X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Fws%2Funiprot%2FUniProtWsTools.java;h=d24b171071a36d01d8e06dd16270ab2351ddc176;hb=7c8aef9df85f80605041ba4d065c798a3cd78d5a;hp=cd8ab4315780f936cf0c1376e548e3dd1676605d;hpb=c78e0c6ccc1b8c7f4e77db43be8d09e2d7c5b78e;p=jalview.git

diff --git a/forester/java/src/org/forester/ws/uniprot/UniProtWsTools.java b/forester/java/src/org/forester/ws/uniprot/UniProtWsTools.java
index cd8ab43..d24b171 100644
--- a/forester/java/src/org/forester/ws/uniprot/UniProtWsTools.java
+++ b/forester/java/src/org/forester/ws/uniprot/UniProtWsTools.java
@@ -45,19 +45,23 @@ public final class UniProtWsTools {
         UNKNOWN, UNIPROT;
     }
     public final static String   BASE_URL           = "http://www.uniprot.org/";
+    public final static String   BASE_EMBL_DB_URL   = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch/embl/";
     private final static String  URL_ENC            = "UTF-8";
     // uniprot/expasy accession number format (6 chars):
     // letter digit letter-or-digit letter-or-digit letter-or-digit digit
+    // ?: => no back-reference
+    // \A => begin of String
+    // \Z => end of String
     private final static Pattern UNIPROT_AC_PATTERN = Pattern
-                                                            .compile( "^.*[^a-zA-Z0-9]?([A-NR-ZOPQ]\\d[A-Z0-9]{3}\\d)[^a-zA-Z0-9]?" );
-    private final static boolean DEBUG              = false;
+                                                            .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d[A-Z0-9]{3}\\d)(?:[^a-zA-Z0-9]|\\Z)" );
+    private final static boolean DEBUG              = true;
 
     private static String encode( final String str ) throws UnsupportedEncodingException {
         return URLEncoder.encode( str.trim(), URL_ENC );
     }
 
     /**
-     * Return null if no match.
+     * Returns null if no match.
      * 
      * @param query
      * @param db 
@@ -208,14 +212,23 @@ public final class UniProtWsTools {
         return taxonomies;
     }
 
-    public static List<String> queryUniprot( final String query, int max_lines_to_return ) throws IOException {
+    public static List<String> queryEmblDb( final String query, final int max_lines_to_return ) throws IOException {
+        return queryDb( query, max_lines_to_return, BASE_EMBL_DB_URL );
+    }
+
+    public static List<String> queryUniprot( final String query, final int max_lines_to_return ) throws IOException {
+        return queryDb( query, max_lines_to_return, BASE_URL );
+    }
+
+    public static List<String> queryDb( final String query, int max_lines_to_return, final String base_url )
+            throws IOException {
         if ( ForesterUtil.isEmpty( query ) ) {
             throw new IllegalArgumentException( "illegal attempt to use empty query " );
         }
         if ( max_lines_to_return < 1 ) {
             max_lines_to_return = 1;
         }
-        final URL url = new URL( BASE_URL + query );
+        final URL url = new URL( base_url + query );
         if ( DEBUG ) {
             System.out.println( "url: " + url.toString() );
         }
@@ -224,6 +237,7 @@ public final class UniProtWsTools {
         String line;
         final List<String> result = new ArrayList<String>();
         while ( ( line = in.readLine() ) != null ) {
+            System.out.println( line );
             result.add( line );
             if ( result.size() > max_lines_to_return ) {
                 break;
@@ -238,4 +252,10 @@ public final class UniProtWsTools {
         final List<String> lines = queryUniprot( "uniprot/" + query + ".txt", max_lines_to_return );
         return UniProtEntry.createInstanceFromPlainText( lines );
     }
+
+    public static SequenceDatabaseEntry obtainEmblEntry( final String query, final int max_lines_to_return )
+            throws IOException {
+        final List<String> lines = queryEmblDb( query, max_lines_to_return );
+        return EbiDbEntry.createInstanceFromPlainText( lines );
+    }
 }