Update app.py

app.py

@@ -25,39 +25,68 @@ import matplotlib.pyplot as plt
 #==========================================================================================================
 
 def find_midi(title, artist):
-  
+
     print('=' * 70)
-    print('Loading MIDI file...')
-    
-    #==================================================
-    
-    print('Searching titles...Please wait...')
-    random.shuffle(AUX_DATA)
-    
-    titles_index = []
-    
-    for A in AUX_DATA:
-      titles_index.append(A[0])
+    print('Req start time: {:%Y-%m-%d %H:%M:%S}'.format(datetime.datetime.now(PDT)))
+    start_time = reqtime.time()
+
+    print('-' * 70)
+    print('Req title:', title)
+    print('Req artist:', artist)
+    print('-' * 70)
+
+
+    input_text = ''
+
+    if title != '':
+        input_text += title
+    if artist != '':
+        input_text += ' by ' + artist
     
-    search_string = ''
+    print('Searching...')
+
+    query_embedding = model.encode([input_text])
     
-    if title != '' and artist != '':
-      search_string = title + ' --- ' + artist
+    # Compute cosine similarity between query and each sentence in the corpus
+    similarities = util.cos_sim(query_embedding, corpus_embeddings)
     
-    else:
-      search_string = title + artist
+    # Find the index of the most similar sentence
+    closest_index = np.argmax(similarities)
+    closest_index_match_ratio = max(similarities[0]).tolist()
+
+    best_corpus_match = all_MIDI_files_names[closest_index]
+
+    print('Done!')
+    print('=' * 70)
     
-    search_match = process.extract(query=search_string, choices=titles_index, limit=1)
-    search_index = titles_index.index(search_match[0][0])
+    print('Match corpus index', closest_index)
+    print('Match corpus ratio', closest_index_match_ratio)
     
-    print('Done!')
     print('=' * 70)
-    print('Selected title:', AUX_DATA[search_index][0])
+    print('Done!')
     print('=' * 70)
 
-    outy = AUX_DATA[search_index][1]
+    song_artist = best_corpus_match[0]
+    zlib_file_name = best_corpus_match[1]
+
+    print('Fetching MIDI score...')
 
-    print('Sample INTs', outy[:12])
+    with open(zlib_file_name, 'rb') as f:
+        compressed_data = f.read()
+    
+    # Decompress the data
+    decompressed_data = zlib.decompress(compressed_data)
+    
+    # Convert the bytes back to a list using pickle
+    scores_data = pickle.loads(decompressed_data)
+    
+    fnames = [f[0] for f in scores_data]
+    
+    fnameidx = fnames.index(song_artist)
+    
+    MIDI_score_data = scores_data[fnameidx][1]
+    
+    print('Sample INTs', MIDI_score_data[:12])
     print('=' * 70)
     
     if len(outy) != 0: