FastaIndexer.java - code915 - My code for INSERM UMR915

Source path: svn/ trunk/ core/ src/ java/ fr/ inserm/ umr915/ core/ bio/ fasta/ FastaIndexer.java

‹r71

r463

package fr.inserm.umr915.core.bio.fasta;

import java.io.BufferedInputStream;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileInputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
import java.util.logging.Logger;

import javax.xml.XMLConstants;
import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.stream.XMLOutputFactory;
import javax.xml.stream.XMLStreamException;
import javax.xml.stream.XMLStreamWriter;

import org.lindenb.io.IOUtils;
import org.lindenb.lang.InvalidXMLException;
import org.lindenb.sw.vocabulary.DC;
import org.lindenb.sw.vocabulary.RDF;
import org.lindenb.util.StringUtils;
import org.lindenb.xml.XMLUtilities;
import org.w3c.dom.Document;
import org.w3c.dom.Element;

public class FastaIndexer
	{
	public static final String NS="uri:fr.inserm.umr915:fastaidx:1.0:";
	
	@SuppressWarnings("unused")
	private static final Logger LOG=Logger.getLogger("fr.inserm.umr915");
	
	
	private FastaIndexer()
		{
		
		}
	 
	
	protected void echo(XMLStreamWriter w,FastaIndex fi) throws XMLStreamException
		{
		w.writeStartElement("fi","Index",FastaIndexer.NS);
		
		w.writeCharacters("\n\t");
		
		w.writeEmptyElement("fi","source",FastaIndexer.NS);
		w.writeAttribute("rdf", RDF.NS, "resource","file://"+fi.getFile().getAbsolutePath());
		
		w.writeCharacters("\n\t");
		
		for(String name: fi.getNames())
			{
			w.writeStartElement("dc","title",DC.NS);
			w.writeCharacters(name);
			w.writeEndElement();
			}
		w.writeCharacters("\n\t");
		
		w.writeStartElement("fi","line-size",FastaIndexer.NS);
		w.writeAttribute("rdf", RDF.NS, "datatype", "http://www.w3.org/2001/XMLSchema#int");
		w.writeCharacters(String.valueOf(fi.lineSize));
		w.writeEndElement();
		
		w.writeCharacters("\n\t");
		
		w.writeStartElement("fi","start",FastaIndexer.NS);
		w.writeAttribute("rdf", RDF.NS, "datatype", "http://www.w3.org/2001/XMLSchema#unsignedLong");
		w.writeCharacters(String.valueOf(fi.getSeqStart()));
		w.writeEndElement();
		
		w.writeCharacters("\n\t");
		
		w.writeStartElement("fi","end",FastaIndexer.NS);
		w.writeAttribute("rdf", RDF.NS, "datatype", "http://www.w3.org/2001/XMLSchema#unsignedLong");
		w.writeCharacters(String.valueOf(fi.getSeqEnd()));
		w.writeEndElement();
		
		w.writeCharacters("\n\t");
		
		w.writeStartElement("fi","length",FastaIndexer.NS);
		w.writeAttribute("rdf", RDF.NS, "datatype", "http://www.w3.org/2001/XMLSchema#int");
		w.writeCharacters(String.valueOf(fi.getLength()));
		w.writeEndElement();
		
		w.writeCharacters("\n");
		
		w.writeEndElement();
		
		w.writeCharacters("\n");
		}
	
	private void build(XMLStreamWriter out,File fastaFile) throws XMLStreamException,IOException
		{
		if(!fastaFile.isAbsolute())
			{
			throw new IOException("Not an absolute file path:"+fastaFile);
			}
		int curr_line_size=0;
		InputStream in=new BufferedInputStream(new FileInputStream(fastaFile));
		FastaIndex currIndex=null;
		long offset=-1L;
		
		while(true)
			{
			int c=in.read();
			if(c!=-1) offset++;
			//end of line or end of file
			if(currIndex!=null && (c==-1 || (c=='\n')))
				{
				currIndex.lineSize=Math.max(currIndex.lineSize,curr_line_size);
				curr_line_size=0;
				}
			
			if(c==-1 || c=='>')
				{
				if(currIndex!=null) echo(out,currIndex);
				
				if(c==-1) break;
				StringBuilder name=new StringBuilder();
				currIndex=new FastaIndex();
				currIndex.source=fastaFile;
				currIndex.nameStart= offset;
				while(true)
					{
					c=in.read();
					if(c==-1) throw new IOException("unfinished name at offset "+offset);
					offset++;
					if(c=='\n') break;
					name.append((char)c);
					}
				
				currIndex.nameEnd=offset;
				String seqName=name.toString().trim();
				currIndex.names.add(seqName);
				if(seqName.toLowerCase().startsWith("chrom"))
					{
					seqName=seqName.substring(5).trim();
					currIndex.names.add(seqName);
					}
				else if(seqName.toLowerCase().startsWith("chr"))
					{
					seqName=seqName.substring(3).trim();
					currIndex.names.add(seqName);
					}
				currIndex.seqStart=offset+1;
				currIndex.seqEnd=offset+1;
				currIndex.sequence_length=0;
				curr_line_size=0;
				}
			else if(currIndex!=null)
				{
				if(c=='\n')
					{
					//already processed ignore
					}
				else if(Character.isLetter(c))
					{
					currIndex.seqEnd= offset+1L;
					currIndex.sequence_length++;
					curr_line_size++;
					}
				else
					{
					throw new IOException("Illegal character "+(char)c+" at "+offset);
					}
				}
			}
		in.close();
		}
	
