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                             = "1.11";
  75     final static private String PRG_DATE                                = "2012.09.15";
  76
  77     private static void argumentsError() {
  78         System.out.println();
  79         System.out.println( decorator.PRG_NAME + " -" + ADVANCED_TABLE_OPTION + " | -f=<c> <phylogenies infile> "
  80                 + "[mapping table file] <phylogenies outfile>" );
  81         System.out.println();
  82         System.out.println( "options:" );
  83         System.out.println();
  84         System.out.println( " -" + ADVANCED_TABLE_OPTION + " : table instead of one to one map (-f=<c>)" );
  85         System.out.println( " -r=<n> : allow to remove up to n characters from the end of the names" );
  86         System.out.println( "          in phylogenies infile if not found (in map) otherwise" );
  87         System.out.println( " -p     : picky, fails if node name not found in mapping table" );
  88         System.out.println( " -" + TREE_NAME_OPTION + "=<s>: name for the phylogeny" );
  89         System.out.println( " -" + TREE_ID_OPTION + "=<s>: identifier for the phylogeny (in the form provider:value)" );
  90         System.out.println( " -" + TREE_DESC_OPTION + "=<s>: description for phylogenies" );
  91         System.out.println();
  92         System.out.println();
  93         System.out.println( "advanced options, only available if -" + ADVANCED_TABLE_OPTION + " is not used:" );
  94         System.out.println();
  95         System.out.println( " -f=<c> : field to be replaced: " + NODE_NAME_FIELD + " : node name" );
  96         System.out.println( "                                " + SEQUENCE_ANNOTATION_DESC
  97                 + " : sequence annotation description" );
  98         System.out.println( "                                " + DS_FILED + " : domain structure" );
  99         System.out.println( "                                " + TAXONOMY_CODE_FIELD + " : taxonomy code" );
 100         System.out.println( "                                " + TAXONOMY_SCIENTIFIC_NAME_FIELD
 101                 + ": taxonomy scientific name" );
 102         System.out.println( "                                " + SEQUENCE_NAME_FIELD + " : sequence name" );
 103         System.out.println( " -k=<n> : key column in mapping table (0 based)," );
 104         System.out.println( "          names of the node to be decorated - default is 0" );
 105         System.out.println( " -v=<n> : value column in mapping table (0 based)," );
 106         System.out.println( "          data which with to decorate - default is 1" );
 107         System.out.println( " -" + EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION
 108                 + "    : to extract bracketed scientific names, e.g. [Nematostella vectensis]" );
 109         System.out.println( " -" + EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION
 110                 + "    : to extract bracketed taxonomic codes, e.g. [NEMVE]" );
 111         System.out.println( " -s=<c> : column separator in mapping file, default is \""
 112                 + decorator.MAPPING_FILE_SEPARATOR_DEFAULT + "\"" );
 113         System.out.println( " -x     : process name \"intelligently\" (only for -f=n)" );
 114         System.out.println( " -" + decorator.PROCESS_SIMILAR_TO_OPTION
 115                 + "    : process name \"intelligently\" and process information after \"similar to\" (only for -f=n)" );
 116         System.out.println( " -c     : cut name after first space (only for -f=n)" );
 117         System.out.println( " -" + decorator.TRIM_AFTER_TILDE_OPTION
 118                 + "     : trim node name to be replaced after tilde" );
 119         System.out.println();
 120         System.exit( -1 );
 121     }
 122
 123     public static void main( final String args[] ) {
 124         ForesterUtil.printProgramInformation( decorator.PRG_NAME, decorator.PRG_VERSION, decorator.PRG_DATE );
 125         if ( ( args.length < 4 ) || ( args.length > 12 ) ) {
 126             decorator.argumentsError();
 127         }
 128         CommandLineArguments cla = null;
 129         try {
 130             cla = new CommandLineArguments( args );
 131         }
 132         catch ( final Exception e ) {
 133             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 134         }
 135         if ( ( cla.getNumberOfNames() < 3 ) || ( cla.getNumberOfNames() > 4 ) ) {
 136             decorator.argumentsError();
 137         }
 138         final File phylogenies_infile = cla.getFile( 0 );
 139         final File mapping_infile = cla.getFile( 1 );
 140         final File phylogenies_outfile = cla.getFile( 2 );
 141         if ( phylogenies_outfile.exists() ) {
 142             ForesterUtil.fatalError( PRG_NAME, "[" + phylogenies_outfile + "] already exists" );
