Changeset 24398

Timestamp:

2011-08-12T16:40:17+12:00 (12 years ago)

Author:

davidb

Message:

Latest round of changes to code, in preparing for demo to Goldsmiths

Location:

gs3-extensions/audioDB/trunk/src/src/java/org/greenstone/gsdl3

Files:

• service/AbstractAudioSearch.java (added)
• service/AbstractGS2AudioSearch.java (added)
• service/GS2AudioDBSearch.java (modified) (8 diffs)
• util/AudioDBDocInfo.java (modified) (1 diff)
• util/AudioDBWrapper.java (modified) (5 diffs)

gs3-extensions/audioDB/trunk/src/src/java/org/greenstone/gsdl3/service/GS2AudioDBSearch.java

@@ -38 +38 @@
 import org.apache.log4j.*;
 
-public class GS2AudioDBSearch extends AbstractGS2Search {
+public class GS2AudioDBSearch extends AbstractGS2AudioSearch {
 
     static Logger logger = Logger.getLogger (org.greenstone.gsdl3.service.GS2AudioDBSearch.class.getName ());
@@ -48 +48 @@
 
     protected static final String AUDIODB_DEFAULT_DIRECTORY = "audioDB";
-    protected static final String ADB_FILENAME = "lhs-features.adb";
+    protected static final String ADB_FILENAME = "lsh-features.adb";
     
     protected AudioDBWrapper audiodb_src = null;
@@ -60 +60 @@
     
      /** do the actual query */
-    protected Element processTextQuery (Element request) {
+    protected Element processAudioQuery (Element request) {
     // MG needs to be synchronized (this inspiration for this class)
     // Since it is not known how concurrent audioDB can be, play it safe for
@@ -68 +68 @@
         // Create a new (empty) result message ('doc' is in ServiceRack.java)
         Element result = this.doc.createElement (GSXML.RESPONSE_ELEM);
-        result.setAttribute (GSXML.FROM_ATT, TEXT_QUERY_SERVICE); // Result looks the same as a text query
+
+        // Rather than QUERY_SERVICE use "TextQuery" 
+        // => makes the result looks the same as a text query
+        result.setAttribute (GSXML.FROM_ATT, "TextQuery"); 
         result.setAttribute (GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
         
@@ -74 +77 @@
         Element param_list = (Element) GSXML.getChildByTagName (request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
         if (param_list == null) {
-        logger.error ("AudioDB TextQuery request had no paramList.");
+        logger.error ("AudioDB AudioQuery request had no paramList.");
         return result;  // Return the empty result
         }
@@ -86 +89 @@
         return result;  // Return the empty result
         }
-        
+
         // If an index hasn't been specified, use the default
         String index = (String) params.get (INDEX_PARAM);
@@ -94 +97 @@
         
         // The location of the audioDB index
-        String index_dir = GSFile.collectionIndexDir (this.site_home, this.cluster_name) +  File.separatorChar + index;
-        
+        String toplevel_index_dir = GSFile.collectionIndexDir (this.site_home, this.cluster_name);
+        String audioDB_index_dir = toplevel_index_dir + File.separatorChar + index;
+        String assoc_index_dir = toplevel_index_dir + File.separatorChar + "assoc";
+
         // set the audioDB query parameters to the values the user has specified
         setStandardQueryParams (params);
         
-        this.audiodb_src.runQuery (index_dir, ADB_FILENAME, query);
+        this.audiodb_src.runQuery (audioDB_index_dir, ADB_FILENAME, assoc_index_dir, query);
         Vector docs = this.audiodb_src.getQueryResult ();
 
         if (docs.isEmpty()) {
         // something has gone wrong
-        GSXML.addError (this.doc, result, "Couldn't query the mg database", GSXML.ERROR_TYPE_SYSTEM);
+        GSXML.addError (this.doc, result, "Couldn't query the audioDB database", GSXML.ERROR_TYPE_SYSTEM);
         return result;
         }
@@ -129 +134 @@
         result.appendChild (document_list);
         for (int d = 0; d < docs.size (); d++) {
-            String doc_id = ((AudioDBDocInfo) docs.elementAt (d)).oid_;
-            float rank = ((AudioDBDocInfo) docs.elementAt (d)).rank_;
-            Element doc_node = createDocNode (doc_id, Float.toString (rank));
+            AudioDBDocInfo adb_doc = (AudioDBDocInfo) docs.elementAt(d);
+
+            String doc_id  = adb_doc.getDocID();
+            double rank    = adb_doc.getTopRank();
+            String offsets = adb_doc.getOffsetList();
+
+            Element doc_node = createDocNode (doc_id, Double.toString (rank));
+            doc_node.setAttribute("frameOffset", offsets);
+
             document_list.appendChild (doc_node);
         }

gs3-extensions/audioDB/trunk/src/src/java/org/greenstone/gsdl3/util/AudioDBDocInfo.java

