app.py

import gradio as gr
import pandas as pd
import os
import sys
from qatch.connectors.sqlite_connector import SqliteConnector
from qatch.generate_dataset.orchestrator_generator import OrchestratorGenerator
from qatch.evaluate_dataset.orchestrator_evaluator import OrchestratorEvaluator
#from predictor.orchestrator_predictor import OrchestratorPredictor
import utils_get_db_tables_info
import utilities as us
import time
import plotly.express as px
import plotly.graph_objects as go
import plotly.colors as pc

with open('style.css', 'r') as file:
    css = file.read()

# DataFrame di default
df_default = pd.DataFrame({
    'Name': ['Alice', 'Bob', 'Charlie'],
    'Age': [25, 30, 35],
    'City': ['New York', 'Los Angeles', 'Chicago']
})

models_path = "models.csv"

# Variabile globale per tenere traccia dei dati correnti
df_current = df_default.copy()

input_data = {
    'input_method': "", 
    'data_path': "",
    'db_name': "", 
    'data': {
        'data_frames': {},    # dictionary of dataframes
        'db': None             # SQLITE3 database object
    },
    'models': []
}

def load_data(file, path, use_default):
    """Carica i dati da un file, un percorso o usa il DataFrame di default."""
    global df_current
    if use_default:
        input_data["input_method"] = 'default'
        input_data["data_path"] = os.path.join(".", "data", "data_interface", "mytable.sqlite")
        input_data["db_name"] = os.path.splitext(os.path.basename(input_data["data_path"]))[0]
        input_data["data"]['data_frames'] = {'MyTable': df_current}

        if( input_data["data"]['data_frames']):
            table2primary_key = {}
            for table_name, df in input_data["data"]['data_frames'].items():
                # Assign primary keys for each table
                table2primary_key[table_name] = 'id'
            input_data["data"]["db"] = SqliteConnector(
                relative_db_path=input_data["data_path"],
                db_name=input_data["db_name"],
                tables= input_data["data"]['data_frames'],
                table2primary_key=table2primary_key
            )

        df_current = df_default.copy()  # Ripristina i dati di default
        return input_data["data"]['data_frames']
    
    selected_inputs = sum([file is not None, bool(path), use_default])
    if selected_inputs > 1:
        return 'Errore: Selezionare solo un metodo di input alla volta.'
    
    if file is not None:
        try:
            input_data["input_method"] = 'uploaded_file'
            input_data["db_name"] = os.path.splitext(os.path.basename(file))[0]
            input_data["data_path"] = os.path.join(".", "data", "data_interface",f"{input_data['db_name']}.sqlite")
            input_data["data"] = us.load_data(file, input_data["db_name"])
            df_current = input_data["data"]['data_frames'].get('MyTable', df_default)  # Carica il DataFrame
            if( input_data["data"]['data_frames']):
                table2primary_key = {}
                for table_name, df in input_data["data"]['data_frames'].items():
                    # Assign primary keys for each table
                    table2primary_key[table_name] = 'id'
                input_data["data"]["db"] = SqliteConnector(
                    relative_db_path=input_data["data_path"],
                    db_name=input_data["db_name"],
                    tables= input_data["data"]['data_frames'],
                    table2primary_key=table2primary_key
                )
            return input_data["data"]['data_frames']
        except Exception as e:
            return f'Errore nel caricamento del file: {e}'
    
    """

    if path:

        if not os.path.exists(path):

            return 'Errore: Il percorso specificato non esiste.'

        try:

            input_data["input_method"] = 'uploaded_file'

            input_data["data_path"] = path

            input_data["db_name"] = os.path.splitext(os.path.basename(path))[0]

            input_data["data"] = us.load_data(input_data["data_path"], input_data["db_name"])

            df_current = input_data["data"]['data_frames'].get('MyTable', df_default)  # Carica il DataFrame

            

            return input_data["data"]['data_frames']

        except Exception as e:

            return f'Errore nel caricamento del file dal percorso: {e}'

    """
    
    
    return input_data["data"]['data_frames']

def preview_default(use_default):
    """Mostra il DataFrame di default se il checkbox è selezionato."""
    if use_default:
        return df_default  # Mostra il DataFrame di default
    return df_current  # Mostra il DataFrame corrente, che potrebbe essere stato modificato

def update_df(new_df):
    """Aggiorna il DataFrame corrente."""
    global df_current  # Usa la variabile globale per aggiornarla
    df_current = new_df
    return df_current

def open_accordion(target):
    # Apre uno e chiude l'altro
    if target == "reset":
        df_current = df_default.copy()
        input_data['input_method'] = ""
        input_data['data_path'] = ""
        input_data['db_name'] = ""
        input_data['data']['data_frames'] = {}
        input_data['data']['db'] = None
        input_data['models'] = []
        return gr.update(open=True), gr.update(open=False, visible=False), gr.update(open=False, visible=False), gr.update(open=False, visible=False), gr.update(open=False, visible=False), gr.update(value=False), gr.update(value=None)
    elif target == "model_selection":
        return gr.update(open=False), gr.update(open=False), gr.update(open=True, visible=True), gr.update(open=False), gr.update(open=False)

# Interfaccia Gradio

with gr.Blocks(theme='d8ahazard/rd_blue', css_paths='style.css') as interface:
    gr.Markdown("# QATCH")
    data_state = gr.State(None)  # Memorizza i dati caricati
    upload_acc = gr.Accordion("Upload your data section", open=True, visible=True)
    select_table_acc = gr.Accordion("Select tables", open=False, visible=False)
    select_model_acc = gr.Accordion("Select models", open=False, visible=False)
    qatch_acc = gr.Accordion("QATCH execution", open=False, visible=False)
    metrics_acc = gr.Accordion("Metrics", open=False, visible=False)
    #metrics_acc = gr.Accordion("Metrics", open=False, visible=False, render=False)



    #################################
    #  PARTE DI INSERIMENTO DEL DB  #
    #################################
    with upload_acc:
        gr.Markdown("## Caricamento dei Dati")

        file_input = gr.File(label="Trascina e rilascia un file", file_types=[".csv", ".xlsx", ".sqlite"])
        with gr.Row():
            default_checkbox = gr.Checkbox(label="Usa DataFrame di default")
        preview_output = gr.DataFrame(interactive=True, visible=True, value=df_default)
        submit_button = gr.Button("Carica Dati", interactive=False)  # Disabilitato di default
        output = gr.JSON(visible=False)  # Output dizionario

        # Funzione per abilitare il bottone se sono presenti dati da caricare
        def enable_submit(file, use_default):
            return gr.update(interactive=bool(file or use_default))

        # Funzione per deselezionare il checkbox se viene caricato un file
        def deselect_default(file):
            if file:
                return gr.update(value=False)
            return gr.update()

        # Abilita il bottone quando i campi di input sono valorizzati
        file_input.change(fn=enable_submit, inputs=[file_input, default_checkbox], outputs=[submit_button])
        default_checkbox.change(fn=enable_submit, inputs=[file_input, default_checkbox], outputs=[submit_button])
        
        # Mostra l'anteprima del DataFrame di default quando il checkbox è selezionato
        default_checkbox.change(fn=preview_default, inputs=[default_checkbox], outputs=[preview_output])
        preview_output.change(fn=update_df, inputs=[preview_output], outputs=[preview_output])

        # Deseleziona il checkbox quando viene caricato un file
        file_input.change(fn=deselect_default, inputs=[file_input], outputs=[default_checkbox])

        def handle_output(file, use_default):
            """Gestisce l'output quando si preme il bottone 'Carica Dati'."""
            result = load_data(file, None, use_default)
            
            if isinstance(result, dict):  # Se result è un dizionario di DataFrame
                if len(result) == 1:  # Se c'è solo una tabella
                    return (
                        gr.update(visible=False),  # Nasconde l'output JSON
                        result,  # Salva lo stato dei dati
                        gr.update(visible=False),  # Nasconde la selezione tabella
                        result,  # Mantiene lo stato dei dati
                        gr.update(interactive=False),  # Disabilita il pulsante di submit
                        gr.update(visible=True, open=True),  # Passa direttamente a select_model_acc
                        gr.update(visible=True, open=False)
                    )
                else:
                    return (
                        gr.update(visible=False),
                        result,
                        gr.update(open=True, visible=True),
                        result,
                        gr.update(interactive=False),
                        gr.update(visible=False),  # Mantiene il comportamento attuale
                        gr.update(visible=True, open=True)
                    )
            else:
                return (
                    gr.update(visible=False),
                    None,
                    gr.update(open=False, visible=True),
                    None,
                    gr.update(interactive=True),
                    gr.update(visible=False),
                    gr.update(visible=True, open=True)
                )

        submit_button.click(
            fn=handle_output,
            inputs=[file_input, default_checkbox], 
            outputs=[output, output, select_table_acc, data_state, submit_button, select_model_acc, upload_acc]
        )



