forester/java/src/org/forester/application/decorator.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: www.phylosoft.org/forester
  25
  26 package org.forester.application;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.util.ArrayList;
  31 import java.util.List;
  32 import java.util.Map;
  33
  34 import org.forester.io.parsers.PhylogenyParser;
  35 import org.forester.io.writers.PhylogenyWriter;
  36 import org.forester.phylogeny.Phylogeny;
  37 import org.forester.phylogeny.data.Identifier;
  38 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  39 import org.forester.phylogeny.factories.PhylogenyFactory;
  40 import org.forester.tools.PhylogenyDecorator;
  41 import org.forester.tools.PhylogenyDecorator.FIELD;
  42 import org.forester.util.BasicTable;
  43 import org.forester.util.BasicTableParser;
  44 import org.forester.util.CommandLineArguments;
  45 import org.forester.util.ForesterUtil;
  46
  47 public final class decorator {
  48
  49     private static final String  SEQUENCE_NAME_FIELD                    = "s";
  50     private static final String  TAXONOMY_CODE_FIELD                    = "c";
  51     private static final String  TAXONOMY_SCIENTIFIC_NAME_FIELD         = "sn";
  52     private static final String  DS_FILED                               = "d";
  53     private static final String  SEQUENCE_ANNOTATION_DESC               = "a";
  54     private static final String  NODE_NAME_FIELD                        = "n";
  55     final static private String  PICKY_OPTION                           = "p";
  56     final static private String  FIELD_OPTION                           = "f";
  57     final static private String  MOVE_DOMAIN_NUMBER_OPTION              = "mdn";       // Hidden expert option.
  58     final static private String  TREE_NAME_OPTION                       = "pn";
  59     final static private String  TREE_ID_OPTION                         = "pi";
  60     final static private String  TREE_DESC_OPTION                       = "pd";
  61     final static private String  EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION = "sn";
  62     final static private String  PROCESS_NAME_INTELLIGENTLY_OPTION      = "x";
  63     final static private String  PROCESS_SIMILAR_TO_OPTION              = "xs";
  64     final static private String  CUT_NAME_AFTER_FIRST_SPACE_OPTION      = "c";
  65     final static private String  ALLOW_REMOVAL_OF_CHARS_OPTION          = "r";
  66     final static private String  ADVANCED_TABLE_OPTION                  = "table";
  67     final static private String  KEY_COLUMN                             = "k";
  68     final static private String  VALUE_COLUMN                           = "v";
  69     final static private String  MAPPING_FILE_SEPARATOR_OPTION          = "s";
  70     final static private String  MAPPING_FILE_SEPARATOR_DEFAULT         = ":";
  71     final static private boolean USE_FIRST_SEPARATOR_ONLY               = true;
  72     final static private String  PRG_NAME                               = "decorator";
  73     final static private String  PRG_VERSION                            = "1.10";
  74     final static private String  PRG_DATE                               = "2009.10.08";
  75
  76     private static void argumentsError() {
  77         System.out.println();
  78         System.out.println( decorator.PRG_NAME + " -" + ADVANCED_TABLE_OPTION + " | -f=<c> <phylogenies infile> "
  79                 + "[mapping table file] <phylogenies outfile>" );
  80         System.out.println();
  81         System.out.println( "options:" );
  82         System.out.println();
  83         System.out.println( " -" + ADVANCED_TABLE_OPTION + " : table instead of one to one map (-f=<c>)" );
  84         System.out.println( " -r=<n> : allow to remove up to n characters from the end of the names" );
  85         System.out.println( "          in phylogenies infile if not found (in map) otherwise" );
  86         System.out.println( " -p     : for picky, fails if node name not found in mapping table, default is off" );
  87         System.out.println( " -" + TREE_NAME_OPTION + "=<s>: name for the phylogeny" );
  88         System.out.println( " -" + TREE_ID_OPTION + "=<s>: identifier for the phylogeny (in the form provider:value)" );
  89         System.out.println( " -" + TREE_DESC_OPTION + "=<s>: description for phylogenies" );
  90         System.out.println();
  91         System.out.println();
  92         System.out.println( "advanced options, only available if -" + ADVANCED_TABLE_OPTION + " is not used:" );
  93         System.out.println();
  94         System.out.println( " -f=<c> : field to be replaced: " + NODE_NAME_FIELD + " : node name" );
  95         System.out.println( "                                " + SEQUENCE_ANNOTATION_DESC
  96                 + " : sequence annotation description" );
  97         System.out.println( "                                " + DS_FILED + " : domain structure" );
  98         System.out.println( "                                " + TAXONOMY_CODE_FIELD + " : taxonomy code" );
  99         System.out.println( "                                " + TAXONOMY_SCIENTIFIC_NAME_FIELD
 100                 + ": taxonomy scientific name" );
 101         System.out.println( "                                " + SEQUENCE_NAME_FIELD + " : sequence name" );
