Make Kaggle authentication lazy for HF Space

src/mlpipeline/components/data_ingestion.py

@@ -1,7 +1,6 @@
 import os
 import shutil
 from pathlib import Path
-from kaggle.api.kaggle_api_extended import KaggleApi
 
 from mlpipeline.entity import DataIngestionConfig, DataIngestionArtifact
 from mlpipeline.logging.logger import get_logger
@@ -14,13 +13,29 @@ logger = get_logger(__name__)
 class DataIngestion:
     def __init__(self, config: DataIngestionConfig):
         self.config = config
-        self.kaggle_api = KaggleApi()
-        self.kaggle_api.authenticate()
+        self.kaggle_api = None
+    
+    def _authenticate_kaggle(self):
+        """Lazy authentication - only when needed"""
+        if self.kaggle_api is None:
+            try:
+                from kaggle.api.kaggle_api_extended import KaggleApi
+                self.kaggle_api = KaggleApi()
+                self.kaggle_api.authenticate()
+                logger.info("Kaggle API authenticated successfully")
+            except Exception as e:
+                raise DataIngestionException(
+                    f"Failed to authenticate with Kaggle API: {e}",
+                    sys
+                )
     
     def download_data(self) -> DataIngestionArtifact:
         try:
             logger.info("Starting data ingestion")
             
+            # Authenticate only when downloading
+            self._authenticate_kaggle()
+            
             os.makedirs(self.config.root_dir, exist_ok=True)
             
             competition_name = "playground-series-s6e2"