1021 add flexable attr control

data_representation/step1_midi2corpus_fined.py

@@ -168,7 +168,7 @@ class CorpusMaker():
       print("length of midi list: ", len(self.midi_list))
       # Use set for faster lookup (O(1) per check)
       processed_files_set = set(processed_files)
-      # self.midi_list = [x for x in self.midi_list if x.name not in processed_files_set]
+      self.midi_list = [x for x in self.midi_list if x.name not in processed_files_set]
       # reverse the list to process the latest files first
       self.midi_list.reverse()
       print(f"length of midi list after filtering: ", len(self.midi_list))

data_representation/step2_corpus2event.py

@@ -61,7 +61,7 @@ class Corpus2Event():
     # remove the corpus files that are already in the out_dir
     # Use set for faster existence checks
     existing_files = set(f.name for f in self.out_dir.glob("*.pkl"))
-    # corpus_list = [corpus for corpus in corpus_list if corpus.name not in existing_files]
+    corpus_list = [corpus for corpus in corpus_list if corpus.name not in existing_files]
     for filepath_name, event in tqdm(map(self._load_single_corpus_and_make_event, corpus_list), total=len(corpus_list)):
       if event is None:
         broken_count += 1