Update app.py

app.py

@@ -1,6 +1,6 @@
 import streamlit as st
 from datasets import load_dataset, concatenate_datasets
-import os
+import json
 
 def load_and_combine_datasets():
     python_codes_dataset = load_dataset('flytech/python-codes-25k', split='train')
@@ -11,10 +11,11 @@ def load_and_combine_datasets():
     
     return combined_dataset
 
-def save_combined_dataset(combined_dataset, dataset_name):
-    combined_dataset.save_to_disk(dataset_name)
-    # Print the location where the dataset is saved
-    st.write(f"Dataset saved at: {dataset_name}")
+def save_combined_dataset_as_jsonl(combined_dataset, file_path):
+    with open(file_path, 'w', encoding='utf-8') as f:
+        for example in combined_dataset:
+            json.dump(example, f, ensure_ascii=False)
+            f.write('\n')
 
 def main():
     st.title("Combined Dataset Viewer")
@@ -25,22 +26,11 @@ def main():
     # Display a subset of the combined dataset
     st.write("Subset of Combined Dataset:", combined_dataset[:10])
 
-    # Add option to save the combined dataset
-    if st.button("Save Combined Dataset"):
-        dataset_name = st.text_input("Enter a name for the combined dataset:")
-        if dataset_name:
-            save_combined_dataset(combined_dataset, dataset_name)
-            st.success(f"Combined dataset saved as '{dataset_name}'!")
-
-    # Add option to download the combined dataset
-    if st.button("Download Combined Dataset"):
-        dataset_name = st.text_input("Enter a name for the combined dataset:")
-        if dataset_name:
-            save_combined_dataset(combined_dataset, dataset_name)
-            filepath = os.path.join(os.getcwd(), dataset_name)
-            filesize = os.path.getsize(filepath) / (1024 * 1024)  # Size in MB
-            st.write(f"Download the combined dataset: [{dataset_name}]({filepath})")
-            st.write(f"File Size: {filesize:.2f} MB")
+    # Add option to save the combined dataset as JSONL
+    if st.button("Save Combined Dataset (JSONL)"):
+        file_path = "combined_dataset.jsonl"
+        save_combined_dataset_as_jsonl(combined_dataset, file_path)
+        st.write(f"Combined dataset saved as JSONL file: {file_path}")
 
 if __name__ == "__main__":
     main()