@@ -19 +19 @@
 package org.greenstone.gsdl3.util;
 
+import java.util.Vector;
+import java.lang.Comparable;
 
-public class AudioDBDocInfo
+public class AudioDBDocInfo implements Comparable<AudioDBDocInfo>
 {
-    public String oid_ = null;
-    public float rank_ = 0;
+    public String oid_;
+    public Vector<Double> rankVector_;
+    public Vector<Integer> offsetVector_;
 
-    public AudioDBDocInfo(String doc_oid, float rank)
+    public AudioDBDocInfo(String doc_oid, double rank, int offset)
     {
     oid_ = doc_oid;
-    rank_ = rank;
+    
+    rankVector_ = new Vector<Double>();
+    rankVector_.add (rank);
+
+    offsetVector_ = new Vector<Integer>();
+    offsetVector_.add(offset);
     }
 
-    public String toString()
+
+    public AudioDBDocInfo(String doc_oid, Vector<Double> rankVector, Vector<Integer> offsetVector)
     {
-    return "" + oid_ + " (" + rank_ + ")";
+    oid_ = doc_oid;
+    rankVector_ = rankVector;
+    offsetVector_ = offsetVector;
+    }
+
+    public String getDocID()
+    {
+    return oid_;
+    }
+
+    public double getTopRank()
+    {
+    return rankVector_.get(0);
+    }
+
+
+    public String getOffsetList() 
+    {
+    StringBuffer all_offsets = new StringBuffer();
+    boolean first = true;
+
+    for (int i=0; i<offsetVector_.size(); i++) {
+        int offset = offsetVector_.get(i);
+        String offsetStr = Integer.toString(offset);
+
+        if (first) {
+        first = false;
+        }
+        else {
+        all_offsets.append(",");
+        }
+
+        all_offsets.append(offsetStr);
+    }
+
+    return all_offsets.toString();
+    }
+
+    public int compareTo(AudioDBDocInfo di)
+    {
+    // based on first entry in rank Vector
+
+    // embodies a descending sort order
+    double lrank = rankVector_.get(0);
+    double rrank = di.rankVector_.get(0);
+
+    if (lrank<rrank) {
+        return 1;
+    }
+    else if (lrank>rrank) {
+        return -1;
+    }
+    else {
+        return 0;
+    }
     }
 }

gs3-extensions/audioDB/trunk/src/src/java/org/greenstone/gsdl3/util/AudioDBWrapper.java

@@ -19 +19 @@
 package org.greenstone.gsdl3.util;
 
-import java.io.File;
+import java.io.*;
 import java.util.Vector;
+import java.util.Collections;
+
+import org.apache.log4j.*;
 
 /** java wrapper class for access to the audioDB executable
@@ -30 +33 @@
 {
     /** the query result, filled in by runQuery */
-    protected Vector query_result_ = null;
+    protected Vector query_result_;
 
     protected int offset_ = 100;
@@ -40 +43 @@
     protected int max_docs_;
 
+    static Logger logger = Logger.getLogger (org.greenstone.gsdl3.util.AudioDBWrapper.class.getName ());
+
     public AudioDBWrapper() {  
-    query_result_ = new Vector();
+    query_result_ = null;
     }
 
@@ -70 +75 @@
     // the following was in MG version, do we need this in audioDB version?
     //public String getQueryParams() {}
+
+
+    protected boolean addQueryResult(boolean first_entry, String doc_id, 
+                  Vector<Double> rankVector, Vector<Integer> offsetVector)
+    {
+
+    if (first_entry) {
+        AudioDBDocInfo audioDB_doc_info = new AudioDBDocInfo(doc_id,rankVector,offsetVector);
+        query_result_.add(audioDB_doc_info);
+        first_entry = false;
+    }
+    else {
+        double rank = rankVector.get(0);
+        int offset = offsetVector.get(0);
+        AudioDBDocInfo audioDB_doc_info = new AudioDBDocInfo(doc_id,rank,offset);
+        
+        query_result_.add(audioDB_doc_info);
+    }
+
+    return first_entry;
+    }
+
 
     /** actually carry out the query.
@@ -79 +106 @@
      * base_dir must end with a file separator (OS dependant)
      */
-    public void runQuery(String index_dir, String adb_file, String query_string) {
+    public void runQuery(String audioDB_index_dir, String adb_file, 
+             String assoc_index_dir, String query_string) {
 
     // combine index_dir with audiodb fileanem
 
-    String full_adb_filename  = index_dir + File.separatorChar + adb_file;
+    String full_adb_filename  = audioDB_index_dir + File.separatorChar + adb_file;
+    String full_chr12_filename = assoc_index_dir + File.separatorChar 
+        + query_string + File.separatorChar + "doc.chr12";
+
+    int num_matches_within_track = 6;
 
     String cmd = "audioDB"
         + " -d " + full_adb_filename
-        + " -Q sequence"
+        + " -Q nsequence"
         + " -p " + offset_ 
-        + " -n 1"
+        + " -n " + num_matches_within_track
         + " -l " + length_
-        + " -r " + "numMatchingDocs"
-        + " -f " + "track.chr12";
-
+        + " -r " + max_docs_
+        + " -f " + full_chr12_filename;
+
+    System.err.println("**** cmd = " + cmd);
+
+    Runtime runtime = Runtime.getRuntime();
+    try {
+        Process audioDB_proc = runtime.exec(cmd);
+        InputStream ais = audioDB_proc.getInputStream();
+        InputStreamReader aisr = new InputStreamReader(ais);
+        BufferedReader abr = new BufferedReader(aisr);
+
+        query_result_ = new Vector();
+
+        boolean first_entry = true;
+        int line_count = 0;
+
+        String root_doc_id = null;
+        Vector<Double> rankVector = new Vector<Double>();
+        Vector<Integer> offsetVector = new Vector<Integer>();
+
+        // Example output
+        //   D8 0.00105175
+        //   1.69786e-16 392 392
+        //   0.00113568 392 673
+        //   0.00127239 392 910
+        //   0.00139736 392 481
+        //   0.00145331 392 303
+        //   D2 0.00429758
+        //   0.00403335 392 865
+        //   0.00411288 392 458
+        //   0.00442461 392 866
+        //   0.00444272 392 864
+        //   0.00447434 392 424
+        // ...
+
+        String line;
+        while ((line = abr.readLine()) != null) {
+        String[] tokens = line.split("\\s+");
+        line_count++;
+
+        if (tokens.length==2) {
+            // processing a top-level doc line
+
+            if (line_count>1) {
+            // struck new top-level entry => store vector vals for previous block
+            first_entry = addQueryResult(first_entry,root_doc_id,rankVector,offsetVector);
+            // and now reset vectors to empty to be ready for next chain of values
+            rankVector = new Vector<Double>();
+            offsetVector = new Vector<Integer>();
+            }
+
+            root_doc_id = tokens[0];
+        }
+        else {
+            // should be 3 items 
+            double euclidean_dist = Double.parseDouble(tokens[0]);
+            int src_frame = Integer.parseInt(tokens[1]);
+            int target_frame = Integer.parseInt(tokens[2]);
+
+            // enforce 1.0 as upper limit due to rounding errors 
+            // in audioDB distance calculations
+            double rank = Math.min(1.0 - euclidean_dist,1.0); 
+
+            if ((line_count==2) && (src_frame==target_frame)) {
+            // Found match with self
+            continue;
+            }
+
+            rankVector.add(rank);
+            offsetVector.add(target_frame);
+        }
+
+        }
+
+        addQueryResult(first_entry,root_doc_id,rankVector,offsetVector);
+
+        abr.close();
+
+        // sort query_result_ on 'rank' field 
+        // note: compareTo() method impelemented to sort into descending order
+
+        Collections.sort(query_result_);
+
+
+    }
+    catch (Exception e) {
+        logger.error("Failed to execute the following command: " + cmd);
+        e.printStackTrace();
+    }
 
     }