JAL-1064 patch for NPE when registerd database names do not resolve to one or more...
[jalview.git] / src / jalview / ws / seqfetcher / ASequenceFetcher.java
index 7b455bf..09a7a95 100644 (file)
@@ -1,10 +1,36 @@
+/*\r
+ * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)\r
+ * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, G Barton, M Clamp, S Searle\r
+ * \r
+ * This file is part of Jalview.\r
+ * \r
+ * Jalview is free software: you can redistribute it and/or\r
+ * modify it under the terms of the GNU General Public License \r
+ * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.\r
+ * \r
+ * Jalview is distributed in the hope that it will be useful, but \r
+ * WITHOUT ANY WARRANTY; without even the implied warranty \r
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR \r
+ * PURPOSE.  See the GNU General Public License for more details.\r
+ * \r
+ * You should have received a copy of the GNU General Public License along with Jalview.  If not, see <http://www.gnu.org/licenses/>.\r
+ */\r
 package jalview.ws.seqfetcher;\r
 \r
 import jalview.datamodel.AlignmentI;\r
+import jalview.datamodel.DBRefEntry;\r
 import jalview.datamodel.SequenceI;\r
+import jalview.util.DBRefUtils;\r
 \r
+import java.util.ArrayList;\r
+import java.util.Collection;\r
 import java.util.Enumeration;\r
+import java.util.HashSet;\r
 import java.util.Hashtable;\r
+import java.util.Iterator;\r
+import java.util.List;\r
+import java.util.Map;\r
+import java.util.Stack;\r
 import java.util.Vector;\r
 \r
 public class ASequenceFetcher\r
@@ -13,7 +39,7 @@ public class ASequenceFetcher
   /**\r
    * set of databases we can retrieve entries from\r
    */\r
-  protected Hashtable FETCHABLEDBS;\r
+  protected Hashtable<String, Map<String, DbSourceProxy>> FETCHABLEDBS;\r
 \r
   public ASequenceFetcher()\r
   {\r
@@ -58,125 +84,143 @@ public class ASequenceFetcher
   public SequenceI[] getSequences(jalview.datamodel.DBRefEntry[] refs)\r
   {\r
     SequenceI[] ret = null;\r
-    Vector rseqs = new Vector();\r
-    Hashtable queries = new Hashtable();\r
+    Vector<SequenceI> rseqs = new Vector();\r
+    Hashtable<String, List<String>> queries = new Hashtable();\r
     for (int r = 0; r < refs.length; r++)\r
     {\r
       if (!queries.containsKey(refs[r].getSource()))\r
       {\r
-        queries.put(refs[r].getSource(), new Vector());\r
+        queries.put(refs[r].getSource(), new ArrayList<String>());\r
       }\r
-      Vector qset = (Vector) queries.get(refs[r].getSource());\r
+      List<String> qset = queries.get(refs[r].getSource());\r
       if (!qset.contains(refs[r].getAccessionId()))\r
       {\r
-        qset.addElement(refs[r].getAccessionId());\r
+        qset.add(refs[r].getAccessionId());\r
       }\r
     }\r
-    Enumeration e = queries.keys();\r
+    Enumeration<String> e = queries.keys();\r
     while (e.hasMoreElements())\r
     {\r
-      Vector query = null;\r
+      List<String> query = null;\r
       String db = null;\r
-      try\r
+      db = e.nextElement();\r
+      query = queries.get(db);\r
+      if (!isFetchable(db))\r
       {\r
-        db = (String) e.nextElement();\r
-        query = (Vector) queries.get(db);\r
-        if (!isFetchable(db))\r
-          throw new Exception(\r
-                  "Don't know how to fetch from this database :" + db);\r
-        DbSourceProxy fetcher = getSourceProxy(db);\r
-        boolean doMultiple = fetcher.getAccessionSeparator() != null; // No\r
-                                                                      // separator\r
-                                                                      // - no\r
-                                                                      // Multiple\r
-                                                                      // Queries\r
-        Enumeration qs = query.elements();\r
-        while (qs.hasMoreElements())\r
+        reportStdError(db, query, new Exception(\r
+                "Don't know how to fetch from this database :" + db));\r
+        continue;\r
+      }\r
+      Iterator<DbSourceProxy> fetchers = getSourceProxy(db).iterator();\r
+      Stack<String> queriesLeft = new Stack<String>();\r
+//      List<String> queriesFailed = new ArrayList<String>();\r
+      queriesLeft.addAll(query);\r
+      while (fetchers.hasNext())\r
