app.py

import gradio as gr
import json
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots
import os
import traceback
from datetime import datetime
from packaging import version

# Color scheme for charts
COLORS = px.colors.qualitative.Plotly

# Line colors for radar charts
line_colors = [
    "#EE4266",
    "#00a6ed",
    "#ECA72C",
    "#B42318",
    "#3CBBB1",
]

# Fill colors for radar charts
fill_colors = [
    "rgba(238,66,102,0.05)",
    "rgba(0,166,237,0.05)",
    "rgba(236,167,44,0.05)",
    "rgba(180,35,24,0.05)",
    "rgba(60,187,177,0.05)",
]

# Language definitions
LANGUAGES = {"English": {
        "clear_charts": "Clear Charts",
        "lang_selector_label": "Language / Язык",
        "description": "This leaderboard allows comparing RAG systems based on generative and retrieval metrics across different question types (simple, comparison, multi-hop, conditional, etc.). <li>Questions are automatically generated from news sources.</li><li>The question dataset is updated regularly, and metrics for open models are recalculated.</li><li>User submissions use the latest calculated metrics for them.</li><li>To recalculate a previously submitted configuration with the latest data version, use the submit_id received during the initial submission via the client (see instructions below).</li>",
        "version_info_template": "## Version {} → {} questions, generated from news sources → {}",
        "gen_metrics_title": "### Generation Metrics",
        "ret_metrics_title": "### Retrieval Metrics",
        "overall_tab_title": "Overall Table",
        "no_data_message": "No data available. Please submit some results.",
        "by_type_tab_title": "By Question Type",
        "category_display_names": {
            "simple": "Simple Questions",
            "set": "Set-based",
            "mh": "Multi-hop",
            "cond": "Conditional",
            "comp": "Comparison"
        },
        "no_data_category_template": "No data available for {} category.",
        "category_performance_template": "#### Performance on {}",
        "citation_title": "### Citation",
        "citation_description": """
```
@misc{chernogorskii2025dragondynamicragbenchmark,
      title={DRAGON: Dynamic RAG Benchmark On News}, 
      author={Fedor Chernogorskii and Sergei Averkiev and Liliya Kudraleeva and Zaven Martirosian and Maria Tikhonova and Valentin Malykh and Alena Fenogenova},
      year={2025},
      eprint={2507.05713},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2507.05713}, 
}
```

""",
        "version_selector_title": "### Version Selection",
        "only_actual_label": "Only actual versions",
        "only_actual_info": "Start counting from the current dataset version",
        "n_versions_label": "Take n last versions",
        "n_versions_info": "Number of versions to calculate metrics for",
        "filter_button": "Apply Filter",
        "info_text": "Click on models in the table to add them to the charts",
        "footer_text": "<footer>DRAGON. Dynamic RAG Benchmark Leaderboard</footer>",
        "radar_gen_title": "Performance on Generation Tasks",
        "radar_ret_title": "Performance on Retrieval Tasks"
    },
    "Русский": {
        "clear_charts": "Очистить графики",
        # "lang_selector_label": "Language",
        "description": "На этом лидерборде можно сравнить RAG системы в разрезе генеративных и поисковых метрик моделей по вопросам разного типа (простые вопросы, сравнения, multi-hop, условные и др.). <li>Вопросы автоматичеки генерируются на основе новостных источников.</li><li>Обновление датасета с вопросами происходит регулярно, при этом пересчитываются все метрики для открытых моделей.</li><li>Для пользовательских сабмитов учитываются последние посчитанные для них метрики.</li><li>Чтобы посчитать ранее отправленную конфигурацию на последней версии данных, используйте submit_id, полученный при первой отправке через клиент (см. инструкцию ниже).</li>",
        "version_info_template": "## Версия {} → {} вопросов, сгенерированных по новостным источникам → {}",
        "gen_metrics_title": "### Генеративные метрики",
        "ret_metrics_title": "### Метрики поиска",
        "overall_tab_title": "Общая таблица",
        "no_data_message": "Нет данных. Пожалуйста, отправьте результаты.",
        "by_type_tab_title": "По типам вопросов",
        "category_display_names": {
            "simple": "Simple",
            "set": "Set",
            "mh": "Multi-hop",
            "cond": "Conditional",
            "comp": "Comparison"
        },
        "no_data_category_template": "Нет данных для категории {}.",
        "category_performance_template": "#### Производительность на {}",
        "citation_title": "### Цитирование",
        "citation_description": """
```
@article{dynamic-rag-benchmark,
  title={Dynamic RAG Benchmark},
  author={RAG Benchmark Team},
  journal={arXiv preprint},
  year={2025},
  url={https://github.com/rag-benchmark}
}
```