	
	
	public static List<FastaIndex> readIndex(File file) throws IOException
		{
		Set<String> titles= new HashSet<String>();
		List<FastaIndex> indexes=new ArrayList<FastaIndex>();
		try {
			DocumentBuilderFactory factory=DocumentBuilderFactory.newInstance();
			factory.setCoalescing(true);
			factory.setExpandEntityReferences(true);
			factory.setValidating(false);
			factory.setIgnoringComments(true);
			factory.setIgnoringElementContentWhitespace(true);
			factory.setNamespaceAware(true);
			DocumentBuilder builder=factory.newDocumentBuilder();
			
			Document dom=builder.parse(file);
			Element root=dom.getDocumentElement();
			if(!XMLUtilities.isA(root, RDF.NS,"RDF"))
				{
				throw new InvalidXMLException(root, "not a rdf:RDF");
				}
			for(Element m:XMLUtilities.elements(root, FastaIndexer.NS, "Index"))
				{
				
				FastaIndex fi=new FastaIndex();
				Element e= XMLUtilities.one(m, FastaIndexer.NS,"line-size");
				fi.lineSize=Integer.parseInt(e.getTextContent());
				
				e= XMLUtilities.one(m, FastaIndexer.NS,"start");
				fi.seqStart=Long.parseLong(e.getTextContent());
				
				e= XMLUtilities.one(m, FastaIndexer.NS,"end");
				fi.seqEnd=Long.parseLong(e.getTextContent());
				
				
				e= XMLUtilities.one(m, FastaIndexer.NS,"length");
				fi.sequence_length=Integer.parseInt(e.getTextContent());
				
				e= XMLUtilities.one(m, FastaIndexer.NS,"source");
				String src= e.getAttributeNS(RDF.NS,"resource");
				if(!src.startsWith("file://")) throw new InvalidXMLException(e,""+src+" should start with file://");
				src=src.substring(7);
					
				if(StringUtils.isBlank(src)) throw new InvalidXMLException(e,"not @rdf:resource");
				fi.source=new File(src);
				if(!fi.source.isAbsolute()) throw new InvalidXMLException(e,"not absolute path in @rdf:resource="+fi.source);
				
				List<Element> names= XMLUtilities.elements(m, DC.NS,"title");
				if(names.isEmpty()) throw new InvalidXMLException(m,"Names missing in "+file);
				for(Element e2: names)
					{
					String name=e2.getTextContent();
					if(titles.contains(name))
						{
						throw new InvalidXMLException(m,"name defined twice:"+name);
						}
					}
				for(Element e2: names)
					{
					String name=e2.getTextContent();
					titles.add(name);
					fi.names.add(name);
					}
				
				
				
				indexes.add(fi);
				}
			return indexes;
			} 
		catch (IOException e)
			{
			throw e;
			}
		catch (Throwable e)
			{
			throw new IOException(e);
			}
		}
	
	public static void main(String[] args)
		{
		int optind=0;
		
		try {
			FastaIndexer app=new FastaIndexer();
			Set<File> files=new HashSet<File>();
			while(optind<args.length)
				{
				if(args[optind].equals("-h"))
					{
					System.err.println("Creates an index (XML format) for some FASTA sequences.\n");
					System.err.println(" -i <file> (optional) read a file containing some absolute path to FASTA");
					System.err.println(" <stdin with file path|files>");
					return;
					}
				else if(args[optind].equals("-i"))
					{
					BufferedReader in=IOUtils.openReader(args[++optind]);
					String line;
					while((line=in.readLine())!=null)
						{
						if(line.startsWith("#") || StringUtils.isBlank(line)) continue;
						files.add(new File(line.trim()));
						}
					in.close();
					}
				else if(args[optind].equals("--"))
					{
					optind++;
					break;
					}
				else if(args[optind].startsWith("-"))
					{
					System.err.println("Unnown option: "+args[optind]);
					return;
					}
				else
					{
					break;
					}
				++optind;
				}
			if(optind==args.length && files.isEmpty())
				{
				BufferedReader in=new BufferedReader(new InputStreamReader(System.in));
				String line;
				while((line=in.readLine())!=null)
					{
					if(line.startsWith("#") || StringUtils.isBlank(line)) continue;
					files.add(new File(line.trim()));
					}
				}
			
			while(optind< args.length)
				{
				String inputName=args[optind++];
				File file=new File(inputName);
				if(!file.isAbsolute())
					{
					System.err.println("Not an absolute file path:"+file);
					return;
					}
				files.add(file);
				}
			
			if(files.isEmpty())
				{
				System.err.println("Files missing");
				return;
				}
			XMLOutputFactory xmlfactory= XMLOutputFactory.newInstance();
			XMLStreamWriter w= xmlfactory.createXMLStreamWriter(System.out,"UTF-8");
			w.writeStartDocument("UTF-8","1.0");
			w.writeStartElement("rdf","RDF",RDF.NS);
			w.writeAttribute("xmlns", XMLConstants.XML_NS_URI, "rdf", RDF.NS);
			w.writeAttribute("xmlns", XMLConstants.XML_NS_URI, "fi", FastaIndexer.NS);
			w.writeAttribute("xmlns", XMLConstants.XML_NS_URI, "dc", DC.NS);
			w.writeCharacters("\n");
			for(File file:files)
				{
				app.build(w,file);
				}
			w.writeEndElement();
			w.writeCharacters("\n");
			w.flush();
			w.close();
			} 
		catch (Exception e)
			{
			e.printStackTrace();
			}
		}
	}

Show details Hide details

Change log

r74 by plindenbaum.u915 on May 18, 2010 Diff

cont

Go to:

Project members, sign in to write a code review

Older revisions

r71 by plindenbaum.u915 on May 17, 2010 Diff

continue

r70 by plindenbaum.u915 on May 17, 2010 Diff

sequence indexing

r68 by plindenbaum.u915 on May 14, 2010 Diff

cont

All revisions of this file

File info

Size: 9540 bytes, 351 lines

View raw file