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: https://sites.google.com/site/cmzmasek/home/software/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.PhylogenyMethods;
  39 import org.forester.phylogeny.PhylogenyMethods.DESCENDANT_SORT_PRIORITY;
  40 import org.forester.phylogeny.data.Identifier;
  41 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  42 import org.forester.phylogeny.factories.PhylogenyFactory;
  43 import org.forester.tools.PhylogenyDecorator;
  44 import org.forester.tools.PhylogenyDecorator.FIELD;
  45 import org.forester.util.BasicTable;
  46 import org.forester.util.BasicTableParser;
  47 import org.forester.util.CommandLineArguments;
  48 import org.forester.util.ForesterUtil;
  49
  50 public final class decorator {
  51
  52     private static final String SEQUENCE_NAME_FIELD                     = "s";
  53     private static final String TAXONOMY_CODE_FIELD                     = "c";
  54     private static final String TAXONOMY_SCIENTIFIC_NAME_FIELD          = "sn";
  55     private static final String DS_FILED                                = "d";
  56     private static final String SEQUENCE_ANNOTATION_DESC                = "a";
  57     private static final String NODE_NAME_FIELD                         = "n";
  58     final static private String PICKY_OPTION                            = "p";
  59     final static private String FIELD_OPTION                            = "f";
  60     final static private String TRIM_AFTER_TILDE_OPTION                 = "t";
  61     final static private String TREE_NAME_OPTION                        = "pn";
  62     final static private String TREE_ID_OPTION                          = "pi";
  63     final static private String TREE_DESC_OPTION                        = "pd";
  64     final static private String MIDPOINT_ROOT_OPTION                    = "mp";
  65     final static private String ORDER_TREE_OPTION                       = "or";
  66     final static private String EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION  = "sn";
  67     final static private String EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION = "tc";
  68     final static private String PROCESS_NAME_INTELLIGENTLY_OPTION       = "x";
  69     final static private String PROCESS_SIMILAR_TO_OPTION               = "xs";
  70     final static private String CUT_NAME_AFTER_FIRST_SPACE_OPTION       = "c";
  71     final static private String ALLOW_REMOVAL_OF_CHARS_OPTION           = "r";
  72     final static private String ADVANCED_TABLE_OPTION                   = "table";
  73     final static private String KEY_COLUMN                              = "k";
  74     final static private String VALUE_COLUMN                            = "v";
  75     final static private String MAPPING_FILE_SEPARATOR_OPTION           = "s";
  76     final static private char   MAPPING_FILE_SEPARATOR_DEFAULT          = '\t';
  77     final static private String PRG_NAME                                = "decorator";
  78     final static private String PRG_VERSION                             = "1.14";
  79     final static private String PRG_DATE                                = "130426";
  80
  81     public static void main( final String args[] ) {
  82         ForesterUtil.printProgramInformation( decorator.PRG_NAME, decorator.PRG_VERSION, decorator.PRG_DATE );
  83         if ( ( args.length < 4 ) || ( args.length > 12 ) ) {
  84             decorator.argumentsError();
  85         }
  86         CommandLineArguments cla = null;
  87         try {
  88             cla = new CommandLineArguments( args );
  89         }
  90         catch ( final Exception e ) {
  91             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  92         }
  93         if ( ( cla.getNumberOfNames() < 3 ) || ( cla.getNumberOfNames() > 4 ) ) {
  94             decorator.argumentsError();
  95         }
  96         final File phylogenies_infile = cla.getFile( 0 );
  97         final File mapping_infile = cla.getFile( 1 );
  98         final File phylogenies_outfile = cla.getFile( 2 );
  99         if ( phylogenies_outfile.exists() ) {
 100             ForesterUtil.fatalError( PRG_NAME, "[" + phylogenies_outfile + "] already exists" );
 101         }
 102         final List<String> allowed_options = new ArrayList<String>();
 103         allowed_options.add( decorator.ADVANCED_TABLE_OPTION );
 104         allowed_options.add( decorator.PICKY_OPTION );
 105         allowed_options.add( decorator.FIELD_OPTION );
 106         allowed_options.add( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION );
 107         allowed_options.add( decorator.PROCESS_SIMILAR_TO_OPTION );
 108         allowed_options.add( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION );
 109         allowed_options.add( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 110         allowed_options.add( decorator.KEY_COLUMN );
 111         allowed_options.add( decorator.VALUE_COLUMN );
 112         allowed_options.add( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 113         allowed_options.add( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION );
 114         allowed_options.add( decorator.EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION );
 115         allowed_options.add( decorator.TREE_NAME_OPTION );
 116         allowed_options.add( decorator.TREE_ID_OPTION );
 117         allowed_options.add( decorator.TREE_DESC_OPTION );
 118         allowed_options.add( decorator.TRIM_AFTER_TILDE_OPTION );
 119         allowed_options.add( decorator.ORDER_TREE_OPTION );
 120         allowed_options.add( decorator.MIDPOINT_ROOT_OPTION );
 121         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 122         if ( dissallowed_options.length() > 0 ) {
 123             ForesterUtil.fatalError( decorator.PRG_NAME, "unknown option(s): " + dissallowed_options );
 124         }
 125         final boolean advanced_table = cla.isOptionSet( decorator.ADVANCED_TABLE_OPTION );
 126         if ( !advanced_table ) {
 127             final List<String> mandatory_options = new ArrayList<String>();
 128             mandatory_options.add( decorator.FIELD_OPTION );
 129             final String missing_options = cla.validateMandatoryOptionsAsString( mandatory_options );
 130             if ( missing_options.length() > 0 ) {
 131                 ForesterUtil.fatalError( decorator.PRG_NAME, "missing option(s): " + missing_options );
 132             }
 133         }
 134         final boolean picky = cla.isOptionSet( decorator.PICKY_OPTION );
 135         char separator = decorator.MAPPING_FILE_SEPARATOR_DEFAULT;
 136         if ( cla.isOptionSet( decorator.MAPPING_FILE_SEPARATOR_OPTION ) ) {
 137             if ( advanced_table ) {
 138                 argumentsError();
 139             }
 140             separator = cla.getOptionValueAsChar( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 141         }
 142         int key_column = 0;
 143         int value_column = 1;
 144         String field_str = "";
 145         FIELD field = FIELD.NODE_NAME;
 146         int numbers_of_chars_allowed_to_remove_if_not_found_in_map = -1;
 147         boolean cut_name_after_space = false;
 148         boolean process_name_intelligently = false;
 149         boolean process_similar_to = false;
 150         boolean extract_bracketed_scientific_name = false;
 151         boolean extract_bracketed_tax_code = false;
 152         boolean trim_after_tilde = false;
 153         boolean order_tree = false;
 154         boolean midpoint_root = false;
 155         String tree_name = "";
 156         String tree_id = "";
 157         String tree_desc = "";
 158         try {
 159             if ( cla.isOptionSet( decorator.TREE_NAME_OPTION ) ) {
 160                 tree_name = cla.getOptionValueAsCleanString( decorator.TREE_NAME_OPTION );
 161             }
 162             if ( cla.isOptionSet( decorator.TREE_ID_OPTION ) ) {
 163                 tree_id = cla.getOptionValueAsCleanString( decorator.TREE_ID_OPTION );
 164             }
 165             if ( cla.isOptionSet( decorator.TREE_DESC_OPTION ) ) {
 166                 tree_desc = cla.getOptionValueAsCleanString( decorator.TREE_DESC_OPTION );
 167             }
 168             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION ) ) {
 169                 if ( advanced_table ) {
 170                     argumentsError();
 171                 }
 172                 extract_bracketed_scientific_name = true;
 173             }
 174             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION ) ) {
 175                 if ( advanced_table ) {
 176                     argumentsError();
 177                 }
 178                 extract_bracketed_tax_code = true;
 179             }
 180             if ( cla.isOptionSet( decorator.KEY_COLUMN ) ) {
 181                 if ( advanced_table ) {
 182                     argumentsError();
 183                 }
 184                 key_column = cla.getOptionValueAsInt( decorator.KEY_COLUMN );
 185             }
 186             if ( cla.isOptionSet( decorator.VALUE_COLUMN ) ) {
 187                 if ( advanced_table ) {
