inprogress

[jalview.git] / forester / java / src / org / forester / util / SequenceIdParser.java

diff --git a/forester/java/src/org/forester/util/SequenceIdParser.java b/forester/java/src/org/forester/util/SequenceIdParser.java
index 0856f34..3b6bc5d 100644 (file)
--- a/forester/java/src/org/forester/util/SequenceIdParser.java
+++ b/forester/java/src/org/forester/util/SequenceIdParser.java
@@ -59,6 +59,9 @@ public final class SequenceIdParser {
     // underscore character ('_'). For example, a RefSeq protein accession is NP_015325. \r
     private final static Pattern REFSEQ_PATTERN                  = Pattern\r
                                                                          .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}_\\d{6,})(?:[^a-zA-Z0-9]|\\Z)" );\r
+    // See: http://web.expasy.org/docs/userman.html#ID_line\r
+    private final static Pattern TREMBL_PATTERN                  = Pattern\r
+                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z][0-9][A-Z0-9]{3}[0-9])(?:[^a-zA-Z0-9]|\\Z)" );\r
 \r
     /**\r
      * Returns null if no match.\r
@@ -73,9 +76,25 @@ public final class SequenceIdParser {
         if ( !ForesterUtil.isEmpty( v ) ) {\r
             return new Identifier( v, Identifier.REFSEQ );\r
         }\r
+        v = parseTrEMBLAccessor( s );\r
+        if ( !ForesterUtil.isEmpty( v ) ) {\r
+            return new Identifier( v, Identifier.SP );\r
+        }\r
         return null;\r
     }\r
 \r
+    public final static boolean isProtein( final String query ) {\r
+        final String r1 = parseRefSeqAccessor( query );\r
+        if ( !ForesterUtil.isEmpty( r1 ) && ( r1.charAt( 1 ) == 'P' ) ) {\r
+            return true;\r
+        }\r
+        final String r2 = parseTrEMBLAccessor( query );\r
+        if ( !ForesterUtil.isEmpty( r2 ) ) {\r
+            return true;\r
+        }\r
+        return GENBANK_PROTEIN_AC_PATTERN.matcher( query ).lookingAt();\r
+    }\r
+\r
     /**\r
      * Returns null if no match.\r
      * \r
@@ -114,6 +133,18 @@ public final class SequenceIdParser {
         return null;\r
     }\r
 \r
+    /**\r
+     * Returns null if no match.\r
+     * \r
+     */\r
+    private final static String parseTrEMBLAccessor( final String query ) {\r
+        final Matcher m = TREMBL_PATTERN.matcher( query );\r
+        if ( m.lookingAt() ) {\r
+            return m.group( 1 );\r
+        }\r
+        return null;\r
+    }\r
+\r
     private SequenceIdParser() {\r
         // Hiding the constructor.\r
     }\r