    ######################################
    #  PARTE DI SELEZIONE DELLE TABELLE  #
    ######################################
    with select_table_acc:
        
        table_selector = gr.CheckboxGroup(choices=[], label="Seleziona le tabelle da visualizzare", value=[])
        table_outputs = [gr.DataFrame(label=f"Tabella {i+1}", interactive=True, visible=False) for i in range(5)]
        selected_table_names = gr.Textbox(label="Tabelle selezionate", visible=False, interactive=False)

        # Bottone di selezione modelli (inizialmente disabilitato)
        open_model_selection = gr.Button("Choose your models", interactive=False)

        def update_table_list(data):
            """Aggiorna dinamicamente la lista delle tabelle disponibili."""
            if isinstance(data, dict) and data:
                table_names = list(data.keys())  # Ritorna solo i nomi delle tabelle
                return gr.update(choices=table_names, value=[])  # Reset delle selezioni
            return gr.update(choices=[], value=[])

        def show_selected_tables(data, selected_tables):
            """Mostra solo le tabelle selezionate dall'utente e abilita il bottone."""
            updates = []
            if isinstance(data, dict) and data:
                available_tables = list(data.keys())  # Nomi effettivamente disponibili
                selected_tables = [t for t in selected_tables if t in available_tables]  # Filtra selezioni valide
                
                tables = {name: data[name] for name in selected_tables}  # Filtra i DataFrame
                
                for i, (name, df) in enumerate(tables.items()):
                    updates.append(gr.update(value=df, label=f"Tabella: {name}", visible=True))

                # Se ci sono meno di 5 tabelle, nascondi gli altri DataFrame
                for _ in range(len(tables), 5):
                    updates.append(gr.update(visible=False))
            else:
                updates = [gr.update(value=pd.DataFrame(), visible=False) for _ in range(5)]

            # Abilitare/disabilitare il bottone in base alle selezioni
            button_state = bool(selected_tables)  # True se almeno una tabella è selezionata, False altrimenti
            updates.append(gr.update(interactive=button_state))  # Aggiorna stato bottone
            
            return updates

        def show_selected_table_names(selected_tables):
            """Mostra i nomi delle tabelle selezionate quando si preme il bottone."""
            if selected_tables:
                return gr.update(value=", ".join(selected_tables), visible=False)
            return gr.update(value="", visible=False)

        # Aggiorna automaticamente la lista delle checkbox quando `data_state` cambia
        data_state.change(fn=update_table_list, inputs=[data_state], outputs=[table_selector])

        # Aggiorna le tabelle visibili e lo stato del bottone in base alle selezioni dell'utente
        table_selector.change(fn=show_selected_tables, inputs=[data_state, table_selector], outputs=table_outputs + [open_model_selection])

        # Mostra la lista delle tabelle selezionate quando si preme "Choose your models"
        open_model_selection.click(fn=show_selected_table_names, inputs=[table_selector], outputs=[selected_table_names])
        open_model_selection.click(open_accordion, inputs=gr.State("model_selection"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc])



    ####################################
    #  PARTE DI SELEZIONE DEL MODELLO  #
    ####################################
    with select_model_acc:
        gr.Markdown("**Model Selection**")

        # Supponiamo che `us.read_models_csv` restituisca anche il percorso dell'immagine
        model_list_dict = us.read_models_csv(models_path)
        model_list = [model["code"] for model in model_list_dict]
        model_images = [model["image_path"] for model in model_list_dict]

        model_checkboxes = []
        rows = []
        