 188                     argumentsError();
 189                 }
 190                 value_column = cla.getOptionValueAsInt( decorator.VALUE_COLUMN );
 191             }
 192             if ( cla.isOptionSet( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION ) ) {
 193                 if ( advanced_table ) {
 194                     argumentsError();
 195                 }
 196                 cut_name_after_space = true;
 197             }
 198             if ( cla.isOptionSet( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION ) ) {
 199                 if ( advanced_table ) {
 200                     argumentsError();
 201                 }
 202                 process_name_intelligently = true;
 203             }
 204             if ( cla.isOptionSet( decorator.PROCESS_SIMILAR_TO_OPTION ) ) {
 205                 if ( advanced_table ) {
 206                     argumentsError();
 207                 }
 208                 process_similar_to = true;
 209             }
 210             if ( cla.isOptionSet( decorator.TRIM_AFTER_TILDE_OPTION ) ) {
 211                 if ( advanced_table ) {
 212                     argumentsError();
 213                 }
 214                 trim_after_tilde = true;
 215             }
 216             if ( cla.isOptionSet( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION ) ) {
 217                 numbers_of_chars_allowed_to_remove_if_not_found_in_map = cla
 218                         .getOptionValueAsInt( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 219             }
 220             if ( cla.isOptionSet( decorator.MIDPOINT_ROOT_OPTION ) ) {
 221                 midpoint_root = true;
 222             }
 223             if ( cla.isOptionSet( decorator.ORDER_TREE_OPTION ) ) {
 224                 order_tree = true;
 225             }
 226             if ( cla.isOptionSet( decorator.FIELD_OPTION ) ) {
 227                 field_str = cla.getOptionValue( decorator.FIELD_OPTION );
 228                 if ( field_str.equals( NODE_NAME_FIELD ) ) {
 229                     field = FIELD.NODE_NAME;
 230                 }
 231                 else if ( field_str.equals( SEQUENCE_ANNOTATION_DESC ) ) {
 232                     field = FIELD.SEQUENCE_ANNOTATION_DESC;
 233                 }
 234                 else if ( field_str.equals( DS_FILED ) ) {
 235                     field = FIELD.DOMAIN_STRUCTURE;
 236                     extract_bracketed_scientific_name = false;
 237                     extract_bracketed_tax_code = false;
 238                 }
 239                 else if ( field_str.equals( TAXONOMY_CODE_FIELD ) ) {
 240                     field = FIELD.TAXONOMY_CODE;
 241                 }
 242                 else if ( field_str.equals( SEQUENCE_NAME_FIELD ) ) {
 243                     field = FIELD.SEQUENCE_NAME;
 244                 }
 245                 else if ( field_str.equals( TAXONOMY_SCIENTIFIC_NAME_FIELD ) ) {
 246                     field = FIELD.TAXONOMY_SCIENTIFIC_NAME;
 247                     extract_bracketed_scientific_name = false;
 248                     extract_bracketed_tax_code = false;
 249                 }
 250                 else {
 251                     ForesterUtil.fatalError( decorator.PRG_NAME, "unknown value for \"" + decorator.FIELD_OPTION
 252                             + "\" option: \"" + field_str + "\"" );
 253                 }
 254             }
 255         }
 256         catch ( final Exception e ) {
 257             ForesterUtil.fatalError( decorator.PRG_NAME, "error in command line: " + e.getMessage() );
 258         }
 259         if ( ( field != FIELD.NODE_NAME ) && ( cut_name_after_space || process_name_intelligently ) ) {
 260             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x or -c option without -f=n" );
 261         }
 262         if ( ( field != FIELD.NODE_NAME ) && process_similar_to ) {
 263             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 264                     + " option without -f=n" );
 265         }
 266         if ( cut_name_after_space && process_name_intelligently ) {
 267             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x and -c option together" );
 268         }
 269         if ( process_similar_to && process_name_intelligently ) {
 270             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 271                     + " and -x option together" );
 272         }
 273         if ( process_similar_to && cut_name_after_space ) {
 274             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 275                     + " and -c option together" );