 102         System.out.println( " -k=<n> : key column in mapping table (0 based)," );
 103         System.out.println( "          names of the node to be decorated - default is 0" );
 104         System.out.println( " -v=<n> : value column in mapping table (0 based)," );
 105         System.out.println( "          data which with to decorate - default is 1" );
 106         System.out.println( " -" + EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION
 107                 + "    : to extract bracketed scientific names" );
 108         System.out.println( " -s=<c> : column separator in mapping file, default is \""
 109                 + decorator.MAPPING_FILE_SEPARATOR_DEFAULT + "\"" );
 110         System.out.println( " -x     : process name \"intelligently\" (only for -f=n)" );
 111         System.out.println( " -" + decorator.PROCESS_SIMILAR_TO_OPTION
 112                 + "    : process name \"intelligently\" and process information after \"similar to\" (only for -f=n)" );
 113         System.out.println( " -c     : cut name after first space (only for -f=n)" );
 114         System.out.println();
 115         System.exit( -1 );
 116     }
 117
 118     public static void main( final String args[] ) {
 119         ForesterUtil.printProgramInformation( decorator.PRG_NAME, decorator.PRG_VERSION, decorator.PRG_DATE );
 120         if ( ( args.length < 4 ) || ( args.length > 12 ) ) {
 121             decorator.argumentsError();
 122         }
 123         CommandLineArguments cla = null;
 124         try {
 125             cla = new CommandLineArguments( args );
 126         }
 127         catch ( final Exception e ) {
 128             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 129         }
 130         if ( ( cla.getNumberOfNames() < 3 ) || ( cla.getNumberOfNames() > 4 ) ) {
 131             decorator.argumentsError();
 132         }
 133         final File phylogenies_infile = cla.getFile( 0 );
 134         final File mapping_infile = cla.getFile( 1 );
 135         final File phylogenies_outfile = cla.getFile( 2 );
 136         if ( phylogenies_outfile.exists() ) {
 137             ForesterUtil.fatalError( PRG_NAME, "[" + phylogenies_outfile + "] already exists" );
 138         }
 139         final List<String> allowed_options = new ArrayList<String>();
 140         allowed_options.add( decorator.ADVANCED_TABLE_OPTION );
 141         allowed_options.add( decorator.PICKY_OPTION );
 142         allowed_options.add( decorator.FIELD_OPTION );
 143         allowed_options.add( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION );
 144         allowed_options.add( decorator.PROCESS_SIMILAR_TO_OPTION );
 145         allowed_options.add( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION );
 146         allowed_options.add( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 147         allowed_options.add( decorator.KEY_COLUMN );
 148         allowed_options.add( decorator.VALUE_COLUMN );
 149         allowed_options.add( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 150         allowed_options.add( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION );
 151         allowed_options.add( decorator.TREE_NAME_OPTION );
 152         allowed_options.add( decorator.TREE_ID_OPTION );
 153         allowed_options.add( decorator.TREE_DESC_OPTION );
 154         allowed_options.add( decorator.MOVE_DOMAIN_NUMBER_OPTION );
 155         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 156         if ( dissallowed_options.length() > 0 ) {
 157             ForesterUtil.fatalError( decorator.PRG_NAME, "unknown option(s): " + dissallowed_options );
 158         }
 159         final boolean advanced_table = cla.isOptionSet( decorator.ADVANCED_TABLE_OPTION );
 160         if ( !advanced_table ) {
 161             final List<String> mandatory_options = new ArrayList<String>();
 162             mandatory_options.add( decorator.FIELD_OPTION );
 163             final String missing_options = cla.validateMandatoryOptionsAsString( mandatory_options );
 164             if ( missing_options.length() > 0 ) {
 165                 ForesterUtil.fatalError( decorator.PRG_NAME, "missing option(s): " + missing_options );
 166             }
 167         }
 168         final boolean picky = cla.isOptionSet( decorator.PICKY_OPTION );
 169         String separator = decorator.MAPPING_FILE_SEPARATOR_DEFAULT;
 170         if ( cla.isOptionSet( decorator.MAPPING_FILE_SEPARATOR_OPTION ) ) {
 171             if ( advanced_table ) {
 172                 argumentsError();
 173             }
 174             separator = cla.getOptionValue( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 175         }
 176         int key_column = 0;
 177         int value_column = 1;
 178         String field_str = "";
 179         FIELD field = FIELD.NODE_NAME;
 180         int numbers_of_chars_allowed_to_remove_if_not_found_in_map = -1;
 181         boolean cut_name_after_space = false;
 182         boolean process_name_intelligently = false;
 183         boolean process_similar_to = false;
 184         boolean extract_bracketed_scientific_name = false;
 185         boolean move_domain_numbers_at_end_to_middle = false;
 186         String tree_name = "";
 187         String tree_id = "";
 188         String tree_desc = "";
 189         try {
 190             if ( cla.isOptionSet( decorator.TREE_NAME_OPTION ) ) {
 191                 tree_name = cla.getOptionValueAsCleanString( decorator.TREE_NAME_OPTION );
 192             }
 193             if ( cla.isOptionSet( decorator.TREE_ID_OPTION ) ) {
 194                 tree_id = cla.getOptionValueAsCleanString( decorator.TREE_ID_OPTION );
 195             }
 196             if ( cla.isOptionSet( decorator.TREE_DESC_OPTION ) ) {
 197                 tree_desc = cla.getOptionValueAsCleanString( decorator.TREE_DESC_OPTION );
 198             }
 199             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION ) ) {
 200                 if ( advanced_table ) {
 201                     argumentsError();
 202                 }
 203                 extract_bracketed_scientific_name = true;
 204             }
 205             if ( cla.isOptionSet( decorator.KEY_COLUMN ) ) {
 206                 if ( advanced_table ) {
 207                     argumentsError();
 208                 }
 209                 key_column = cla.getOptionValueAsInt( decorator.KEY_COLUMN );
 210             }
 211             if ( cla.isOptionSet( decorator.VALUE_COLUMN ) ) {
 212                 if ( advanced_table ) {
 213                     argumentsError();
 214                 }
 215                 value_column = cla.getOptionValueAsInt( decorator.VALUE_COLUMN );
 216             }
 217             if ( cla.isOptionSet( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION ) ) {
 218                 if ( advanced_table ) {
 219                     argumentsError();
 220                 }
 221                 cut_name_after_space = true;
 222             }
 223             if ( cla.isOptionSet( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION ) ) {
 224                 if ( advanced_table ) {
 225                     argumentsError();
 226                 }
 227                 process_name_intelligently = true;
 228             }
 229             if ( cla.isOptionSet( decorator.PROCESS_SIMILAR_TO_OPTION ) ) {
 230                 if ( advanced_table ) {
 231                     argumentsError();
 232                 }
 233                 process_similar_to = true;
 234             }
 235             if ( cla.isOptionSet( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION ) ) {
 236                 numbers_of_chars_allowed_to_remove_if_not_found_in_map = cla
 237                         .getOptionValueAsInt( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 238             }
 239             if ( cla.isOptionSet( decorator.MOVE_DOMAIN_NUMBER_OPTION ) ) {
 240                 move_domain_numbers_at_end_to_middle = true;
 241             }
 242             if ( cla.isOptionSet( decorator.FIELD_OPTION ) ) {
 243                 field_str = cla.getOptionValue( decorator.FIELD_OPTION );
 244                 if ( field_str.equals( NODE_NAME_FIELD ) ) {
 245                     field = FIELD.NODE_NAME;
 246                 }
 247                 else if ( field_str.equals( SEQUENCE_ANNOTATION_DESC ) ) {
 248                     field = FIELD.SEQUENCE_ANNOTATION_DESC;
 249                 }
 250                 else if ( field_str.equals( DS_FILED ) ) {
 251                     field = FIELD.DOMAIN_STRUCTURE;
 252                     extract_bracketed_scientific_name = false;
 253                 }
 254                 else if ( field_str.equals( TAXONOMY_CODE_FIELD ) ) {
 255                     field = FIELD.TAXONOMY_CODE;
 256                 }
 257                 else if ( field_str.equals( SEQUENCE_NAME_FIELD ) ) {
 258                     field = FIELD.SEQUENCE_NAME;
 259                 }
 260                 else if ( field_str.equals( TAXONOMY_SCIENTIFIC_NAME_FIELD ) ) {
 261                     field = FIELD.TAXONOMY_SCIENTIFIC_NAME;
 262                     extract_bracketed_scientific_name = false;
 263                 }
 264                 else {
 265                     ForesterUtil.fatalError( decorator.PRG_NAME, "unknown value for \"" + decorator.FIELD_OPTION
 266                             + "\" option: \"" + field_str + "\"" );
 267                 }
 268             }
 269         }
 270         catch ( final Exception e ) {
 271             ForesterUtil.fatalError( decorator.PRG_NAME, "error in command line: " + e.getMessage() );
 272         }
 273         if ( ( field != FIELD.NODE_NAME ) && ( cut_name_after_space || process_name_intelligently ) ) {
 274             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x or -c option without -f=n" );
 275         }
 276         if ( ( field != FIELD.NODE_NAME ) && process_similar_to ) {
 277             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 278                     + " option without -f=n" );
 279         }
 280         if ( cut_name_after_space && process_name_intelligently ) {
 281             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x and -c option together" );
 282         }
 283         if ( process_similar_to && process_name_intelligently ) {
 284             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 285                     + " and -x option together" );
 286         }
 287         if ( process_similar_to && cut_name_after_space ) {
