Update app.py

app.py

@@ -1,107 +1,77 @@
-import os
+!pip install gradio google-generativeai
+
 import gradio as gr
 import google.generativeai as genai
 
-# 🔑 Load Gemini API key from Hugging Face Secrets
-GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
-if not GEMINI_API_KEY:
-    raise ValueError("❌ GEMINI_API_KEY not found. Please set it in Hugging Face Space → Settings → Secrets.")
-
-# Configure Gemini
+# 🔑 Configure Gemini
+GEMINI_API_KEY = "YOUR_API_KEY_HERE"
 genai.configure(api_key=GEMINI_API_KEY)
 
 # Initialize model
 model = genai.GenerativeModel("gemini-2.5-pro", generation_config={"temperature": 0})
 
-
 # ---------- STEP 1: TRANSCRIPTION ----------
 def transcribe_files(qp_file, ms_file, ans_file):
-    if ans_file is None:
-        return "⚠️ Please upload an Answer Sheet."
-
+    # Upload Answer Sheet
     uploaded_as = genai.upload_file(path=ans_file.name, display_name="Answer Sheet")
 
     transcription_instructions = """
     Persona:
     You are an expert transcriptionist specializing in scientific and mathematical documents.
-    Your task is to transcribe the provided handwritten student solutions into Markdown with LaTeX.
-
-    Rules:
-    - Use LaTeX for all math ($ ... $ inline, $$ ... $$ for display).
-    - Do NOT correct mistakes. Just transcribe faithfully.
+    Your task is to transcribe the provided handwritten student solutions into Markdown+LaTeX.
+    Follow these rules:
+    - Use LaTeX for all math ($ ... $ or $$ ... $$).
+    - Do not correct mistakes, just transcribe.
     - Ignore strikethroughs.
-    - Use **bold** for question numbering (e.g., **1.**, **2a.**).
-    - Preserve step-by-step structure (use align* for multi-line equations).
-    - If a symbol is illegible, mark it as [unclear].
+    - Use **bold** for question numbering.
+    - Preserve step-by-step derivations.
     """
 
     response = model.generate_content([transcription_instructions, uploaded_as])
-
-    # Direct extraction (like in your notebook)
     transcription = getattr(response, "text", None)
-    if not transcription:
+    if not transcription and response.candidates:
         transcription = response.candidates[0].content.parts[0].text
-
-    return transcription
-
+    return transcription or "No transcription generated."
 
 # ---------- STEP 2: GRADING ----------
 def grade_files(qp_file, ms_file, ans_file, transcription):
-    if not transcription or transcription.strip() == "":
-        return "⚠️ Please transcribe first before grading."
-
-    if qp_file is None or ms_file is None:
-        return "⚠️ Please upload Question Paper and Marking Scheme."
-
+    # Upload QP and MS
     uploaded_qp = genai.upload_file(path=qp_file.name, display_name="Question Paper")
     uploaded_ms = genai.upload_file(path=ms_file.name, display_name="Marking Scheme")
 
     grading_system = """
     Instructions to Examiners:
-    - M: Method marks (must be earned before dependent A marks).
-    - A: Accuracy marks (only if method is correct).
-    - FT: Follow-through rules apply if earlier mistakes were made.
-    - Do NOT award marks not explicitly in the markscheme.
-    - Identify errors as:
-        * Conceptual Error
-        * Silly Mistake
-        * None
-
-    Output:
-    - For each part: show awarded marks + justification.
-    - Explain reasoning step by step.
-    - Conclude with the total score.
+    - M: Method marks
+    - A: Accuracy marks
+    - FT: Follow-through rules
+    - Apply marking strictly as per scheme.
     """
 
     response = model.generate_content([
-        f"You are an official examiner. Grade the student's transcription using the rules:\n{grading_system}",
+        f"You are an examiner. Grade the transcription using the rules:\n{grading_system}",
         uploaded_qp,
         uploaded_ms,
         transcription
     ])
-
-    # Direct extraction (like in your notebook)
     grading = getattr(response, "text", None)
-    if not grading:
+    if not grading and response.candidates:
         grading = response.candidates[0].content.parts[0].text
-
-    return grading
-
+    return grading or "No grading generated."
 
 # ---------- GRADIO UI ----------
 with gr.Blocks() as demo:
     gr.Markdown("## 📘 Automated Transcription & Grading System")
 
     with gr.Row():
-        qp = gr.File(label="📄 Upload Question Paper (PDF)", file_types=[".pdf"])
-        ms = gr.File(label="📑 Upload Marking Scheme (PDF)", file_types=[".pdf"])
-        ans = gr.File(label="📝 Upload Answer Sheet (PDF)", file_types=[".pdf"])
+        qp = gr.File(label="Upload Question Paper (PDF)")
+        ms = gr.File(label="Upload Marking Scheme (PDF)")
+        ans = gr.File(label="Upload Answer Sheet (PDF)")
 
     transcribe_btn = gr.Button("🔍 Transcribe Answer Sheet")
-    transcription_output = gr.Textbox(label="🖋️ Transcription", lines=20)
+    transcription_output = gr.Textbox(label="Transcription", lines=20)
 
     grade_btn = gr.Button("✅ Grade Answers")
-    grading_output = gr.Textbox(label="📊 Grading Result", lines=20)
+    grading_output = gr.Textbox(label="Grading Result", lines=20)
 
     transcribe_btn.click(fn=transcribe_files, inputs=[qp, ms, ans], outputs=transcription_output)
     grade_btn.click(fn=grade_files, inputs=[qp, ms, ans, transcription_output], outputs=grading_output)