Шаблон для цитирования нашего бенча.
""",
        "version_selector_title": "### Выбор версий",
        "only_actual_label": "Только актуальные версии",
        "only_actual_info": "Считать, начиная с актуальной версии датасета",
        "n_versions_label": "Взять n последних версий",
        "n_versions_info": "Количество версий для подсчета метрик",
        "filter_button": "Применить фильтр",
        "info_text": "Кликайте на модели в таблице, чтобы добавить их в графики",
        "footer_text": "<footer>DRAGON. Dynamic RAG Benchmark Leaderboard</footer>",
        "radar_gen_title": "Производительность на Генеративных Заданиях",
        "radar_ret_title": "Производительность на Поисковых Заданиях"
    }
}
DEFAULT_LANG = "English"

# Define the question categories
QUESTION_CATEGORIES = ["simple", "set", "mh", "cond", "comp"]
METRIC_TYPES = ["retrieval", "generation"]

def load_results():
    """Load results from the results.json file."""
    try:
        # Get the directory of the current script
        script_dir = os.path.dirname(os.path.abspath(__file__))
        # Build the path to results.json
        results_path = os.path.join(script_dir, 'results.json')
        
        print(f"Loading results from: {results_path}")
        
        with open(results_path, 'r', encoding='utf-8') as f:
            results = json.load(f)
            print(f"Successfully loaded results with {len(results.get('items', {}))} version(s)")
            return results
    except FileNotFoundError:
        # Return empty structure if file doesn't exist
        print(f"Results file not found, creating empty structure")
        return {"items": {}, "last_version": "1.0", "n_questions": "0"}
    except Exception as e:
        print(f"Error loading results: {e}")
        print(traceback.format_exc())
        return {"items": {}, "last_version": "1.0", "n_questions": "0"}

def filter_and_process_results(results, n_versions, only_actual_versions):
    """Filter results by version and process them for display."""
    if not results or "items" not in results:
        return pd.DataFrame(), [], [], []

    all_items = results["items"]
    
    # Get all versions and sort them
    all_versions_sorted = sorted([version.parse(v_str) for v_str in all_items.keys()], reverse=True)
    
    # Filter versions to consider based on n_versions slider
    versions_to_consider = all_versions_sorted[:n_versions]
    versions_to_consider_str = {str(v) for v in versions_to_consider}

    rows = []
    for version_str, version_items in all_items.items():
        if version_str not in versions_to_consider_str:
            continue
            
        for guid, item in version_items.items():
            config = item.get("config", {})
            model_name = item.get("model_name", "N/A")
            metrics = item.get("metrics", {})
            judge_metrics = metrics.get("judge", {})
            
            row = {
                'Model': f"{model_name} ({guid[:6]})",
                'Embeddings': config.get('embedding_model', 'N/A'),
                'Top k': config.get('retrieval_config', {}).get('top_k', 'N/A'),
                # 'Judge': round(judge_metrics.get("judge_total_score", 0.0) / 2, 4),
                'Version': version_str,
                'Last Updated': item.get("timestamp", ""),
                'guid': guid
            }
            
            if row['Last Updated']:
                try:
                    dt = datetime.fromisoformat(row['Last Updated'].replace('Z', '+00:00'))
                    row['Last Updated'] = dt.strftime("%Y-%m-%d")
                except (ValueError, TypeError):
                    pass
            
            category_sums = {mtype: 0.0 for mtype in METRIC_TYPES}
            category_counts = {mtype: 0 for mtype in METRIC_TYPES}

            for category in QUESTION_CATEGORIES:
                if category in metrics:
                    for metric_type in METRIC_TYPES:
                        if metric_type in metrics[category]:
                            metric_values = metrics[category][metric_type]
                            if metric_values and len(metric_values) > 0:
                                avg_value = sum(metric_values.values()) / len(metric_values)
                                col_name = f"{category}_{metric_type}"
                                row[col_name] = round(avg_value, 4)
                                category_sums[metric_type] += avg_value
                                category_counts[metric_type] += 1
            
            for metric_type in METRIC_TYPES:
                if category_counts[metric_type] > 0:
                    avg = category_sums[metric_type] / category_counts[metric_type]
                    row[f"{metric_type}_avg"] = round(avg, 4)
            
            rows.append(row)

    df = pd.DataFrame(rows)
    
    # Get lists of metrics for each category
    category_metrics = []
    if not df.empty:
        for category in QUESTION_CATEGORIES:
            metrics_list = []
            for metric_type in METRIC_TYPES:
                col_name = f"{category}_{metric_type}"
                if col_name in df.columns:
                    metrics_list.append(col_name)
            if metrics_list:
                category_metrics.append((category, metrics_list))

    # Define retrieval and generation columns for radar charts
    retrieval_metrics = []
    generation_metrics = []
    if not df.empty:
        retrieval_metrics = [f"{category}_retrieval" for category, _ in category_metrics if f"{category}_retrieval" in df.columns]
        generation_metrics = [f"{category}_generation" for category, _ in category_metrics if f"{category}_generation" in df.columns]
    
    return df, retrieval_metrics, generation_metrics, category_metrics

def create_radar_chart(df, selected_models, metrics, title, name_col="Model"):
    """Create a radar chart for the selected models and metrics."""
    if not metrics or len(selected_models) == 0:
        # Return empty figure if no metrics or models selected
        fig = go.Figure()
        fig.update_layout(
            title=title,
            title_font_size=16,
            height=400,
            width=500,
            margin=dict(l=30, r=30, t=50, b=30)
        )
        return fig
    
