only process if at least 2 files

model_run.sbatch

@@ -6,8 +6,8 @@
 #SBATCH -n 5                             # use 5 tasks
 #SBATCH --cpus-per-task=1                # use 1 thread per taks
 #SBATCH -N 1                             # request slots on 1 node
-#SBATCH --output=/scratch/grzonkow/model_0102.txt         # capture output
-#SBATCH --error=/scratch/grzonkow/err_0102.txt          # and error streams
+#SBATCH --output=/scratch/grzonkow/model_new.txt         # capture output
+#SBATCH --error=/scratch/grzonkow/err_new.txt          # and error streams
 
 
 

preprocess.py

@@ -121,6 +121,8 @@ def get_paths_diff_charts(song_dirs, indexed_vocabulary, device):
         all_paths_old_len = len(all_paths)
         audio_exists = False
         list_items = os.listdir(song_dir)
+        # no runthrough, need at least 2 files, sm and audio file
+        if len(list_items) >= 2:
             for item in os.listdir(song_dir):
                 if item.endswith('.sm'):
                     opened_file = simfile.open(f"{song_dir}/{item}")
@@ -136,10 +138,10 @@ def get_paths_diff_charts(song_dirs, indexed_vocabulary, device):
                     # sample notes over time, return list of NoteData and [] if empy (for now)
                     converted_charts = []
 
-                if idx > 2500:
-                    print(
-                        f"File: {item}, Number_charts: {len(charts)}, Song: {idx}, Processed charts: {num_charts}, "
-                        f"Elapsed time: {elapsed_second_step}, Time: {now}")
+                    #if idx > 2500:
+                    #    print(
+                    #        f"File: {item}, Number_charts: {len(charts)}, Song: {idx}, Processed charts: {num_charts}, "
+                    #        f"Elapsed time: {elapsed_second_step}, Time: {now}")
                     for i, chart in enumerate(charts):
                         num_charts += 1
                         notes_per_time = process_chart(opened_file, sm_file, chart, indexed_vocabulary, device)
@@ -148,10 +150,10 @@ def get_paths_diff_charts(song_dirs, indexed_vocabulary, device):
                         if max_len_charts < len(notes_per_time):
                             max_len_charts = len(notes_per_time)
                         converted_charts.append(notes_per_time)
-                    if idx > 2500:
-                        print(
-                            f"File: {item}, Chart: {i}, "
-                            f"Elapsed time: {elapsed_second_step}, Time: {now}")
+                       # if idx > 2500:
+                       #     print(
+                       #         f"File: {item}, Chart: {i}, "
+                       #         f"Elapsed time: {elapsed_second_step}, Time: {now}")
                     all_difficulties.extend(difficulties)
                     all_charts.extend(converted_charts)
                     assert (len(converted_charts) == len(difficulties))