forester/java/src/org/forester/io/parsers/phyloxml/PhyloXmlParser.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2009 Christian M. Zmasek
   6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   7 // All rights reserved
   8 //
   9 // This library is free software; you can redistribute it and/or
  10 // modify it under the terms of the GNU Lesser General Public
  11 // License as published by the Free Software Foundation; either
  12 // version 2.1 of the License, or (at your option) any later version.
  13 //
  14 // This library is distributed in the hope that it will be useful,
  15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 // Lesser General Public License for more details.
  18 //
  19 // You should have received a copy of the GNU Lesser General Public
  20 // License along with this library; if not, write to the Free Software
  21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  22 //
  23 // Contact: phylosoft @ gmail . com
  24 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
  25
  26 package org.forester.io.parsers.phyloxml;
  27
  28 import java.io.BufferedReader;
  29 import java.io.File;
  30 import java.io.FileInputStream;
  31 import java.io.IOException;
  32 import java.io.InputStream;
  33 import java.io.InputStreamReader;
  34 import java.io.Reader;
  35 import java.io.StringReader;
  36 import java.net.URL;
  37 import java.util.Date;
  38 import java.util.Enumeration;
  39 import java.util.zip.ZipEntry;
  40 import java.util.zip.ZipFile;
  41 import java.util.zip.ZipInputStream;
  42
  43 import javax.xml.parsers.ParserConfigurationException;
  44 import javax.xml.parsers.SAXParser;
  45 import javax.xml.parsers.SAXParserFactory;
  46
  47 import org.forester.io.parsers.PhylogenyParser;
  48 import org.forester.io.parsers.util.PhylogenyParserException;
  49 import org.forester.phylogeny.Phylogeny;
  50 import org.forester.util.ForesterConstants;
  51 import org.forester.util.ForesterUtil;
  52 import org.xml.sax.InputSource;
  53 import org.xml.sax.SAXException;
  54 import org.xml.sax.SAXNotRecognizedException;
  55 import org.xml.sax.SAXNotSupportedException;
  56 import org.xml.sax.SAXParseException;
  57 import org.xml.sax.XMLReader;
  58 import org.xml.sax.helpers.DefaultHandler;
  59
  60 public class PhyloXmlParser implements PhylogenyParser {
  61
  62     final public static String   JAXP_SCHEMA_LANGUAGE                       = "http://java.sun.com/xml/jaxp/properties/schemaLanguage";
  63     final public static String   W3C_XML_SCHEMA                             = "http://www.w3.org/2001/XMLSchema";
  64     final public static String   JAXP_SCHEMA_SOURCE                         = "http://java.sun.com/xml/jaxp/properties/schemaSource";
  65     final public static String   SAX_FEATURES_VALIDATION                    = "http://xml.org/sax/features/validation";
  66     final public static String   APACHE_FEATURES_VALIDATION_SCHEMA          = "http://apache.org/xml/features/validation/schema";
  67     final public static String   APACHE_FEATURES_VALIDATION_SCHEMA_FULL     = "http://apache.org/xml/features/validation/schema-full-checking";
  68     final public static String   APACHE_PROPERTIES_SCHEMA_EXTERNAL_LOCATION = "http://apache.org/xml/properties/schema/external-schemaLocation";
  69     final static private boolean TIME                                       = false;
  70     private Object               _source;
  71     private boolean              _valid;
  72     private boolean              _zipped_inputstream;
  73     private int                  _error_count;
  74     private int                  _warning_count;
  75     private String               _schema_location;
  76     private StringBuffer         _error_messages;
  77     private StringBuffer         _warning_messages;
  78
  79     private PhyloXmlParser() {
  80         init();
  81         reset();
  82     }
  83
  84     public int getErrorCount() {
  85         return _error_count;
  86     }
  87
  88     public StringBuffer getErrorMessages() {
  89         return _error_messages;
  90     }
  91
  92     private Reader getReaderFromZipFile() throws IOException {
  93         Reader reader = null;
  94         final ZipFile zip_file = new ZipFile( getSource().toString() );
  95         final Enumeration<?> zip_file_entries = zip_file.entries();
  96         while ( zip_file_entries.hasMoreElements() ) {
  97             final ZipEntry zip_file_entry = ( ZipEntry ) zip_file_entries.nextElement();
  98             if ( !zip_file_entry.isDirectory() && ( zip_file_entry.getSize() > 0 ) ) {
  99                 final InputStream is = zip_file.getInputStream( zip_file_entry );
 100                 reader = new InputStreamReader( is, ForesterConstants.UTF_8 );
 101                 break;
 102             }
 103         }
 104         try {
 105             zip_file.close();
 106         }
 107         catch ( final Exception e ) {
 108             // Ignore
