// Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
//
// Contact: phylosoft @ gmail . com
-// WWW: www.phylosoft.org/forester
+// WWW: https://sites.google.com/site/cmzmasek/home/software/forester
package org.forester.application;
import org.forester.io.parsers.PhylogenyParser;
import org.forester.io.parsers.nexus.NexusPhylogeniesParser;
import org.forester.io.parsers.nhx.NHXParser;
+import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
+import org.forester.io.parsers.util.ParserUtils;
import org.forester.io.writers.PhylogenyWriter;
import org.forester.phylogeny.Phylogeny;
import org.forester.phylogeny.PhylogenyMethods;
+import org.forester.phylogeny.PhylogenyMethods.DESCENDANT_SORT_PRIORITY;
import org.forester.phylogeny.PhylogenyNode;
import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
import org.forester.phylogeny.factories.PhylogenyFactory;
import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
import org.forester.util.CommandLineArguments;
import org.forester.util.ForesterUtil;
-import org.forester.util.ForesterUtil.PhylogenyNodeField;
public class phyloxml_converter {
final static private String ORDER_SUBTREES = "o";
final static private String NO_TREE_LEVEL_INDENDATION = "ni";
final static private String REPLACE_UNDER_SCORES = "ru";
+ final static private String IGNORE_QUOTES = "iqs";
final static private String PRG_NAME = "phyloxml_converter";
- final static private String PRG_VERSION = "1.30";
- final static private String PRG_DATE = "2011.03.01";
+ final static private String PRG_VERSION = "1.301";
+ final static private String PRG_DATE = "2012.08.31";
final static private String E_MAIL = "phylosoft@gmail.com";
final static private String WWW = "www.phylosoft.org/forester/";
final static private boolean SPECIAL = false;
- public static void main( final String args[] ) {
+ public static void main( final String args[] ) throws PhyloXmlDataFormatException {
ForesterUtil.printProgramInformation( PRG_NAME, PRG_VERSION, PRG_DATE, E_MAIL, WWW );
CommandLineArguments cla = null;
try {
allowed_options.add( REPLACE_UNDER_SCORES );
allowed_options.add( EXTRACT_TAXONOMY );
allowed_options.add( EXTRACT_TAXONOMY_PF );
+ allowed_options.add( IGNORE_QUOTES );
if ( cla.getNumberOfNames() != 2 ) {
System.out.println();
System.out.println( "[" + PRG_NAME + "] incorrect number of arguments" );
System.exit( -1 );
}
final String field_option_value = cla.getOptionValue( FIELD_OPTION );
- PhylogenyNodeField field = null;
+ PhylogenyMethods.PhylogenyNodeField field = null;
if ( field_option_value.equals( FIELD_CLADE_NAME ) ) {
- field = PhylogenyNodeField.CLADE_NAME;
+ field = PhylogenyMethods.PhylogenyNodeField.CLADE_NAME;
}
else if ( field_option_value.equals( FIELD_TAXONOMY_CODE ) ) {
- field = PhylogenyNodeField.TAXONOMY_CODE;
+ field = PhylogenyMethods.PhylogenyNodeField.TAXONOMY_CODE;
}
else if ( field_option_value.equals( FIELD_TAXONOMY_SCI_NAME ) ) {
- field = PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME;
+ field = PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME;
}
else if ( field_option_value.equals( FIELD_TAXONOMY_COMM_NAME ) ) {
- field = PhylogenyNodeField.TAXONOMY_COMMON_NAME;
+ field = PhylogenyMethods.PhylogenyNodeField.TAXONOMY_COMMON_NAME;
}
else if ( field_option_value.equals( FIELD_SEQUENCE_GENE_NAME ) ) {
- field = PhylogenyNodeField.SEQUENCE_NAME;
+ field = PhylogenyMethods.PhylogenyNodeField.SEQUENCE_NAME;
}
else if ( field_option_value.equals( FIELD_SEQUENCE_SYMBOL ) ) {
- field = PhylogenyNodeField.SEQUENCE_SYMBOL;
+ field = PhylogenyMethods.PhylogenyNodeField.SEQUENCE_SYMBOL;
}
else if ( field_option_value.equals( FIELD_UNIPROT_TAXONOMY_ID_SPLIT_1 ) ) {
- field = PhylogenyNodeField.TAXONOMY_ID_UNIPROT_1;
+ field = PhylogenyMethods.PhylogenyNodeField.TAXONOMY_ID_UNIPROT_1;
}
else if ( field_option_value.equals( FIELD_UNIPROT_TAXONOMY_ID_SPLIT_2 ) ) {
- field = PhylogenyNodeField.TAXONOMY_ID_UNIPROT_2;
+ field = PhylogenyMethods.PhylogenyNodeField.TAXONOMY_ID_UNIPROT_2;
}
else if ( field_option_value.equals( FIELD_DUMMY ) ) {
}
ForesterUtil.fatalError( PRG_NAME, "unknown value for -\"" + FIELD_OPTION + "\" option: \""
+ field_option_value + "\"" );
}
+ boolean ignore_quotes = false;
+ if ( cla.isOptionSet( IGNORE_QUOTES ) ) {
+ ignore_quotes = true;
+ }
boolean int_values_are_boots = false;
if ( cla.isOptionSet( INTERNAL_NAMES_ARE_BOOT_SUPPPORT ) ) {
int_values_are_boots = true;
Phylogeny[] phys = null;
