import dash
from dash import dcc, html, Input, Output, State
import dash_bootstrap_components as dbc
import plotly.express as px
import pandas as pd
import dash_daq as daq
import plotly.graph_objects as go
import base64
from openai import OpenAI
from dash.exceptions import PreventUpdate
import os
# os.environ["TRANSFORMERS_CACHE"] = "/app/models"
# os.environ["HF_HOME"] = "/app/models"
import utils as u
import requests
import bs4
import re
from langchain_huggingface import HuggingFaceEmbeddings

info_text = """
### Funcionalidades:

- A secção das definições permite-lhe escolher o tipo gráfico de dispersão: 1D, 2D, ou 3D.
- A ferramenta classifica artigos consoante quatro eixos: viés político, fiabilidade, objetividade e legibilidade.
- Todos os eixos podem ser escolhidos na secção das definições, bem como o seu intervalo de valores.
- Um filtro de fontes está também disponível para melhor análise.
- Na última parte da secção, um botão permite agrupar os dados e calcular a média por fonte, em vez dos dados de nível de artigo.
- O gráfico em si oferece duas funções interativas:
    - Dados específicos da notícia são mostrados com a passagem do rato por cima do respetivo ponto.
    - Ao clicar num ponto, o corpo do artigo é exibido numa janela pop-up.
- Por baixo do gráfico, um artigo pode ser adicionado para classificação em tempo real através de um ficheiro .txt ou upload de URL. Certifique-se de que qualquer artigo de uma URL é de uma fontes indicadas e não está bloqueado a subscrição.

### Classificação:

- A classificação de cada eixo é realizada usando um modelo de linguagem de grande escala (LLM).
- O modelo gera um descritor para cada artigo e para cada eixo.
- Exemplos de descritores podem ser encontrados nos dados de passagem dos pontos já presentes no gráfico.
- O descritor gerado para cada eixo é então comparado com uma base de dados de descritores obtidos a partir de um conjunto de dados etiquetados.
- Com base na similaridade com os descritores na base de dados, o valor do eixo é calculado para cada eixo de forma independente.
- Para o eixo de legibilidade, o valor é obtido usando métricas de legibilidade estabelecidas e mapeando-as para a escala do eixo.
"""


# Dash app
app = dash.Dash(__name__, title='Media Bias Chart', external_stylesheets=[dbc.themes.BOOTSTRAP])
server = app.server

data = pd.read_csv("./demo_data.csv")
data = data[["Título", "Texto", "Fonte", "Descritor de Viés Político", "Descritor de Fiabilidade", "Descritor de Objetividade", "Viés Político", "Fiabilidade", "Objetividade", "Legibilidade"]]

embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-mpnet-base-v2")
political_bias_db, reliability_db, objectivity_db = u.setup_db()

app.layout = html.Div([

    dcc.Store(id="data-store", data=data.to_dict("records"), storage_type="session"),
    html.Div([

        # dcc.Store(id='news-data', data=data.to_dict('records')),

        html.H2("Definições", style={'textAlign': 'center'}),

        # Dropdown for selecting chart type
        html.Label("Selecione o tipo de gráfico:"),
        dcc.Dropdown(
            id="chart-type",
            options=[
                {"label": "Gráfico de Dispersão 1D", "value": "1D"},
                {"label": "Gráfico de Dispersão 2D", "value": "2D"},
                {"label": "Gráfico de Dispersão 3D", "value": "3D"}
            ],
            value="2D",
            clearable=False,
            style={'marginBottom': '40px'}
        ),

        # Dropdowns for selecting axes
        html.Div([
            html.Label("Selecione o eixo X:"),
            dcc.Dropdown(
                id="x-axis",
                options=[{"label": col, "value": col} for col in data.columns[6:]],
                value="Viés Político",
                clearable=False,
                style={'marginBottom': '10px'},
            ),
            dcc.RangeSlider(
                id="x-filter",
                min=-100,
                max=100,
                step=5,
                marks={-100: '-100', -50: '-50', 0: '0', 50: '50', 100: '100'},
                value=[-100, 100],
                allowCross=False,
                className="range-slider"
            )
        ], id="x-axis-container", style={'marginBottom': '20px'}),

