comparison query.cpp @ 462:f689510baaf4 api-inversion

Simplify audioDB::query_loop_points.

Using the new functions audiodb_track_id_datum() and
audiodb_datum_qpointers(), much of the body of the method disappears.
Of course, we've probably introduced some inefficiencies and extra
memory copies, but I'm fairly sure that this method is going to be
dominated by disk i/o time anyway, so it doesn't matter.

 
   if(qstate->exact_evaluation_queue->size() == 0) {
     return;
   }
 
-  // Compute database info.  FIXME: we more than likely don't need
-  // very much of the database so write a new function to build these
-  // values per-track or, even better, per-point
-  if(!(adb->header->flags & O2_FLAG_LARGE_ADB)) {
-    if(audiodb_set_up_dbpointers(adb, spec, &dbpointers)) {
-      error("failed to set up dbpointers");
-    }
-  }
-
-  // We are guaranteed that the order of points is sorted by:
-  // {trackID, spos, qpos} so we can be relatively efficient in
-  // initialization of track data.  We assume that points usually
-  // don't overlap, so we will use exhaustive dot product evaluation
-  // instead of memoization of partial sums which is used for
-  // exhaustive brute-force evaluation from smaller databases in
-  // query_loop()
+  /* We are guaranteed that the order of points is sorted by:
+   * {trackID, spos, qpos} so we can be relatively efficient in
+   * initialization of track data.  We assume that points usually
+   * don't overlap, so we will use exhaustive dot product evaluation
+   * (instead of memoization of partial sums, as in query_loop()). */
   double dist;
-  size_t data_buffer_size = 0;
-  double *data_buffer = 0;
-  Uns32T trackOffset = 0;
-  Uns32T trackIndexOffset = 0;
-  Uns32T currentTrack = 0x80000000; // Initialize with a value outside of track index range
+  double *dbdata = 0, *dbdata_pointer;
+  Uns32T currentTrack = 0x80000000; // KLUDGE: Initialize with a value outside of track index range
   Uns32T npairs = qstate->exact_evaluation_queue->size();
-  while(npairs--){
+  while(npairs--) {
     PointPair pp = qstate->exact_evaluation_queue->top();
-    // Large ADB track data must be loaded here for sPower
-    if(adb->header->flags & O2_FLAG_LARGE_ADB) {
-      trackOffset=0;
-      trackIndexOffset=0;
-      if(currentTrack!=pp.trackID){
-	char* prefixedString = new char[O2_MAXFILESTR];
-	char* tmpStr = prefixedString;
-	// On currentTrack change, allocate and load track data
-	currentTrack=pp.trackID;
-	SAFE_DELETE_ARRAY(dbpointers.l2norm_data);
-	SAFE_DELETE_ARRAY(dbpointers.power_data);
-	if(infid>0)
-	  close(infid);
-	// Open and check dimensions of feature file
-	strncpy(prefixedString, featureFileNameTable+pp.trackID*O2_FILETABLE_ENTRY_SIZE, O2_MAXFILESTR);
-	prefix_name((char ** const) &prefixedString, adb_feature_root);
-	if (prefixedString!=tmpStr)
-	  delete[] tmpStr;
-	initInputFile(prefixedString);
-	// Load the feature vector data for current track into data_buffer
-	if(audiodb_read_data(adb, infid, pp.trackID, &data_buffer, &data_buffer_size))
-          error("failed to read data");
-	// Load power and calculate power and l2norm sequence sums
-	init_track_aux_data(pp.trackID, data_buffer, &dbpointers.l2norm_data, &dbpointers.l2norm, &dbpointers.power_data, &dbpointers.power);
-      }
-    }
-    else{
-      // These offsets are w.r.t. the entire database of feature vectors and auxillary variables
-      trackOffset = (*adb->track_offsets)[pp.trackID];
-      trackIndexOffset = trackOffset/(adb->header->dim * sizeof(double)); // num vectors offset
-    }    
+    if(currentTrack != pp.trackID) {
+      SAFE_DELETE_ARRAY(dbdata);
+      SAFE_DELETE_ARRAY(dbpointers.l2norm_data);
+      SAFE_DELETE_ARRAY(dbpointers.power_data);
+      SAFE_DELETE_ARRAY(dbpointers.mean_duration);
+      currentTrack = pp.trackID;
+      adb_datum_t d = {0};
+      if(audiodb_track_id_datum(adb, pp.trackID, &d)) {
+        error("failed to get datum");
+      }
+      if(audiodb_datum_qpointers(&d, sequence_length, &dbdata, &dbdata_pointer, &dbpointers)) {
+        audiodb_free_datum(&d);
+        error("failed to get dbpointers");
+      }
+      audiodb_free_datum(&d);
+    }
     Uns32T qPos = usingQueryPoint?0:pp.qpos;// index for query point
-    Uns32T sPos = trackIndexOffset+pp.spos; // index into l2norm table
+    Uns32T sPos = pp.spos; // index into l2norm table
     // Test power thresholds before computing distance
     if( ( (!power_refine) || audiodb_powers_acceptable(&spec->refine, qpointers->power[qPos], dbpointers.power[sPos])) &&
-	( qPos<qpointers->nvectors-sequence_length+1 && pp.spos<(*adb->track_lengths)[pp.trackID]-sequence_length+1 ) ){
-      // Non-large ADB track data is loaded inside power test for efficiency
-      if(!(adb->header->flags & O2_FLAG_LARGE_ADB) && (currentTrack!=pp.trackID) ){
-	// On currentTrack change, allocate and load track data
-	currentTrack=pp.trackID;
-	lseek(dbfid, adb->header->dataOffset + trackOffset, SEEK_SET);
-	if(audiodb_read_data(adb, dbfid, currentTrack, &data_buffer, &data_buffer_size))
-          error("failed to read data");
-      }
+	( qPos<qpointers->nvectors-sequence_length+1 && sPos<(*adb->track_lengths)[pp.trackID]-sequence_length+1 ) ){
       // Compute distance
-      dist = audiodb_dot_product(query + qPos*adb->header->dim, data_buffer + pp.spos*adb->header->dim, adb->header->dim*sequence_length);
+      dist = audiodb_dot_product(query + qPos*adb->header->dim, dbdata + sPos*adb->header->dim, adb->header->dim*sequence_length);
       double qn = qpointers->l2norm[qPos];
       double sn = dbpointers.l2norm[sPos];
       switch(spec->params.distance) {
       case ADB_DISTANCE_EUCLIDEAN_NORMED:
 	dist = 2 - (2/(qn*sn))*dist;

       }
     }
     qstate->exact_evaluation_queue->pop();
   }
   // Cleanup
+  SAFE_DELETE_ARRAY(dbdata);
   SAFE_DELETE_ARRAY(dbpointers.l2norm_data);
   SAFE_DELETE_ARRAY(dbpointers.power_data);
   SAFE_DELETE_ARRAY(dbpointers.mean_duration);
   delete qstate->exact_evaluation_queue;
 }