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.parsers.util.ParserUtils;
  36 import org.forester.io.writers.PhylogenyWriter;
  37 import org.forester.phylogeny.Phylogeny;
  38 import org.forester.phylogeny.data.Identifier;
  39 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  40 import org.forester.phylogeny.factories.PhylogenyFactory;
  41 import org.forester.tools.PhylogenyDecorator;
  42 import org.forester.tools.PhylogenyDecorator.FIELD;
  43 import org.forester.util.BasicTable;
  44 import org.forester.util.BasicTableParser;
  45 import org.forester.util.CommandLineArguments;
  46 import org.forester.util.ForesterUtil;
  47
  48 public final class decorator {
  49
  50     private static final String SEQUENCE_NAME_FIELD                     = "s";
  51     private static final String TAXONOMY_CODE_FIELD                     = "c";
  52     private static final String TAXONOMY_SCIENTIFIC_NAME_FIELD          = "sn";
  53     private static final String DS_FILED                                = "d";
  54     private static final String SEQUENCE_ANNOTATION_DESC                = "a";
  55     private static final String NODE_NAME_FIELD                         = "n";
  56     final static private String PICKY_OPTION                            = "p";
  57     final static private String FIELD_OPTION                            = "f";
  58     final static private String TRIM_AFTER_TILDE_OPTION                 = "t";
  59     final static private String TREE_NAME_OPTION                        = "pn";
  60     final static private String TREE_ID_OPTION                          = "pi";
  61     final static private String TREE_DESC_OPTION                        = "pd";
  62     final static private String EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION  = "sn";
  63     final static private String EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION = "tc";
  64     final static private String PROCESS_NAME_INTELLIGENTLY_OPTION       = "x";
  65     final static private String PROCESS_SIMILAR_TO_OPTION               = "xs";
  66     final static private String CUT_NAME_AFTER_FIRST_SPACE_OPTION       = "c";
  67     final static private String ALLOW_REMOVAL_OF_CHARS_OPTION           = "r";
  68     final static private String ADVANCED_TABLE_OPTION                   = "table";
  69     final static private String KEY_COLUMN                              = "k";
  70     final static private String VALUE_COLUMN                            = "v";
  71     final static private String MAPPING_FILE_SEPARATOR_OPTION           = "s";
  72     final static private String MAPPING_FILE_SEPARATOR_DEFAULT          = ": ";
  73     final static private String PRG_NAME                                = "decorator";
  74     final static private String PRG_VERSION                             = "1.12";
  75     final static private String PRG_DATE                                = "2012.12.21";
  76
  77     public static void main( final String args[] ) {
  78         ForesterUtil.printProgramInformation( decorator.PRG_NAME, decorator.PRG_VERSION, decorator.PRG_DATE );
  79         if ( ( args.length < 4 ) || ( args.length > 12 ) ) {
  80             decorator.argumentsError();
  81         }
  82         CommandLineArguments cla = null;
  83         try {
  84             cla = new CommandLineArguments( args );
  85         }
  86         catch ( final Exception e ) {
  87             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  88         }
  89         if ( ( cla.getNumberOfNames() < 3 ) || ( cla.getNumberOfNames() > 4 ) ) {
  90             decorator.argumentsError();
  91         }
  92         final File phylogenies_infile = cla.getFile( 0 );
  93         final File mapping_infile = cla.getFile( 1 );
  94         final File phylogenies_outfile = cla.getFile( 2 );
  95         if ( phylogenies_outfile.exists() ) {
  96             ForesterUtil.fatalError( PRG_NAME, "[" + phylogenies_outfile + "] already exists" );
  97         }
  98         final List<String> allowed_options = new ArrayList<String>();
  99         allowed_options.add( decorator.ADVANCED_TABLE_OPTION );
 100         allowed_options.add( decorator.PICKY_OPTION );
 101         allowed_options.add( decorator.FIELD_OPTION );
 102         allowed_options.add( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION );
 103         allowed_options.add( decorator.PROCESS_SIMILAR_TO_OPTION );
 104         allowed_options.add( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION );
 105         allowed_options.add( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 106         allowed_options.add( decorator.KEY_COLUMN );
 107         allowed_options.add( decorator.VALUE_COLUMN );
 108         allowed_options.add( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 109         allowed_options.add( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION );
 110         allowed_options.add( decorator.EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION );
 111         allowed_options.add( decorator.TREE_NAME_OPTION );
 112         allowed_options.add( decorator.TREE_ID_OPTION );
 113         allowed_options.add( decorator.TREE_DESC_OPTION );
 114         allowed_options.add( decorator.TRIM_AFTER_TILDE_OPTION );
 115         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 116         if ( dissallowed_options.length() > 0 ) {
 117             ForesterUtil.fatalError( decorator.PRG_NAME, "unknown option(s): " + dissallowed_options );
 118         }