        html.Div([
            html.Label("Selecione o eixo Y:"),
            dcc.Dropdown(
                id="y-axis",
                options=[{"label": col, "value": col} for col in data.columns[6:]],
                value="Fiabilidade",
                clearable=False,
                style={'marginBottom': '10px'},
            ),
            dcc.RangeSlider(
                id="y-filter",
                min=-100,
                max=100,
                step=5,
                marks={-100: '-100', -50: '-50', 0: '0', 50: '50', 100: '100'},
                value=[-100, 100],
                allowCross=False,
                className="range-slider"
            )
        ], id="y-axis-container", style={'marginBottom': '20px'}),

        html.Div([
            html.Label("Selecione o eixo Z:"),
            dcc.Dropdown(
                id="z-axis",
                options=[{"label": col, "value": col} for col in data.columns[6:]],
                value="Objetividade",
                clearable=False,
                style={'marginBottom': '10px'},
            ),
            dcc.RangeSlider(
                id="z-filter",
                min=-100,
                max=100,
                step=5,
                marks={-100: '-100', -50: '-50', 0: '0', 50: '50', 100: '100'},
                value=[-100, 100],
                allowCross=False,
                className="range-slider"
            )
        ], id="z-axis-container", style={'marginBottom': '30px'}),

        # Checkbox filter for sources
        html.Label("Filtrar por Fonte:"),
        dcc.Dropdown(
            id="source-filter",
            options=([{"label": "Selecionar Todos", "value": "ALL"}] +
                    [{"label": src, "value": src} for src in data["Fonte"].unique()]),
            value=data["Fonte"].unique().tolist(),
            multi=True,
            placeholder="Selectione a fonte...",
            style={'font-family' : 'Arial', 'marginBottom': '20px'} 
        ),

        # Toggle button to group by source
        daq.ToggleSwitch(
            id="group-toggle",
            label="Agrupar por fonte (Média)",
            size=60
        ),
    ],
        style={
        'width': '25%', 'padding': '20px', 'backgroundColor': '#f8f9fa',
        'position': 'fixed', 'height': '100vh', 'overflowY': 'auto', 'font-family': 'Calibri'
        }
    ),

    html.Div([
        html.Div(
            dbc.Button("Informação", id="open-info", size="lg", n_clicks=0,
                       style={
                        "backgroundColor": "#E5ECF6",
                        "color": "black",
                        "border": "none",
                        # "font-weight": "bold"
                    }),
            className="d-flex justify-content-end"
        ),
    
        dbc.Modal(
            [
                # dbc.ModalHeader(dbc.ModalTitle("How to Use This App")),
                dbc.ModalBody(dcc.Markdown(info_text)),
                dbc.ModalFooter(
                    dbc.Button("Fechar", id="close-info", className="ms-auto", n_clicks=0)
                ),
            ],
            id="modal-info",
            is_open=False,
        ),

        # Graph
        html.H1("Classificação dos Media Portugueses", style={'textAlign': 'center', 'font-family': 'Calibri', 'font-weight': 'bold'}),
        dcc.Graph(id="news-plot", style={'height': '800px'}),

        dbc.Modal(
            [
                dbc.ModalHeader(dbc.ModalTitle(id="modal-title")),
                dbc.ModalBody(id='modal-body'),
                dbc.ModalFooter(
                    dbc.Button("Fechar", id="close", className="ms-auto", n_clicks=0)
                ),
            ],
            id="modal",
            is_open=False,
        ),

        dcc.Loading(
            id="upload-loading",
            type="circle",
            fullscreen=False,
            children=[

                # for article input in .txt
                dcc.Upload(
                    id='upload-article',
                    children=html.Div([
                        '📄 Arraste ou Selecione um Artigo para Classificar (formato .txt)'
                    ]),
                    style={
                        'width': '100%',
                        'height': '60px',
                        'lineHeight': '60px',
                        'borderWidth': '1px',
                        'borderStyle': 'dashed',
                        'borderRadius': '5px',
                        'textAlign': 'center',
                        'margin': '10px 0',
                        'font-family': 'Calibri',
                        'margin-bottom': '30px' 
                    },
                    accept='.txt'
                ),