+      {\r
+        List<String> queriesMade = new ArrayList<String>();\r
+        HashSet queriesFound = new HashSet<String>();\r
+        try\r
         {\r
-          StringBuffer qsb = new StringBuffer();\r
-          do\r
-          {\r
-            qsb.append((String) qs.nextElement());\r
-            if (qs.hasMoreElements() && doMultiple) // and not reached limit for\r
-                                                    // multiple queries at one\r
-                                                    // time for this source\r
-            {\r
-              qsb.append(fetcher.getAccessionSeparator());\r
-            }\r
-          } while (doMultiple && qs.hasMoreElements());\r
-          \r
-          AlignmentI seqset=null;\r
-          try {\r
-            // create a fetcher and go to it\r
-            seqset = fetcher.getSequenceRecords(qsb.toString());\r
-          } catch (Exception ex)\r
-          {\r
-            System.err.println("Failed to retrieve the following from "+db);\r
-            System.err.println(qsb);\r
-            ex.printStackTrace(System.err);\r
-          }\r
-          // TODO: Merge alignment together - perhaps\r
-          if (seqset != null)\r
+          DbSourceProxy fetcher = fetchers.next();\r
+          boolean doMultiple = fetcher.getAccessionSeparator() != null; // No\r
+          // separator\r
+          // - no\r
+          // Multiple\r
+          // Queries\r
+          while (!queriesLeft.isEmpty())\r
           {\r
-            SequenceI seqs[] = seqset.getSequencesArray();\r
-            if (seqs != null)\r
+            StringBuffer qsb = new StringBuffer();\r
+            do\r
             {\r
-              for (int is = 0; is < seqs.length; is++)\r
+              if (qsb.length() > 0)\r
               {\r
-                rseqs.addElement(seqs[is]);\r
-                seqs[is] = null;\r
+                qsb.append(fetcher.getAccessionSeparator());\r
               }\r
+              String q = queriesLeft.pop();\r
+              queriesMade.add(q);\r
+              qsb.append(q);\r
+            } while (doMultiple && !queriesLeft.isEmpty());\r
+\r
+            AlignmentI seqset = null;\r
+            try\r
+            {\r
+              // create a fetcher and go to it\r
+              seqset = fetcher.getSequenceRecords(qsb.toString()); // ,\r
+                      // queriesFailed);\r
+            } catch (Exception ex)\r
+            {\r
+              System.err.println("Failed to retrieve the following from "\r
+                      + db);\r
+              System.err.println(qsb);\r
+              ex.printStackTrace(System.err);\r
             }\r
-            else\r
+            // TODO: Merge alignment together - perhaps\r
+            if (seqset != null)\r
             {\r
-              if (fetcher.getRawRecords() != null)\r
+              SequenceI seqs[] = seqset.getSequencesArray();\r
+              if (seqs != null)\r
               {\r
-                System.out.println("# Retrieved from " + db + ":"\r
-                        + qs.toString());\r
-                StringBuffer rrb = fetcher.getRawRecords();\r
-                /*\r
-                 * for (int rr = 0; rr<rrb.length; rr++) {\r
-                 */\r
-                String hdr;\r
-                // if (rr<qs.length)\r
-                // {\r
-                hdr = "# " + db + ":" + qsb.toString();\r
-                /*\r
-                 * } else { hdr = "# part "+rr; }\r
-                 */\r
-                System.out.println(hdr);\r
-                if (rrb != null)\r
-                  System.out.println(rrb);\r
-                System.out.println("# end of " + hdr);\r
+                for (int is = 0; is < seqs.length; is++)\r
+                {\r
+                  rseqs.addElement(seqs[is]);\r
+                  DBRefEntry[] frefs = DBRefUtils.searchRefs(seqs[is]\r
+                          .getDBRef(), new DBRefEntry(db, null, null));\r
+                  if (frefs != null)\r
+                  {\r
+                    for (DBRefEntry dbr : frefs)\r
+                    {\r
+                      queriesFound.add(dbr.getAccessionId());\r
+                      queriesMade.remove(dbr.getAccessionId());\r
+                    }\r
+                  }\r