try {
final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
- final PhylogenyParser parser = ForesterUtil.createParserDependingOnFileType( infile, true );
+ final PhylogenyParser parser = ParserUtils.createParserDependingOnFileType( infile, true );
if ( parser instanceof NHXParser ) {
- if ( ( field != PhylogenyNodeField.TAXONOMY_CODE )
- && ( field != PhylogenyNodeField.TAXONOMY_COMMON_NAME )
- && ( field != PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME ) ) {
+ if ( ( field != PhylogenyMethods.PhylogenyNodeField.TAXONOMY_CODE )
+ && ( field != PhylogenyMethods.PhylogenyNodeField.TAXONOMY_COMMON_NAME )
+ && ( field != PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME ) ) {
if ( extr_taxonomy_pf_only ) {
( ( NHXParser ) parser )
- .setTaxonomyExtraction( ForesterUtil.TAXONOMY_EXTRACTION.PFAM_STYLE_ONLY );
+ .setTaxonomyExtraction( NHXParser.TAXONOMY_EXTRACTION.PFAM_STYLE_STRICT );
replace_underscores = false;
}
else if ( extr_taxonomy ) {
- ( ( NHXParser ) parser ).setTaxonomyExtraction( ForesterUtil.TAXONOMY_EXTRACTION.YES );
+ ( ( NHXParser ) parser )
+ .setTaxonomyExtraction( NHXParser.TAXONOMY_EXTRACTION.PFAM_STYLE_RELAXED );
replace_underscores = false;
}
}
else {
- ( ( NHXParser ) parser ).setTaxonomyExtraction( ForesterUtil.TAXONOMY_EXTRACTION.NO );
+ ( ( NHXParser ) parser ).setTaxonomyExtraction( NHXParser.TAXONOMY_EXTRACTION.NO );
}
( ( NHXParser ) parser ).setReplaceUnderscores( replace_underscores );
- ( ( NHXParser ) parser ).setIgnoreQuotes( false );
+ ( ( NHXParser ) parser ).setIgnoreQuotes( ignore_quotes );
}
else if ( parser instanceof NexusPhylogeniesParser ) {
( ( NexusPhylogeniesParser ) parser ).setReplaceUnderscores( replace_underscores );
}
if ( int_values_are_boots ) {
for( final Phylogeny phy : phys ) {
- ForesterUtil.transferInternalNamesToBootstrapSupport( phy );
+ PhylogenyMethods.transferInternalNamesToBootstrapSupport( phy );
}
}
if ( field != null ) {
for( final Phylogeny phy : phys ) {
- ForesterUtil.transferNodeNameToField( phy, field );
+ PhylogenyMethods.transferNodeNameToField( phy, field, false );
}
}
if ( midpoint_reroot ) {
}
if ( order_subtrees ) {
for( final Phylogeny phy : phys ) {
- phy.orderAppearance( true );
+ PhylogenyMethods.orderAppearance( phy.getRoot(), true, true, DESCENDANT_SORT_PRIORITY.TAXONOMY );
+ phy.externalNodesHaveChanged();
+ phy.clearHashIdToNodeMap();
+ phy.recalculateNumberOfExternalDescendants( true );
}
}
try {
System.out.println();
System.out.println( " options: " );
System.out.println( " -" + INTERNAL_NAMES_ARE_BOOT_SUPPPORT
- + " : internal names in NH or NHX tree are bootstrap support values" );
- System.out.println( " -" + REPLACE_UNDER_SCORES + ": replace all underscores with spaces" );
- System.out.println( " -" + MIDPOINT_REROOT + " : midpoint reroot" );
- System.out.println( " -" + ORDER_SUBTREES + " : order subtrees" );
+ + " : internal names in NH or NHX tree are bootstrap support values" );
+ System.out.println( " -" + REPLACE_UNDER_SCORES + " : replace all underscores with spaces" );
+ System.out.println( " -" + MIDPOINT_REROOT + " : midpoint reroot" );
+ System.out.println( " -" + ORDER_SUBTREES + " : order subtrees" );
System.out
.println( " -"
+ EXTRACT_TAXONOMY
- + ": extract taxonomy to taxonomy code from \"seqname_TAXON\"-style names (cannot be used with the following field options: "
+ + " : extract taxonomy to taxonomy code from \"seqname_TAXON\"-style names (cannot be used with the following field options: "
+ FIELD_TAXONOMY_CODE + ", " + FIELD_TAXONOMY_COMM_NAME + ", " + FIELD_TAXONOMY_SCI_NAME + ")" );
System.out
.println( " -"
+ EXTRACT_TAXONOMY_PF
- + ": extract taxonomy to taxonomy code from Pfam (\"seqname_TAXON/x-y\") style names only (cannot be used with the following field options: "
+ + " : extract taxonomy to taxonomy code from Pfam (\"seqname_TAXON/x-y\") style names only (cannot be used with the following field options: "
+ FIELD_TAXONOMY_CODE + ", " + FIELD_TAXONOMY_COMM_NAME + ", " + FIELD_TAXONOMY_SCI_NAME + ")" );
- System.out.println( " -" + NO_TREE_LEVEL_INDENDATION + ": no tree level indendation in phyloXML output" );
+ System.out.println( " -" + NO_TREE_LEVEL_INDENDATION + " : no tree level indendation in phyloXML output" );
+ System.out.println( " -" + IGNORE_QUOTES + ": ignore quotes and whitespace (e.g. \"a b\" becomes ab)" );
System.out.println();
}
}
git://source.jalview.org / jalview.git / blobdiff