 288             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 289                     + " and -c option together" );
 290         }
 291         Phylogeny[] phylogenies = null;
 292         try {
 293             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 294             final PhylogenyParser pp = ForesterUtil.createParserDependingOnFileType( phylogenies_infile, true );
 295             phylogenies = factory.create( phylogenies_infile, pp );
 296         }
 297         catch ( final Exception e ) {
 298             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to read phylgenies from [" + phylogenies_infile
 299                     + "] [" + e.getMessage() + "]" );
 300         }
 301         Map<String, String> map = null;
 302         if ( !advanced_table ) {
 303             BasicTable<String> mapping_table = null;
 304             try {
 305                 mapping_table = BasicTableParser.parse( mapping_infile, separator, decorator.USE_FIRST_SEPARATOR_ONLY );
 306             }
 307             catch ( final Exception e ) {
 308                 ForesterUtil.fatalError( decorator.PRG_NAME,
 309                                          "failed to read [" + mapping_infile + "] [" + e.getMessage() + "]" );
 310             }
 311             if ( ( key_column < 0 ) || ( key_column >= mapping_table.getNumberOfColumns() ) ) {
 312                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for key column" );
 313             }
 314             if ( ( value_column < 0 ) || ( value_column >= mapping_table.getNumberOfColumns() ) ) {
 315                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for value column" );
 316             }
 317             map = mapping_table.getColumnsAsMap( key_column, value_column );
 318         }
 319         if ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id )
 320                 || !ForesterUtil.isEmpty( tree_desc ) ) {
 321             if ( ( phylogenies.length > 1 )
 322                     && ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id ) ) ) {
 323                 ForesterUtil.fatalError( decorator.PRG_NAME,
 324                                          "attempt to set same name or id on more than one phylogeny" );
 325             }
 326             if ( !ForesterUtil.isEmpty( tree_name ) ) {
 327                 phylogenies[ 0 ].setName( tree_name );
 328             }
 329             if ( !ForesterUtil.isEmpty( tree_id ) ) {
 330                 final String[] s_ary = tree_id.split( ":" );
 331                 phylogenies[ 0 ].setIdentifier( new Identifier( s_ary[ 1 ], s_ary[ 0 ] ) );
 332             }
 333             if ( !ForesterUtil.isEmpty( tree_desc ) ) {
 334                 for( int i = 0; i < phylogenies.length; ++i ) {
 335                     phylogenies[ i ].setDescription( tree_desc );
 336                 }
 337             }
 338         }
 339         try {
 340             if ( advanced_table ) {
 341                 Map<String, Map<String, String>> table = null;
 342                 try {
 343                     table = PhylogenyDecorator.parseMappingTable( mapping_infile );
 344                 }
 345                 catch ( final IOException e ) {
 346                     ForesterUtil.fatalError( decorator.PRG_NAME,
 347                                              "failed to read \"" + mapping_infile + "\" [" + e.getMessage() + "]" );
 348                 }
 349                 PhylogenyDecorator.decorate( phylogenies,
 350                                              table,
 351                                              picky,
 352                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 353             }
 354             else {
 355                 PhylogenyDecorator.decorate( phylogenies,
 356                                              map,
 357                                              field,
 358                                              extract_bracketed_scientific_name,
 359                                              picky,
 360                                              cut_name_after_space,
 361                                              process_name_intelligently,
 362                                              process_similar_to,
 363                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 364                                              move_domain_numbers_at_end_to_middle );
 365             }
 366         }
 367         catch ( final NullPointerException e ) {
 368             ForesterUtil.unexpectedFatalError( decorator.PRG_NAME, e );
 369         }
 370         catch ( final Exception e ) {
 371             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to map [" + e + "]" );
 372         }
 373         try {
 374             final PhylogenyWriter w = new PhylogenyWriter();
 375             w.toPhyloXML( phylogenies, 0, phylogenies_outfile, ForesterUtil.getLineSeparator() );
 376         }
 377         catch ( final IOException e ) {
 378             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to write output [" + e.getMessage() + "]" );
 379         }
 380         System.out.println();
 381         ForesterUtil.programMessage( PRG_NAME, "wrote: " + phylogenies_outfile );
 382         ForesterUtil.programMessage( PRG_NAME, "OK." );
 383     }
 384 }