne@68: 				resultsFile=b;
jpayne@68: 			}else if(a.equals("makedirectories") || a.equals("mkdirs") || a.equals("mkdir")){
jpayne@68: 				makeDirectories=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("tree") || a.equals("taxtree")){
jpayne@68: 				taxTreeFile=b;
jpayne@68: 			}else if(parser.parse(arg, a, b)){//Parse standard flags in the parser
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(prefix==null){prefix="";}
jpayne@68: 		if("auto".equalsIgnoreCase(taxTreeFile)){taxTreeFile=TaxTree.defaultTreeFile();}
jpayne@68: 		
jpayne@68: 		{//Process parser fields
jpayne@68: 			Parser.processQuality();
jpayne@68: 			
jpayne@68: 			maxReads=parser.maxReads;
jpayne@68: 			
jpayne@68: 			overwrite=parser.overwrite;
jpayne@68: 			
jpayne@68: 			in1=parser.in1;
jpayne@68: 			
jpayne@68: 			extin=parser.extin;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		if(outPath==null || outPath.trim().length()==0){outPath="";}
jpayne@68: 		else{
jpayne@68: 			outPath=outPath.trim().replace('\\', '/').replaceAll("/+", "/");
jpayne@68: 			if(!outPath.endsWith("/")){outPath=outPath+"/";}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		assert(FastaReadInputStream.settingsOK());
jpayne@68: 		
jpayne@68: 		//Ensure there is an input file
jpayne@68: 		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
jpayne@68: 		
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: 		//Ensure output files can be written
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, false, false, resultsFile)){
jpayne@68: 			outstream.println(resultsFile);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+resultsFile+"\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		if(!Tools.testInputFiles(false, true, in1)){
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(!Tools.testForDuplicateFiles(true, in1, resultsFile)){
jpayne@68: 			throw new RuntimeException("\nSome file names were specified multiple times.\n");
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		//Create input FileFormat objects
jpayne@68: 		ffin1=FileFormat.testInput(in1, FileFormat.FASTA, extin, true, true);
jpayne@68: 		
jpayne@68: 		tree=TaxTree.loadTaxTree(taxTreeFile, outstream, true, false);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Outer Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	public void makeDirectoryTree(String root, boolean writeNames){
jpayne@68: 		for(TaxNode node : tree.nodes){
jpayne@68: 			if(node!=null){
jpayne@68: 				String dir=tree.toDir(node, root);
jpayne@68: 				File df=new File(dir);
jpayne@68: 				if(!df.exists()){df.mkdirs();}
jpayne@68: 				if(writeNames){
jpayne@68: 					try {
jpayne@68: 						String fname=node.simpleName()+".name";
jpayne@68: 						File nf=new File(fname);
jpayne@68: 						if(!nf.exists()){
jpayne@68: 							ReadWrite.writeString(node.name, dir+fname);
jpayne@68: 						}
jpayne@68: 					} catch (Exception e) {
jpayne@68: 						// TODO Auto-generated catch block
jpayne@68: 						e.printStackTrace();
jpayne@68: 					}
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Create read streams and process all data */
jpayne@68: 	public void process(Timer t){
jpayne@68: 		
jpayne@68: 		Timer t2=new Timer();
jpayne@68: 		if(makeDirectories){
jpayne@68: 			makeDirectoryTree(outPath, true);
jpayne@68: 			t2.stop("Finished making directories. ");
jpayne@68: 			t2.start();
jpayne@68: 		}
jpayne@68: 		processInner();
jpayne@68: 		t2.stop();
jpayne@68: 		t2.stop("Finished writing data. ");
jpayne@68: 		
jpayne@68: 		//Do anything necessary after processing
jpayne@68: 		
jpayne@68: 		if(resultsFile!=null){
jpayne@68: 			TextStreamWriter tsw=new TextStreamWriter(resultsFile, overwrite, false, false);
jpayne@68: 			tsw.start();
jpayne@68: 			for(TaxNode tn : nodes.keySet()){
jpayne@68: 				Long data=nodes.get(tn);
jpayne@68: 				if(data==null){data=0L;}
jpayne@68: 				tsw.println(tn.id+"\t"+data+"\t"+tn.levelStringExtended(false)+"\t"+tn.name);
jpayne@68: 			}
jpayne@68: 			errorState|=tsw.poisonAndWait();
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Report timing and results
jpayne@68: 		{
jpayne@68: 			t.stop();
jpayne@68: 			
jpayne@68: 			//Calculate units per nanosecond
jpayne@68: 			double rpnano=readsProcessed/(double)(t.elapsed);
jpayne@68: 			double lpnano=linesProcessed/(double)(t.elapsed);
jpayne@68: 			double bpnano=basesProcessed/(double)(t.elapsed);
jpayne@68: 			
jpayne@68: 			//Add "k" and "m" for large numbers
jpayne@68: 			String rpstring=Tools.padKM(readsProcessed, 8);
jpayne@68: 			String lpstring=Tools.padKM(linesProcessed, 8);
jpayne@68: 			String bpstring=Tools.padKM(basesProcessed, 8);
jpayne@68: 
jpayne@68: 			String li="Lines In:               \t"+linesProcessed+" lines";
jpayne@68: 			String lo="Lines Out:              \t"+linesOut+" lines";
jpayne@68: 			while(lo.length()<li.length()){lo=lo+" ";}
jpayne@68: 
jpayne@68: 			String ri="Reads In:               \t"+readsProcessed+" reads";
jpayne@68: 			String ro="Reads Out:              \t"+readsOut+" reads";
jpayne@68: 			while(ro.length()<ri.length()){ro=ro+" ";}
jpayne@68: 
