app.py

import gradio as gr
from gradio.components import Dataframe
from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns
import pandas as pd
from apscheduler.schedulers.background import BackgroundScheduler
from huggingface_hub import snapshot_download
import os
from gradio.themes import Soft

from src.about import (
    CITATION_BUTTON_LABEL,
    CITATION_BUTTON_TEXT,
    EVALUATION_QUEUE_TEXT,
    INTRODUCTION_TEXT,
    LLM_BENCHMARKS_TEXT,
    TITLE,
    Tasks
)
from src.display.css_html_js import custom_css
from src.display.utils import (
    BENCHMARK_COLS,
    COLS,
    EVAL_COLS,
    EVAL_TYPES,
    AutoEvalColumn,
    auto_eval_column_attrs,
    Language,
)
from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN, LOCAL_MODE
from src.populate import get_evaluation_queue_df, get_leaderboard_df
from src.submission.submit import add_new_eval


def restart_space():
    """Restart the Hugging Face space"""
    if LOCAL_MODE:
        print("Running in local mode, skipping space restart")
        return
        
    try:
        API.restart_space(repo_id=REPO_ID)
    except Exception as e:
        print(f"Failed to restart space: {e}")
        print("Continuing without restart")

### Space initialisation
def initialize_data_directories():
    """Initialize directories for assessment data"""
    # Create local directories if they don't exist
    os.makedirs(EVAL_REQUESTS_PATH, exist_ok=True)
    os.makedirs(EVAL_RESULTS_PATH, exist_ok=True)
    
    if LOCAL_MODE:
        print("Running in local mode, using local directories only")
        return
    
    # Try to download from HF if not in local mode
    try:
        print(f"Downloading request data from {QUEUE_REPO} to {EVAL_REQUESTS_PATH}")
        snapshot_download(
            repo_id=QUEUE_REPO, local_dir=EVAL_REQUESTS_PATH, repo_type="dataset", 
            tqdm_class=None, etag_timeout=30, token=TOKEN
        )
    except Exception as e:
        print(f"Failed to download request data: {e}")
        print("Using local data only")
    
    try:
        print(f"Downloading result data from {RESULTS_REPO} to {EVAL_RESULTS_PATH}")
        snapshot_download(
            repo_id=RESULTS_REPO, local_dir=EVAL_RESULTS_PATH, repo_type="dataset", 
            tqdm_class=None, etag_timeout=30, token=TOKEN
        )
    except Exception as e:
        print(f"Failed to download result data: {e}")
        print("Using local data only")

# Initialize data
initialize_data_directories()

# Load data for leaderboard
LEADERBOARD_DF = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS)

# Extract unique languages for filtering
def get_unique_languages(df):
    """Extract all unique individual languages from the Language column"""
    if df.empty or auto_eval_column_attrs.language.name not in df.columns:
        return []
    
    all_languages = set()
    for value in df[auto_eval_column_attrs.language.name].unique():
        if isinstance(value, str):
            if "/" in value:
                languages = [lang.strip() for lang in value.split("/")]
                all_languages.update(languages)
            else:
                all_languages.add(value.strip())
    
    return sorted(list(all_languages))

# Create a mapping for language filtering
UNIQUE_LANGUAGES = get_unique_languages(LEADERBOARD_DF)

# Create a special column for individual language filtering
if not LEADERBOARD_DF.empty:
    # Create a column that contains all individual languages as a list
    LEADERBOARD_DF["_languages_list"] = LEADERBOARD_DF[auto_eval_column_attrs.language.name].apply(
        lambda x: [lang.strip() for lang in str(x).split("/")] if pd.notna(x) else []
    )
    
    # Create a text version of Active Maintenance for checkboxgroup filtering
    LEADERBOARD_DF["_maintenance_filter"] = LEADERBOARD_DF[auto_eval_column_attrs.availability.name].apply(
        lambda x: "Active" if x else "Inactive"
    )

# Load queue data
(
    finished_eval_queue_df,
    running_eval_queue_df,
    pending_eval_queue_df,
    rejected_eval_queue_df,
) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS)

def init_leaderboard(dataframe):
    """Initialize the leaderboard component"""
    if dataframe is None or dataframe.empty:
        # Create an empty dataframe with the expected columns
        all_columns = COLS + [task.value.col_name for task in Tasks]
        empty_df = pd.DataFrame(columns=pd.Index(all_columns))
        print("Warning: Leaderboard DataFrame is empty. Using empty dataframe.")
        dataframe = empty_df
        
    # Create filter columns list with proper typing
    filter_columns = []
    
    # 1. Library types
    filter_columns.append(ColumnFilter(auto_eval_column_attrs.library_type.name, type="checkboxgroup", label="Library types"))
    