 109         }
 110         return reader;
 111     }
 112
 113     private String getSchemaLocation() {
 114         return _schema_location;
 115     }
 116
 117     private Object getSource() {
 118         return _source;
 119     }
 120
 121     public int getWarningCount() {
 122         return _warning_count;
 123     }
 124
 125     public StringBuffer getWarningMessages() {
 126         return _warning_messages;
 127     }
 128
 129     private void init() {
 130         setZippedInputstream( false );
 131     }
 132
 133     public boolean isValid() {
 134         return _valid;
 135     }
 136
 137     private boolean isZippedInputstream() {
 138         return _zipped_inputstream;
 139     }
 140
 141     @Override
 142     public Phylogeny[] parse() throws IOException, PhylogenyParserException {
 143         reset();
 144         final PhyloXmlHandler handler = new PhyloXmlHandler();
 145         final SAXParserFactory factory = SAXParserFactory.newInstance();
 146         factory.setNamespaceAware( true );
 147         try {
 148             if ( !ForesterUtil.isEmpty( getSchemaLocation() ) ) {
 149                 factory.setFeature( SAX_FEATURES_VALIDATION, true );
 150                 factory.setFeature( APACHE_FEATURES_VALIDATION_SCHEMA, true );
 151                 factory.setFeature( APACHE_FEATURES_VALIDATION_SCHEMA_FULL, true );
 152             }
 153         }
 154         catch ( final SAXNotRecognizedException e ) {
 155             e.printStackTrace();
 156             throw new PhylogenyParserException( "sax not recognized exception: " + e.getLocalizedMessage() );
 157         }
 158         catch ( final SAXNotSupportedException e ) {
 159             e.printStackTrace();
 160             throw new PhylogenyParserException( "sax not supported exception: " + e.getLocalizedMessage() );
 161         }
 162         catch ( final ParserConfigurationException e ) {
 163             e.printStackTrace();
 164             throw new PhylogenyParserException( "parser configuration exception: " + e.getLocalizedMessage() );
 165         }
 166         catch ( final Exception e ) {
 167             e.printStackTrace();
 168             throw new PhylogenyParserException( "error while configuring sax parser: " + e.getLocalizedMessage() );
 169         }
 170         try {
 171             final SAXParser parser = factory.newSAXParser();
 172             if ( !ForesterUtil.isEmpty( getSchemaLocation() ) ) {
 173                 parser.setProperty( JAXP_SCHEMA_LANGUAGE, W3C_XML_SCHEMA );
 174                 parser.setProperty( JAXP_SCHEMA_SOURCE, getSchemaLocation() );
 175                 parser.setProperty( APACHE_PROPERTIES_SCHEMA_EXTERNAL_LOCATION, getSchemaLocation() );
 176             }
 177             final XMLReader xml_reader = parser.getXMLReader();
 178             xml_reader.setContentHandler( handler );
 179             xml_reader.setErrorHandler( new PhyloXmlParserErrorHandler() );
 180             long start_time = 0;
 181             if ( TIME ) {
 182                 start_time = new Date().getTime();
 183             }
 184             if ( getSource() instanceof File ) {
 185                 if ( !getSource().toString().toLowerCase().endsWith( ".zip" ) ) {
 186                     final InputStream is = new FileInputStream( (File) getSource() );
 187                     final InputStreamReader isr = new InputStreamReader( is, ForesterConstants.UTF_8 );
 188                     xml_reader.parse( new InputSource( new BufferedReader( isr ) ) );
 189                 }
 190                 else {
 191                     final Reader reader = getReaderFromZipFile();
 192                     if ( reader == null ) {
 193                         throw new PhylogenyParserException( "zip file \"" + getSource()
 194                                                             + "\" appears not to contain any entries" );
 195                     }
 196                     xml_reader.parse( new InputSource( new BufferedReader( reader ) ) );
 197                 }
 198             }
 199             else if ( getSource() instanceof InputSource ) {
 200                 final InputSource is = ( InputSource ) getSource();
 201                 is.setEncoding( ForesterConstants.UTF_8 );
 202                 xml_reader.parse( is );
 203             }
 204             else if ( getSource() instanceof InputStream ) {
 205                 if ( !isZippedInputstream() ) {
 206                     final InputStream is = ( InputStream ) getSource();
 207                     xml_reader.parse( new InputSource( new BufferedReader( new InputStreamReader( is, ForesterConstants.UTF_8 ) ) ) );
 208                 }
 209                 else {
 210                     final ZipInputStream zip_is = new ZipInputStream( ( InputStream ) getSource() );
 211                     zip_is.getNextEntry();
 212                     xml_reader.parse( new InputSource( new BufferedReader( new InputStreamReader( zip_is, ForesterConstants.UTF_8 ) ) ) );
