public final class ParserUtils {
- final public static String TAX_CODE = "(?:[A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA|CAP";
+ final public static String TAX_CODE = "(?:[A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA";
+ final public static Pattern TAXOMONY_CODE_PATTERN_A = Pattern.compile( "(?:\\b|_)(" + TAX_CODE + ")\\b" );
+ final public static Pattern TAXOMONY_CODE_PATTERN_BRACKETED = Pattern.compile( "\\[(" + TAX_CODE + ")\\]" );
+ final public static Pattern TAXOMONY_CODE_PATTERN_PFR = Pattern.compile( "(?:\\b|_)[a-zA-Z0-9]{3,}_("
+ + TAX_CODE + ")\\b" );
final public static Pattern TAXOMONY_SN_PATTERN = Pattern
- .compile( "(?:\\b|_)[a-zA-Z0-9]{3,}_([A-Z][a-z]+_[a-z]{2,}(?:_[a-z][a-z0-9_]+)?)\\b" );
+ .compile( "(?:\\b|_)[a-zA-Z0-9]{3,}_([A-Z][a-z]{2,30}_[a-z]{3,30}(?:_[a-z][a-z0-9_]+)?)\\b" );
+ final public static Pattern TAXOMONY_SN_PATTERN_SN = Pattern
+ .compile( "\\b([A-Z][a-z]{2,30}[_ ][a-z]{3,30}(?:[_ ][a-z]{3,30})?)(?:\\b|_)?" );
+ final public static Pattern TAXOMONY_SN_PATTERN_STRAIN_1 = Pattern
+ .compile( "\\b([A-Z][a-z]{2,30}[_ ][a-z]{3,30}[_ ](?:str|subsp|var)[a-z]{0,5}\\.?[_ ]\\S{1,60})(?:\\b|_)" );
+ final public static Pattern TAXOMONY_SN_PATTERN_STRAIN_2 = Pattern
+ .compile( "\\b([A-Z][a-z]{2,30}[_ ][a-z]{3,30}[_ ]\\((?:str|subsp|var)[a-z]{0,5}\\.?[_ ]\\S{1,60}\\))(?:\\b|_)?" );
+ final public static Pattern TAXOMONY_SN_PATTERN_SP = Pattern
+ .compile( "\\b([A-Z][a-z]{2,30}[_ ]sp\\.)(?:\\b|_)?" );
+
+ final public static Pattern TAXOMONY_SN_PATTERN_GENUS = Pattern.compile( "([A-Z][a-z]{2,30})" );
final private static Pattern TAXOMONY_CODE_PATTERN_PFS = Pattern.compile( "(?:\\b|_)[A-Z0-9]{4,}_("
+ TAX_CODE + ")/\\d+-\\d+\\b" );
- final public static Pattern TAXOMONY_CODE_PATTERN_PFR = Pattern.compile( "(?:\\b|_)[a-zA-Z0-9]{3,}_("
- + TAX_CODE + ")\\b" );
- final public static Pattern TAXOMONY_CODE_PATTERN_A = Pattern.compile( "(?:\\b|_)(" + TAX_CODE + ")\\b" );
- final public static Pattern TAXOMONY_CODE_PATTERN_4 = Pattern.compile( "\\[(" + TAX_CODE + ")\\]" );
- final public static Pattern TAXOMONY_CODE_PATTERN_6 = Pattern
- .compile( "\\[([A-Z9][A-Z]{2}[A-Z0-9]{3})\\]" );
- final private static Pattern TAXOMONY_UNIPROT_ID_PATTERN_A = Pattern.compile( "(?:\\b|_)(\\d{1,7})\\b" );
+ final private static Pattern TAXOMONY_UNIPROT_ID_PATTERN_PFR = Pattern
+ .compile( "(?:\\b|_)[A-Z0-9]{1,}_(\\d{1,7})\\b" );
final private static Pattern TAXOMONY_UNIPROT_ID_PATTERN_PFS = Pattern
.compile( "(?:\\b|_)[A-Z0-9]{4,}_(\\d{1,7})/\\d+-\\d+\\b" );
- final private static Pattern TAXOMONY_UNIPROT_ID_PATTERN_PFR = Pattern
- .compile( "(?:\\b|_)[a-zA-Z0-9]{3,}_(\\d{1,7})\\b" );
final public static PhylogenyParser createParserDependingFileContents( final File file,
final boolean phyloxml_validate_against_xsd )
PhylogenyParser parser = null;
final String first_line = ForesterUtil.getFirstLine( file ).trim().toLowerCase();
if ( first_line.startsWith( "<" ) ) {
- parser = new PhyloXmlParser();
+ parser = PhyloXmlParser.createPhyloXmlParser();
if ( phyloxml_validate_against_xsd ) {
final ClassLoader cl = PhyloXmlParser.class.getClassLoader();
final URL xsd_url = cl.getResource( ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE );
if ( parser == null ) {
parser = createParserDependingFileContents( file, phyloxml_validate_against_xsd );
}
- return parser;
- }
-
- /**
- * Return null if it can not guess the parser to use based on name suffix.
