fix default input

wimudp/data_poisoning/dirty_label/dataset_filter.py

@@ -22,3 +22,7 @@ def process_csv_file(concept_a: str = None, concept_c_action: str = None, rows_n
 
 def filter_caption_len(row: pd.Series, concept_a: str, concept_c_action: str) -> bool:
     return concept_c_action in row["caption"] and concept_a not in row["caption"]
+
+if __name__ == "__main__":
+    df = process_csv_file()
+    df.head(ROWS_NUMBER).to_csv(CSV_CONCEPT_C_FILE)

wimudp/data_poisoning/generate_dirty_label.py

@@ -5,8 +5,11 @@
 
 def main():
     concept_a = input("Enter value for CONCEPT_A (or press Enter to use default): ").strip()
+    concept_a = None if concept_a == "" else concept_a
+
     concept_a_action = input("Enter value for CONCEPT_A_ACTION (or press Enter to use default): ").strip()
-
+    concept_a_action = None if concept_a_action == "" else concept_a_action
+
     rows_number = input("Enter value for ROWS_NUMBER (or press Enter to use default): ").strip()
     rows_number = int(rows_number) if rows_number.isdigit() else None
 

wimudp/data_poisoning/generate_nightshade.py

@@ -4,15 +4,21 @@
 
 def main():
     concept_a = input("Enter value for CONCEPT_A (or press Enter to use default): ").strip()
+    concept_a = None if concept_a == "" else concept_a
+
     concept_c_action = input("Enter value for CONCEPT_C_ACTION (or press Enter to use default): ").strip()
-    rows_number = input("Enter value for ROWS_NUMBER (or press Enter to use default): ").strip()
+    concept_c_action = None if concept_c_action == "" else concept_c_action
 
+    rows_number = input("Enter value for ROWS_NUMBER (or press Enter to use default): ").strip()
     rows_number = int(rows_number) if rows_number.isdigit() else None
 
     concept_c = input("Enter value for CONCEPT_C (or press Enter to use default): ").strip()
+    concept_c = None if concept_c == "" else concept_c
+
     concept_c_action_extractor = input("Enter value for CONCEPT_C_ACTION (or press Enter to use default): ").strip()
-    samples_number = input("Enter value for SAMPLES_NUMBER (or press Enter to use default): ").strip()
+    concept_c_action_extractor = None if concept_c_action_extractor == "" else concept_c_action_extractor
 
+    samples_number = input("Enter value for SAMPLES_NUMBER (or press Enter to use default): ").strip()
     samples_number = int(samples_number) if samples_number.isdigit() else None
 
     display_stage("Filtering data")

wimudp/data_poisoning/nightshade/data_extractor.py

@@ -56,3 +56,8 @@ def check_audio_file(youtube_id: str) -> bool:
     file_path = os.path.join(os.getcwd(), AUDIOS_SAMPLES_DIR, f"{youtube_id}.wav")
 
     return os.path.exists(file_path)
+
+if __name__ == "__main__":
+    samples = get_samples()
+
+    samples.to_csv(CSV_NS_SAMPLES_FILE, index=False)