 119         final boolean advanced_table = cla.isOptionSet( decorator.ADVANCED_TABLE_OPTION );
 120         if ( !advanced_table ) {
 121             final List<String> mandatory_options = new ArrayList<String>();
 122             mandatory_options.add( decorator.FIELD_OPTION );
 123             final String missing_options = cla.validateMandatoryOptionsAsString( mandatory_options );
 124             if ( missing_options.length() > 0 ) {
 125                 ForesterUtil.fatalError( decorator.PRG_NAME, "missing option(s): " + missing_options );
 126             }
 127         }
 128         final boolean picky = cla.isOptionSet( decorator.PICKY_OPTION );
 129         String separator = decorator.MAPPING_FILE_SEPARATOR_DEFAULT;
 130         if ( cla.isOptionSet( decorator.MAPPING_FILE_SEPARATOR_OPTION ) ) {
 131             if ( advanced_table ) {
 132                 argumentsError();
 133             }
 134             separator = cla.getOptionValue( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 135         }
 136         int key_column = 0;
 137         int value_column = 1;
 138         String field_str = "";
 139         FIELD field = FIELD.NODE_NAME;
 140         int numbers_of_chars_allowed_to_remove_if_not_found_in_map = -1;
 141         boolean cut_name_after_space = false;
 142         boolean process_name_intelligently = false;
 143         boolean process_similar_to = false;
 144         boolean extract_bracketed_scientific_name = false;
 145         boolean extract_bracketed_tax_code = false;
 146         boolean trim_after_tilde = false;
 147         String tree_name = "";
 148         String tree_id = "";
 149         String tree_desc = "";
 150         try {
 151             if ( cla.isOptionSet( decorator.TREE_NAME_OPTION ) ) {
 152                 tree_name = cla.getOptionValueAsCleanString( decorator.TREE_NAME_OPTION );
 153             }
 154             if ( cla.isOptionSet( decorator.TREE_ID_OPTION ) ) {
 155                 tree_id = cla.getOptionValueAsCleanString( decorator.TREE_ID_OPTION );
 156             }
 157             if ( cla.isOptionSet( decorator.TREE_DESC_OPTION ) ) {
 158                 tree_desc = cla.getOptionValueAsCleanString( decorator.TREE_DESC_OPTION );
 159             }
 160             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION ) ) {
 161                 if ( advanced_table ) {
 162                     argumentsError();
 163                 }
 164                 extract_bracketed_scientific_name = true;
 165             }
 166             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION ) ) {
 167                 if ( advanced_table ) {
 168                     argumentsError();
 169                 }
 170                 extract_bracketed_tax_code = true;
 171             }
 172             if ( cla.isOptionSet( decorator.KEY_COLUMN ) ) {
 173                 if ( advanced_table ) {
 174                     argumentsError();
 175                 }
 176                 key_column = cla.getOptionValueAsInt( decorator.KEY_COLUMN );
 177             }
 178             if ( cla.isOptionSet( decorator.VALUE_COLUMN ) ) {
 179                 if ( advanced_table ) {
 180                     argumentsError();
 181                 }
 182                 value_column = cla.getOptionValueAsInt( decorator.VALUE_COLUMN );
 183             }
 184             if ( cla.isOptionSet( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION ) ) {
 185                 if ( advanced_table ) {
 186                     argumentsError();
 187                 }
 188                 cut_name_after_space = true;
 189             }
 190             if ( cla.isOptionSet( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION ) ) {
 191                 if ( advanced_table ) {
 192                     argumentsError();
 193                 }
 194                 process_name_intelligently = true;
 195             }
 196             if ( cla.isOptionSet( decorator.PROCESS_SIMILAR_TO_OPTION ) ) {
 197                 if ( advanced_table ) {
 198                     argumentsError();
 199                 }
 200                 process_similar_to = true;
 201             }
 202             if ( cla.isOptionSet( decorator.TRIM_AFTER_TILDE_OPTION ) ) {
 203                 if ( advanced_table ) {
 204                     argumentsError();
 205                 }
 206                 trim_after_tilde = true;
 207             }
 208             if ( cla.isOptionSet( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION ) ) {
 209                 numbers_of_chars_allowed_to_remove_if_not_found_in_map = cla
 210                         .getOptionValueAsInt( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 211             }
 212             if ( cla.isOptionSet( decorator.FIELD_OPTION ) ) {
 213                 field_str = cla.getOptionValue( decorator.FIELD_OPTION );
 214                 if ( field_str.equals( NODE_NAME_FIELD ) ) {
 215                     field = FIELD.NODE_NAME;
 216                 }
 217                 else if ( field_str.equals( SEQUENCE_ANNOTATION_DESC ) ) {
 218                     field = FIELD.SEQUENCE_ANNOTATION_DESC;
 219                 }
 220                 else if ( field_str.equals( DS_FILED ) ) {
 221                     field = FIELD.DOMAIN_STRUCTURE;
 222                     extract_bracketed_scientific_name = false;
 223                     extract_bracketed_tax_code = false;
 224                 }
 225                 else if ( field_str.equals( TAXONOMY_CODE_FIELD ) ) {