 276         }
 277         if ( extract_bracketed_scientific_name && extract_bracketed_tax_code ) {
 278             argumentsError();
 279         }
 280         Phylogeny[] phylogenies = null;
 281         try {
 282             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 283             final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( phylogenies_infile, true );
 284             phylogenies = factory.create( phylogenies_infile, pp );
 285         }
 286         catch ( final Exception e ) {
 287             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to read phylgenies from [" + phylogenies_infile
 288                     + "] [" + e.getMessage() + "]" );
 289         }
 290         Map<String, String> map = null;
 291         if ( !advanced_table ) {
 292             BasicTable<String> mapping_table = null;
 293             try {
 294                 mapping_table = BasicTableParser.parse( mapping_infile, separator, true, false );
 295             }
 296             catch ( final Exception e ) {
 297                 ForesterUtil.fatalError( decorator.PRG_NAME,
 298                                          "failed to read [" + mapping_infile + "] [" + e.getMessage() + "]" );
 299             }
 300             if ( ( key_column < 0 ) || ( key_column >= mapping_table.getNumberOfColumns() ) ) {
 301                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for key column" );
 302             }
 303             if ( ( value_column < 0 ) || ( value_column >= mapping_table.getNumberOfColumns() ) ) {
 304                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for value column" );
 305             }
 306             map = mapping_table.getColumnsAsMap( key_column, value_column );
 307         }
 308         if ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id )
 309                 || !ForesterUtil.isEmpty( tree_desc ) ) {
 310             if ( ( phylogenies.length > 1 )
 311                     && ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id ) ) ) {
 312                 ForesterUtil.fatalError( decorator.PRG_NAME,
 313                                          "attempt to set same name or id on more than one phylogeny" );
 314             }
 315             if ( !ForesterUtil.isEmpty( tree_name ) ) {
 316                 phylogenies[ 0 ].setName( tree_name );
 317             }
 318             if ( !ForesterUtil.isEmpty( tree_id ) ) {
 319                 final String[] s_ary = tree_id.split( ":" );
 320                 phylogenies[ 0 ].setIdentifier( new Identifier( s_ary[ 1 ], s_ary[ 0 ] ) );
 321             }
 322             if ( !ForesterUtil.isEmpty( tree_desc ) ) {
 323                 for( final Phylogeny phylogenie : phylogenies ) {
 324                     phylogenie.setDescription( tree_desc );
 325                 }
 326             }
 327         }
 328         try {
 329             if ( advanced_table ) {
 330                 Map<String, Map<String, String>> table = null;
 331                 try {
 332                     table = PhylogenyDecorator.parseMappingTable( mapping_infile );
 333                 }
 334                 catch ( final IOException e ) {
 335                     ForesterUtil.fatalError( decorator.PRG_NAME,
 336                                              "failed to read \"" + mapping_infile + "\" [" + e.getMessage() + "]" );
 337                 }
 338                 PhylogenyDecorator.decorate( phylogenies,
 339                                              table,
 340                                              picky,
 341                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 342             }
 343             else {
 344                 PhylogenyDecorator.decorate( phylogenies,
 345                                              map,
 346                                              field,
 347                                              extract_bracketed_scientific_name,
 348                                              extract_bracketed_tax_code,
 349                                              picky,
 350                                              cut_name_after_space,
 351                                              process_name_intelligently,
 352                                              process_similar_to,
 353                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 354                                              trim_after_tilde );
 355             }
 356         }
 357         catch ( final NullPointerException e ) {
 358             ForesterUtil.unexpectedFatalError( decorator.PRG_NAME, e );
 359         }
 360         catch ( final Exception e ) {
 361             ForesterUtil.fatalError( decorator.PRG_NAME, e.getLocalizedMessage() );
 362         }
 363         if ( midpoint_root || order_tree ) {
 364             for( final Phylogeny phy : phylogenies ) {
 365                 if ( midpoint_root ) {
 366                     PhylogenyMethods.midpointRoot( phy );
 367                 }
 368                 if ( order_tree ) {
 369                     PhylogenyMethods.orderAppearance( phy.getRoot(), true, true, DESCENDANT_SORT_PRIORITY.TAXONOMY );
 370                 }
 371             }
 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
 385     private static void argumentsError() {
 386         System.out.println();
 387         System.out.println( decorator.PRG_NAME + " -" + ADVANCED_TABLE_OPTION + " | -f=<c> <phylogenies infile> "
 388                 + "[mapping table file] <phylogenies outfile>" );
 389         System.out.println();
 390         System.out.println( "options:" );
 391         System.out.println();
 392         System.out.println( " -" + ADVANCED_TABLE_OPTION + " : table instead of one to one map (-f=<c>)" );
 393         System.out.println( " -r=<n> : allow to remove up to n characters from the end of the names" );
 394         System.out.println( "          in phylogenies infile if not found (in map) otherwise" );
 395         System.out.println( " -p     : picky, fails if node name not found in mapping table" );
 396         System.out.println( " -" + TREE_NAME_OPTION + "=<s>: name for the phylogeny" );
 397         System.out.println( " -" + TREE_ID_OPTION + "=<s>: identifier for the phylogeny (in the form provider:value)" );
 398         System.out.println( " -" + TREE_DESC_OPTION + "=<s>: description for phylogenies" );
 399         System.out.println();
 400         System.out.println();
 401         System.out.println( "advanced options, only available if -" + ADVANCED_TABLE_OPTION + " is not used:" );
 402         System.out.println();
 403         System.out.println( " -f=<c> : field to be replaced: " + NODE_NAME_FIELD + " : node name" );
 404         System.out.println( "                                " + SEQUENCE_ANNOTATION_DESC
 405                 + " : sequence annotation description" );
 406         System.out.println( "                                " + DS_FILED + " : domain structure" );
 407         System.out.println( "                                " + TAXONOMY_CODE_FIELD + " : taxonomy code" );
 408         System.out.println( "                                " + TAXONOMY_SCIENTIFIC_NAME_FIELD
 409                 + ": taxonomy scientific name" );
 410         System.out.println( "                                " + SEQUENCE_NAME_FIELD + " : sequence name" );
 411         System.out.println( " -k=<n> : key column in mapping table (0 based)," );
 412         System.out.println( "          names of the node to be decorated - default is 0" );
 413         System.out.println( " -v=<n> : value column in mapping table (0 based)," );
 414         System.out.println( "          data which with to decorate - default is 1" );
 415         System.out.println( " -" + EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION
 416                 + "    : to extract bracketed scientific names, e.g. [Nematostella vectensis]" );
 417         System.out.println( " -" + EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION
 418                 + "    : to extract bracketed taxonomic codes, e.g. [NEMVE]" );
 419         System.out.println( " -s=<c> : column separator in mapping file, default is \""
 420                 + decorator.MAPPING_FILE_SEPARATOR_DEFAULT + "\"" );
 421         System.out.println( " -x     : process name \"intelligently\" (only for -f=n)" );
 422         System.out.println( " -" + decorator.PROCESS_SIMILAR_TO_OPTION
 423                 + "    : process name \"intelligently\" and process information after \"similar to\" (only for -f=n)" );
 424         System.out.println( " -c     : cut name after first space (only for -f=n)" );
 425         System.out.println( " -" + decorator.TRIM_AFTER_TILDE_OPTION
 426                 + "     : trim node name to be replaced after tilde" );
 427         System.out.println( " -" + decorator.MIDPOINT_ROOT_OPTION + "     : to midpoint-root the tree" );
 428         System.out.println( " -" + decorator.ORDER_TREE_OPTION + "     : to order tree branches" );
 429         System.out.println();
 430         System.exit( -1 );
 431     }
 432 }