Update app.py

app.py

@@ -5,11 +5,8 @@ import os
 from dotenv import load_dotenv
 
 # --- Configuration ---
-# Load environment variables from a .env file if it exists (for local development)
 load_dotenv()
 
-# It's recommended to set your GOOGLE_API_KEY as a secret in your Hugging Face Space settings.
-# The key should be named GOOGLE_API_KEY.
 try:
     api_key = os.environ.get("GOOGLE_API_KEY")
     if not api_key:
@@ -18,19 +15,16 @@ try:
     model = genai.GenerativeModel('gemini-2.5-flash')
 except Exception as e:
     print(f"Error configuring GenerativeAI: {e}")
-    # Handle the case where the API key is not set, e.g., by disabling the interface or showing an error.
     model = None
 
 # --- Data Loading ---
 try:
     df = pd.read_csv('qp_with_ms_images (1).csv')
-    # Clean up column names that might have extra spaces
     df.columns = df.columns.str.strip()
 except FileNotFoundError:
     print("Error: 'qp_with_ms_images (1).csv' not found. Please upload the data file to your Space.")
-    df = pd.DataFrame() # Create an empty DataFrame to avoid errors
+    df = pd.DataFrame()
 
-# --- Provided Topics List ---
 TOPICS = """
 SL 1.1 - Operations with numbers in the form a × 10k where 1 < a < 10 and k is an integer.
 SL 1.2 - Arithmetic sequences and series. Use of the formulae for the nth term and the sum of the first n terms of the sequence. Use of sigma notation for sums of arithmetic sequences. Applications. Analysis, interpretation and prediction where a model is not perfectly arithmetic in real life.
@@ -125,57 +119,69 @@ AHL 5.19 - Maclaurin series to obtain expansions for eˣ, sinx, cosx, ln(1+x), (
 def generate_test(total_marks, topic_name, level):
     """
     Generates a test paper using the Gemini LLM based on user inputs.
+    Prioritizes the most recent questions from the topic.
     """
     if model is None or df.empty:
         return "Error: Application is not configured correctly. Please check API key and data file."
 
-    # Filter DataFrame based on level
+    # --- Filter and sort by latest date ---
     if level == 'SL':
         filtered_df = df[df['Level'] == 'SL'].copy()
-    else: # AHL can use both SL and AHL questions
+    else:
         filtered_df = df.copy()
 
-    # Prepare data for the prompt (convert relevant columns to string)
-    # Using a sample to keep the prompt size manageable if the filtered data is too large
+    # Sort by date descending (newest first)
+    if 'Date' in filtered_df.columns:
+        try:
+            filtered_df['Date'] = pd.to_datetime(filtered_df['Date'], errors='coerce')
+            filtered_df = filtered_df.sort_values(by='Date', ascending=False)
+        except Exception:
+            pass  # If Date parsing fails, skip sorting
+
+    # Limit dataset size for LLM input
     if len(filtered_df) > 300:
-        filtered_df = filtered_df.sample(n=300, random_state=42)
+        filtered_df = filtered_df.head(300)
 
     question_data = filtered_df[[
         'Question Number', 'Date', 'Topic', 'Level', 'Max Marks', 'QP Content'
     ]].to_string(index=False)
 
-    # --- Prompt for the LLM ---
-   prompt = f"""
-You are an expert test creator for the International Baccalaureate (IB) Mathematics program.
-Your task is to create a test paper based on the user's request using a provided question bank.
-
-**User Request:**
-- Total Marks: {total_marks}
-- Topic: "{topic_name}"
-- Level: {level}
-
-**Instructions:**
-1. Carefully select a combination of questions from the provided **QUESTION BANK** that best fit the requested topic and level.
-2. The total marks of all selected questions should be as close as possible to the requested **Total Marks ({total_marks})**.
-3. When selecting questions, **prioritize the most recent questions** (based on the 'Date' field) within the requested topic and level.
-4. If multiple questions have similar relevance or marks, **prefer newer questions** over older ones.
-5. Present the final test in a clean, well-formatted markdown format. It should look like a real test paper.
-6. For each question, you **MUST** include its ID (from the 'Question Number' column), Date, Topic code, and the maximum marks.
-7. Display the full question content, including all sub-questions and any included diagrams or formulas, exactly as it appears in the 'QP Content' column.
-8. Use the provided **TOPICS LIST** to understand which topic codes fall under the user's requested topic name. For example, if the user requests 'Calculus', you should look for questions with topic codes like SL 5.1, SL 5.2, ..., AHL 5.19.
-
-**TOPICS LIST:**
----
-{TOPICS}
----
-
-**QUESTION BANK:**
----
-{question_data}
----
-
-Now, create the test based on the user's request, ensuring that the **latest available questions** for the topic are prioritized.
-"""
+    # --- Enhanced Prompt (with latest question priority) ---
+    prompt = f"""
+    You are an expert test creator for the International Baccalaureate (IB) Mathematics program.
+    Your task is to create a test paper based on the user's request using a provided question bank.
+
+    **User Request:**
+    - Total Marks: {total_marks}
+    - Topic: "{topic_name}"
+    - Level: {level}
+
+    **Instructions:**
+    1. Carefully select a combination of questions from the provided **QUESTION BANK** that best fit the requested topic and level.
+    2. The total marks of all selected questions should be as close as possible to the requested **Total Marks ({total_marks})**.
+    3. When choosing questions, **prioritize the latest (most recent) questions based on the 'Date' field**.
+    4. If multiple questions fit equally well, choose the newest ones first.
+    5. Present the final test in a clean, well-formatted markdown format. It should look like a real test paper.
+    6. For each question, include:
+        - **Question ID** (from 'Question Number')
+        - **Date**
+        - **Topic Code**
+        - **Maximum Marks**
+    7. Display the full question content exactly as it appears in 'QP Content'.
+    8. Use the provided **TOPICS LIST** to identify relevant topic codes.
+
+    **TOPICS LIST:**
+    ---
+    {TOPICS}
+    ---
+
+    **QUESTION BANK (sorted by most recent first):**
+    ---
+    {question_data}
+    ---
+
+    Now, create the test based on the user's request, making sure to use the **latest available questions** first.
+    """
 
     try:
         response = model.generate_content(prompt)
@@ -185,29 +191,23 @@ Now, create the test based on the user's request, ensuring that the **latest ava
 
 # --- Gradio Interface ---
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 📝 IB Mathematics Test Generator")
+    gr.Markdown("# 📝 IB Mathematics Test Generator (Latest Question Priority)")
     gr.Markdown(
-        "Enter the desired total marks, a topic name, and the level (SL/AHL) to generate a test paper "
-        "using a bank of past questions."
+        "Enter the desired total marks, a topic name, and the level (SL/AHL). "
+        "This version prioritizes the **latest available questions** when generating your test."
     )
 
     with gr.Row():
         with gr.Column(scale=1):
             total_marks_input = gr.Slider(
-                minimum=10,
-                maximum=150,
-                step=1,
-                value=50,
-                label="Total Marks for the Test"
+                minimum=10, maximum=150, step=1, value=50, label="Total Marks for the Test"
             )
             topic_input = gr.Textbox(
                 label="Topic Name",
                 placeholder="e.g., Trigonometry, Calculus, Algebra, Vectors..."
             )
             level_input = gr.Radio(
-                choices=['SL', 'AHL'],
-                label="Select Level",
-                value='AHL'
+                choices=['SL', 'AHL'], label="Select Level", value='AHL'
             )
             generate_button = gr.Button("Generate Test", variant="primary")
 
@@ -221,4 +221,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()
+# --- Provided Topics List ---
+
+