    # Filter dataframe for selected models
    filtered_df = df[df['Model'].isin(selected_models)]
    
    if filtered_df.empty:
        # Return empty figure if no data
        fig = go.Figure()
        fig.update_layout(
            title=title,
            title_font_size=16,
            height=400,
            width=500,
            margin=dict(l=30, r=30, t=50, b=30)
        )
        return fig
    
    # Limit to top 5 models for better visualization (similar to inspiration file)
    if len(filtered_df) > 5:
        filtered_df = filtered_df.head(5)
    
    # Prepare data for radar chart
    categories = [m.split('_', 1)[0] for m in metrics]  # Get category name (simple, set, etc.)
    
    fig = go.Figure()
    
    # Process in reverse order to match inspiration file
    for i, (_, row) in enumerate(filtered_df.iterrows()):
        values = [row[m] for m in metrics]
        # Close the loop for radar chart
        values.append(values[0])
        categories_loop = categories + [categories[0]]
        
        fig.add_trace(go.Scatterpolar(
            name=row[name_col],
            r=values,
            theta=categories_loop,
            showlegend=True,
            mode="lines",
            line=dict(width=2, color=line_colors[i % len(line_colors)]),
            fill="toself",
            fillcolor=fill_colors[i % len(fill_colors)]
        ))
    
    fig.update_layout(
        font=dict(size=13, color="black"),
        template="plotly_white",
        polar=dict(
            radialaxis=dict(
                visible=True,
                gridcolor="black",
                linecolor="rgba(0,0,0,0)",
                gridwidth=1,
                showticklabels=False,
                ticks="",
                range=[0, 1]  # Ensure consistent range for scores
            ),
            angularaxis=dict(
                gridcolor="black", 
                gridwidth=1.5, 
                linecolor="rgba(0,0,0,0)"
            ),
        ),
        legend=dict(
            orientation="h",
            yanchor="bottom",
            y=-0.35,
            xanchor="center",
            x=0.4,
            itemwidth=30,
            font=dict(size=13),
            entrywidth=0.6,
            entrywidthmode="fraction",
        ),
        margin=dict(l=0, r=16, t=30, b=30),
        autosize=True,
    )
    
    return fig

def create_summary_df(df, retrieval_metrics, generation_metrics):
    """Create a summary dataframe with averaged metrics for display."""
    if df.empty:
        return pd.DataFrame()
    
    summary_df = df.copy()
    
    # Add retrieval average
    if retrieval_metrics:
        retrieval_avg = summary_df[retrieval_metrics].mean(axis=1).round(4)
        summary_df['Retrieval (avg)'] = retrieval_avg
    
    # Add generation average
    if generation_metrics:
        generation_avg = summary_df[generation_metrics].mean(axis=1).round(4)
        summary_df['Generation (avg)'] = generation_avg
    
    # Add total score if all three columns exist
    if 'Retrieval (avg)' in summary_df.columns and 'Generation (avg)' in summary_df.columns:
    # if 'Retrieval (avg)' in summary_df.columns and 'Generation (avg)' in summary_df.columns and 'Judge' in summary_df.columns:
        # summary_df['Total Score'] = summary_df[['Retrieval (avg)', 'Generation (avg)', 'Judge']].mean(axis=1).round(4)
        summary_df['Total Score'] = summary_df[['Retrieval (avg)', 'Generation (avg)']].mean(axis=1).round(4)
        summary_df = summary_df.sort_values('Total Score', ascending=False)
    
    # Select columns for display
    summary_cols = ['Model', 'Embeddings', 'Top k']
    # if 'Judge' in summary_df.columns:
    #     summary_cols.append('Judge')
    if 'Retrieval (avg)' in summary_df.columns:
        summary_cols.append('Retrieval (avg)')
    if 'Generation (avg)' in summary_df.columns:
        summary_cols.append('Generation (avg)')
    if 'Total Score' in summary_df.columns:
        summary_cols.append('Total Score')
    if 'Version' in summary_df.columns:
        summary_cols.append('Version')
    if 'Last Updated' in summary_df.columns:
        summary_cols.append('Last Updated')
    
    return summary_df[summary_cols]

def create_category_df(df, category, retrieval_col, generation_col):
    """Create a dataframe for a specific category with detailed metrics."""
    if df.empty or retrieval_col not in df.columns or generation_col not in df.columns:
        return pd.DataFrame()
    
    category_df = df.copy()
    
    # Calculate total score for this category
    category_df[f'Score'] = (category_df[retrieval_col] + category_df[generation_col]).round(4)
    