 143         }
 144         final List<String> allowed_options = new ArrayList<String>();
 145         allowed_options.add( decorator.ADVANCED_TABLE_OPTION );
 146         allowed_options.add( decorator.PICKY_OPTION );
 147         allowed_options.add( decorator.FIELD_OPTION );
 148         allowed_options.add( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION );
 149         allowed_options.add( decorator.PROCESS_SIMILAR_TO_OPTION );
 150         allowed_options.add( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION );
 151         allowed_options.add( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 152         allowed_options.add( decorator.KEY_COLUMN );
 153         allowed_options.add( decorator.VALUE_COLUMN );
 154         allowed_options.add( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 155         allowed_options.add( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION );
 156         allowed_options.add( decorator.TREE_NAME_OPTION );
 157         allowed_options.add( decorator.TREE_ID_OPTION );
 158         allowed_options.add( decorator.TREE_DESC_OPTION );
 159         allowed_options.add( decorator.TRIM_AFTER_TILDE_OPTION );
 160         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 161         if ( dissallowed_options.length() > 0 ) {
 162             ForesterUtil.fatalError( decorator.PRG_NAME, "unknown option(s): " + dissallowed_options );
 163         }
 164         final boolean advanced_table = cla.isOptionSet( decorator.ADVANCED_TABLE_OPTION );
 165         if ( !advanced_table ) {
 166             final List<String> mandatory_options = new ArrayList<String>();
 167             mandatory_options.add( decorator.FIELD_OPTION );
 168             final String missing_options = cla.validateMandatoryOptionsAsString( mandatory_options );
 169             if ( missing_options.length() > 0 ) {
 170                 ForesterUtil.fatalError( decorator.PRG_NAME, "missing option(s): " + missing_options );
 171             }
 172         }
 173         final boolean picky = cla.isOptionSet( decorator.PICKY_OPTION );
 174         String separator = decorator.MAPPING_FILE_SEPARATOR_DEFAULT;
 175         if ( cla.isOptionSet( decorator.MAPPING_FILE_SEPARATOR_OPTION ) ) {
 176             if ( advanced_table ) {
 177                 argumentsError();
 178             }
 179             separator = cla.getOptionValue( decorator.MAPPING_FILE_SEPARATOR_OPTION );
 180         }
 181         int key_column = 0;
 182         int value_column = 1;
 183         String field_str = "";
 184         FIELD field = FIELD.NODE_NAME;
 185         int numbers_of_chars_allowed_to_remove_if_not_found_in_map = -1;
 186         boolean cut_name_after_space = false;
 187         boolean process_name_intelligently = false;
 188         boolean process_similar_to = false;
 189         boolean extract_bracketed_scientific_name = false;
 190         boolean extract_bracketed_tax_code = false;
 191         boolean trim_after_tilde = false;
 192         String tree_name = "";
 193         String tree_id = "";
 194         String tree_desc = "";
 195         try {
 196             if ( cla.isOptionSet( decorator.TREE_NAME_OPTION ) ) {
 197                 tree_name = cla.getOptionValueAsCleanString( decorator.TREE_NAME_OPTION );
 198             }
 199             if ( cla.isOptionSet( decorator.TREE_ID_OPTION ) ) {
 200                 tree_id = cla.getOptionValueAsCleanString( decorator.TREE_ID_OPTION );
 201             }
 202             if ( cla.isOptionSet( decorator.TREE_DESC_OPTION ) ) {
 203                 tree_desc = cla.getOptionValueAsCleanString( decorator.TREE_DESC_OPTION );
 204             }
 205             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_SCIENTIC_NAME_OPTION ) ) {
 206                 if ( advanced_table ) {
 207                     argumentsError();
 208                 }
 209                 extract_bracketed_scientific_name = true;
 210             }
 211             if ( cla.isOptionSet( decorator.EXTRACT_BRACKETED_TAXONOMIC_CODE_OPTION ) ) {
 212                 if ( advanced_table ) {
 213                     argumentsError();
 214                 }
 215                 extract_bracketed_tax_code = true;
 216             }
 217             if ( cla.isOptionSet( decorator.KEY_COLUMN ) ) {
 218                 if ( advanced_table ) {
 219                     argumentsError();
 220                 }
 221                 key_column = cla.getOptionValueAsInt( decorator.KEY_COLUMN );
 222             }
 223             if ( cla.isOptionSet( decorator.VALUE_COLUMN ) ) {
 224                 if ( advanced_table ) {
 225                     argumentsError();
 226                 }
 227                 value_column = cla.getOptionValueAsInt( decorator.VALUE_COLUMN );