                # for article input in url format
                html.Div([
                html.Label([
                    html.Span("Ou Insira um Link ", style={'font-size': '18px', 'font-family': 'Calibri'}),
                    html.Span("(Apenas para Expresso, Público, Eco Sapo, e Diário de Notícias)", style={'font-size': '12px', 'font-family': 'Calibri'})
                ]),
                dcc.Input(
                    id='url-input',
                    type='url',
                    placeholder='https://exemplo.com/artigo-notícias',
                    style={
                        'width': '98.3%',
                        'padding': '10px',
                        'margin': '10px 0',
                        'border': '1px solid #ccc',
                        'borderRadius': '5px',
                        'font-family': 'Calibri'
                        }
                    ),
                html.Button('Submeter Link', id='submit-url-button', n_clicks=0, style={
                    'margin': '10px 0',
                    'font-family': 'Calibri'
                    }),
                ]),

                html.Div(id='upload-feedback', style={'margin': '10px 0'}),
            ]
        )
    ],
    style={'marginLeft': '27%', 'padding': '20px'}
    #style={'width': '75%', 'padding': '20px', 'backgroundColor': '#f8f9fa', 'position': 'fixed', 'left': '25%', 'top': '0', 'bottom': '0', 'overflowY': 'auto'}
    )
])
   
# to update the chart dynamically
@app.callback(
    Output("news-plot", "figure", allow_duplicate=True),
    [Input("chart-type", "value"),
     Input("x-axis", "value"),
     Input("y-axis", "value"),
     Input("z-axis", "value"),
     Input("x-filter", "value"),
     Input("y-filter", "value"),
     Input("z-filter", "value"),
     Input("source-filter", "value"),
     Input("group-toggle", "value")],
     State("data-store", "data"),
     prevent_initial_call=True
)
def update_chart(chart_type, x_axis, y_axis, z_axis, x_range, y_range, z_range, selected_sources, group_toggle, data_records):

    data = pd.DataFrame(data_records)

    axis_extremities ={
        "Viés Político" : ["Enviesado à Esquerda", "Enviesado à Direita", -100, 100],
        "Fiabilidade" : ["Não fiável", "Fiável", -100, 100],
        "Objetividade" : ["Baseado em Opinião", "Factual/Objetivo", -100, 100],
        "Legibilidade" : ["Difícil de ler", "Fácil de ler", 0, 100]
    }

    all_sources = data["Fonte"].unique().tolist()
    if set(selected_sources) == set(all_sources) or not selected_sources:
        filtered_data = data
    else:
        filtered_data = data[data["Fonte"].isin(selected_sources)]

    if group_toggle:
        filtered_data_1 = filtered_data[filtered_data["Fonte"] != "Utilizador"].groupby("Fonte").mean().reset_index()
        filtered_data_2 = filtered_data[filtered_data["Fonte"] == "Utilizador"][["Fonte", "Viés Político", "Fiabilidade", "Objetividade", "Legibilidade"]]
        filtered_data = pd.concat([filtered_data_1, filtered_data_2], ignore_index=True)
        filtered_data["Título"] = filtered_data["Fonte"]

    # Apply range filters for each axis
    filtered_data = filtered_data[(filtered_data[x_axis] >= x_range[0]) & 
                                 (filtered_data[x_axis] <= x_range[1])]
    
    # Only apply Y and Z filters for the relevant chart types
    if chart_type in ["2D", "3D"]:
        filtered_data = filtered_data[(filtered_data[y_axis] >= y_range[0]) & 
                                     (filtered_data[y_axis] <= y_range[1])]
    
    if chart_type == "3D":
        filtered_data = filtered_data[(filtered_data[z_axis] >= z_range[0]) & 
                                     (filtered_data[z_axis] <= z_range[1])]

    if chart_type == "1D":

        to_hover = ["Fonte"]

        if not group_toggle:
            if x_axis != "Legibilidade":
                to_hover.append(f"Descritor de {x_axis}")

        # 1D plot
        filtered_data['Custom Size'] = filtered_data["Fonte"].apply(lambda x: 0 if x == "Utilizador" else 12)
        fig = px.scatter(filtered_data, x=x_axis, y=[0] * len(filtered_data), #text="Título",
                         # color="reliability score", # size="size_fixed",
                         color="Fonte",
                         color_continuous_scale="Viridis",
                         title=f"Dispersão 1D: {x_axis}",
                         hover_name="Título",
                         hover_data={col: True for col in to_hover} | {"Custom Size": False},
                         size='Custom Size',
                         size_max=10,
                         opacity=0.8)
        
        fig.update_yaxes(visible=False, showticklabels=False)  # Hide Y-axis
        # fig.update_traces(marker=dict(opacity=0.7, line=dict(width=1, color='black')))

        if group_toggle:
            fig.update_traces(marker=dict(opacity=0, size=0)) # to remove the original marker dots
            for i, row in filtered_data.iterrows():
                img_path = u.get_source_image(row["Fonte"])
                img_data = u.get_img_data(img_path)
                
                fig.add_layout_image(
                    dict(
                        source=f'data:image/png;base64,{img_data}',
                        x=row[x_axis],
                        y=0,
                        xref="x",
                        yref="y",
                        sizex=6,
                        sizey=6,
                        xanchor="center",
                        yanchor="middle",
                        opacity=0.8,
                    )
                )
        else:
            for _, row in filtered_data.iterrows():
                if row["Fonte"] == "Utilizador":
                    img_data = u.get_img_data("logos/x.png")
                    
                    fig.add_layout_image(
                        dict(
                            source=f'data:image/png;base64,{img_data}',
                            x=row[x_axis],
                            y=0,
                            xref="x",
                            yref="y",
                            sizex=7,
                            sizey=7,
                            xanchor="center",
                            yanchor="middle"
                        )
                    )
        
        # X-axis left annotation (outside)
        fig.add_annotation(
            x=0.13,   # Far left in paper coords
            y=0,   # Bottom
            xref="paper",
            yref="paper",
            text=f"⬅️ {axis_extremities[x_axis][0]}",
            showarrow=False,
            xanchor="right",
            yanchor="top",
            font=dict(size=12),
            yshift=-25  # Further outside the plot
        )

        # X-axis right annotation (outside)
        fig.add_annotation(
            x=0.87,   # Far right
            y=0,
            xref="paper",
            yref="paper",
            text=f"{axis_extremities[x_axis][1]} ➡️",
            showarrow=False,
            xanchor="left",
            yanchor="top",
            font=dict(size=12),
            yshift=-25
        )


        fig.update_layout(
            xaxis=dict(
                range=[axis_extremities[x_axis][2], axis_extremities[x_axis][3]]
            ))
    
    elif chart_type == "2D":

        to_hover = ["Fonte"]

        if not group_toggle:
            if x_axis != "Legibilidade":
                to_hover.append(f"Descritor de {x_axis}")
            if y_axis != "Legibilidade":
                to_hover.append(f"Descritor de {y_axis}")

        # 2D plot
        filtered_data['Custom Size'] = filtered_data["Fonte"].apply(lambda x: 0 if x == "Utilizador" else 10)
        fig = px.scatter(filtered_data, x=x_axis, y=y_axis, # text="Título",
            # color="reliability score", # size="size_fixed",
            color_continuous_scale="Viridis",
            color="Fonte",
            title=f"Dispersão 2D: {x_axis} vs {y_axis}",
            hover_name="Título",
            size='Custom Size',
            hover_data={col: True for col in to_hover} | {"Custom Size": False},
            size_max=10,
            opacity=0.8)

        if group_toggle:
            fig.update_traces(marker=dict(opacity=0, size=0)) # to remove the original marker dots
            for i, row in filtered_data.iterrows():
                img_path = u.get_source_image(row["Fonte"])
                img_data = u.get_img_data(img_path)
                
                fig.add_layout_image(
                    dict(
                        source=f'data:image/png;base64,{img_data}',
                        x=row[x_axis],
                        y=row[y_axis],
                        xref="x",
                        yref="y",
                        sizex=10,
                        sizey=10,
                        xanchor="center",
                        yanchor="middle",
                        opacity=0.8,
                    )
                )
        else:
            for _, row in filtered_data.iterrows():
                if row["Fonte"] == "Utilizador":
                    img_data = u.get_img_data("logos/x.png")
                    
                    fig.add_layout_image(
                        dict(
                            source=f'data:image/png;base64,{img_data}',
                            x=row[x_axis],
                            y=row[y_axis],
                            xref="x",
                            yref="y",
                            sizex=7,
                            sizey=7,
                            xanchor="center",
                            yanchor="middle"
                        )
                    )
        
        # X-axis left annotation (outside)
        fig.add_annotation(
            x=0.15,   # Far left in paper coords
            y=0,   # Bottom
            xref="paper",
            yref="paper",
            text=f"⬅️ {axis_extremities[x_axis][0]}",
            showarrow=False,
            xanchor="right",
            yanchor="top",
            font=dict(size=12),
            yshift=-25  # Further outside the plot
        )