    # Sort by total score
    category_df = category_df.sort_values(f'Score', ascending=False)
    
    # Select columns for display
    category_cols = ['Model', 'Embeddings', retrieval_col, generation_col, f'Score']
    
    # Rename columns for display
    category_df = category_df[category_cols].rename(columns={
        retrieval_col: 'Retrieval',
        generation_col: 'Generation'
    })
    
    return category_df

# Load initial data
results = load_results()
last_version = results.get("last_version", "1.0")
n_questions = results.get("n_questions", "100")
date_title = results.get("date_title", "---")

# Initial data processing
df, retrieval_metrics, generation_metrics, category_metrics = filter_and_process_results(
    results, n_versions=1, only_actual_versions=True
)

# Pre-generate charts for initial display
default_models = df['Model'].head(5).tolist() if not df.empty else []
initial_gen_chart_title = LANGUAGES[DEFAULT_LANG]["radar_gen_title"]
initial_ret_chart_title = LANGUAGES[DEFAULT_LANG]["radar_ret_title"]
initial_gen_chart = create_radar_chart(df, default_models, generation_metrics, initial_gen_chart_title)
initial_ret_chart = create_radar_chart(df, default_models, retrieval_metrics, initial_ret_chart_title, name_col='Embeddings')

# Create summary dataframe
summary_df = create_summary_df(df, retrieval_metrics, generation_metrics)

with gr.Blocks(css="""
    .title-container { 
        text-align: center; 
        margin-bottom: 10px; 
    }
    .description-text {
        text-align: left;
        padding: 10px;
        margin-bottom: 0px;
    }
    .version-info {
        text-align: center;
        padding: 10px;
        background-color: #f0f0f0;
        border-radius: 8px;
        margin-bottom: 15px;
    }
    .version-selector {
        padding: 15px;
        border: 1px solid #ddd;
        border-radius: 8px;
        margin-bottom: 20px;
        background-color: #f9f9f9;
        height: 100%;
    }
    .citation-block {
        padding: 15px;
        border: 1px solid #ddd;
        border-radius: 8px;
        margin-bottom: 20px;
        background-color: #f9f9f9;
        font-family: monospace;
        font-size: 14px;
        overflow-x: auto;
        height: 100%;
    }
    .flex-row-container {
        display: flex;
        justify-content: space-between;
        gap: 20px;
        width: 100%;
    }
    .charts-container {
        display: flex;
        gap: 20px;
        margin-bottom: 20px;
    }
    .chart-box {
        flex: 1;
        border: 1px solid #eee;
        border-radius: 8px;
        padding: 10px;
        background-color: white;
        min-height: 550px; /* Increased height to accommodate legend at bottom */
    }
    .metrics-table {
        border: 1px solid #eee;
        border-radius: 8px;
        padding: 15px;
        background-color: white;
    }
    .info-text {
        font-size: 0.9em;
        font-style: italic; 
        color: #666;
        margin-top: 5px;
    }
    footer {
        text-align: center;
        margin-top: 30px;
        font-size: 0.9em;
        color: #666;
    }
    /* Style for selected rows */
    table tbody tr.selected {
        background-color: rgba(25, 118, 210, 0.1) !important;
        border-left: 3px solid #1976d2;
    }
    /* Add this class via JavaScript */
    .gr-table tbody tr.selected td:first-child {
        font-weight: bold;
        color: #1976d2;
    }
    .category-tab {
        padding: 10px;
    }
    .chart-title {
        font-size: 1.2em;
        font-weight: bold;
        margin-bottom: 10px;
        text-align: center;
    }
    .clear-charts-button {
        display: flex;
        justify-content: center;
        margin-top: 10px;
        margin-bottom: 20px;
    }
    .lang-selector {
        width: fit-content; /* Adjust width to content */
        margin-left: auto;  /* Push to the right */
        margin-right: 0;   /* Keep it flush right */
        margin-bottom: 15px; /* Keep bottom margin */
        padding: 10px;
        background-color: #f9f9f9;
        border-radius: 8px;
        border: none;
        padding: 0 !important;
    }
    .lang-selector .form {
        border: none !important;
    }
""") as demo:
    current_lang_dict = gr.State(LANGUAGES[DEFAULT_LANG])
    current_language = gr.State(DEFAULT_LANG)

    with gr.Row(elem_classes=["title-container"]):
        #title with emoji connected with dragon
        main_title_md = gr.Markdown("# 🐉 DRAGON. Dynamic RAG Benchmark On News")

    # Language Selector
    with gr.Row(elem_classes=["lang-selector"]):
        lang_selector = gr.Radio(
            list(LANGUAGES.keys()),
            label="",
            value=DEFAULT_LANG,
            interactive=True
        )

    # Description
    with gr.Row(elem_classes=["description-text"]):
        description_md = gr.Markdown(value=LANGUAGES[DEFAULT_LANG]["description"])

    # Version info
    with gr.Row(elem_classes=["version-info"]):
        version_info_md = gr.Markdown(
            value=LANGUAGES[DEFAULT_LANG]["version_info_template"].format(last_version, n_questions, date_title)
        )

    # Radar Charts
    with gr.Row(elem_classes=["charts-container"]):
        with gr.Column(elem_classes=["chart-box"]):
            gen_chart_title_md = gr.Markdown(
                value=LANGUAGES[DEFAULT_LANG]["gen_metrics_title"], elem_classes=["chart-title"]
            )
            generation_chart = gr.Plot(value=initial_gen_chart)

        with gr.Column(elem_classes=["chart-box"]):
            ret_chart_title_md = gr.Markdown(
                value=LANGUAGES[DEFAULT_LANG]["ret_metrics_title"], elem_classes=["chart-title"]
            )
            retrieval_chart = gr.Plot(value=initial_ret_chart)

    # Clear Charts Button
    with gr.Row(elem_classes=["clear-charts-button"]):
        clear_charts_btn = gr.Button(
            value=LANGUAGES[DEFAULT_LANG]["clear_charts"],
            variant="secondary"
        )

    # Metrics table with tabs
    with gr.Tabs(elem_classes=["metrics-table"]) as metrics_tabs:
        with gr.TabItem(label=LANGUAGES[DEFAULT_LANG]["overall_tab_title"]) as summary_tab:
            selected_models = gr.State(default_models)
            empty_data_md = gr.Markdown(
                value=LANGUAGES[DEFAULT_LANG]["no_data_message"],
                visible=df.empty # Initially visible only if df is empty
            )
            # Initialize metrics_table even if empty, but maybe hide it
            metrics_table = gr.DataFrame(
                value=summary_df if not df.empty else pd.DataFrame(),
                headers=summary_df.columns.tolist() if not df.empty else [],
                datatype=["str"] * (len(summary_df.columns) if not df.empty else 0),
                row_count=(min(10, len(summary_df)) if not summary_df.empty else 0),
                col_count=(len(summary_df.columns) if not summary_df.empty else 0),
                interactive=False,
                wrap=True,
                visible=not df.empty # Initially visible only if df is not empty
            )

        with gr.TabItem(label=LANGUAGES[DEFAULT_LANG]["by_type_tab_title"]) as category_main_tab:
            category_tabs = gr.Tabs()
            category_tables = {}
            category_tab_items = {} # Store TabItem components
            category_no_data_mds = {} # Store "no data" Markdowns
            category_title_mds = {} # Store category title Markdowns

            # Get initial display names
            initial_category_display_names = LANGUAGES[DEFAULT_LANG]["category_display_names"]

            with category_tabs:
                for category, _ in category_metrics:
                    display_name = initial_category_display_names.get(category, category.capitalize())
                    if f"{category}_retrieval" in df.columns and f"{category}_generation" in df.columns:
                        with gr.TabItem(label=display_name, elem_classes=["category-tab"]) as tab_item:
                            category_tab_items[category] = tab_item # Store the TabItem

                            # Create dataframe for this category
                            category_df = create_category_df(df, category, f"{category}_retrieval", f"{category}_generation")

                            category_no_data_mds[category] = gr.Markdown(
                                value=LANGUAGES[DEFAULT_LANG]["no_data_category_template"].format(display_name),
                                visible=category_df.empty
                            )
                            category_title_mds[category] = gr.Markdown(
                                value=LANGUAGES[DEFAULT_LANG]["category_performance_template"].format(display_name),
                                visible=not category_df.empty
                            )
                            category_tables[category] = gr.DataFrame(
                                value=category_df if not category_df.empty else pd.DataFrame(),
                                headers=category_df.columns.tolist() if not category_df.empty else [],
                                datatype=["str"] * (len(category_df.columns) if not category_df.empty else 0),
                                row_count=(min(10, len(category_df)) if not category_df.empty else 0),
                                col_count=(len(category_df.columns) if not category_df.empty else 0),
                                interactive=False,
                                wrap=True,
                                visible=not category_df.empty
                            )

    # Version selector and Citation block in a flex container
    with gr.Row():
        # Citation block (left side)
        with gr.Column(scale=1, elem_classes=["citation-block"]):
            citation_title_md = gr.Markdown(value=LANGUAGES[DEFAULT_LANG]["citation_title"])
            citation_desc_md = gr.Markdown(value=LANGUAGES[DEFAULT_LANG]["citation_description"])