 228             }
 229             if ( cla.isOptionSet( decorator.CUT_NAME_AFTER_FIRST_SPACE_OPTION ) ) {
 230                 if ( advanced_table ) {
 231                     argumentsError();
 232                 }
 233                 cut_name_after_space = true;
 234             }
 235             if ( cla.isOptionSet( decorator.PROCESS_NAME_INTELLIGENTLY_OPTION ) ) {
 236                 if ( advanced_table ) {
 237                     argumentsError();
 238                 }
 239                 process_name_intelligently = true;
 240             }
 241             if ( cla.isOptionSet( decorator.PROCESS_SIMILAR_TO_OPTION ) ) {
 242                 if ( advanced_table ) {
 243                     argumentsError();
 244                 }
 245                 process_similar_to = true;
 246             }
 247             if ( cla.isOptionSet( decorator.TRIM_AFTER_TILDE_OPTION ) ) {
 248                 if ( advanced_table ) {
 249                     argumentsError();
 250                 }
 251                 trim_after_tilde = true;
 252             }
 253             if ( cla.isOptionSet( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION ) ) {
 254                 numbers_of_chars_allowed_to_remove_if_not_found_in_map = cla
 255                         .getOptionValueAsInt( decorator.ALLOW_REMOVAL_OF_CHARS_OPTION );
 256             }
 257             if ( cla.isOptionSet( decorator.FIELD_OPTION ) ) {
 258                 field_str = cla.getOptionValue( decorator.FIELD_OPTION );
 259                 if ( field_str.equals( NODE_NAME_FIELD ) ) {
 260                     field = FIELD.NODE_NAME;
 261                 }
 262                 else if ( field_str.equals( SEQUENCE_ANNOTATION_DESC ) ) {
 263                     field = FIELD.SEQUENCE_ANNOTATION_DESC;
 264                 }
 265                 else if ( field_str.equals( DS_FILED ) ) {
 266                     field = FIELD.DOMAIN_STRUCTURE;
 267                     extract_bracketed_scientific_name = false;
 268                     extract_bracketed_tax_code = false;
 269                 }
 270                 else if ( field_str.equals( TAXONOMY_CODE_FIELD ) ) {
 271                     field = FIELD.TAXONOMY_CODE;
 272                 }
 273                 else if ( field_str.equals( SEQUENCE_NAME_FIELD ) ) {
 274                     field = FIELD.SEQUENCE_NAME;
 275                 }
 276                 else if ( field_str.equals( TAXONOMY_SCIENTIFIC_NAME_FIELD ) ) {
 277                     field = FIELD.TAXONOMY_SCIENTIFIC_NAME;
 278                     extract_bracketed_scientific_name = false;
 279                     extract_bracketed_tax_code = false;
 280                 }
 281                 else {
 282                     ForesterUtil.fatalError( decorator.PRG_NAME, "unknown value for \"" + decorator.FIELD_OPTION
 283                             + "\" option: \"" + field_str + "\"" );
 284                 }
 285             }
 286         }
 287         catch ( final Exception e ) {
 288             ForesterUtil.fatalError( decorator.PRG_NAME, "error in command line: " + e.getMessage() );
 289         }
 290         if ( ( field != FIELD.NODE_NAME ) && ( cut_name_after_space || process_name_intelligently ) ) {
 291             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x or -c option without -f=n" );
 292         }
 293         if ( ( field != FIELD.NODE_NAME ) && process_similar_to ) {
 294             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 295                     + " option without -f=n" );
 296         }
 297         if ( cut_name_after_space && process_name_intelligently ) {
 298             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -x and -c option together" );
 299         }
 300         if ( process_similar_to && process_name_intelligently ) {
 301             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 302                     + " and -x option together" );
 303         }
 304         if ( process_similar_to && cut_name_after_space ) {
 305             ForesterUtil.fatalError( decorator.PRG_NAME, "attempt to use -" + decorator.PROCESS_SIMILAR_TO_OPTION
 306                     + " and -c option together" );
 307         }
 308         if ( extract_bracketed_scientific_name && extract_bracketed_tax_code ) {
 309             argumentsError();
 310         }
 311         Phylogeny[] phylogenies = null;
 312         try {
 313             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 314             final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( phylogenies_infile, true );
 315             phylogenies = factory.create( phylogenies_infile, pp );
 316         }
 317         catch ( final Exception e ) {
 318             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to read phylgenies from [" + phylogenies_infile