- *
- * @param filename
- * @return
- */
- final public static PhylogenyParser createParserDependingOnSuffix( final String filename,
- final boolean phyloxml_validate_against_xsd ) {
- PhylogenyParser parser = null;
- final String filename_lc = filename.toLowerCase();
- if ( filename_lc.endsWith( ".tol" ) || filename_lc.endsWith( ".tolxml" ) || filename_lc.endsWith( ".tol.zip" ) ) {
- parser = new TolParser();
- }
- else if ( filename_lc.endsWith( ".xml" ) || filename_lc.endsWith( ".px" ) || filename_lc.endsWith( "phyloxml" )
- || filename_lc.endsWith( ".zip" ) ) {
- parser = new PhyloXmlParser();
- if ( phyloxml_validate_against_xsd ) {
- final ClassLoader cl = PhyloXmlParser.class.getClassLoader();
- final URL xsd_url = cl.getResource( ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE );
- if ( xsd_url != null ) {
- ( ( PhyloXmlParser ) parser ).setValidateAgainstSchema( xsd_url.toString() );
- }
- else {
- if ( ForesterConstants.RELEASE ) {
- throw new RuntimeException( "failed to get URL for phyloXML XSD from jar file from ["
- + ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE + "]" );
- }
- }
+ if ( ( parser != null ) && file.toString().toLowerCase().endsWith( ".zip" ) ) {
+ if ( parser instanceof PhyloXmlParser ) {
+ ( ( PhyloXmlParser ) parser ).setZippedInputstream( true );
+ }
+ else if ( parser instanceof TolParser ) {
+ ( ( TolParser ) parser ).setZippedInputstream( true );
}
- }
- else if ( filename_lc.endsWith( ".nexus" ) || filename_lc.endsWith( ".nex" ) || filename_lc.endsWith( ".nx" ) ) {
- parser = new NexusPhylogeniesParser();
- }
- else if ( filename_lc.endsWith( ".nhx" ) || filename_lc.endsWith( ".nh" ) || filename_lc.endsWith( ".newick" )
- || filename_lc.endsWith( ".nwk" ) ) {
- parser = new NHXParser();
}
return parser;
}
throws FileNotFoundException, IOException {
final String lc_filename = url.getFile().toString().toLowerCase();
PhylogenyParser parser = createParserDependingOnSuffix( lc_filename, phyloxml_validate_against_xsd );
- if ( ( parser != null ) && lc_filename.endsWith( ".zip" ) ) {
- if ( parser instanceof PhyloXmlParser ) {
- ( ( PhyloXmlParser ) parser ).setZippedInputstream( true );
- }
- else if ( parser instanceof TolParser ) {
- ( ( TolParser ) parser ).setZippedInputstream( true );
- }
- }
if ( parser == null ) {
final String first_line = ForesterUtil.getFirstLine( url ).trim().toLowerCase();
if ( first_line.startsWith( "<" ) ) {
- parser = new PhyloXmlParser();
+ parser = PhyloXmlParser.createPhyloXmlParser();
if ( phyloxml_validate_against_xsd ) {
final ClassLoader cl = PhyloXmlParser.class.getClassLoader();
final URL xsd_url = cl.getResource( ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE );
parser = new NHXParser();
}
}
+ if ( ( parser != null ) && lc_filename.endsWith( ".zip" ) ) {
+ if ( parser instanceof PhyloXmlParser ) {
+ ( ( PhyloXmlParser ) parser ).setZippedInputstream( true );
+ }
+ else if ( parser instanceof TolParser ) {
