Update app.py

app.py

@@ -5,31 +5,31 @@ import gradio as gr
 
 client = OpenAI()
 
-def generate_questions(category):
-  if category == "":
-    category = "general knowledge"
-  print(category)
-  response = client.chat.completions.create(
-    model="gpt-4o-mini",
-    messages=[
-      {
-        "role": "user",
-        "content": [
-          {
-            "type": "text",
-            "text": "Break the category \""+category+"\" into 6 subcategories, and for each subcategory create 5 True/False questions ranging from a question a Beginner would know to a question only an Expert would know. There should be as many True as False, and the structure of the questions should not make it obvious which is the answer. The harder questions should be tricky and trick non-experts into saying the wrong thing. Provide the correct answers and a field with a 1 sentence explanation. This will total out to 30 questions. Output just a JSON, nothing else. Here's an example JSON output for \"nutrition\":\n\n```json\n{\n  \"Macronutrients\": [\n    {\n      \"question\": \"Protein is one of the three primary macronutrients.\",\n      \"answer\": True,\n      \"explanation\": \"Protein is one of the three primary macronutrients, along with carbohydrates and fats.\"\n    },\n    {\n      \"question\": \"Carbohydrates are the body's main source of energy.\",\n      \"answer\": True,\n      \"explanation\": \"Carbohydrates are typically the body's preferred energy source.\"\n    },\n    {\n      \"question\": \"Fats have the same caloric content per gram as carbohydrates.\",\n      \"answer\": False,\n      \"explanation\": \"Fats have 9 calories per gram, while carbohydrates have 4 calories per gram.\"\n    },\n    {\n      \"question\": \"All proteins are equally effective for muscle growth.\",\n      \"answer\": False,\n      \"explanation\": \"Different proteins have varying amino acid profiles and bioavailability, affecting their effectiveness.\"\n    },\n    {\n      \"question\": \"Omega-3 fatty acids are a type of fat that can reduce inflammation.\",\n      \"answer\": True,\n      \"explanation\": \"Omega-3 fatty acids, found in foods like fish, are known to have anti-inflammatory properties.\"\n    }\n  ],\n  \"Micronutrients\": [\n    { ..."
-          }
-        ]
-      }
-    ],
-    response_format={ "type": "json_object" },
-    temperature=1,
-    max_tokens=4071,
-    top_p=1,
-    frequency_penalty=0,
-    presence_penalty=0
-  )
-  return json.loads(response.choices[0].message.content)
+def generate_questions(category, num_categories, num_questions):
+    if category == "":
+        category = "general knowledge"
+    print(category)
+    response = client.chat.completions.create(
+        model="gpt-4o-mini",
+        messages=[
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "text",
+                        "text": f"Break the category \"{category}\" into {num_categories} subcategories, and for each subcategory create {num_questions} True/False questions ranging from a question a Beginner would know to a question only an Expert would know. There should be as many True as False, and the structure of the questions should not make it obvious which is the answer. The harder questions should be tricky and trick non-experts into saying the wrong thing. Provide the correct answers and a field with a 1 sentence explanation. This will total out to {num_categories * num_questions} questions. Output just a JSON, nothing else. Below is an example JSON output for \"nutrition\" where 6 categories of 5 questions were requested, but remember, for you, there should be a total of {num_categories * num_questions} questions with {num_categories} categories and {num_questions} questions per category.\n\n```json\n{{\n  \"Macronutrients\": [\n    {{\n      \"question\": \"Protein is one of the three primary macronutrients.\",\n      \"answer\": True,\n      \"explanation\": \"Protein is one of the three primary macronutrients, along with carbohydrates and fats.\"\n    }},\n    {{\n      \"question\": \"Carbohydrates are the body's main source of energy.\",\n      \"answer\": True,\n      \"explanation\": \"Carbohydrates are typically the body's preferred energy source.\"\n    }},\n    {{\n      \"question\": \"Fats have the same caloric content per gram as carbohydrates.\",\n      \"answer\": False,\n      \"explanation\": \"Fats have 9 calories per gram, while carbohydrates have 4 calories per gram.\"\n    }},\n    {{\n      \"question\": \"All proteins are equally effective for muscle growth.\",\n      \"answer\": False,\n      \"explanation\": \"Different proteins have varying amino acid profiles and bioavailability, affecting their effectiveness.\"\n    }},\n    {{\n      \"question\": \"Omega-3 fatty acids are a type of fat that can reduce inflammation.\",\n      \"answer\": True,\n      \"explanation\": \"Omega-3 fatty acids, found in foods like fish, are known to have anti-inflammatory properties.\"\n    }}\n  ],\n  \"Micronutrients\": [\n    {{ ...\"}}"
+                    }
+                ]
+            }
+        ],
+        response_format={"type": "json_object"},
+        temperature=1,
+        max_tokens=4071,
+        top_p=1,
+        frequency_penalty=0,
+        presence_penalty=0
+    )
+    return json.loads(response.choices[0].message.content)
 
 # Function to calculate Metaknowledge score
 def calculate_meta_cog_score(df):
@@ -151,10 +151,8 @@ def analyze_results(df, overall_scores, subcategory_scores):
 
     return f"## Analysis of Results\n\n{analysis}\n\n## Detailed Questions and Answers\n\n{question_details}"
 
-
-
 # Modify the submit_answer function to include analysis
-def submit_answer(category, questions, index, user_answer, confidence, user_answers):
+def submit_answer(category, num_categories, num_questions, questions, index, user_answer, confidence, user_answers):
     question_data = questions[index]
     subcategory = question_data["subcategory"]
 
@@ -181,27 +179,31 @@ def submit_answer(category, questions, index, user_answer, confidence, user_answ
         return "", index, gr.update(visible=False), user_answers, results_df, gr.update(visible=False), gr.update(visible=False), gr.update(value=analysis, visible=True)
     else:
         question_text, _, _, visible = display_current_question(questions, index)
-        return question_text, index, gr.update(visible=True), user_answers, gr.update(visible=False), gr.update(visible=True), gr.update(visible=True), gr.update(visible=False)
+        return question_text, index, gr.update(visible=True), user_answers, gr.update(visible=False), gr.update(visible=True, value=0.5), gr.update(visible=True, value=None), gr.update(visible=False)
 
 # Gradio UI setup
 with gr.Blocks(theme="soft") as app:
-    gr.Markdown("""## Metaknowledge Test
-    See how well you know what you know. Enter a category and a 30-question quiz will be generated. Answer the True/False questions about it, and scores will then be calculated on your knowledge of the category, and on your ***knowledge of your knowledge*** of the category.""")
-    category_input = gr.Textbox(label="Category", placeholder="general knowledge")
+    gr.Markdown("""## Metaknowledge Quiz
+    See how well you know what you know. Enter a category and a quiz will be generated. Answer the True/False questions about it, and scores will then be calculated on your knowledge of the category, and on your ***knowledge of your knowledge*** of the category.""")
+    with gr.Row():
+      category_input = gr.Textbox(label="Category", placeholder="general knowledge", scale=4)
+      num_categories_input = gr.Number(label="Number of Categories", value=6, scale=1)
+      num_questions_input = gr.Number(label="Questions per Category", value=5, scale=1)
+      total_questions_display = gr.Number(label="Total Questions in Quiz", value=30, scale=1)
     submit_category = gr.Button("Generate Quiz")
     question_area = gr.Markdown(visible=False)
     answer_area = gr.Radio(["True", "False"], label="Your Answer", visible=False)
     confidence_slider = gr.Slider(0, 1, label="Confidence Level", value=0.5, visible=False)
     submit_answer_btn = gr.Button("Submit Answer", visible=False)
     result_area = gr.DataFrame(label="Results", visible=False)
-    loading_text = gr.Textbox(label="Generating Test...", visible=False)
+    loading_text = gr.Textbox(label="Generating Quiz...", visible=False)
     analysis_area = gr.Markdown(visible=False)  # Add this line for analysis area
     questions_state = gr.State()
     index_state = gr.State(0)
     user_answers_state = gr.State([])
 
-    def on_generate_quiz(category):
-        questions_data = generate_questions(category)
+    def on_generate_quiz(category, num_categories, num_questions):
+        questions_data = generate_questions(category, num_categories, num_questions)
 
         questions = []
         for subcategory, qs in questions_data.items():
@@ -211,6 +213,7 @@ with gr.Blocks(theme="soft") as app:
 
         import random
         random.shuffle(questions)
+        print(len(questions))
 
         index = 0
         question_text, _, _, visible = display_current_question(questions, index)
@@ -220,8 +223,8 @@ with gr.Blocks(theme="soft") as app:
             index,
             [],
             gr.update(visible=visible),
-            gr.update(visible=True),
-            gr.update(visible=True),
+            gr.update(visible=True, value=None),
+            gr.update(visible=True, value=0.5),
             gr.update(visible=True),
             gr.update(visible=False),
             gr.update(visible=False),
@@ -234,16 +237,36 @@ with gr.Blocks(theme="soft") as app:
     def display_loading():
         return gr.update(visible=True)
 
+    def update_total_questions(num_categories, num_questions):
+        total_questions = num_categories * num_questions
+        return gr.update(value=total_questions)
+
     def display_results(index, questions):
         if index >= len(questions):
-          return gr.update(visible=True)
+            return gr.update(visible=True)
+
+    num_categories_input.change(update_total_questions, inputs=[num_categories_input, num_questions_input], outputs=[total_questions_display])
+    num_questions_input.change(update_total_questions, inputs=[num_categories_input, num_questions_input], outputs=[total_questions_display])
 
     submit_category.click(remove_button, inputs=[], outputs=[submit_category])
     submit_category.click(display_loading, inputs=[], outputs=[loading_text])
 
+    def make_uninteractive():
+        return (
+            gr.update(interactive=False),
+            gr.update(interactive=False),
+            gr.update(interactive=False)
+        )
+
+    submit_category.click(
+        make_uninteractive,
+        inputs=[],
+        outputs=[category_input, num_categories_input, num_questions_input]
+    )
+
     submit_category.click(
         on_generate_quiz,
-        inputs=[category_input],
+        inputs=[category_input, num_categories_input, num_questions_input],
         outputs=[
             question_area,
             questions_state,
@@ -262,11 +285,11 @@ with gr.Blocks(theme="soft") as app:
 
     submit_answer_btn.click(
         submit_answer,
-        inputs=[category_input, questions_state, index_state, answer_area, confidence_slider, user_answers_state],
+        inputs=[category_input, num_categories_input, num_questions_input, questions_state, index_state, answer_area, confidence_slider, user_answers_state],
         outputs=[question_area, index_state, submit_answer_btn, user_answers_state, result_area, confidence_slider, answer_area, analysis_area]  # Add analysis_area here
     )
 
     submit_answer_btn.click(display_results, inputs=[index_state, questions_state], outputs=[result_area])
-
+    
 # Launch the app
 app.launch(share=False)