 226                     field = FIELD.TAXONOMY_CODE;
 227                 }
 228                 else if ( field_str.equals( SEQUENCE_NAME_FIELD ) ) {
 229                     field = FIELD.SEQUENCE_NAME;
 230                 }
 231                 else if ( field_str.equals( TAXONOMY_SCIENTIFIC_NAME_FIELD ) ) {
 232                     field = FIELD.TAXONOMY_SCIENTIFIC_NAME;
 233                     extract_bracketed_scientific_name = false;
 234                     extract_bracketed_tax_code = false;
 235                 }
 236                 else {
 237                     ForesterUtil.fatalError( decorator.PRG_NAME, "unknown value for \"" + decorator.FIELD_OPTION
 238                             + "\" option: \"" + field_str + "\"" );
 239                 }
 240             }
 241         }
 242         catch ( final Exception e ) {
 243             ForesterUtil.fatalError( decorator.PRG_NAME, "error in command line: " + e.getMessage() );
 244         }
 245         if ( ( field != FIELD.NODE_NAME ) && ( cut_name_after_space || process_name_intelligently ) ) {
 246             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x or -c option without -f=n" );
 247         }
 248         if ( ( field != FIELD.NODE_NAME ) && process_similar_to ) {
 249             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 250                     + " option without -f=n" );
 251         }
 252         if ( cut_name_after_space && process_name_intelligently ) {
 253             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x and -c option together" );
 254         }
 255         if ( process_similar_to && process_name_intelligently ) {
 256             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 257                     + " and -x option together" );
 258         }
 259         if ( process_similar_to && cut_name_after_space ) {
 260             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 261                     + " and -c option together" );
 262         }
 263         if ( extract_bracketed_scientific_name && extract_bracketed_tax_code ) {
 264             argumentsError();
 265         }
 266         Phylogeny[] phylogenies = null;
 267         try {
 268             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 269             final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( phylogenies_infile, true );
 270             phylogenies = factory.create( phylogenies_infile, pp );
 271         }
 272         catch ( final Exception e ) {
 273             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to read phylgenies from [" + phylogenies_infile
 274                     + "] [" + e.getMessage() + "]" );
 275         }
 276         Map<String, String> map = null;
 277         if ( !advanced_table ) {
 278             BasicTable<String> mapping_table = null;
 279             try {
 280                 mapping_table = BasicTableParser.parse( mapping_infile, separator, false, true );
 281             }
 282             catch ( final Exception e ) {
 283                 ForesterUtil.fatalError( decorator.PRG_NAME,
 284                                          "failed to read [" + mapping_infile + "] [" + e.getMessage() + "]" );
 285             }
 286             if ( ( key_column < 0 ) || ( key_column >= mapping_table.getNumberOfColumns() ) ) {
 287                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for key column" );
 288             }
 289             if ( ( value_column < 0 ) || ( value_column >= mapping_table.getNumberOfColumns() ) ) {
 290                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for value column" );
 291             }
 292             map = mapping_table.getColumnsAsMap( key_column, value_column );
 293         }
 294         if ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id )
 295                 || !ForesterUtil.isEmpty( tree_desc ) ) {
 296             if ( ( phylogenies.length > 1 )
 297                     && ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id ) ) ) {
 298                 ForesterUtil.fatalError( decorator.PRG_NAME,
 299                                          "attempt to set same name or id on more than one phylogeny" );
 300             }
 301             if ( !ForesterUtil.isEmpty( tree_name ) ) {
 302                 phylogenies[ 0 ].setName( tree_name );
 303             }
 304             if ( !ForesterUtil.isEmpty( tree_id ) ) {
 305                 final String[] s_ary = tree_id.split( ":" );
 306                 phylogenies[ 0 ].setIdentifier( new Identifier( s_ary[ 1 ], s_ary[ 0 ] ) );
 307             }
 308             if ( !ForesterUtil.isEmpty( tree_desc ) ) {
 309                 for( final Phylogeny phylogenie : phylogenies ) {
 310                     phylogenie.setDescription( tree_desc );
 311                 }
 312             }
 313         }
 314         try {
 315             if ( advanced_table ) {
 316                 Map<String, Map<String, String>> table = null;
 317                 try {
 318                     table = PhylogenyDecorator.parseMappingTable( mapping_infile );
 319                 }
 320                 catch ( final IOException e ) {
 321                     ForesterUtil.fatalError( decorator.PRG_NAME,
 322                                              "failed to read \"" + mapping_infile + "\" [" + e.getMessage() + "]" );
 323                 }
 324                 PhylogenyDecorator.decorate( phylogenies,