+ ( ( TolParser ) parser ).setZippedInputstream( true );
+ }
+ }
return parser;
}
if ( m.find() ) {
return m.group( 1 ).replace( '_', ' ' );
}
+ final Matcher m_str1 = TAXOMONY_SN_PATTERN_STRAIN_1.matcher( name );
+ if ( m_str1.find() ) {
+ return m_str1.group( 1 ).replace( '_', ' ' );
+ }
+ final Matcher m_str2 = TAXOMONY_SN_PATTERN_STRAIN_2.matcher( name );
+ if ( m_str2.find() ) {
+ return m_str2.group( 1 ).replace( '_', ' ' );
+ }
+ final Matcher m_sn = TAXOMONY_SN_PATTERN_SN.matcher( name );
+
+ if ( m_sn.find() ) {
+ return m_sn.group( 1 ).replace( '_', ' ' );
+ }
+
+ final Matcher m_sp = TAXOMONY_SN_PATTERN_SP.matcher( name );
+
+ if ( m_sp.find() ) {
+ return m_sp.group( 1 ).replace( '_', ' ' );
+ }
return null;
}
if ( m.find() ) {
return m.group( 1 );
}
- else if ( taxonomy_extraction == TAXONOMY_EXTRACTION.AGGRESSIVE ) {
- m = TAXOMONY_UNIPROT_ID_PATTERN_A.matcher( name );
- if ( m.find() ) {
- return m.group( 1 );
- }
- }
+ //else if ( taxonomy_extraction == TAXONOMY_EXTRACTION.AGGRESSIVE ) {
+ // m = TAXOMONY_UNIPROT_ID_PATTERN_A.matcher( name );
+ // if ( m.find() ) {
+ // return m.group( 1 );
+ // }
+ //}
}
return null;
}
public final static Phylogeny[] readPhylogenies( final String file_name ) throws FileNotFoundException, IOException {
return readPhylogenies( new File( file_name ) );
}
+
+ /**
+ * Return null if it can not guess the parser to use based on name suffix.
+ *
+ * @param filename
+ * @return
+ */
+ final private static PhylogenyParser createParserDependingOnSuffix( final String filename,
+ final boolean phyloxml_validate_against_xsd ) {
+ PhylogenyParser parser = null;
+ final String filename_lc = filename.toLowerCase();
+ if ( filename_lc.endsWith( ".tol" ) || filename_lc.endsWith( ".tolxml" ) || filename_lc.endsWith( ".tol.zip" ) ) {
+ parser = new TolParser();
+ }
+ else if ( filename_lc.endsWith( ".xml" ) || filename_lc.endsWith( "phyloxml" ) || filename_lc.endsWith( ".zip" ) ) {
+ parser = PhyloXmlParser.createPhyloXmlParser();
+ if ( phyloxml_validate_against_xsd ) {
+ final ClassLoader cl = PhyloXmlParser.class.getClassLoader();
+ final URL xsd_url = cl.getResource( ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE );
+ if ( xsd_url != null ) {
+ ( ( PhyloXmlParser ) parser ).setValidateAgainstSchema( xsd_url.toString() );
+ }
+ else {
+ if ( ForesterConstants.RELEASE ) {
+ throw new RuntimeException( "failed to get URL for phyloXML XSD from jar file from ["
+ + ForesterConstants.LOCAL_PHYLOXML_XSD_RESOURCE + "]" );
+ }
+ }
+ }
+ }
+ else if ( filename_lc.endsWith( ".nexus" ) || filename_lc.endsWith( ".nex" ) || filename_lc.endsWith( ".nx" ) ) {
+ parser = new NexusPhylogeniesParser();
+ }
+ else if ( filename_lc.endsWith( ".nhx" ) || filename_lc.endsWith( ".nh" ) || filename_lc.endsWith( ".newick" )
+ || filename_lc.endsWith( ".nwk" ) ) {
+ parser = new NHXParser();
+ }
+ return parser;
+ }
}
git://source.jalview.org / jalview.git / blobdiff