comparison query.cpp @ 461:2b8cfec91ed7 api-inversion

Factor out some bits of audiodb_query_spec_qpointers()

New functions audiodb_track_id_datum() and audiodb_datum_qpointers()
will be useful in places where audioDB::init_track_aux_data is currently
being called.  We should be able to simplify audioDB::query_loop_points
quite a lot this way...

   if (!timesFile->eof()) {
     error("too many timepoints in times file", timesFileName);
   }
 }
 
+int audiodb_track_id_datum(adb_t *adb, uint32_t track_id, adb_datum_t *d) {
+  off_t track_offset = (*adb->track_offsets)[track_id];
+  if(adb->header->flags & O2_FLAG_LARGE_ADB) {
+    /* create a reference/insert, then use adb_insert_create_datum() */
+    adb_reference_t reference = {0};
+    char features[MAXSTR], power[MAXSTR], times[MAXSTR];
+    lseek(adb->fd, adb->header->dataOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
+    /* FIXME: learn not to worry and love the bomb^Wbuffer overflow */
+    read(adb->fd, features, MAXSTR);
+    reference.features = features;
+    if(adb->header->flags & O2_FLAG_POWER) {
+      lseek(adb->fd, adb->header->powerTableOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
+      read(adb->fd, power, MAXSTR);
+      reference.power = power;
+    }
+    if(adb->header->flags & O2_FLAG_TIMES) {
+      lseek(adb->fd, adb->header->timesTableOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
+      read(adb->fd, times, MAXSTR);
+      reference.times = times;
+    }
+    audiodb_insert_create_datum(&reference, d);
+  } else {
+    /* initialize from sources of data that we already have */
+    d->nvectors = (*adb->track_lengths)[track_id];
+    d->dim = adb->header->dim;
+    d->key = (*adb->keys)[track_id].c_str();
+    /* read out stuff from the database tables */
+    d->data = (double *) malloc(d->nvectors * d->dim * sizeof(double));
+    lseek(adb->fd, adb->header->dataOffset + track_offset, SEEK_SET);
+    read(adb->fd, d->data, d->nvectors * d->dim * sizeof(double));
+    if(adb->header->flags & O2_FLAG_POWER) {
+      d->power = (double *) malloc(d->nvectors * sizeof(double));
+      lseek(adb->fd, adb->header->powerTableOffset + track_offset / d->dim, SEEK_SET);
+      read(adb->fd, d->power, d->nvectors * sizeof(double));
+    }
+    if(adb->header->flags & O2_FLAG_TIMES) {
+      d->times = (double *) malloc(2 * d->nvectors * sizeof(double));
+      lseek(adb->fd, adb->header->timesTableOffset + track_offset / d->dim, SEEK_SET);
+      read(adb->fd, d->times, 2 * d->nvectors * sizeof(double));
+    }
+  }
+  return 0;
+}
+
+int audiodb_datum_qpointers(adb_datum_t *d, uint32_t sequence_length, double **vector_data, double **vector, adb_qpointers_internal_t *qpointers) {
+  uint32_t nvectors = d->nvectors;
+
+  qpointers->nvectors = nvectors;
+
+  size_t vector_size = nvectors * sizeof(double) * d->dim;
+  *vector_data = new double[vector_size];
+  memcpy(*vector_data, d->data, vector_size);
+
+  qpointers->l2norm_data = new double[vector_size / d->dim];
+  audiodb_l2norm_buffer(*vector_data, d->dim, nvectors, qpointers->l2norm_data);
+  audiodb_sequence_sum(qpointers->l2norm_data, nvectors, sequence_length);
+  audiodb_sequence_sqrt(qpointers->l2norm_data, nvectors, sequence_length);
+
+  if(d->power) {
+    qpointers->power_data = new double[vector_size / d->dim];
+    memcpy(qpointers->power_data, d->power, vector_size / d->dim);
+    audiodb_sequence_sum(qpointers->power_data, nvectors, sequence_length);
+    audiodb_sequence_average(qpointers->power_data, nvectors, sequence_length);
+  }
+
+  if(d->times) {
+    qpointers->mean_duration = new double[1];
+    *qpointers->mean_duration = 0;
+    for(unsigned int k = 0; k < nvectors; k++) {
+      *qpointers->mean_duration += d->times[2*k+1] - d->times[2*k];
+    }
+    *qpointers->mean_duration /= nvectors;
+  }
+
+  *vector = *vector_data;
+  qpointers->l2norm = qpointers->l2norm_data;
+  qpointers->power = qpointers->power_data;
+  return 0;
+}
+
 int audiodb_query_spec_qpointers(adb_t *adb, adb_query_spec_t *spec, double **vector_data, double **vector, adb_qpointers_internal_t *qpointers) {
   adb_datum_t *datum;
   adb_datum_t d = {0};
-  uint32_t nvectors;
   uint32_t sequence_length;
   uint32_t sequence_start;
 
   datum = spec->qid.datum;
   sequence_length = spec->qid.sequence_length;

   } else if (datum->key) {
     uint32_t track_id;
     if((track_id = audiodb_key_index(adb, datum->key)) == (uint32_t) -1) {
       return 1;
     }
-    off_t track_offset = (*adb->track_offsets)[track_id];
-    
-    if(adb->header->flags & O2_FLAG_LARGE_ADB) {
-      /* create a reference/insert, then use adb_insert_create_datum() */
-      adb_reference_t reference = {0};
-      char features[MAXSTR], power[MAXSTR], times[MAXSTR];
-      lseek(adb->fd, adb->header->dataOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
-      /* FIXME: learn not to worry and love the bomb^Wbuffer overflow */
-      read(adb->fd, features, MAXSTR);
-      reference.features = features;
-      if(adb->header->flags & O2_FLAG_POWER) {
-        lseek(adb->fd, adb->header->powerTableOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
-        read(adb->fd, power, MAXSTR);
-        reference.power = power;
-      }
-      if(adb->header->flags & O2_FLAG_TIMES) {
-        lseek(adb->fd, adb->header->timesTableOffset + track_id * O2_FILETABLE_ENTRY_SIZE, SEEK_SET);
-        read(adb->fd, times, MAXSTR);
-        reference.times = times;
-      }
-      audiodb_insert_create_datum(&reference, &d);
-    } else {
-      /* initialize from sources of data that we already have */
-      d.nvectors = (*adb->track_lengths)[track_id];
-      d.dim = adb->header->dim;
-      d.key = datum->key;
-      /* read out stuff from the database tables */
-      d.data = (double *) malloc(d.nvectors * d.dim * sizeof(double));
-      lseek(adb->fd, adb->header->dataOffset + track_offset, SEEK_SET);
-      read(adb->fd, d.data, d.nvectors * d.dim * sizeof(double));
-      if(adb->header->flags & O2_FLAG_POWER) {
-        d.power = (double *) malloc(d.nvectors * sizeof(double));
-        lseek(adb->fd, adb->header->powerTableOffset + track_offset / d.dim, SEEK_SET);
-        read(adb->fd, d.power, d.nvectors * sizeof(double));
-      }
-      if(adb->header->flags & O2_FLAG_TIMES) {
-        d.times = (double *) malloc(2 * d.nvectors * sizeof(double));
-        lseek(adb->fd, adb->header->timesTableOffset + track_offset / d.dim, SEEK_SET);
-        read(adb->fd, d.times, 2 * d.nvectors * sizeof(double));
-      }
-    }
+    audiodb_track_id_datum(adb, track_id, &d);
   } else {
     return 1;
   }
 
-  /* Now we have a full(ish) datum, compute all the qpointery stuff
-     that we care about (l2norm/power/mean duration).  (This bit could
-     conceivably become a new function) */
-  nvectors = d.nvectors;
   /* FIXME: check the overflow logic here */
-  if(sequence_start + sequence_length > nvectors) {
-    /* is there something to free?  goto error */
+  if(sequence_start + sequence_length > d.nvectors) {
+    if(datum != &d) {
+      audiodb_free_datum(&d);
+    }
     return 1;
   }
 
-  qpointers->nvectors = nvectors;
-
-  size_t vector_size = nvectors * sizeof(double) * d.dim;
-  *vector_data = new double[vector_size];
-  memcpy(*vector_data, d.data, vector_size);
-
-  qpointers->l2norm_data = new double[vector_size / d.dim];
-  audiodb_l2norm_buffer(*vector_data, d.dim, nvectors, qpointers->l2norm_data);
-  audiodb_sequence_sum(qpointers->l2norm_data, nvectors, sequence_length);
-  audiodb_sequence_sqrt(qpointers->l2norm_data, nvectors, sequence_length);
-
-  if(d.power) {
-    qpointers->power_data = new double[vector_size / d.dim];
-    memcpy(qpointers->power_data, d.power, vector_size / d.dim);
-    audiodb_sequence_sum(qpointers->power_data, nvectors, sequence_length);
-    audiodb_sequence_average(qpointers->power_data, nvectors, sequence_length);
-  }
-
-  if(d.times) {
-    qpointers->mean_duration = new double[1];
-    *qpointers->mean_duration = 0;
-    for(unsigned int k = 0; k < nvectors; k++) {
-      *qpointers->mean_duration += d.times[2*k+1] - d.times[2*k];
-    }
-    *qpointers->mean_duration /= nvectors;
-  }
-
-
-  /* Finally, set up the moving qpointers. */
+  audiodb_datum_qpointers(&d, sequence_length, vector_data, vector, qpointers);
+
+  /* Finally, if applicable, set up the moving qpointers. */
   if(spec->qid.flags & ADB_QUERY_ID_FLAG_EXHAUSTIVE) {
-    *vector = *vector_data;
-    qpointers->l2norm = qpointers->l2norm_data;
-    qpointers->power = qpointers->power_data;
+    /* the qpointers are already at the start, and so correct. */
   } else {
+    /* adjust the qpointers to point to the correct place in the sequence */
     *vector = *vector_data + spec->qid.sequence_start * d.dim;
     qpointers->l2norm = qpointers->l2norm_data + spec->qid.sequence_start;
     if(d.power) {
       qpointers->power = qpointers->power_data + spec->qid.sequence_start;
     }
-    /* FIXME: this is a little bit ugly.  No, a lot ugly.  But at the
-     * moment this is how query_loop() knows when to stop, so for
-     * now... */
     qpointers->nvectors = sequence_length;
   }
 
   /* Clean up: free any bits of datum that we have ourselves
    * allocated. */