 213                 }
 214             }
 215             else if ( getSource() instanceof StringBuffer ) {
 216                 final StringReader string_reader = new StringReader( getSource().toString() );
 217                 xml_reader.parse( new InputSource( string_reader ) );
 218             }
 219             else {
 220                 throw new PhylogenyParserException( "phyloXML parser: attempt to parse object of unsupported type: \""
 221                         + getSource().getClass() + "\"" );
 222             }
 223             if ( TIME ) {
 224                 System.out.println( "[TIME] phyloXML parsing: " + ( new Date().getTime() - start_time ) + "ms." );
 225             }
 226         }
 227         catch ( final SAXException sax_exception ) {
 228             throw new PhylogenyParserException( "failed to parse [" + getSource() + "]: "
 229                     + sax_exception.getLocalizedMessage() );
 230         }
 231         catch ( final ParserConfigurationException parser_config_exception ) {
 232             throw new PhylogenyParserException( "failed to parse [" + getSource()
 233                                                 + "]. Problem with XML parser configuration: " + parser_config_exception.getLocalizedMessage() );
 234         }
 235         catch ( final IOException e ) {
 236             throw new PhylogenyParserException( "problem with input source: " + e.getLocalizedMessage() );
 237         }
 238         catch ( final Exception e ) {
 239             throw new PhylogenyParserException( e.getLocalizedMessage() );
 240         }
 241         catch ( final Error err ) {
 242             err.printStackTrace();
 243             throw new PhylogenyParserException( "severe error: " + err.getLocalizedMessage() );
 244         }
 245         final Phylogeny[] ps = new Phylogeny[ handler.getPhylogenies().size() ];
 246         int i = 0;
 247         for( final Phylogeny phylogeny : handler.getPhylogenies() ) {
 248             ps[ i++ ] = phylogeny;
 249         }
 250         return ps;
 251     }
 252
 253     private void reset() {
 254         _valid = true;
 255         _error_count = 0;
 256         _warning_count = 0;
 257         _error_messages = new StringBuffer();
 258         _warning_messages = new StringBuffer();
 259     }
 260
 261     @Override
 262     public void setSource( final Object source ) {
 263         _source = source;
 264     }
 265
 266     public void setValidateAgainstSchema( final String schema_location ) {
 267         _schema_location = schema_location;
 268     }
 269
 270     public void setZippedInputstream( final boolean zipped_inputstream ) {
 271         _zipped_inputstream = zipped_inputstream;
 272     }
 273
 274     public static PhyloXmlParser createPhyloXmlParserXsdValidating() {
 275         final PhyloXmlParser xml_parser = new PhyloXmlParser();
 276         final ClassLoader cl = PhyloXmlParser.class.getClassLoader();
 277         final URL xsd_url = cl.getResource( ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE );
 278         if ( xsd_url != null ) {
 279             xml_parser.setValidateAgainstSchema( xsd_url.toString() );
 280         }
 281         else {
 282             throw new RuntimeException( "failed to get URL for phyloXML XSD from jar file from ["
 283                     + ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE + "]" );
 284         }
 285         return xml_parser;
 286     }
 287
 288     public static PhyloXmlParser createPhyloXmlParser() {
 289         final PhyloXmlParser xml_parser = new PhyloXmlParser();
 290         return xml_parser;
 291     }
 292
 293     private class PhyloXmlParserErrorHandler extends DefaultHandler {
 294
 295         @Override
 296         public void error( final SAXParseException e ) {
 297             ++_error_count;
 298             _valid = false;
 299             throw new PhyloXmlException( "phyloXML error at line " + e.getLineNumber() + ": \n"
 300                     + e.getLocalizedMessage() );
 301         }
 302
 303         @Override
 304         public void fatalError( final SAXParseException e ) {
 305             ++_error_count;
 306             _valid = false;
 307             throw new PhyloXmlException( "fatal XML error at line " + e.getLineNumber() + ": \n"
 308                     + e.getLocalizedMessage() );
 309         }
 310
 311         @Override
 312         public void warning( final SAXParseException e ) {
 313             ++_warning_count;
 314             if ( _error_messages.length() > 1 ) {
 315                 _error_messages.append( ForesterUtil.LINE_SEPARATOR );
 316             }
 317             _warning_messages.append( "[line: " + e.getLineNumber() + "] " + e.getMessage() );
 318         }
 319     }
 320
 321     @Override
 322     public String getName() {
 323         return "phyloXML Parser";
 324     }
 325 }