Update app.py

app.py

@@ -12,7 +12,7 @@ import json
 import io
 from tqdm import tqdm
 import subprocess
-from huggingface_hub import snapshot_download, upload_file
+from huggingface_hub import snapshot_download, upload_file, HfApi, create_repo
 
 # Function to download a Parquet file from a specified URL
 def download_parquet(url, local_path):
@@ -454,14 +454,33 @@ def generate_failed_items_str(indices):
 
 # Function to upload the output file to Hugging Face
 def upload_output_to_huggingface(output_file_path, repo_name, token):
-    upload_file(
-        path_or_fileobj=output_file_path,
-        path_in_repo=output_file_path,
-        repo_id=repo_name,
-        repo_type="dataset",
-        token=token
-    )
-    print(f"Uploaded {output_file_path} to Hugging Face repository: {repo_name}")
+    api = HfApi()
+    
+    # Check if the repository exists
+    try:
+        api.repo_info(repo_id=repo_name, repo_type="dataset", token=token)
+    except Exception as e:
+        if "404" in str(e):
+            # Create the repository if it doesn't exist
+            create_repo(repo_id=repo_name, repo_type="dataset", token=token)
+            print(f"Created repository: {repo_name}")
+        else:
+            print(f"Failed to check repository existence: {e}")
+            return
+
+    # Upload the file to the repository
+    try:
+        upload_file(
+            path_or_fileobj=output_file_path,
+            path_in_repo=output_file_path,
+            repo_id=repo_name,
+            repo_type="dataset",
+            token=token
+        )
+        print(f"Uploaded {output_file_path} to Hugging Face repository: {repo_name}")
+    except Exception as e:
+        print(f"Failed to upload {output_file_path} to Hugging Face: {e}")
+        raise
 
 def translate_dataset(train_url, local_parquet_path, input_file_path, output_file_path, raw_file_path, range_specification, model_type, output_dir, output_repo_name, token, translator, tokenizer):
     try: