Update handler.py

handler.py

@@ -256,17 +256,43 @@ def _run_align_script(bo_fn, en_fn, output_dir):
     logging.info(f"Total time taken for Aligning: {total_time} mins")
     return output_fn
 def align(text_pair):
+    
     logging.info(f"Running aligner for TM{text_pair['text_id']}...")
     with TemporaryDirectory() as tmpdir:
         output_dir = Path(tmpdir)
-        bo_fn = download_file(text_pair["bo_file_url"], output_fn=output_dir / "bo.tx")
-        en_fn = download_file(text_pair["en_file_url"], output_fn=output_dir / "en.tx")
-        print("bo_fn: ", bo_fn)
-        print("en_fn: ", en_fn)
-        aligned_fn = _run_align_script(bo_fn, en_fn, output_dir)
-        print("aligned_fn: ", aligned_fn)
-        repo_url = create_tm(aligned_fn, text_pair=text_pair)
-        return {"tm_repo_url": repo_url}
+        
+        # Download files and validate them
+        bo_fn = download_file(text_pair["bo_file_url"], output_fn=output_dir / "bo.txt")
+        en_fn = download_file(text_pair["en_file_url"], output_fn=output_dir / "en.txt")
+
+        # Check if files are downloaded correctly
+        if not bo_fn.exists() or bo_fn.stat().st_size == 0:
+            logging.error(f"Failed to download or empty file: {bo_fn}")
+            return {"error": "Failed to download Tibetan file or file is empty"}
+
+        if not en_fn.exists() or en_fn.stat().st_size == 0:
+            logging.error(f"Failed to download or empty file: {en_fn}")
+            return {"error": "Failed to download English file or file is empty"}
+
+        # Log content of files for verification
+        logging.info(f"Content of {bo_fn.name}: {bo_fn.read_text()[:100]}...")
+        logging.info(f"Content of {en_fn.name}: {en_fn.read_text()[:100]}...")
+
+        # Run alignment script
+        try:
+            aligned_fn = _run_align_script(bo_fn, en_fn, output_dir)
+            logging.info(f"Alignment script output file: {aligned_fn}")
+        except Exception as e:
+            logging.error(f"Alignment script error: {e}")
+            return {"error": f"Alignment script failed: {e}"}
+
+        # Create TM repository
+        try:
+            repo_url = create_tm(aligned_fn, text_pair=text_pair)
+            return {"tm_repo_url": repo_url}
+        except Exception as e:
+            logging.error(f"Error in creating TM repository: {e}")
+            return {"error": f"Error in repository creation: {e}"}
 
 class EndpointHandler():
     def __init__(self, path=""):