jpayne@68: package gff;
jpayne@68: 
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.Arrays;
jpayne@68: 
jpayne@68: import fileIO.ByteStreamWriter;
jpayne@68: import shared.Shared;
jpayne@68: import shared.Tools;
jpayne@68: import structures.ByteBuilder;
jpayne@68: 
jpayne@68: public class GbffFeature {
jpayne@68: 
jpayne@68: 	public GbffFeature(final ArrayList<byte[]> lines0, final String typeString, final String accessionString){
jpayne@68: 		accession=accessionString;
jpayne@68: 		setType(typeString);
jpayne@68: 		parseSlow(lines0);
jpayne@68: 		if(type==rRNA){
jpayne@68: 			setSubtype();
jpayne@68: 		}
jpayne@68: 		if(stop<start){error=true;}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private void parseSlow(final ArrayList<byte[]> lines0){
jpayne@68: 		ArrayList<byte[]> lines=fixLines(lines0);
jpayne@68: 		parseStartStop(lines.get(0));
jpayne@68: 		for(int i=1; i<lines.size(); i++){
jpayne@68: 			byte[] line=lines.get(i);
jpayne@68: 			if(Tools.startsWith(line, "product=")){
jpayne@68: 				product=parseLine(line);
jpayne@68: 			}else if(Tools.startsWith(line, "locus_tag=")){
jpayne@68: 				locus_tag=parseLine(line);
jpayne@68: 			}else if(Tools.equals(line, "pseudo")){
jpayne@68: 				pseudo=true;
jpayne@68: 			}
jpayne@68: 			
jpayne@68: //			else if(Tools.startsWith(line, "ID=")){
jpayne@68: //				id=parseLine(line);
jpayne@68: //			}else if(Tools.startsWith(line, "Name=")){
jpayne@68: //				name=parseLine(line);
jpayne@68: //			}
jpayne@68: 		}
jpayne@68: //		System.err.println("\nvvvvv");
jpayne@68: //		for(byte[] line : lines0){
jpayne@68: //			System.err.println("'"+new String(line)+"'");
jpayne@68: //		}
jpayne@68: //		for(byte[] line : lines){
jpayne@68: //			System.err.println("'"+new String(line)+"'");
jpayne@68: //		}
jpayne@68: //		System.err.println("^^^^^");
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	ArrayList<byte[]> fixLines(ArrayList<byte[]> lines){
jpayne@68: 		ArrayList<byte[]> fixed=new ArrayList<byte[]>();
jpayne@68: 		ByteBuilder bb=new ByteBuilder();
jpayne@68: 		for(byte[] line : lines){
jpayne@68: 			if(bb.length()>0 && line[21]=='/'){
jpayne@68: 				fixed.add(bb.toBytes());
jpayne@68: 				bb.clear();
jpayne@68: 			}
jpayne@68: 			append(bb, line);
jpayne@68: 		}
jpayne@68: 		if(bb.length()>0){
jpayne@68: 			fixed.add(bb.toBytes());
jpayne@68: 			bb.clear();
jpayne@68: 		}
jpayne@68: 		return fixed;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void append(ByteBuilder bb, byte[] line){
jpayne@68: 		assert(line[20]==' ');
jpayne@68: 		assert(line.length>21);
jpayne@68: //		assert(line[21]!=' ') : "'"+new String(line)+"'";
jpayne@68: 		if(line[21]=='/'){
jpayne@68: 			bb.append(line, 22, line.length-22);
jpayne@68: 		}else{
jpayne@68: //			System.err.println(line.length+", "+21+", "+(line.length-21+1)+"\n'"+new String(line)+"'");
jpayne@68: 			if(bb.length>0){bb.append(' ');}
jpayne@68: 			bb.append(line, 21, line.length-21);
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void setType(String typeString){
jpayne@68: 		int x=Tools.find(typeString, typeStrings);
jpayne@68: 		assert(x>=0) : x+", "+typeString;
jpayne@68: 		type=x;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void parseStartStop(final byte[] line0){
jpayne@68: 		byte[] line=line0;
jpayne@68: 		
jpayne@68: 		if(line[0]=='c'){
jpayne@68: 			assert(Tools.startsWith(line, "complement("));
jpayne@68: 			line=Arrays.copyOfRange(line, 11, line.length-1);
jpayne@68: 			strand=Shared.MINUS;
jpayne@68: 		}
jpayne@68: 		if(line[0]=='j'){
jpayne@68: 			assert(Tools.startsWith(line, "join("));
jpayne@68: 			line=Arrays.copyOfRange(line, 5, line.length-1);
jpayne@68: 			strand=Shared.MINUS;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		int i=0;
jpayne@68: 		for(start=0; i<line.length; i++){
jpayne@68: 			int x=line[i];
jpayne@68: 			if(x=='.'){break;}
jpayne@68: 			else if(x!='<'){
jpayne@68: 				if(Tools.isDigit(x)){
jpayne@68: 					start=start*10+(x-'0');
jpayne@68: 				}else{
jpayne@68: 					//if(!error){System.err.println(new String(line0)+"\n"+new String(line));}
jpayne@68: 					error=true;
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: //		while(line[i]=='.'){i++;} //Not needed
