app.py

# Venus Annotation System
# Author: Du Mingzhe ([email protected])
# Date: 2024-09-25

import uuid
import streamlit as st
import streamlit_ext as ste
from code_editor import code_editor
from datasets import load_dataset, Dataset


st.title(":blue[Venus] Annotation System 🪐")

# Step 1: Load the problem set
my_bar = st.progress(0, text="Loading the problem set...")

my_bar.progress(10, text="Loading [Elfsong/Venus] datasets...")
if "raw_ds" not in st.session_state.keys():
    st.session_state["raw_ds"] = load_dataset("Elfsong/Venus", "python3")
raw_ds = st.session_state["raw_ds"]

my_bar.progress(55, text="Loading [Elfsong/venus_case] datasets...")
if "case_ds" not in st.session_state.keys():
    st.session_state["case_ds"] = load_dataset("Elfsong/venus_case", "python3")
case_ds = st.session_state["case_ds"]

my_bar.progress(90, text="Filtering out the cases that already exist...")
if "candidates" not in st.session_state.keys():
    case_ds_ids = set(case_ds['train']['question_id'])
    candidates = [raw_ds['train'][i] for i in range(len(raw_ds['train'])) if raw_ds['train'][i]['question_id'] not in case_ds_ids]
    st.session_state["candidates"] = candidates
candidates = st.session_state["candidates"]

my_bar.progress(100, text="System Initialized Successfully 🚀")
    
# Step 2: Select the problem
candidates_dict = {}
for candidate in candidates:
    candidate_name = str(candidate['question_id']) + '.' + str(candidate['name']) + ' [' + str(candidate['difficulty']).upper() + ']'
    candidates_dict[candidate_name] = candidate
option = ste.selectbox("Select a problem here", candidates_dict.keys())
example = candidates_dict[option]

tab1, tab2, tab3 = st.tabs(["Problem Description", "Canonical Solution", "Test Cases Generator"])

with tab1:
    st.html(example['content'])
with tab2:
    solutions_displayed = 0
    for solution in example['rt_list']:
        if "Solution" in solution['code']:
            st.write(f"Canonical Solution {solutions_displayed + 1}")
            st.code(solution['code'])
            solutions_displayed += 1
        if solutions_displayed >= 3:
            break
with tab3:
    editor_buttons = [{
        "name": "Submit", 
        "feather": "Play",
        "primary": True, 
        "hasText": True, 
        "showWithIcon": True, 
        "commands": ["submit"], 
        "style": {"bottom": "0.44rem","right": "0.4rem"}
    }]
    predefined_code = "def generate_test_cases():\n\tpass\n\ndef serialize_input():\n\tpass\n\ndef deserialize_input():\n\tpass\n\ndef serialize_output():\n\tpass\n\ndef deserialize_output():\n\tpass"
    response_dict = code_editor(predefined_code, lang="python", height=20, options={"wrap": False}, buttons=editor_buttons)
    if response_dict['type'] == 'submit':
        new_ds = Dataset.from_list([{
            "question_id": example['question_id'],
            "test_case_functions": response_dict['text'],
        }])
        
        ds_name = str(uuid.uuid1())
        new_ds.push_to_hub(f"Elfsong/Venus_Anotation", f'python3-{ds_name}')
        st.write("Thanks for your contribution! 🌟")