+                  seqs[is] = null;\r
+                }\r
+              }\r
+              else\r
+              {\r
+                if (fetcher.getRawRecords() != null)\r
+                {\r
+                  System.out.println("# Retrieved from " + db + ":"\r
+                          + qsb.toString());\r
+                  StringBuffer rrb = fetcher.getRawRecords();\r
+                  /*\r
+                   * for (int rr = 0; rr<rrb.length; rr++) {\r
+                   */\r
+                  String hdr;\r
+                  // if (rr<qs.length)\r
+                  // {\r
+                  hdr = "# " + db + ":" + qsb.toString();\r
+                  /*\r
+                   * } else { hdr = "# part "+rr; }\r
+                   */\r
+                  System.out.println(hdr);\r
+                  if (rrb != null)\r
+                    System.out.println(rrb);\r
+                  System.out.println("# end of " + hdr);\r
+                }\r
+\r
               }\r
             }\r
+\r
           }\r
+        } catch (Exception ex)\r
+        {\r
+          reportStdError(db, queriesMade, ex);\r
         }\r
-      } catch (Exception ex)\r
-      {\r
-        System.err\r
-                .println("Failed to retrieve the following references from "\r
-                        + db);\r
-        Enumeration qv = query.elements();\r
-        int n = 0;\r
-        while (qv.hasMoreElements())\r
+        if (queriesMade.size() > 0)\r
         {\r
-          System.err.print(" " + qv.nextElement() + ";");\r
-          if (n++ > 10)\r
-          {\r
-            System.err.println();\r
-            n = 0;\r
-          }\r
+          System.out.println("# Adding " + queriesMade.size()\r
+                  + " ids back to queries list for searching again (" + db\r
+                  + ".");\r
+          queriesLeft.addAll(queriesMade);\r
         }\r
-        System.err.println();\r
-        ex.printStackTrace();\r
       }\r
     }\r
     if (rseqs.size() > 0)\r
     {\r
       ret = new SequenceI[rseqs.size()];\r
       Enumeration sqs = rseqs.elements();\r
-      int si=0;\r
+      int si = 0;\r
       while (sqs.hasMoreElements())\r
       {\r
         SequenceI s = (SequenceI) sqs.nextElement();\r
@@ -187,6 +231,26 @@ public class ASequenceFetcher
     return ret;\r
   }\r
 \r
+  public void reportStdError(String db, List<String> queriesMade,\r
+          Exception ex)\r
+  {\r
+\r
+    System.err.println("Failed to retrieve the following references from "\r
+            + db);\r
+    int n = 0;\r
+    for (String qv : queriesMade)\r
+    {\r
+      System.err.print(" " + qv + ";");\r
+      if (n++ > 10)\r
+      {\r
+        System.err.println();\r
+        n = 0;\r
+      }\r
+    }\r
+    System.err.println();\r
+    ex.printStackTrace();\r
+  }\r
+\r
   /**\r
    * Retrieve an instance of the proxy for the given source\r
    * \r
@@ -195,10 +259,169 @@ public class ASequenceFetcher
    *          retrieval of specific DB source/version combinations.\r
    * @return an instance of DbSourceProxy for that db.\r
    */\r
-  public DbSourceProxy getSourceProxy(String db)\r
+  public List<DbSourceProxy> getSourceProxy(String db)\r
   {\r
-    DbSourceProxy dbs = (DbSourceProxy) FETCHABLEDBS.get(db);\r
+    List<DbSourceProxy> dbs;\r
+    Map<String,DbSourceProxy> dblist = FETCHABLEDBS.get(db);\r
+    if (dblist==null) {return new ArrayList<DbSourceProxy>();};\r
+    if (dblist.size()>1) {\r
+      DbSourceProxy[] l=dblist.values().toArray(new DbSourceProxy[0]);\r
+      int i=0;\r
+      String[] nm=new String[l.length];\r
+      for (DbSourceProxy s:l)\r
+      {\r
+        nm[i++]=s.getDbName().toLowerCase();\r
+      }\r
+      jalview.util.QuickSort.sort(nm,l);\r
+      dbs = new ArrayList<DbSourceProxy>();\r
+      for (i=l.length-1;i>=0; i--)\r
+      {\r
+        dbs.add(l[i]);\r
+      }\r
+    } else {\r
+      dbs = new ArrayList<DbSourceProxy>(dblist.values());\r
+    }\r
     return dbs;\r
   }\r
 \r
-}
\ No newline at end of file
+  /**\r
+   * constructs and instance of the proxy and registers it as a valid\r
+   * dbrefsource\r
+   * \r
+   * @param dbSourceProxy\r
+   *          reference for class implementing\r
+   *          jalview.ws.seqfetcher.DbSourceProxy\r
+   * @throws java.lang.IllegalArgumentException\r