jpayne@68: 			outstream.println(ri+"\t"+basesProcessed+" bases");
jpayne@68: 			outstream.println(ro+"\t"+basesOut+" bases");
jpayne@68: 			outstream.println(li);
jpayne@68: 			outstream.println(lo);
jpayne@68: 			outstream.println();
jpayne@68: 			
jpayne@68: 			outstream.println("Time:                         \t"+t);
jpayne@68: 			outstream.println("Reads Processed:    "+rpstring+" \t"+String.format(Locale.ROOT, "%.2fk reads/sec", rpnano*1000000));
jpayne@68: 			outstream.println("Lines Processed:    "+lpstring+" \t"+String.format(Locale.ROOT, "%.2fk reads/sec", lpnano*1000000));
jpayne@68: 			outstream.println("Bases Processed:    "+bpstring+" \t"+String.format(Locale.ROOT, "%.2fm bases/sec", bpnano*1000));
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Throw an exception of there was an error in a thread
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: 	/** Iterate through the reads */
jpayne@68: 	void processInner(){
jpayne@68: 		ByteFile bf=ByteFile.makeByteFile(ffin1);
jpayne@68: 		TaxNode currentNode=null;
jpayne@68: 		long currentSize=0;
jpayne@68: 		ByteStreamWriter bsw=null;
jpayne@68: 		for(byte[] line=bf.nextLine(); line!=null; line=bf.nextLine()){
jpayne@68: 			linesProcessed++;
jpayne@68: 			if(line.length>0){
jpayne@68: 				final boolean header=(line[0]=='>');
jpayne@68: 				if(header){
jpayne@68: 					if(maxReads>0 && readsProcessed>=maxReads){break;}
jpayne@68: 					readsProcessed++;
jpayne@68: 					if(currentNode!=null){nodes.put(currentNode, nodes.get(currentNode)+currentSize);}
jpayne@68: 					
jpayne@68: 					final TaxNode tn=tree.parseNodeFromHeader(new String(line, 1, line.length-1), false);
jpayne@68: 					
jpayne@68: 					if(tn==null || tn!=currentNode){
jpayne@68: 						if(bsw!=null){errorState=bsw.poisonAndWait()|errorState; bsw=null;}
jpayne@68: 					}
jpayne@68: 					if(tn!=null && tn!=currentNode){
jpayne@68: 						String dir=tree.toDir(tn, outPath);
jpayne@68: 						final boolean found=nodes.containsKey(tn);
jpayne@68: 						if(!found){nodes.put(tn, 0L);}
jpayne@68: 						FileFormat ff=FileFormat.testOutput(dir+prefix+tn.id+".fa.gz", FileFormat.FASTA, null, true, overwrite && !found, found, false);
jpayne@68: 						bsw=new ByteStreamWriter(ff);
jpayne@68: 						bsw.start();
jpayne@68: 					}
jpayne@68: 					
jpayne@68: 					currentNode=tn;
jpayne@68: 					currentSize=0;
jpayne@68: 					if(bsw!=null){readsOut++;}
jpayne@68: 				}else{
jpayne@68: 					basesProcessed+=line.length;
jpayne@68: 					currentSize+=line.length;
jpayne@68: 				}
jpayne@68: 				if(bsw!=null){
jpayne@68: 					linesOut++;
jpayne@68: 					if(!header){basesOut+=line.length;}
jpayne@68: 					bsw.println(line);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		if(bsw!=null){
jpayne@68: 			errorState=bsw.poisonAndWait()|errorState; bsw=null;
jpayne@68: 			if(currentNode!=null){nodes.put(currentNode, nodes.get(currentNode)+currentSize);}
jpayne@68: 		}
jpayne@68: 		bf.close();
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------            Fields            ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file path */
jpayne@68: 	private String in1=null;
jpayne@68: 
jpayne@68: 	/** Primary output file path */
jpayne@68: 	private String outPath=null;
jpayne@68: 	
jpayne@68: 	private String prefix;
jpayne@68: 	
jpayne@68: 	/** Override input file extension */
jpayne@68: 	private String extin=null;
jpayne@68: 	
jpayne@68: 	/** For listing what is present in the output */
jpayne@68: 	public String resultsFile=null;
jpayne@68: 	
jpayne@68: 	public String taxTreeFile=null;
jpayne@68: 	
jpayne@68: 	public boolean makeDirectories=true;
jpayne@68: 	
jpayne@68: 	public LinkedHashMap<TaxNode, Long> nodes=new LinkedHashMap<TaxNode, Long>();
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Number of reads processed */
jpayne@68: 	protected long readsProcessed=0;
jpayne@68: 	/** Number of lines processed */
jpayne@68: 	protected long linesProcessed=0;
jpayne@68: 	/** Number of bases processed */
jpayne@68: 	protected long basesProcessed=0;
jpayne@68: 
jpayne@68: 	/** Number of reads out */
jpayne@68: 	public long readsOut=0;
jpayne@68: 	/** Number of lines out */
jpayne@68: 	public long linesOut=0;
jpayne@68: 	/** Number of bases out */
jpayne@68: 	public long basesOut=0;
jpayne@68: 
jpayne@68: 	/** Quit after processing this many input reads; -1 means no limit */
jpayne@68: 	private long maxReads=-1;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Final Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file */
jpayne@68: 	private final FileFormat ffin1;
jpayne@68: 	
jpayne@68: 	private final TaxTree tree;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Common Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Print status messages to this output stream */
jpayne@68: 	private PrintStream outstream=System.err;
jpayne@68: 	/** Print verbose messages */
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	/** True if an error was encountered */
jpayne@68: 	public boolean errorState=false;
jpayne@68: 	/** Overwrite existing output files */
jpayne@68: 	private boolean overwrite=true;
jpayne@68: 	
jpayne@68: }