package prok;

import java.io.File;
import java.io.PrintStream;
import java.util.ArrayList;

import fileIO.ByteFile;
import fileIO.ByteFile1;
import fileIO.ByteFile2;
import fileIO.FileFormat;
import fileIO.ReadWrite;
import shared.Parse;
import shared.Parser;
import shared.PreParser;
import shared.Shared;
import shared.Timer;
import shared.Tools;
import stream.ConcurrentGenericReadInputStream;
import stream.ConcurrentReadInputStream;
import stream.ConcurrentReadOutputStream;
import stream.FASTQ;
import stream.FastaReadInputStream;
import stream.Read;
import structures.ListNum;
import tax.TaxNode;
import tax.TaxTree;
import tracker.ReadStats;

/**
 * Removes unwanted sequences from Silva, particularly bacteria flagged as euks due to name misidentification.
 * 
 * @author Brian Bushnell
 * @date January 27, 2020
 *
 */
public class FilterSilva {
	
	public static void main(String[] args){
		//Start a timer immediately upon code entrance.
		Timer t=new Timer();

		//Create an instance of this class
		FilterSilva x=new FilterSilva(args);

		//Run the object
		x.process(t);
		
		//Close the print stream if it was redirected
		Shared.closeStream(x.outstream);
	}
	
	public FilterSilva(String[] args){
		
		{//Preparse block for help, config files, and outstream
			PreParser pp=new PreParser(args, getClass(), false);
			args=pp.args;
			outstream=pp.outstream;
		}
		
		//Set shared static variables
		Shared.capBuffers(4); //Only for singlethreaded programs
		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
		ReadWrite.setZipThreads(Shared.threads());
		FASTQ.TEST_INTERLEAVED=FASTQ.FORCE_INTERLEAVED=false;
		TaxTree.SILVA_MODE=true;
		
		Parser parser=new Parser();
		for(int i=0; i<args.length; i++){
			String arg=args[i];
			String[] split=arg.split("=");
			String a=split[0].toLowerCase();
			String b=split.length>1 ? split[1] : null;
			if(b!=null && b.equalsIgnoreCase("null")){b=null;}

			if(a.equals("verbose")){
				verbose=Parse.parseBoolean(b);
				ByteFile1.verbose=verbose;
				ByteFile2.verbose=verbose;
				stream.FastaReadInputStream.verbose=verbose;
				ConcurrentGenericReadInputStream.verbose=verbose;
				stream.FastqReadInputStream.verbose=verbose;
				ReadWrite.verbose=verbose;
			}else if(a.equals("tree") || a.equals("treefile")){
				treeFile=b;
			}else if(parser.parse(arg, a, b)){
				//do nothing
			}
			
			else if(parser.in1==null && i==0 && Tools.looksLikeInputStream(arg)){
				parser.in1=arg;
			}else if(parser.out1==null && i==1 && !arg.contains("=")){
				parser.out1=arg;
			}else{
				outstream.println("Unknown parameter "+args[i]);
				assert(false) : "Unknown parameter "+args[i];
				//				throw new RuntimeException("Unknown parameter "+args[i]);
			}
		}
		
		{//Process parser fields
			Parser.processQuality();
			
			maxReads=parser.maxReads;
			
			overwrite=ReadStats.overwrite=parser.overwrite;
			append=ReadStats.append=parser.append;
			
			in1=parser.in1;

			out1=parser.out1;
			
			extin=parser.extin;
			extout=parser.extout;
		}
		
		assert(FastaReadInputStream.settingsOK());
		
		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
		if(!ByteFile.FORCE_MODE_BF1 && !ByteFile.FORCE_MODE_BF2){
			ByteFile.FORCE_MODE_BF2=false;
			ByteFile.FORCE_MODE_BF1=true;
		}

		if(out1!=null && out1.equalsIgnoreCase("null")){out1=null;}
		
		if(!Tools.testOutputFiles(overwrite, append, false, out1)){
			outstream.println((out1==null)+", "+out1);
			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+out1+"\n");
		}
		
		ffout1=FileFormat.testOutput(out1, FileFormat.FASTQ, extout, true, overwrite, append, false);

		ffin1=FileFormat.testInput(in1, FileFormat.FASTQ, extin, true, true);
		
		tree=TaxTree.loadTaxTree(treeFile, outstream, true, false);
	}
	
	void process(Timer t){
		
		final ConcurrentReadInputStream cris;
		{
			cris=ConcurrentReadInputStream.getReadInputStream(maxReads, true, ffin1, null);
			cris.start();
			if(verbose){outstream.println("Started cris");}
		}
		boolean paired=cris.paired();
//		if(verbose){
			if(!ffin1.samOrBam()){outstream.println("Input is being processed as "+(paired ? "paired" : "unpaired"));}
//		}

		final ConcurrentReadOutputStream ros;
		if(out1!=null){
			final int buff=4;

			assert(!out1.equalsIgnoreCase(in1) && !out1.equalsIgnoreCase(in1)) : "Input file and output file have same name.";
			
			ros=ConcurrentReadOutputStream.getStream(ffout1, null, null, null, buff, null, false);
			ros.start();
		}else{ros=null;}
		
		long readsProcessed=0, readsOut=0;
		long basesProcessed=0, basesOut=0;
		
		{
			
			ListNum<Read> ln=cris.nextList();
			ArrayList<Read> reads=(ln!=null ? ln.list : null);
			
//			outstream.println("Fetched "+reads);
			
			if(reads!=null && !reads.isEmpty()){
				Read r=reads.get(0);
				assert((ffin1==null || ffin1.samOrBam()) || (r.mate!=null)==cris.paired());
			}

			while(ln!=null && reads!=null && reads.size()>0){//ln!=null prevents a compiler potential null access warning
				
				final ArrayList<Read> listOut=new ArrayList<Read>(reads.size());
				for(int idx=0; idx<reads.size(); idx++){
					final Read r1=reads.get(idx);
					
					final int initialLength1=r1.length();
					final boolean keep=process(r1);
					if(keep){
						listOut.add(r1);
						readsOut++;
						basesOut+=r1.length();
					}
					
					readsProcessed++;
					basesProcessed+=initialLength1;
				}
				
				if(ros!=null){ros.add(listOut, ln.id);}

				cris.returnList(ln);
				ln=cris.nextList();
				reads=(ln!=null ? ln.list : null);
			}
			if(ln!=null){
				cris.returnList(ln.id, ln.list==null || ln.list.isEmpty());
			}
		}
		
		errorState|=ReadStats.writeAll();
		
		errorState|=ReadWrite.closeStreams(cris, ros);
		
		t.stop();
		outstream.println(Tools.timeReadsBasesProcessed(t, readsProcessed, basesProcessed, 8));
		outstream.println(Tools.readsBasesOut(readsProcessed, basesProcessed, readsOut, basesOut, 8, false));
		
		if(errorState){
			throw new RuntimeException(getClass().getName()+" terminated in an error state; the output may be corrupt.");
		}
	}
	
	/*--------------------------------------------------------------*/
	
	private boolean process(Read r){
		TaxNode tn=tree.parseNodeFromHeader(r.id, true);
		if(tn==null){return false;}
		if(tree.isEukaryote(tn.id)){
			if(r.id.contains(";Chloroplast;") || r.id.contains("Mitochondria")){return false;}
			if(r.id.contains("Bacteria;") || r.id.contains("Archaea;")){return false;}
			//Unclear how to remove 16S from euks at this point; best to do it later.
			//Might be worth adding the SSU type to the sequence header, though, e.g. 16S/18S/m16S/m18S/c16S
		}
		return true;
	}
	
	/*--------------------------------------------------------------*/
	
	private String in1=null;

	private String out1=null;
	
	private String extin=null;
	private String extout=null;
	
	private String treeFile="auto";
	
	/*--------------------------------------------------------------*/

	private long maxReads=-1;
	
	/*--------------------------------------------------------------*/
	
	private final FileFormat ffin1;

	private final FileFormat ffout1;
	
	private final TaxTree tree;
	
	/*--------------------------------------------------------------*/
	
	private PrintStream outstream=System.err;
	public static boolean verbose=false;
	public boolean errorState=false;
	private boolean overwrite=true;
	private boolean append=false;
	
}