+   *           if class does not implement jalview.ws.seqfetcher.DbSourceProxy\r
+   */\r
+  protected void addDBRefSourceImpl(Class dbSourceProxy)\r
+          throws java.lang.IllegalArgumentException\r
+  {\r
+    DbSourceProxy proxy = null;\r
+    try\r
+    {\r
+      Object proxyObj = dbSourceProxy.getConstructor(null)\r
+              .newInstance(null);\r
+      if (!DbSourceProxy.class.isInstance(proxyObj))\r
+      {\r
+        throw new IllegalArgumentException(\r
+                dbSourceProxy.toString()\r
+                        + " does not implement the jalview.ws.seqfetcher.DbSourceProxy");\r
+      }\r
+      proxy = (DbSourceProxy) proxyObj;\r
+    } catch (IllegalArgumentException e)\r
+    {\r
+      throw e;\r
+    } catch (Exception e)\r
+    {\r
+      // Serious problems if this happens.\r
+      throw new Error("DBRefSource Implementation Exception", e);\r
+    }\r
+    addDbRefSourceImpl(proxy);\r
+  }\r
+\r
+  /**\r
+   * add the properly initialised DbSourceProxy object 'proxy' to the list of\r
+   * sequence fetchers\r
+   * \r
+   * @param proxy\r
+   */\r
+  protected void addDbRefSourceImpl(DbSourceProxy proxy)\r
+  {\r
+    if (proxy != null)\r
+    {\r
+      if (FETCHABLEDBS == null)\r
+      {\r
+        FETCHABLEDBS = new Hashtable<String, Map<String,DbSourceProxy>>();\r
+      }\r
+      Map<String,DbSourceProxy> slist = FETCHABLEDBS.get(proxy.getDbSource());\r
+      if (slist == null)\r
+      {\r
+        FETCHABLEDBS.put(proxy.getDbSource(),\r
+                slist = new Hashtable<String,DbSourceProxy>());\r
+      }\r
+      slist.put(proxy.getDbName(),proxy);\r
+    }\r
+  }\r
+\r
+  /**\r
+   * test if the database handler for dbName contains the given dbProperty\r
+   * when a dbName resolves to a set of proxies - this method will return the result of the test for the first instance.\r
+   * TODO implement additional method to query all sources for a db to find one with a particular property\r
+   * @param dbName\r
+   * @param dbProperty\r
+   * @return true if proxy has the given property\r
+   */\r
+  public boolean hasDbSourceProperty(String dbName, String dbProperty)\r
+  {\r
+    // TODO: decide if invalidDbName exception is thrown here.\r
+\r
+    List<DbSourceProxy> proxies = getSourceProxy(dbName);\r
+    if (proxies != null)\r
+    {\r
+      for (DbSourceProxy proxy : proxies)\r
+      {\r
+        if (proxy.getDbSourceProperties() != null)\r
+        {\r
+          return proxy.getDbSourceProperties().containsKey(dbProperty);\r
+        }\r
+      }\r
+    }\r
+    return false;\r
+  }\r
+\r
+  /**\r
+   * select sources which are implemented by instances of the given class\r
+   * \r
+   * @param class that implements DbSourceProxy\r
+   * @return null or vector of source names for fetchers\r
+   */\r
+  public String[] getDbInstances(Class class1)\r
+  {\r
+    if (!jalview.ws.seqfetcher.DbSourceProxy.class.isAssignableFrom(class1))\r
+    {\r
+      throw new Error(\r
+              "Implmentation Error - getDbInstances must be given a class that implements jalview.ws.seqfetcher.DbSourceProxy (was given '"\r
+                      + class1 + "')");\r
+    }\r
+    if (FETCHABLEDBS == null)\r
+    {\r
+      return null;\r
+    }\r
+    String[] sources = null;\r
+    Vector src = new Vector();\r
+    Enumeration dbs = FETCHABLEDBS.keys();\r
+    while (dbs.hasMoreElements())\r
+    {\r
+      String dbn = (String) dbs.nextElement();\r
+      for (DbSourceProxy dbp : FETCHABLEDBS.get(dbn).values())\r
+      {\r
+        if (class1.isAssignableFrom(dbp.getClass()))\r
+        {\r
+          src.addElement(dbn);\r
+        }\r
+      }\r
+    }\r
+    if (src.size() > 0)\r
+    {\r
+      src.copyInto(sources = new String[src.size()]);\r
+    }\r
+    return sources;\r
+  }\r
+  public DbSourceProxy[] getDbSourceProxyInstances(\r
+          Class class1)\r
+  {\r
+    ArrayList<DbSourceProxy> prlist=new ArrayList<DbSourceProxy>();\r
+    for (String fetchable:getSupportedDb())\r
+    for (DbSourceProxy pr:getSourceProxy(fetchable))\r
+    {\r
+      if (class1.isInstance(pr)) {prlist.add(pr);}\r
+    }\r
+    if (prlist.size()==0)\r
+    {\r
+      return null;\r
+    }\r
+    return prlist.toArray(new DbSourceProxy[0]);\r
+  }\r
+\r
+}\r