        # Creazione dinamica di checkbox con immagini (3 per riga)
        for i in range(0, len(model_list), 3):
            with gr.Row():
                cols = []
                for j in range(3):
                    if i + j < len(model_list):
                        model = model_list[i + j]
                        image_path = model_images[i + j]
                        with gr.Column():
                            gr.Image(image_path, show_label=False)
                            checkbox = gr.Checkbox(label=model, value=False)
                            model_checkboxes.append(checkbox)
                            cols.append(checkbox)
                rows.append(cols)

        selected_models_output = gr.JSON(visible=False)

        # Funzione per ottenere i modelli selezionati
        def get_selected_models(*model_selections):
            selected_models = [model for model, selected in zip(model_list, model_selections) if selected]
            input_data['models'] = selected_models
            button_state = bool(selected_models)  # True se almeno un modello è selezionato, False altrimenti
            return selected_models, gr.update(open=True, visible=True), gr.update(interactive=button_state)

        # Bottone di submit (inizialmente disabilitato)
        submit_models_button = gr.Button("Submit Models", interactive=False)

        # Collegamento dei checkbox agli eventi di selezione
        for checkbox in model_checkboxes:
            checkbox.change(
                fn=get_selected_models, 
                inputs=model_checkboxes, 
                outputs=[selected_models_output, select_model_acc, submit_models_button]
            )

        submit_models_button.click(
            fn=lambda *args: (get_selected_models(*args), gr.update(open=False, visible=True), gr.update(open=True, visible=True)),
            inputs=model_checkboxes,
            outputs=[selected_models_output, select_model_acc, qatch_acc]
        )
        
        reset_data = gr.Button("Back to upload data section")
        reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc, default_checkbox, file_input])


    ###############################
    #  PARTE DI ESECUZIONE QATCH  #
    ###############################
    with qatch_acc:
        def change_text(text):
            return text
        def qatch_flow():
            orchestrator_generator = OrchestratorGenerator()
            #TODO add to target_df column target_df["columns_used"], tables selection
            #print(input_data['data']['db'])
            target_df = orchestrator_generator.generate_dataset(connector=input_data['data']['db'])
            
            schema_text = utils_get_db_tables_info.utils_extract_db_schema_as_string(
                db_id = input_data["db_name"], 
                base_path = input_data["data_path"], 
                normalize=False, 
                sql=None
            )

            # TODO QUERY PREDICTION
            predictions_dict = {model: pd.DataFrame(columns=['id', 'question', 'predicted_sql', 'time', 'query', 'db_path']) for model in model_list}
            metrics_conc = pd.DataFrame()
            for model in input_data["models"]:
                for index, row in target_df.iterrows():
                    if len(target_df) != 0: load_value = f"##Loading... {round((index + 1) / len(target_df) * 100, 2)}%"
                    else: load_value = "##Loading..."
                    question = row['query']
                    #yield gr.Textbox(question), gr.Textbox(), *[predictions_dict[model] for model in input_data["models"]], None
                    yield gr.Markdown(value=load_value), gr.Textbox(question), gr.Textbox(), metrics_conc, *[predictions_dict[model] for model in model_list]
                    start_time = time.time()

                    # Simulazione della predizione
                    time.sleep(0.03)
                    prediction = "Prediction_placeholder"

                    # Esegui la predizione reale qui
                    # prediction = predictor.run(model, schema_text, question)

                    end_time = time.time()
                    # Crea una nuova riga come dataframe
                    new_row = pd.DataFrame([{
                        'id': index,
                        'question': question,
                        'predicted_sql': prediction,
                        'time': end_time - start_time,
                        'query': row["query"],
                        'db_path': input_data["data_path"]
                    }]).dropna(how="all")  # Rimuove solo righe completamente vuote
                    #TODO con un for
                    for col in target_df.columns:
                        if col not in new_row.columns:
                            new_row[col] = row[col]
                    # Aggiorna il dataframe corrispondente al modello man mano
                    if not new_row.empty:
                        predictions_dict[model] = pd.concat([predictions_dict[model], new_row], ignore_index=True)
                    #yield gr.Textbox(), gr.Textbox(prediction), *[predictions_dict[model] for model in input_data["models"]], None
                    yield gr.Markdown(value=load_value), gr.Textbox(), gr.Textbox(prediction), metrics_conc, *[predictions_dict[model] for model in model_list]

