Better error handling when sleep duration is less than 30 mins

src/asleep/get_sleep.py

@@ -310,23 +310,30 @@ def main():
     (binary_y, all_sleep_wins_df, sleep_wins_long_per_day_df, master_acc,
      master_npids) = get_sleep_windows(data2model, times, non_wear, args)
 
-    y_pred, test_pids = start_sleep_net(
-        master_acc, master_npids, args.outdir,
-        args.model_weight_path, local_repo_path=args.local_repo_path,
-        device_id=args.pytorch_device)
-    sleepnet_output = binary_y
+    if len(master_npids) <= 0:
+        print("No sleep windows >30 mins detected. Exiting...")
+        print("Non-wear time has been written to %s" % non_wear_path)
+        print("Current sleep classification has been written to %s" %
+              os.path.join(args.outdir, 'ssl_sleep.npy'))
+        exit()
+    else:
+        y_pred, test_pids = start_sleep_net(
+            master_acc, master_npids, args.outdir,
+            args.model_weight_path, local_repo_path=args.local_repo_path,
+            device_id=args.pytorch_device)
+        sleepnet_output = binary_y
 
-    for block_id in range(len(all_sleep_wins_df)):
-        start_t = all_sleep_wins_df.iloc[block_id]['start']
-        end_t = all_sleep_wins_df.iloc[block_id]['end']
+        for block_id in range(len(all_sleep_wins_df)):
+            start_t = all_sleep_wins_df.iloc[block_id]['start']
+            end_t = all_sleep_wins_df.iloc[block_id]['end']
 
-        time_filter = (times >= start_t) & (times < end_t)
+            time_filter = (times >= start_t) & (times < end_t)
 
-        # get the corresponding sleepnet predictions
-        sleepnet_pred = y_pred[test_pids == block_id]
+            # get the corresponding sleepnet predictions
+            sleepnet_pred = y_pred[test_pids == block_id]
 
-        # fill the sleepnet predictions back to the original dataframe
-        sleepnet_output[time_filter] = sleepnet_pred
+            # fill the sleepnet predictions back to the original dataframe
+            sleepnet_output[time_filter] = sleepnet_pred
 
     # 3. Skip this step if predictions already exist
     # Output pandas dataframe