 325                                              table,
 326                                              picky,
 327                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 328             }
 329             else {
 330                 PhylogenyDecorator.decorate( phylogenies,
 331                                              map,
 332                                              field,
 333                                              extract_bracketed_scientific_name,
 334                                              extract_bracketed_tax_code,
 335                                              picky,
 336                                              cut_name_after_space,
 337                                              process_name_intelligently,
 338                                              process_similar_to,
 339                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 340                                              trim_after_tilde );
 341             }
 342         }
 343         catch ( final NullPointerException e ) {
 344             ForesterUtil.unexpectedFatalError( decorator.PRG_NAME, e );
 345         }
 346         catch ( final Exception e ) {
 347             ForesterUtil.fatalError( decorator.PRG_NAME, e.getLocalizedMessage() );
 348         }
 349         try {
 350             final PhylogenyWriter w = new PhylogenyWriter();
 351             w.toPhyloXML( phylogenies, 0, phylogenies_outfile, ForesterUtil.getLineSeparator() );
 352         }
 353         catch ( final IOException e ) {
 354             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to write output [" + e.getMessage() + "]" );
 355         }
 356         System.out.println();
 357         ForesterUtil.programMessage( PRG_NAME, "wrote: " + phylogenies_outfile );
 358         ForesterUtil.programMessage( PRG_NAME, "OK." );
 359     }
 360
 361     private static void argumentsError() {
 362         System.out.println();
 363         System.out.println( decorator.PRG_NAME + " -" + ADVANCED_TABLE_OPTION + " | -f=<c> <phylogenies infile> "
 364                 + "[mapping table file] <phylogenies outfile>" );
 365         System.out.println();
 366         System.out.println( "options:" );
 367         System.out.println();
 368         System.out.println( " -" + ADVANCED_TABLE_OPTION + " : table instead of one to one map (-f=<c>)" );
 369         System.out.println( " -r=<n> : allow to remove up to n characters from the end of the names" );
 370         System.out.println( "          in phylogenies infile if not found (in map) otherwise" );
 371         System.out.println( " -p     : picky, fails if node name not found in mapping table" );
 372         System.out.println( " -" + TREE_NAME_OPTION + "=<s>: name for the phylogeny" );
 373         System.out.println( " -" + TREE_ID_OPTION + "=<s>: identifier for the phylogeny (in the form provider:value)" );
 374         System.out.println( " -" + TREE_DESC_OPTION + "=<s>: description for phylogenies" );
 375         System.out.println();
 376         System.out.println();
 377         System.out.println( "advanced options, only available if -" + ADVANCED_TABLE_OPTION + " is not used:" );
 378         System.out.println();
 379         System.out.println( " -f=<c> : field to be replaced: " + NODE_NAME_FIELD + " : node name" );
 380         System.out.println( "                                " + SEQUENCE_ANNOTATION_DESC
 381                 + " : sequence annotation description" );
 382         System.out.println( "                                " + DS_FILED + " : domain structure" );
 383         System.out.println( "                                " + TAXONOMY_CODE_FIELD + " : taxonomy code" );
 384         System.out.println( "                                " + TAXONOMY_SCIENTIFIC_NAME_FIELD
 385                 + ": taxonomy scientific name" );
 386         System.out.println( "                                " + SEQUENCE_NAME_FIELD + " : sequence name" );
 387         System.out.println( " -k=<n> : key column in mapping table (0 based)," );
 388         System.out.println( "          names of the node to be decorated - default is 0" );
 389         System.out.println( " -v=<n> : value column in mapping table (0 based)," );
 390         System.out.println( "          data which with to decorate - default is 1" );
 391         System.out.println( " -" + EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION
 392                 + "    : to extract bracketed scientific names, e.g. [Nematostella vectensis]" );
 393         System.out.println( " -" + EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION
 394                 + "    : to extract bracketed taxonomic codes, e.g. [NEMVE]" );
 395         System.out.println( " -s=<c> : column separator in mapping file, default is \""
 396                 + decorator.MAPPING_FILE_SEPARATOR_DEFAULT + "\"" );
 397         System.out.println( " -x     : process name \"intelligently\" (only for -f=n)" );
 398         System.out.println( " -" + decorator.PROCESS_SIMILAR_TO_OPTION
 399                 + "    : process name \"intelligently\" and process information after \"similar to\" (only for -f=n)" );
 400         System.out.println( " -c     : cut name after first space (only for -f=n)" );
 401         System.out.println( " -" + decorator.TRIM_AFTER_TILDE_OPTION
 402                 + "     : trim node name to be replaced after tilde" );
 403         System.out.println();
 404         System.exit( -1 );
 405     }
 406 }