        # Version selector (right side)
        with gr.Column(scale=1, elem_classes=["version-selector"]):
            version_selector_title_md = gr.Markdown(value=LANGUAGES[DEFAULT_LANG]["version_selector_title"])
            with gr.Column():
                with gr.Row():
                    with gr.Column(scale=3):
                        only_actual_versions = gr.Checkbox(
                            label=LANGUAGES[DEFAULT_LANG]["only_actual_label"],
                            value=True,
                            info=LANGUAGES[DEFAULT_LANG]["only_actual_info"]
                        )
                    with gr.Column(scale=5):
                        n_versions_slider = gr.Slider(
                            minimum=1,
                            maximum=5,
                            value=1,
                            step=1,
                            label=LANGUAGES[DEFAULT_LANG]["n_versions_label"],
                            info=LANGUAGES[DEFAULT_LANG]["n_versions_info"]
                        )
                with gr.Row():
                    filter_btn = gr.Button(value=LANGUAGES[DEFAULT_LANG]["filter_button"], variant="primary")

                info_text_md = gr.Markdown(
                    value=LANGUAGES[DEFAULT_LANG]["info_text"],
                    elem_classes=["info-text"]
                )

    # Footer
    with gr.Row():
        footer_md = gr.Markdown(value=LANGUAGES[DEFAULT_LANG]["footer_text"])

    # Handle row selection for radar charts
    def update_charts(evt: gr.SelectData, selected_models, current_lang):
        try:
            # Get current data with the latest filters applied in update_data
            current_df = df # Use the globally updated df
            current_ret_metrics = retrieval_metrics
            current_gen_metrics = generation_metrics

            # Debug info
            print(f"Selection event: {evt}, type: {type(evt)}")

            selected_model = None

            # Extract the selected model based on the row index
            try:
                component = evt.target
                row_idx = evt.index[0] if isinstance(evt.index, list) else evt.index
                print(f"Row index: {row_idx}, Component: {component}")

                # Determine what type of data we're dealing with and extract model name
                if component is metrics_table:
                    # Summary table was clicked
                    current_summary_df = create_summary_df(current_df, current_ret_metrics, current_gen_metrics)
                    if isinstance(current_summary_df, pd.DataFrame) and not current_summary_df.empty and 0 <= row_idx < len(current_summary_df):
                         selected_model = current_summary_df.iloc[row_idx]['Model']
                    print(f"Selected from summary table: {selected_model}")
                else:
                    # Check if it's a category table
                    for category, table in category_tables.items():
                        if component is table:
                            category_df = create_category_df(
                                current_df,
                                category,
                                f"{category}_retrieval",
                                f"{category}_generation"
                            )
                            if isinstance(category_df, pd.DataFrame) and not category_df.empty and 0 <= row_idx < len(category_df):
                                selected_model = category_df.iloc[row_idx]['Model']
                            print(f"Selected from {category} table: {selected_model}")
                            break

                 # Fallback if model not found yet (should not happen often with explicit checks)
                if selected_model is None and hasattr(evt, 'value') and evt.value:
                    selected_model = evt.value[0] # Assuming model name is the first column value in the selected cell data
                    print(f"Selected model using fallback evt.value: {selected_model}")

            except IndexError:
                 print(f"IndexError: row_idx {row_idx} out of bounds for the component's data.")
                 # Potentially return current state without changes
                 gen_chart = create_radar_chart(current_df, selected_models, current_gen_metrics, LANGUAGES[current_lang]["radar_gen_title"])
                 ret_chart = create_radar_chart(current_df, selected_models, current_ret_metrics, LANGUAGES[current_lang]["radar_ret_title"], name_col='Embeddings')
                 return selected_models, gen_chart, ret_chart
            except Exception as e:
                print(f"Error extracting model name: {e}")
                traceback.print_exc()

            # If we found a model name, toggle its selection
            if selected_model:
                print(f"Selected model: {selected_model}")
                available_models = current_df['Model'].tolist() if not current_df.empty else []

                if selected_model in available_models:
                    new_selected_models = selected_models[:] # Create a copy
                    if selected_model in new_selected_models:
                        new_selected_models.remove(selected_model)
                    else:
                        new_selected_models.append(selected_model)

                    # Ensure only models from the current dataframe are included
                    new_selected_models = [model for model in new_selected_models if model in available_models]

                    # If no models are selected after filtering, select the top available model
                    if not new_selected_models and available_models:
                        new_selected_models = [available_models[0]]

                    selected_models = new_selected_models # Update the state
                else:
                    print(f"Model {selected_model} not found in current dataframe")

            # Create radar charts using the current dataframe and metrics
            gen_chart = create_radar_chart(current_df, selected_models, current_gen_metrics, LANGUAGES[current_lang]["radar_gen_title"])
            ret_chart = create_radar_chart(current_df, selected_models, current_ret_metrics, LANGUAGES[current_lang]["radar_ret_title"], name_col='Embeddings')

            return selected_models, gen_chart, ret_chart
        except Exception as e:
            print(f"Error in update_charts: {e}")
            print(traceback.format_exc())
            # Return potentially existing chart values if error occurs
            current_gen_chart = create_radar_chart(df, selected_models, generation_metrics, LANGUAGES[current_lang]["radar_gen_title"])
            current_ret_chart = create_radar_chart(df, selected_models, retrieval_metrics, LANGUAGES[current_lang]["radar_ret_title"], name_col='Embeddings')
            return selected_models, current_gen_chart, current_ret_chart