    # 2. Programming Language (checkboxgroup - OR filtering)
    filter_columns.append(ColumnFilter(auto_eval_column_attrs.language.name, type="checkboxgroup", label="Programming Language"))
    
    # 3. GitHub Stars
    filter_columns.append(ColumnFilter(
        auto_eval_column_attrs.stars.name,
        type="slider",
        min=0,
        max=50000,
        label="GitHub Stars",
    ))
    
    # 4. Maintenance Status (checkboxgroup - separate from languages)
    filter_columns.append(ColumnFilter("_maintenance_filter", type="checkboxgroup", label="Maintenance Status"))
    
    # Hide columns
    hidden_columns = [getattr(auto_eval_column_attrs, field).name for field in AutoEvalColumn.model_fields if getattr(auto_eval_column_attrs, field).hidden]
    hidden_columns.extend(["_languages_list", "_maintenance_filter", "_original_language"])  # Hide helper columns
    
    return Leaderboard(
        value=dataframe,
        datatype="markdown",
        select_columns=SelectColumns(
            default_selection=[getattr(auto_eval_column_attrs, field).name for field in AutoEvalColumn.model_fields if getattr(auto_eval_column_attrs, field).displayed_by_default],
            cant_deselect=[getattr(auto_eval_column_attrs, field).name for field in AutoEvalColumn.model_fields if getattr(auto_eval_column_attrs, field).never_hidden],
            label="Select Columns to Display:",
        ),
        search_columns=[auto_eval_column_attrs.library.name, auto_eval_column_attrs.license_name.name],
        hide_columns=hidden_columns,
        filter_columns=filter_columns, # type: ignore
        bool_checkboxgroup_label="Filter libraries",
        interactive=False,
    )


demo = gr.Blocks(css=custom_css, theme=Soft())
# demo = gr.Blocks(css=custom_css, theme=Soft(font=["sans-serif"], font_mono=["monospace"]))
with demo:
    gr.HTML(TITLE)
    gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")

    with gr.Tabs(elem_classes="tab-buttons") as tabs:
        with gr.TabItem("🏅 Vulnerability Leaderboard", elem_id="vulnerability-leaderboard-tab", id=0):
            leaderboard = init_leaderboard(LEADERBOARD_DF)

        with gr.TabItem("📝 About", elem_id="about-tab", id=2):
            gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")

        with gr.TabItem("🚀 Submit Library", elem_id="submit-library-tab", id=3):
            with gr.Column():
                with gr.Row():
                    gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")

                with gr.Column():
                    with gr.Accordion(
                        f"✅ Completed Assessments ({len(finished_eval_queue_df)})",
                        open=False,
                    ):
                        with gr.Row():
                            finished_eval_table = Dataframe(
                                value=finished_eval_queue_df,
                                headers=EVAL_COLS,
                                datatype=EVAL_TYPES,
                                row_count=5,
                            )
                    with gr.Accordion(
                        f"🔄 In Progress Assessments ({len(running_eval_queue_df)})",
                        open=False,
                    ):
                        with gr.Row():
                            running_eval_table = Dataframe(
                                value=running_eval_queue_df,
                                headers=EVAL_COLS,
                                datatype=EVAL_TYPES,
                                row_count=5,
                            )

                    with gr.Accordion(
                        f"⏳ Pending Assessment Queue ({len(pending_eval_queue_df)})",
                        open=False,
                    ):
                        with gr.Row():
                            pending_eval_table = Dataframe(
                                value=pending_eval_queue_df,
                                headers=EVAL_COLS,
                                datatype=EVAL_TYPES,
                                row_count=5,
                            )
            with gr.Row():
                gr.Markdown("# ✉️✨ Submit a library for vulnerability assessment", elem_classes="markdown-text")

            with gr.Row():
                with gr.Column():
                    library_name_textbox = gr.Textbox(label="Library name")

            submit_button = gr.Button("Submit for Assessment")
            submission_result = gr.Markdown()
            submit_button.click(
                add_new_eval,
                [
                    library_name_textbox,
                ],
                submission_result,
            )

    with gr.Row():
        with gr.Accordion("📙 Citation", open=True):
            citation_button = gr.Code(
                value=CITATION_BUTTON_TEXT,
                label=CITATION_BUTTON_LABEL,
                lines=14,
                elem_id="citation-button",
                language="yaml",
            )

# Only schedule space restarts if not in local mode
if not LOCAL_MODE:
    scheduler = BackgroundScheduler()
    scheduler.add_job(restart_space, "interval", seconds=1800)
    scheduler.start()

# Launch the app
demo.queue(default_concurrency_limit=40).launch(show_error=True)