from dash import Dash, html, dcc, dash_table
import dash_bootstrap_components as dbc
import pandas as pd
import plotly.express as px
from datasets import load_dataset

# -----------------------------
# Carregar dataset do Hugging Face
# -----------------------------
ds = load_dataset("nvidia/Nemotron-Image-Training-v3", "aokvqa_1")
df = ds["train"].to_pandas()

# Limitar para evitar dashboard pesado
df_preview = df.head(20).copy()

# Criar dataframe de resumo
resumo_df = pd.DataFrame({
    "Métrica": [
        "Total de linhas",
        "Total de colunas"
    ],
    "Valor": [
        len(df),
        len(df.columns)
    ]
})

colunas_df = pd.DataFrame({
    "Colunas": list(df.columns)
})

grafico_df = pd.DataFrame({
    "Tipo": ["Registos"],
    "Quantidade": [len(df)]
})

# -----------------------------
# Criar app
# -----------------------------
app = Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
server = app.server

# -----------------------------
# Layout
# -----------------------------
app.layout = dbc.Container([
    dbc.Row([
        dbc.Col([
            html.H1("Dashboard do Dataset Nemotron", className="text-center my-4"),
            html.P(
                "Dataset carregado com Hugging Face datasets: nvidia/Nemotron-Image-Training-v3 / aokvqa_1",
                className="text-center text-muted"
            )
        ])
    ]),

    dbc.Row([
        dbc.Col(
            dbc.Card(
                dbc.CardBody([
                    html.H5("Total de registos", className="card-title"),
                    html.H2(f"{len(df)}", className="text-primary")
                ])
            ),
            md=4
        ),
        dbc.Col(
            dbc.Card(
                dbc.CardBody([
                    html.H5("Total de colunas", className="card-title"),
                    html.H2(f"{len(df.columns)}", className="text-success")
                ])
            ),
            md=4
        ),
        dbc.Col(
            dbc.Card(
                dbc.CardBody([
                    html.H5("Subset", className="card-title"),
                    html.H2("aokvqa_1", className="text-danger")
                ])
            ),
            md=4
        ),
    ], className="mb-4"),

    dbc.Row([
        dbc.Col([
            dcc.Graph(
                figure=px.bar(
                    grafico_df,
                    x="Tipo",
                    y="Quantidade",
                    title="Quantidade de registos no dataset"
                )
            )
        ], md=6),

        dbc.Col([
            html.H4("Colunas disponíveis"),
            dash_table.DataTable(
                columns=[{"name": i, "id": i} for i in colunas_df.columns],
                data=colunas_df.to_dict("records"),
                style_table={"overflowX": "auto", "height": "350px", "overflowY": "auto"},
                style_cell={"textAlign": "left", "padding": "8px"},
                style_header={"fontWeight": "bold", "backgroundColor": "#f8f9fa"},
                page_size=10
            )
        ], md=6)
    ], className="mb-4"),

    dbc.Row([
        dbc.Col([
            html.H4("Primeiras 20 linhas"),
            dash_table.DataTable(
                columns=[{"name": i, "id": i} for i in df_preview.columns],
                data=df_preview.astype(str).to_dict("records"),
                style_table={"overflowX": "auto"},
                style_cell={
                    "textAlign": "left",
                    "padding": "8px",
                    "maxWidth": "250px",
                    "whiteSpace": "normal"
                },
                style_header={
                    "fontWeight": "bold",
                    "backgroundColor": "#f8f9fa"
                },
                page_size=20
            )
        ])
    ])
], fluid=True)

# -----------------------------
# Run app
# -----------------------------
if __name__ == "__main__":
    app.run(host="0.0.0.0", port=7860, debug=False)