Update app.py

app.py

@@ -4,7 +4,7 @@ import traceback
 import tempfile
 import os
 import gradio as gr
-from typing import Tuple, Optional, Any  # Added Any for file_data_binary
+from typing import Tuple, Optional, Any # Any for file_data_binary for now, though bytes is expected
 
 # Import config first to ensure JAVA_HOME is set early
 import config
@@ -14,18 +14,26 @@ from main_analyzer import analyze_pdf
 # import language_tool_python
 
 
-def process_upload(file_data_binary: Optional[Any]) -> Tuple[str, Optional[str]]:
-    if file_data_binary is None or not hasattr(file_data_binary, 'read'):
-        if isinstance(file_data_binary, bytes):
-            pass  # Good, it's bytes
-        elif file_data_binary is None:
-            return json.dumps({"error": "No file uploaded or file data is None"}, indent=2), None
-        elif not hasattr(file_data_binary, 'read'):
-            return json.dumps({"error": f"Unexpected file data type: {type(file_data_binary)}"}), None
+def process_upload(file_data_binary: Optional[bytes]) -> Tuple[str, Optional[str]]: # Explicitly Optional[bytes]
+    if not isinstance(file_data_binary, bytes):
+        if file_data_binary is None:
+            error_msg = "No file uploaded or file data is None."
+        else:
+            error_msg = f"Unexpected file data type: {type(file_data_binary)}. Expected bytes."
+        return json.dumps({"error": error_msg}, indent=2), None
 
+    temp_pdf_path = None
     try:
-        print(f"App: Processing uploaded file...")
-        results_dict, _ = analyze_pdf(file_data_binary)
+        # Create a temporary file to store the uploaded PDF bytes
+        # delete=False is used because analyze_pdf will open it by path.
+        # We are responsible for deleting it in the finally block.
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_file:
+            temp_pdf_path = temp_file.name
+            temp_file.write(file_data_binary)
+        # The file is closed when exiting the 'with' block, but still exists due to delete=False.
+
+        print(f"App: Processing PDF via temporary file: {temp_pdf_path}")
+        results_dict, _ = analyze_pdf(temp_pdf_path) # Pass the path to the temporary file
         results_json = json.dumps(results_dict, indent=2, ensure_ascii=False)
         return results_json, None
 
@@ -33,6 +41,14 @@ def process_upload(file_data_binary: Optional[Any]) -> Tuple[str, Optional[str]]
         print(f"Error in process_upload: {e}\n{traceback.format_exc()}")
         error_message = json.dumps({"error": str(e), "traceback": traceback.format_exc()}, indent=2)
         return error_message, None
+    finally:
+        # Clean up the temporary file if it was created
+        if temp_pdf_path and os.path.exists(temp_pdf_path):
+            try:
+                os.remove(temp_pdf_path)
+                print(f"App: Cleaned up temporary PDF file: {temp_pdf_path}")
+            except Exception as e_clean:
+                print(f"App: Error cleaning up temporary PDF file {temp_pdf_path}: {e_clean}")
 
 
 def create_interface():
@@ -41,7 +57,7 @@ def create_interface():
             file_input = gr.File(
                 label="Upload PDF",
                 file_types=[".pdf"],
-                type="binary"
+                type="binary"  # This ensures file_data_binary is bytes
             )
 
         with gr.Row():
@@ -86,4 +102,4 @@ if __name__ == "__main__":
     app_interface.launch(
         share=False,
         # server_port=7860
-    )
+    )