 319                     + "] [" + e.getMessage() + "]" );
 320         }
 321         Map<String, String> map = null;
 322         if ( !advanced_table ) {
 323             BasicTable<String> mapping_table = null;
 324             try {
 325                 mapping_table = BasicTableParser.parse( mapping_infile, separator, false, true );
 326             }
 327             catch ( final Exception e ) {
 328                 ForesterUtil.fatalError( decorator.PRG_NAME,
 329                                          "failed to read [" + mapping_infile + "] [" + e.getMessage() + "]" );
 330             }
 331             if ( ( key_column < 0 ) || ( key_column >= mapping_table.getNumberOfColumns() ) ) {
 332                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for key column" );
 333             }
 334             if ( ( value_column < 0 ) || ( value_column >= mapping_table.getNumberOfColumns() ) ) {
 335                 ForesterUtil.fatalError( decorator.PRG_NAME, "illegal value for value column" );
 336             }
 337             map = mapping_table.getColumnsAsMap( key_column, value_column );
 338         }
 339         if ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id )
 340                 || !ForesterUtil.isEmpty( tree_desc ) ) {
 341             if ( ( phylogenies.length > 1 )
 342                     && ( !ForesterUtil.isEmpty( tree_name ) || !ForesterUtil.isEmpty( tree_id ) ) ) {
 343                 ForesterUtil.fatalError( decorator.PRG_NAME,
 344                                          "attempt to set same name or id on more than one phylogeny" );
 345             }
 346             if ( !ForesterUtil.isEmpty( tree_name ) ) {
 347                 phylogenies[ 0 ].setName( tree_name );
 348             }
 349             if ( !ForesterUtil.isEmpty( tree_id ) ) {
 350                 final String[] s_ary = tree_id.split( ":" );
 351                 phylogenies[ 0 ].setIdentifier( new Identifier( s_ary[ 1 ], s_ary[ 0 ] ) );
 352             }
 353             if ( !ForesterUtil.isEmpty( tree_desc ) ) {
 354                 for( int i = 0; i < phylogenies.length; ++i ) {
 355                     phylogenies[ i ].setDescription( tree_desc );
 356                 }
 357             }
 358         }
 359         try {
 360             if ( advanced_table ) {
 361                 Map<String, Map<String, String>> table = null;
 362                 try {
 363                     table = PhylogenyDecorator.parseMappingTable( mapping_infile );
 364                 }
 365                 catch ( final IOException e ) {
 366                     ForesterUtil.fatalError( decorator.PRG_NAME,
 367                                              "failed to read \"" + mapping_infile + "\" [" + e.getMessage() + "]" );
 368                 }
 369                 PhylogenyDecorator.decorate( phylogenies,
 370                                              table,
 371                                              picky,
 372                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 373             }
 374             else {
 375                 PhylogenyDecorator.decorate( phylogenies,
 376                                              map,
 377                                              field,
 378                                              extract_bracketed_scientific_name,
 379                                              extract_bracketed_tax_code,
 380                                              picky,
 381                                              cut_name_after_space,
 382                                              process_name_intelligently,
 383                                              process_similar_to,
 384                                              numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 385                                              trim_after_tilde );
 386             }
 387         }
 388         catch ( final NullPointerException e ) {
 389             ForesterUtil.unexpectedFatalError( decorator.PRG_NAME, e );
 390         }
 391         catch ( final Exception e ) {
 392             ForesterUtil.fatalError( decorator.PRG_NAME, e.getLocalizedMessage() );
 393         }
 394         try {
 395             final PhylogenyWriter w = new PhylogenyWriter();
 396             w.toPhyloXML( phylogenies, 0, phylogenies_outfile, ForesterUtil.getLineSeparator() );
 397         }
 398         catch ( final IOException e ) {
 399             ForesterUtil.fatalError( decorator.PRG_NAME, "failed to write output [" + e.getMessage() + "]" );
 400         }
 401         System.out.println();
 402         ForesterUtil.programMessage( PRG_NAME, "wrote: " + phylogenies_outfile );
 403         ForesterUtil.programMessage( PRG_NAME, "OK." );
 404     }
 405 }