Update app.py

app.py

@@ -7,7 +7,7 @@ import re
 import base64
 import mimetypes
 from datasets import Dataset
-from huggingface_hub import HfApi, login, get_token
+from huggingface_hub import HfApi, get_token
 import huggingface_hub
 import os
 from mistralai import Mistral
@@ -18,7 +18,7 @@ logger = logging.getLogger(__name__)
 
 # --- Mistral OCR Setup ---
 api_key = os.environ.get("MISTRAL_API_KEY")
-hf_token_global = None  # Store HF token globally
+hf_token_global = None
 client = None
 
 if not api_key:
@@ -112,18 +112,33 @@ def perform_ocr_file(file_obj: Any) -> Tuple[str, str, Dict[str, str]]:
         uploaded_file_id = None
 
         if file_ext == '.pdf':
-            with open(file_path, "rb") as f:
+            try:
+                with open(file_path, "rb") as f:
+                    file_content = f.read()
+                
                 logger.info(f"Uploading PDF {file_name} to Mistral...")
-                uploaded_pdf = client.files.upload(file=(file_name, f), purpose="ocr")
+                files = {
+                    "file": (file_name, file_content, "application/pdf")
+                }
+                uploaded_pdf = client.files.upload(
+                    file=files["file"],
+                    purpose="ocr"
+                )
                 uploaded_file_id = uploaded_pdf.id
+                logger.info(f"PDF uploaded successfully. File ID: {uploaded_file_id}")
+
                 signed_url_response = client.files.get_signed_url(file_id=uploaded_file_id)
                 ocr_response = client.ocr.process(
                     model="mistral-ocr-latest",
                     document={"type": "document_url", "document_url": signed_url_response.url},
                     include_image_base64=True
                 )
-            if uploaded_file_id:
-                client.files.delete(file_id=uploaded_file_id)
+            finally:
+                if uploaded_file_id:
+                    try:
+                        client.files.delete(file_id=uploaded_file_id)
+                    except Exception as delete_err:
+                        logger.warning(f"Failed to delete temporary file {uploaded_file_id}: {delete_err}")
 
         elif file_ext in ['.png', '.jpg', '.jpeg', '.webp', '.bmp']:
             with open(file_path, "rb") as f: