form implementation

app.py

@@ -1,11 +1,12 @@
 import streamlit as st
 import pandas as pd
-import os
 from huggingface_hub import HfApi, hf_hub_download
 
+# Constants
 HF_REPO = "giobin/MAIA_human_assessment_annotations"
 CSV_FILENAME = "user_selections.csv"
 
+# Function to assign samples to users
 def assign_samples(csv_path):
     df = pd.read_csv(csv_path)
     group_1 = df[(df["pool_pos"] == 1) & (~df["question_category"].str.endswith("_B"))].head(10)
@@ -22,6 +23,7 @@ def assign_samples(csv_path):
         "Raffaella": group_3,
     }
 
+# Function to load existing annotations from Hugging Face Hub
 def load_existing_annotations():
     try:
         file_path = hf_hub_download(HF_REPO, CSV_FILENAME, repo_type="dataset", token=st.secrets["HF_TOKEN"])
@@ -29,42 +31,44 @@ def load_existing_annotations():
     except Exception:
         return pd.DataFrame(columns=["username", "id"])
 
+# Load datasets
 csv_file = "static/mc.csv"
 assignments = assign_samples(csv_file)
 existing_annotations = load_existing_annotations()
 
+# Valid users
 valid_users = list(assignments.keys())
 
+# Initialize session state variables
 if "username" not in st.session_state:
     st.session_state.username = None
 if "index" not in st.session_state:
     st.session_state.index = 0
 if "results" not in st.session_state:
     st.session_state.results = []
-if "selected_answer" not in st.session_state:
-    st.session_state.selected_answer = None
-if "not_enough_info" not in st.session_state:
-    st.session_state.not_enough_info = False
-
-def update_name():
-    st.session_state.username = st.session_state.selected_user
-    st.session_state.index = 0
 
+# User selection form
 if st.session_state.username is None:
     with st.form("user_form"):
         st.write("### Select Your Name")
         selected_user = st.selectbox("Choose your name:", valid_users, key="selected_user")
-        submit_button = st.form_submit_button("Start", on_click=update_name)
+        submit_button = st.form_submit_button("Start")
+        if submit_button:
+            st.session_state.username = selected_user
+            st.session_state.index = 0
     st.stop()
 
+# Retrieve assigned dataset and filter out already labeled samples
 full_dataset = assignments[st.session_state.username].reset_index(drop=True)
 user_labeled_ids = existing_annotations[existing_annotations["username"] == st.session_state.username]["id"].tolist()
 dataset = full_dataset[~full_dataset["id"].isin(user_labeled_ids)].reset_index(drop=True)
 
-if dataset.empty:
+# Check if all samples are labeled
+if st.session_state.index >= len(dataset):
     st.write("### Great! You have completed your assignment. 🎉")
     st.stop()
 
+# Function to push updated annotations to Hugging Face Hub
 def push_to_hf_hub(csv_path):
     api = HfApi()
     try:
@@ -74,11 +78,8 @@ def push_to_hf_hub(csv_path):
     except Exception as e:
         print(f"Error pushing to HF: {e}")
 
-def save_choice():
-    if st.session_state.selected_answer is None:
-        st.warning("Please select an answer before proceeding.")
-        return
-    
+# Function to save user choice
+def save_choice(selected_answer, not_enough_info):
     sample = dataset.iloc[st.session_state.index]
     st.session_state.results.append({
         "username": st.session_state.username,
@@ -86,59 +87,55 @@ def save_choice():
         "video_id": sample["video_id"],
         "answer1": sample["answer1"],
         "answer2": sample["answer2"],
-        "selected_answer": st.session_state.selected_answer,
+        "selected_answer": selected_answer,
         "target": sample["target"],
-        "not_enough_info": st.session_state.not_enough_info
+        "not_enough_info": not_enough_info
     })
-    
     st.session_state.index += 1
-    st.session_state.selected_answer = None
-    st.session_state.not_enough_info = False
-    
+
+    # Save results and push to Hugging Face Hub if all samples are labeled
     if st.session_state.index >= len(dataset):
         st.write("### Great! You have completed your assignment. 🎉")
         result_df = pd.DataFrame(st.session_state.results)
         csv_path = "user_selections.csv"
-        
         if not existing_annotations.empty:
             result_df = pd.concat([existing_annotations, result_df]).drop_duplicates(subset=["username", "id"], keep="last")
-        
         result_df.to_csv(csv_path, index=False)
         push_to_hf_hub(csv_path)
         st.stop()
-    return
 
+# Display current sample
 sample = dataset.iloc[st.session_state.index]
 
+# Page title and user information
 st.markdown("<h1 style='text-align: center; font-size: 50px;'>MAIA Sample</h1>", unsafe_allow_html=True)
 st.markdown(f"<h3 style='text-align: center;'>User: {st.session_state.username}</h3>", unsafe_allow_html=True)
 st.write("\n\n")
 
+# Instructions
 st.markdown("""
 ### Instructions:
-- Look at the video thumbnail, do not play it!
-- Select one of the two answers.
-- If you think the frame does not provide enough information, select the checkbox.
+- Look at the video thumbnail; do not play it.
+- Select the correct description (A or B).
+- If the frame does not provide enough information to answer the question, select the checkbox.
 - Click 'Next' to proceed.
 """)
 st.write("---")
 
+# Display video thumbnail
 st.video(sample["video_url"])
 
-st.markdown("<h4 style='text-align: center; margin-top: 20px;'>Scegli la descrizione corretta tra A e B</h4>", unsafe_allow_html=True)
-
-col1, col2 = st.columns(2)
-with col1:
-    if st.button(f"A: {sample.get('answer1', 'No answer1 available')}"):
-        st.session_state.selected_answer = 0
-with col2:
-    if st.button(f"B: {sample.get('answer2', 'No answer2 available')}"):
-        st.session_state.selected_answer = 1
-
-st.markdown("<div style='text-align: center;'>", unsafe_allow_html=True)
-st.session_state.not_enough_info = st.checkbox("The frame does not provide enough information to answer the question.")
-st.markdown("</div>", unsafe_allow_html=True)
-st.write("\n")
-
-if st.button("Next"):
-    save_choice()
+# Form for user input
+with st.form("annotation_form"):
+    # Exclusive choice between A and B
+    selected_answer = st.radio(
+        "Choose the correct description:",
+        options=[0, 1],
+        format_func=lambda x: f"A: {sample['answer1']}" if x == 0 else f"B: {sample['answer2']}",
+        key="selected_answer"
+    )
+
+    # Independent checkbox for insufficient information
+    not_enough_info = st.checkbox("The frame does not provide enough information to answer the question.", key="not_enough
+::contentReference[oaicite:12]{index=12}
+