    # Use custom event handler for row selection
    # Make sure to pass current_language state
    metrics_table.select(
        fn=update_charts,
        inputs=[selected_models, current_language],
        outputs=[selected_models, generation_chart, retrieval_chart]
    )

    # Add selection handlers for category tables too
    for category_table in category_tables.values():
        category_table.select(
            fn=update_charts,
            inputs=[selected_models, current_language],
            outputs=[selected_models, generation_chart, retrieval_chart]
        )

    # Handle version filter changes
    def update_data(n_versions, only_actual, current_selected_models, current_lang):
        try:
            # Update global data (df, metrics)
            global df, retrieval_metrics, generation_metrics
            new_df, new_ret_metrics, new_gen_metrics, new_category_metrics = filter_and_process_results(
                results, n_versions=n_versions, only_actual_versions=only_actual
            )
            # Update global references
            df = new_df
            retrieval_metrics = new_ret_metrics
            generation_metrics = new_gen_metrics

            available_models = df['Model'].tolist() if not df.empty else []

            # Filter selected models
            filtered_selected_models = [model for model in current_selected_models if model in available_models]
            if not filtered_selected_models and available_models:
                filtered_selected_models = available_models[:min(5, len(available_models))]

            # Create charts with localized titles
            gen_chart_val = create_radar_chart(df, filtered_selected_models, generation_metrics, LANGUAGES[current_lang]["radar_gen_title"])
            ret_chart_val = create_radar_chart(df, filtered_selected_models, retrieval_metrics, LANGUAGES[current_lang]["radar_ret_title"], name_col='Embeddings')

            # Create summary dataframe
            summary_df_val = create_summary_df(df, retrieval_metrics, generation_metrics)

            # Prepare outputs for tables and charts
            outputs = {
                metrics_table: gr.update(value=summary_df_val if not summary_df_val.empty else pd.DataFrame(), visible=not summary_df_val.empty),
                empty_data_md: gr.update(visible=summary_df_val.empty),
                generation_chart: gen_chart_val,
                retrieval_chart: ret_chart_val,
                selected_models: filtered_selected_models
            }

            # Update category tables
            current_category_display_names = LANGUAGES[current_lang]["category_display_names"]
            for category in category_tables.keys():
                if f"{category}_retrieval" in df.columns and f"{category}_generation" in df.columns:
                    category_df_val = create_category_df(df, category, f"{category}_retrieval", f"{category}_generation")
                    display_name = current_category_display_names.get(category, category.capitalize())

                    outputs[category_tables[category]] = gr.update(value=category_df_val if not category_df_val.empty else pd.DataFrame(), visible=not category_df_val.empty)
                    outputs[category_no_data_mds[category]] = gr.update(visible=category_df_val.empty)
                    outputs[category_title_mds[category]] = gr.update(visible=not category_df_val.empty)
                else:
                    # Hide table and titles if data for category doesn't exist with current filters
                    outputs[category_tables[category]] = gr.update(value=pd.DataFrame(), visible=False)
                    outputs[category_no_data_mds[category]] = gr.update(visible=True) # Show 'no data' instead? Or just hide all? Let's hide title too.
                    outputs[category_title_mds[category]] = gr.update(visible=False)


            # Return updates in the correct order based on outputs list
            output_list = [outputs[metrics_table], outputs[empty_data_md], outputs[generation_chart], outputs[retrieval_chart], outputs[selected_models]]
            for category in category_tables.keys():
                 output_list.extend([
                     outputs[category_tables[category]],
                     outputs[category_no_data_mds[category]],
                     outputs[category_title_mds[category]]
                 ])

            return output_list
        except Exception as e:
            print(f"Error in update_data: {e}")
            print(traceback.format_exc())
            # Return original values in case of error; construct a list of Nones matching output structure
            num_category_outputs = len(category_tables.keys()) * 3
            return [gr.update()]*5 + [gr.update()]*num_category_outputs # Return no changes

    # Define filter button outputs
    filter_outputs = [metrics_table, empty_data_md, generation_chart, retrieval_chart, selected_models]
    for category in category_tables.keys():
         filter_outputs.extend([category_tables[category], category_no_data_mds[category], category_title_mds[category]])

    filter_btn.click(
        fn=update_data,
        inputs=[n_versions_slider, only_actual_versions, selected_models, current_language], # Pass language
        outputs=filter_outputs
    )

    # Function to clear charts
    def clear_charts_localized(current_lang): # Pass language
        empty_models = []
        # Create empty charts with localized titles
        empty_gen_chart = create_radar_chart(df, empty_models, generation_metrics, LANGUAGES[current_lang]["radar_gen_title"])
        empty_ret_chart = create_radar_chart(df, empty_models, retrieval_metrics, LANGUAGES[current_lang]["radar_ret_title"], name_col='Embeddings')
        return empty_models, empty_gen_chart, empty_ret_chart

