Update app.py

app.py

@@ -6,7 +6,7 @@ def run_pipeline(pdf_file, system_prompt, max_step, learning_rate, epochs, model
     # Construct job input
     data = {
         "input": {
-            "pdf_file": pdf_file,
+            "pdf_file": pdf_file.name,
             "system_prompt": system_prompt,
             "max_step": max_step,
             "learning_rate": learning_rate,
@@ -26,39 +26,43 @@ def run_pipeline(pdf_file, system_prompt, max_step, learning_rate, epochs, model
         )
         stdout, stderr = process.communicate()
 
-        if process.returncode != 0:
-            return {"status": "error", "details": f"Handler failed: {stderr.strip()}"}
-
-        # Parse the handler output
-        try:
-            handler_output = json.loads(stdout)
-            if handler_output.get("status") == "success":
-                # Extract relevant fields
-                model_name = handler_output.get("model_name", "N/A")
-                evaluation_results = handler_output.get("evaluation_results", {})
-                semantic_score = evaluation_results.get("average_semantic_score", "N/A")
-                bleu_score = evaluation_results.get("average_bleu_score", "N/A")
-
-                return {
-                    "status": "success",
-                    "model_name": model_name,
-                    "evaluation_results": {
-                        "average_semantic_score": semantic_score,
-                        "average_bleu_score": bleu_score
-                    },
-                }
-            else:
-                # Return error from handler
-                return handler_output
-
-        except json.JSONDecodeError:
-            return {"status": "error", "details": f"Invalid JSON from handler: {stdout.strip()}"}
+        # Extract JSON object from the output
+        output_lines = stdout.splitlines()
+        handler_output = None
+        for line in output_lines:
+            try:
+                parsed_line = json.loads(line)
+                if isinstance(parsed_line, dict) and "status" in parsed_line:
+                    handler_output = parsed_line
+                    break
+            except json.JSONDecodeError:
+                continue
+
+        if handler_output is None:
+            return {"status": "error", "details": f"No valid JSON found in output: {stdout}"}
+
+        # Check the status in the parsed JSON
+        if handler_output.get("status") == "success":
+            # Extract and format the result
+            model_name = handler_output.get("model_name", "N/A")
+            processing_time = handler_output.get("processing_time", "N/A")
+            evaluation_results = handler_output.get("evaluation_results", {})
+
+            return {
+                "model_name": model_name,
+                "processing_time": processing_time,
+                "evaluation_results": evaluation_results
+            }
+        else:
+            # Return error details from the handler output
+            return handler_output
 
     except FileNotFoundError:
         return {"status": "error", "details": "Handler script not found"}
     except Exception as e:
         return {"status": "error", "details": str(e)}
 
+
 # Define Gradio interface
 with gr.Blocks(css='''
     .gradio-container {
@@ -132,7 +136,7 @@ with gr.Blocks(css='''
     # Layout structure with improved spacing
     with gr.Row():
         with gr.Column(scale=2):
-            pdf_file = gr.Textbox(label="PDF Blob Name", placeholder="Enter the PDF file", value="finetuning_pipeline_testing_V1")
+            pdf_file = gr.File(label="Upload PDF File", file_types=[".pdf"])  # Updated to accept file uploads
         with gr.Column(scale=3):
             system_prompt = gr.Textbox(label="System Prompt", placeholder="Enter system instructions or context", value="You are a helpful assistant that provides detailed information based on the provided text.")
         with gr.Column(scale=2):