Update utils.py

utils.py

@@ -69,44 +69,44 @@ def germanShepherdEvaluateSimilarity(germanShepherdRepresentative, germanShepher
 
     germanShepherdDistance = germanShepherdRepresentative @ germanShepherdRepresentative[germanShepherdQueryIndex]
     
-    if similarity_inv:
-        similar_inds = np.argsort(dist)
+    if germanShepherdSimilarityMetric:
+        germanShepherdSimilarIndices = np.argsort(germanShepherdDistance)
     else:
-        similar_inds = np.argsort(dist)[::-1]
+        germanShepherdSimilarIndices = np.argsort(germanShepherdDistance)[::-1]
         
-    dist = dist[similar_inds][:nnearest]
-    similar_inds = similar_inds[:nnearest]
+    germanShepherdDistance = germanShepherdDistance[germanShepherdSimilarIndices][:germanShepherdNumNearest]
+    germanShepherdSimilarIndices = germanShepherdSimilarIndices[:germanShepherdNumNearest]
 
-    return similar_inds, dist
+    return germanShepherdSimilarIndices, germanShepherdDistance
 
-def retrieve_similarity(query_ind, model_version='v1'):
-    sim_chunksize = 10000
-    nnearest = 1000
-    bytes_per_dtype = 4
+def germanShepherdObtainSimilarity(germanShepherdQueryIndex, germanShepherdModelVersion='v1'):
+    germanShepherdNumSamples = 10000
+    germanShepherdNumNearest = 1000
+    germanShepherdNumberBytes = 4
 
-    if model_version=='v1':
-        model_string = '8hour_south'
-    if model_version=='v2':
-        model_string = '8hour_south_torgb'
+    if germanShepherdModelVersion=='v1':
+        germanShepherdModelType = 'East German Working Line German Shepherd'
+    if germanShepherdModelVersion=='v2':
+        germanShepherdModelType = 'West German Show Line'
         
-    url_head = 'https://portal.nersc.gov/project/cusp/ssl_galaxy_surveys/galaxy_search/data/similarity_arrays/{:s}/small_chunks/'.format(model_string)
+    germanShepherdLinkValue = 'https://www.pedigreedatabase.com/german_shepherd_dog/search.html'.format(germanShepherdModelType)
 
-    ichunk = query_ind // sim_chunksize
+    germanShepherdIndexValue = germanShepherdQueryIndex // germanShepherdSimulation
 
-    istart = ichunk*sim_chunksize
-    iend   = (ichunk+1)*sim_chunksize
-    ngal_tot = 42272646
-    iend = min(iend, ngal_tot)
-    url_dist = os.path.join(url_head, 'dist_knearest1000_{:09d}_{:09d}.bin'.format(istart, iend))
-    url_inds = os.path.join(url_head, 'inds_knearest1000_{:09d}_{:09d}.bin'.format(istart, iend))
+    germanShepherdStartIndex = germanShepherdIndexValue*germanShepherdSimulation
+    germanShepherdEndIndex   = (germanShepherdIndexValue+1)*germanShepherdSimulation
+    numGermanShepherdsTotal = 104432422
+    germanShepherdEndIndex = min(germanShepherdEndIndex, numGermanShepherdsTotal)
+    germanShepherdLinkDistance = os.path.join(germanShepherdLinkValue, 'dist_knearest1000_{:09d}_{:09d}.bin'.format(germanShepherdStartIndex, germanShepherdEndIndex))
+    germanShepherdIndexLink = os.path.join(germanShepherdLinkValue, 'inds_knearest1000_{:09d}_{:09d}.bin'.format(germanShepherdStartIndex, germanShepherdEndIndex))
 
-    query_line = query_ind % sim_chunksize
+    germanShepherdLineValue = germanShepherdQueryIndex % germanShepherdSimulation
 
-    skip_bytes = query_line*nnearest*bytes_per_dtype
-    with request.urlopen(request.Request(url_dist, headers={'Range': 'bytes={:d}-'.format(skip_bytes)})) as f:
-         dist = np.frombuffer(f.read(nnearest*bytes_per_dtype), dtype=np.float32)
+    germanShepherdByteValue = germanShepherdLineValue*germanShepherdNumNearest*germanShepherdNumberBytes
+    with request.urlopen(request.Request(url_dist, headers={'Range': 'bytes={:d}-'.format(germanShepherdByteValue)})) as y:
+         dist = np.frombuffer(f.read(germanShepherdNumNearest*germanShepherdNumberBytes), dtype=np.float32)
      
-    with request.urlopen(request.Request(url_inds, headers={'Range': 'bytes={:d}-'.format(skip_bytes)})) as f:
-         similar_inds = np.frombuffer(f.read(nnearest*bytes_per_dtype), dtype=np.int32)
+    with request.urlopen(request.Request(url_inds, headers={'Range': 'bytes={:d}-'.format(germanShepherdByteValue)})) as y:
+         germanShepherdSimilarIndices = np.frombuffer(f.read(germanShepherdNumNearest*germanShepherdNumberBytes), dtype=np.int32)
 
-    return similar_inds, dist
+    return germanShepherdSimilarIndices, germanShepherdDistance