    # Connect clear charts button
    clear_charts_btn.click(
        fn=clear_charts_localized,
        inputs=[current_language], # Pass language
        outputs=[selected_models, generation_chart, retrieval_chart]
    )

    # Function to update language-specific elements
    def update_language(selected_lang):
        lang_dict = LANGUAGES[selected_lang]
        category_display_names = lang_dict.get("category_display_names", {})

        updates = {
            current_language: selected_lang, # Update the state holding the language key
            current_lang_dict: lang_dict, # Update the state holding the translations
            # lang_selector: gr.update(label=lang_dict["lang_selector_label"]),
            description_md: gr.update(value=lang_dict["description"]),
            version_info_md: gr.update(value=lang_dict["version_info_template"].format(last_version, n_questions, date_title)),
            gen_chart_title_md: gr.update(value=lang_dict["gen_metrics_title"]),
            ret_chart_title_md: gr.update(value=lang_dict["ret_metrics_title"]),
            clear_charts_btn: gr.update(value=lang_dict["clear_charts"]),
            summary_tab: gr.update(label=lang_dict["overall_tab_title"]),
            empty_data_md: gr.update(value=lang_dict["no_data_message"]),
            category_main_tab: gr.update(label=lang_dict["by_type_tab_title"]),
            citation_title_md: gr.update(value=lang_dict["citation_title"]),
            citation_desc_md: gr.update(value=lang_dict["citation_description"]),
            version_selector_title_md: gr.update(value=lang_dict["version_selector_title"]),
            only_actual_versions: gr.update(label=lang_dict["only_actual_label"], info=lang_dict["only_actual_info"]),
            n_versions_slider: gr.update(label=lang_dict["n_versions_label"], info=lang_dict["n_versions_info"]),
            filter_btn: gr.update(value=lang_dict["filter_button"]),
            info_text_md: gr.update(value=lang_dict["info_text"]),
            footer_md: gr.update(value=lang_dict["footer_text"]),
            # Update category tab labels and conditional text templates
            **{tab_item: gr.update(label=category_display_names.get(category, category.capitalize()))
               for category, tab_item in category_tab_items.items()},
            **{no_data_md: gr.update(value=lang_dict["no_data_category_template"].format(category_display_names.get(category, category.capitalize())))
               for category, no_data_md in category_no_data_mds.items()},
            **{title_md: gr.update(value=lang_dict["category_performance_template"].format(category_display_names.get(category, category.capitalize())))
               for category, title_md in category_title_mds.items()},
             # Update chart titles dynamically by re-plotting (needed if chart titles change)
             generation_chart: create_radar_chart(df, selected_models.value, generation_metrics, lang_dict["radar_gen_title"]),
             retrieval_chart: create_radar_chart(df, selected_models.value, retrieval_metrics, lang_dict["radar_ret_title"], name_col='Embeddings')
        }

        # Return updates in the correct order based on outputs list below
        output_list = [
            updates[current_language], updates[current_lang_dict],
            updates[description_md], updates[version_info_md], updates[gen_chart_title_md], updates[ret_chart_title_md],
            updates[clear_charts_btn], updates[summary_tab], updates[empty_data_md], updates[category_main_tab],
            updates[citation_title_md], updates[citation_desc_md], updates[version_selector_title_md],
            updates[only_actual_versions], updates[n_versions_slider], updates[filter_btn], updates[info_text_md],
            updates[footer_md], updates[generation_chart], updates[retrieval_chart]
        ]
        # Add category tab items, no_data markdown, and title markdown updates
        for category in category_tables.keys(): # Use category_tables as the source of truth for existing categories
             if category in category_tab_items: output_list.append(updates[category_tab_items[category]])
             if category in category_no_data_mds: output_list.append(updates[category_no_data_mds[category]])
             if category in category_title_mds: output_list.append(updates[category_title_mds[category]])

        return output_list

    # Define the outputs for the language selector change event
    lang_outputs = [
        current_language, current_lang_dict, description_md, version_info_md,
        gen_chart_title_md, ret_chart_title_md, clear_charts_btn, summary_tab, empty_data_md,
        category_main_tab, citation_title_md, citation_desc_md, version_selector_title_md,
        only_actual_versions, n_versions_slider, filter_btn, info_text_md, footer_md,
        generation_chart, retrieval_chart # Charts need to be updated too if their titles change
    ]
    # Add category tab items, no_data markdown, and title markdown to outputs
    for category in category_tables.keys():
         if category in category_tab_items: lang_outputs.append(category_tab_items[category])
         if category in category_no_data_mds: lang_outputs.append(category_no_data_mds[category])
         if category in category_title_mds: lang_outputs.append(category_title_mds[category])


    # Connect language selector change event
    lang_selector.change(
        fn=update_language,
        inputs=[lang_selector],
        outputs=lang_outputs
    )

if __name__ == "__main__":
    demo.launch()