            #END 
            evaluator = OrchestratorEvaluator()
            for model in input_data["models"]:
                metrics_df_model = evaluator.evaluate_df(
                    df=predictions_dict[model],
                    target_col_name="query",              #'<target_column_name>',
                    prediction_col_name="predicted_sql",  #'<prediction_column_name>',
                    db_path_name= "db_path",              #'<db_path_column_name>'
                )
                metrics_df_model['model'] = model
                metrics_conc = pd.concat([metrics_conc, metrics_df_model], ignore_index=True)
                
                if 'valid_efficiency_score' not in metrics_conc.columns:
                    metrics_conc['valid_efficiency_score'] = metrics_conc['VES']

            yield gr.Markdown(), gr.Textbox(), gr.Textbox(), metrics_conc, *[predictions_dict[model] for model in model_list]

        #Loading Bar
        with gr.Row():
            #progress = gr.Progress()
            variable = gr.Markdown()

        #NL -> MODEL -> Generated Quesy
        with gr.Row():
            with gr.Column():
                question_display = gr.Textbox()
            with gr.Column():
                gr.Image()
            with gr.Column():
                prediction_display = gr.Textbox()
        
        dataframe_per_model = {}

        with gr.Tabs() as model_tabs:
            #for model in input_data["models"]:
            for model in model_list:
                #TODO fix model tabs
                with gr.TabItem(model):
                    gr.Markdown(f"**Results for {model}**")
                    dataframe_per_model[model] = gr.DataFrame()


        #question_display.change(fn=change_text, inputs=[gr.State(question)], outputs=[question_display])
        selected_models_display = gr.JSON(label="Modelli selezionati")
        metrics_df = gr.DataFrame(visible=False)
        metrics_df_out= gr.DataFrame(visible=False)
        
        submit_models_button.click(
            fn=qatch_flow,
            inputs=[],
            outputs=[variable, question_display, prediction_display, metrics_df] + list(dataframe_per_model.values())
        )

        submit_models_button.click(
            fn=lambda: gr.update(value=input_data), 
            outputs=[selected_models_display]
        )
        #Funziona per METRICS
        metrics_df.change(fn=change_text, inputs=[metrics_df], outputs=[metrics_df_out])
    
        # def change_tab(selected_models_output, model_tabs):
        #     for model in model_list:
        #         if model in selected_models_output:
        #             pass#model_tabs[model].visible = True
        #         else:
        #             pass#model_tabs[model].visible = False
        #     return model_tabs

        # selected_models_output.change(fn=change_tab, inputs=[selected_models_output, model_tabs], outputs=[])

        proceed_to_metrics_button = gr.Button("Proceed to Metrics")
        proceed_to_metrics_button.click(
            fn=lambda: (gr.update(open=False, visible=True), gr.update(open=True, visible=True)), 
            outputs=[qatch_acc, metrics_acc]
        )
        
        reset_data = gr.Button("Back to upload data section")
        reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc, default_checkbox, file_input])
         
         
         
    #######################################
    #  METRICS VISUALIZATION SECTION      #
    #######################################
    with metrics_acc:
        #confirmation_text = gr.Markdown("## Metrics successfully loaded")

        data_path = 'test_results.csv'

        @gr.render(inputs=metrics_df_out)
        def function_metrics(metrics_df_out):
            def load_data_csv_es():
                return pd.read_csv(data_path)
                #return metrics_df_out
            
            def calculate_average_metrics(df, selected_metrics):
                df['avg_metric'] = df[selected_metrics].mean(axis=1)
                return df

            def generate_model_colors():
                """Generates a unique color map for models in the dataset."""
                df = load_data_csv_es()
                unique_models = df['model'].unique()  # Extract unique models
                num_models = len(unique_models)
                
                # Use the Plotly color scale (you can change it if needed)
                color_palette = pc.qualitative.Plotly  # ['#636EFA', '#EF553B', '#00CC96', ...]
                
                # If there are more models than colors, cycle through them
                colors = {model: color_palette[i % len(color_palette)] for i, model in enumerate(unique_models)}
                
                return colors

            MODEL_COLORS = generate_model_colors()

            # BAR CHART FOR AVERAGE METRICS WITH UPDATE FUNCTION
            def plot_metric(df, selected_metrics, group_by, selected_models):
                df = df[df['model'].isin(selected_models)]
                df = calculate_average_metrics(df, selected_metrics)
                
                # Ensure the group_by value is always valid
                if group_by not in [["tbl_name", "model"], ["model"]]:
                    group_by = ["tbl_name", "model"]  # Default
                
                avg_metrics = df.groupby(group_by)['avg_metric'].mean().reset_index()
                
                fig = px.bar(
                    avg_metrics, 
                    x=group_by[0], 
                    y='avg_metric', 
                    color='model',  
                    color_discrete_map=MODEL_COLORS, 
                    barmode='group',
                    title=f'Average metric per {group_by[0]} 📊',
                    labels={group_by[0]: group_by[0].capitalize(), 'avg_metric': 'Average Metric'},
                    template='plotly_dark'
                )
                
                return fig

            def update_plot(selected_metrics, group_by, selected_models):
                df = load_data_csv_es()
                return plot_metric(df, selected_metrics, group_by, selected_models)


            # RADAR CHART FOR AVERAGE METRICS PER MODEL WITH UPDATE FUNCTION
            def plot_radar(df, selected_models):
                # Filter only selected models
                df = df[df['model'].isin(selected_models)]
                
                # Select relevant metrics
                selected_metrics = ["cell_precision", "cell_recall", "execution_accuracy", "tuple_cardinality", "tuple_constraint"]

                # Compute average metrics per test_category and model
                df = calculate_average_metrics(df, selected_metrics)
                avg_metrics = df.groupby(['model', 'test_category'])['avg_metric'].mean().reset_index()

                # Check if data is available
                if avg_metrics.empty:
                    print("Error: No data available to compute averages.")
                    return go.Figure()

                fig = go.Figure()
                categories = avg_metrics['test_category'].unique()

                for model in selected_models:
                    model_data = avg_metrics[avg_metrics['model'] == model]

                    # Build a list of values for each category (if a value is missing, set it to 0)
                    values = [
                        model_data[model_data['test_category'] == cat]['avg_metric'].values[0] 
                        if cat in model_data['test_category'].values else 0 
                        for cat in categories
                    ]

                    fig.add_trace(go.Scatterpolar(
                        r=values,
                        theta=categories,
                        fill='toself',
                        name=model,
                        line=dict(color=MODEL_COLORS.get(model, "gray"))
                    ))

                fig.update_layout(
                    polar=dict(radialaxis=dict(visible=True, range=[0, max(avg_metrics['avg_metric'].max(), 0.5)])), # Set the radar range
                    title='❇️ Radar Plot of Metrics per Model (Average per Category) ❇️ ',
                    template='plotly_dark',
                    width=700, height=700
                )

                return fig

            def update_radar(selected_models):
                df = load_data_csv_es()
                return plot_radar(df, selected_models)


            # LINE CHART FOR CUMULATIVE TIME WITH UPDATE FUNCTION
            def plot_cumulative_flow(df, selected_models):
                df = df[df['model'].isin(selected_models)]
                
                fig = go.Figure()
                
                for model in selected_models:
                    model_df = df[df['model'] == model].copy()
                    
                    # Calculate cumulative time
                    model_df['cumulative_time'] = model_df['time'].cumsum()
                    
                    # Calculate cumulative number of queries over time
                    model_df['cumulative_queries'] = range(1, len(model_df) + 1)
                    
                    # Select a color for the model
                    color = MODEL_COLORS.get(model, "gray")  # Assigned model color
                    fillcolor = color.replace("rgb", "rgba").replace(")", ", 0.2)")  # 🔹 Makes the area semi-transparent

                    #color = f"rgba({hash(model) % 256}, {hash(model * 2) % 256}, {hash(model * 3) % 256}, 1)"
                    
                    fig.add_trace(go.Scatter(
                        x=model_df['cumulative_time'],
                        y=model_df['cumulative_queries'], 
                        mode='lines+markers',
                        name=model,
                        line=dict(width=2, color=color)
                    ))
                    
                    # Adds the underlying colored area (same color but transparent)
                    """

                    fig.add_trace(go.Scatter(

                        x=model_df['cumulative_time'],

                        y=model_df['cumulative_queries'],

                        fill='tozeroy',

                        mode='none',

                        showlegend=False,  # Hides the area in the legend

                        fillcolor=fillcolor

                    ))

                    """

                fig.update_layout(
                    title="Cumulative Query Flow Chart 📈",
                    xaxis_title="Cumulative Time (s)",
                    yaxis_title="Number of Queries Completed",
                    template='plotly_dark',
                    legend_title="Models"
                )
                
                return fig

            def update_query_rate(selected_models):
                df = load_data_csv_es()
                return plot_cumulative_flow(df, selected_models)


            # RANKING FOR THE TOP 3 MODELS WITH UPDATE FUNCTION
            def ranking_text(df, selected_models, ranking_type):
                #df = load_data_csv_es()
                df = df[df['model'].isin(selected_models)]
                df['valid_efficiency_score'] = pd.to_numeric(df['valid_efficiency_score'], errors='coerce')
                if ranking_type == "valid_efficiency_score":
                    rank_df = df.groupby('model')['valid_efficiency_score'].mean().reset_index()
                    #rank_df = df.groupby('model')['valid_efficiency_score'].mean().reset_index()
                    ascending_order = False  # Higher is better
                elif ranking_type == "time":
                    rank_df = df.groupby('model')['time'].sum().reset_index()
                    rank_df["Ranking Value"] = rank_df["time"].round(2).astype(str) + " s"  # Adds "s" for seconds
                    ascending_order = True  # For time, lower is better
                elif ranking_type == "metrics":
                    selected_metrics = ["cell_precision", "cell_recall", "execution_accuracy", "tuple_cardinality", "tuple_constraint"]
                    df = calculate_average_metrics(df, selected_metrics)
                    rank_df = df.groupby('model')['avg_metric'].mean().reset_index()
                    ascending_order = False  # Higher is better

                if ranking_type != "time":
                    rank_df.rename(columns={rank_df.columns[1]: "Ranking Value"}, inplace=True)
                    rank_df["Ranking Value"] = rank_df["Ranking Value"].round(2)  # Round values except for time

                # Sort based on the selected criterion
                rank_df = rank_df.sort_values(by="Ranking Value", ascending=ascending_order).reset_index(drop=True)

                # Select only the top 3 models
                rank_df = rank_df.head(3)

                # Add medal icons for the top 3
                medals = ["🥇", "🥈", "🥉"]
                rank_df.insert(0, "Rank", medals[:len(rank_df)])

                # Build the formatted ranking string
                ranking_str = "## 🏆 Model Ranking\n"
                for _, row in rank_df.iterrows():
                    ranking_str += f"<span style='font-size:18px;'>{row['Rank']} {row['model']} ({row['Ranking Value']})</span><br>\n"
                
                return ranking_str

            def update_ranking_text(selected_models, ranking_type):
                df = load_data_csv_es()
                return ranking_text(df, selected_models, ranking_type)


            # RANKING FOR THE 3 WORST RESULTS WITH UPDATE FUNCTION
            def worst_cases_text(df, selected_models):
                df = df[df['model'].isin(selected_models)]
                
                selected_metrics = ["cell_precision", "cell_recall", "execution_accuracy", "tuple_cardinality", "tuple_constraint"]
                df = calculate_average_metrics(df, selected_metrics)

                worst_cases_df = df.groupby(['model', 'tbl_name', 'test_category', 'question', 'query', 'predicted_sql'])['avg_metric'].mean().reset_index()
                
                worst_cases_df = worst_cases_df.sort_values(by="avg_metric", ascending=True).reset_index(drop=True)
                
                worst_cases_top_3 = worst_cases_df.head(3)

                worst_cases_top_3["avg_metric"] = worst_cases_top_3["avg_metric"].round(2)

                worst_str = "## ❌ Top 3 Worst Cases\n"
                medals = ["🥇", "🥈", "🥉"]

                for i, row in worst_cases_top_3.iterrows():
                    worst_str += (
                        f"<span style='font-size:18px;'><b>{medals[i]} {row['model']} - {row['tbl_name']} - {row['test_category']}</b> ({row['avg_metric']})</span>  \n"
                        f"<span style='font-size:16px;'>- <b>Question:</b> {row['question']}</span>  \n"
                        f"<span style='font-size:16px;'>- <b>Original Query:</b> `{row['query']}`</span>  \n"
                        f"<span style='font-size:16px;'>- <b>Predicted SQL:</b> `{row['predicted_sql']}`</span>  \n\n"
                    )

                return worst_str

            def update_worst_cases_text(selected_models):
                df = load_data_csv_es()
                return worst_cases_text(df, selected_models)


            metrics = ["cell_precision", "cell_recall", "execution_accuracy", "tuple_cardinality", "tuple_constraint"]
            group_options = {
                "Table": ["tbl_name", "model"],
                "Model": ["model"]
            }

            df_initial = load_data_csv_es()
            models = df_initial['model'].unique().tolist()

            #with gr.Blocks(theme=gr.themes.Default(primary_hue='blue')) as demo:
            gr.Markdown("""## 📊 Model Performance Analysis 📊

            Select one or more metrics to calculate the average and visualize histograms and radar plots.

            """)
            
            # Options selection section
            with gr.Row():
                
                metric_multiselect = gr.CheckboxGroup(choices=metrics, label="Select metrics", value=metrics)
                model_multiselect = gr.CheckboxGroup(choices=models, label="Select models", value=models)
                group_radio = gr.Radio(choices=list(group_options.keys()), label="Select grouping", value="Model")

            output_plot = gr.Plot()

            query_rate_plot = gr.Plot(value=update_query_rate(models))
            
            with gr.Row():  
                with gr.Column(scale=1):
                    radar_plot = gr.Plot(value=update_radar(models))
                    
                with gr.Column(scale=1):
                    ranking_type_radio = gr.Radio(
                        ["valid_efficiency_score", "time", "metrics"], 
                        label="Choose ranking criteria", 
                        value="valid_efficiency_score"
                    )
                    ranking_text_display = gr.Markdown(value=update_ranking_text(models, "valid_efficiency_score"))
                    worst_cases_display = gr.Markdown(value=update_worst_cases_text(models))

            # Callback functions for updating charts
            def on_change(selected_metrics, selected_group, selected_models):
                return update_plot(selected_metrics, group_options[selected_group], selected_models)

            def on_radar_change(selected_models):
                return update_radar(selected_models)
            
            #metrics_df_out.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
            metric_multiselect.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
            group_radio.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
            model_multiselect.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
            model_multiselect.change(update_radar, inputs=model_multiselect, outputs=radar_plot)
            model_multiselect.change(update_ranking_text, inputs=[model_multiselect, ranking_type_radio], outputs=ranking_text_display)
            ranking_type_radio.change(update_ranking_text, inputs=[model_multiselect, ranking_type_radio], outputs=ranking_text_display)
            model_multiselect.change(update_worst_cases_text, inputs=model_multiselect, outputs=worst_cases_display)
            model_multiselect.change(update_query_rate, inputs=[model_multiselect], outputs=query_rate_plot)

            reset_data = gr.Button("Back to upload data section")
            reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc, default_checkbox, file_input])
            
            # Hidden button to force UI refresh on load
            force_update_button = gr.Button("", visible=False)

            # State variable to track first load
            load_trigger = gr.State(value=True)

            # Function to force initial load
            def force_update(is_first_load):
                if is_first_load:
                    return (
                        update_plot(metrics, group_options["Model"], models),
                        update_query_rate(models),
                        update_radar(models),
                        update_ranking_text(models, "valid_efficiency_score"),
                        update_worst_cases_text(models),
                        False  # Change state to prevent continuous reloads
                    )
                return gr.update(), gr.update(), gr.update(), gr.update(), gr.update(), False

            # The invisible button forces chart loading only the first time
            force_update_button.click(
                fn=force_update,
                inputs=[load_trigger],
                outputs=[output_plot, query_rate_plot, radar_plot, ranking_text_display, worst_cases_display, load_trigger]
            )

            # Simulate button click when UI loads
            with gr.Blocks() as demo:
                demo.load(
                    lambda: force_update(True),
                    outputs=[output_plot, query_rate_plot, radar_plot, ranking_text_display, worst_cases_display, load_trigger]
                )  

interface.launch()