jpayne@68: 		for(stop=0; i<line.length; i++){
jpayne@68: 			int x=line[i];
jpayne@68: 			if(x=='.' || x==','){
jpayne@68: 				stop=0;
jpayne@68: 			}else if(x==' '){
jpayne@68: 				//do nothing; line wrap
jpayne@68: 			}else if(x!='>'){
jpayne@68: 				if(Tools.isDigit(x)){
jpayne@68: 					stop=stop*10+(x-'0');
jpayne@68: 				}else{
jpayne@68: 					//if(!error){System.err.println(new String(line0)+"\n"+new String(line));}
jpayne@68: 					error=true;
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	String parseLine(byte[] line){
jpayne@68: 		String[] split=Tools.equalsPattern.split(new String(line));
jpayne@68: 		String s=split[1];
jpayne@68: 		return s.substring(1, s.length()-1);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void setSubtype(){
jpayne@68: 		subtype=-1;
jpayne@68: 		if(product==null){return;}
jpayne@68: 		String[] split=Tools.spacePattern.split(product);
jpayne@68: 		subtype=Tools.find(split[0], typeStrings);
jpayne@68: //		assert(false) : type+", "+subtype+", "+split[0]+", "+this.toString()+"\n"+product;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public void toGff(ByteStreamWriter bsw) {
jpayne@68: 		ByteBuilder bb=bsw.getBuffer();
jpayne@68: 		appendGff(bb);
jpayne@68: 		bb.nl();
jpayne@68: 		bsw.flushBuffer(false);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public ByteBuilder appendGff(ByteBuilder bb) {
jpayne@68: //		bsw.print("#seqid	source	type	start	end	score	strand	phase	attributes\n".getBytes());
jpayne@68: 		bb.append(accession).tab();
jpayne@68: 		bb.append('.').tab();
jpayne@68: 		bb.append((pseudo && type==GENE) ? "pseudogene" : typeStringsGff[type]).tab();
jpayne@68: 		bb.append(start).tab();
jpayne@68: 		bb.append(stop).tab();
jpayne@68: 		bb.append('.').tab();
jpayne@68: 		bb.append(Shared.strandCodes2[strand]).tab();
jpayne@68: 		bb.append('.').tab();
jpayne@68: 		
jpayne@68: 		boolean attributes=false;
jpayne@68: //		if(id!=null){
jpayne@68: //			bb.append("ID=").append(id);
jpayne@68: //			attributes=true;
jpayne@68: //		}
jpayne@68: //		if(name!=null){
jpayne@68: //			if(attributes){bb.append(';');}
jpayne@68: //			bb.append("Name=").append(name);
jpayne@68: //			attributes=true;
jpayne@68: //		}
jpayne@68: 		if(product!=null){
jpayne@68: 			if(attributes){bb.append(';');}
jpayne@68: 			bb.append("product=").append(product);
jpayne@68: 			attributes=true;
jpayne@68: 		}
jpayne@68: 		if(locus_tag!=null){
jpayne@68: 			if(attributes){bb.append(';');}
jpayne@68: 			bb.append("locus_tag=").append(locus_tag);
jpayne@68: 			attributes=true;
jpayne@68: 		}
jpayne@68: 		if(subtype>-1){
jpayne@68: 			if(attributes){bb.append(';');}
jpayne@68: 			bb.append("subtype=").append(typeStringsGff[subtype]);
jpayne@68: 			attributes=true;
jpayne@68: 		}
jpayne@68: 		if(!attributes){bb.append('.');}
jpayne@68: 		return bb;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	
jpayne@68: 	@Override
jpayne@68: 	public String toString(){
jpayne@68: 		return appendGff(new ByteBuilder()).toString();
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	public int type=-1;
jpayne@68: 	public int subtype=-1;
jpayne@68: 	//TODO: could have coding amino, for tRNA
jpayne@68: 	public String product;
jpayne@68: 	public String locus_tag;
jpayne@68: //	public String id;
jpayne@68: //	public String name;
jpayne@68: 	
jpayne@68: 	public int start;
jpayne@68: 	public int stop;
jpayne@68: 	public byte strand=Shared.PLUS;
jpayne@68: 	public String accession;
jpayne@68: 	public boolean pseudo=false;
jpayne@68: 	public boolean error=false;
jpayne@68: 
jpayne@68: 	public static final String[] typeStrings={"gene", "CDS", "rRNA", "tRNA", "ncRNA", "repeat_region", 
jpayne@68: 			"5'UTR", "3'UTR", "intron", "exon", "5S", "16S", "23S"};
jpayne@68: 	public static final String[] typeStringsGff={"gene", "CDS", "rRNA", "tRNA", "ncRNA", "repeat_region", 
jpayne@68: 			"five_prime_UTR", "three_prime_UTR", "intron", "exon", "5S", "16S", "23S"};
jpayne@68: 	
jpayne@68: 	//types
jpayne@68: 	public static final int GENE=0, CDS=1, rRNA=2, tRNA=3, ncRNA=4, repeat_region=5, UTR5=6, UTR3=7, intron=8, exon=9;
jpayne@68: 	//subtypes
jpayne@68: 	public static final int r5S=10, r16S=11, r23S=12;
jpayne@68: 	
jpayne@68: }