        # X-axis right annotation (outside)
        fig.add_annotation(
            x=0.85,   # Far right
            y=0,
            xref="paper",
            yref="paper",
            text=f"{axis_extremities[x_axis][1]} ➡️",
            showarrow=False,
            xanchor="left",
            yanchor="top",
            font=dict(size=12),
            yshift=-25
        )

        # Y-axis bottom annotation (outside)
        fig.add_annotation(
            x=0,
            y=0,   # Bottom
            xref="paper",
            yref="paper",
            text=f"⬅️ {axis_extremities[y_axis][0]}",
            showarrow=False,
            xanchor="right",
            yanchor="bottom",
            font=dict(size=12),
            xshift=-40,  # Move left outside
            textangle=-90
        )

        # Y-axis top annotation (outside)
        fig.add_annotation(
            x=0,
            y=1,   # Top
            xref="paper",
            yref="paper",
            text=f"{axis_extremities[y_axis][1]} ➡️",
            showarrow=False,
            xanchor="right",
            yanchor="top",
            font=dict(size=12),
            xshift=-40,
            textangle=-90
        )

        fig.update_layout(
            xaxis=dict(
                range=[axis_extremities[x_axis][2], axis_extremities[x_axis][3]]
            ),
            yaxis=dict(
                range=[axis_extremities[y_axis][2], axis_extremities[y_axis][3]]
            )
        )

    elif chart_type == "3D":

        to_hover = ["Fonte"]

        if not group_toggle:
            if x_axis != "Legibilidade":
                to_hover.append(f"Descritor de {x_axis}")
            if y_axis != "Legibilidade":
                to_hover.append(f"Descritor de {y_axis}")
            if z_axis != "Legibilidade":
                to_hover.append(f"Descritor de {z_axis}")
        
        user_upload = "Utilizador"
        highlight_df = filtered_data[filtered_data["Fonte"] == user_upload]
        other_df = filtered_data[filtered_data["Fonte"] != user_upload]

        source_list = other_df["Fonte"].unique()
        color_map = {source: f"hsl({i * 360 / len(source_list)}, 70%, 50%)" for i, source in enumerate(source_list)}
        color_array = other_df["Fonte"].map(color_map)

        fig = go.Figure()

        # 3D plot
        if group_toggle:
            fig = px.scatter_3d(other_df, x=x_axis, y=y_axis, z=z_axis,
                color="Fonte",  # Color points by source
                color_continuous_scale="Viridis",
                # symbol="Fonte",
                title=f"Dispersão 3D: {x_axis} vs {y_axis} vs {z_axis}",
                hover_name='Título')
    
            fig.update_traces(marker=dict(size=10, opacity=0.8, line=dict(width=1, color='black')))
            
            highlight_trace = px.scatter_3d(highlight_df, x=x_axis, y=y_axis, z=z_axis,
                                            color_discrete_sequence=["black"],
                                            hover_name="Título")

            highlight_trace.update_traces(
                marker=dict(size=10),
                name="Utilizador",
                showlegend=True)
            for trace in highlight_trace.data:
                fig.add_trace(trace)

        else:
            fig = px.scatter_3d(other_df, x=x_axis, y=y_axis, z=z_axis, 
                    color="Fonte",
                    color_continuous_scale="Viridis",
                    title=f"Dispersão 3D: {x_axis} vs {y_axis} vs {z_axis}",
                    hover_name="Título",
                    hover_data=to_hover)
            fig.update_traces(marker=dict(size=5, opacity=0.8))

            highlight_trace = px.scatter_3d(highlight_df, x=x_axis, y=y_axis, z=z_axis,
                                            color_discrete_sequence=["black"],
                                            hover_name="Título",
                                            hover_data=to_hover)

            highlight_trace.update_traces(
                marker=dict(size=10),
                name="Utilizador",
                showlegend=True)
            for trace in highlight_trace.data:
                fig.add_trace(trace)

        fig.update_layout(
            scene=dict(
                xaxis=dict(
                    title=x_axis,
                    range=[axis_extremities[x_axis][2], axis_extremities[x_axis][3]]
                ),
                yaxis=dict(
                    title=y_axis,
                    range=[axis_extremities[y_axis][2], axis_extremities[y_axis][3]]
                ),
                zaxis=dict(
                    title=z_axis,
                    range=[axis_extremities[z_axis][2], axis_extremities[z_axis][3]]
                )
            )
        )
        
    return fig

# # To update the sources in the source filter
# @app.callback(
#     Output("source-filter", "options"),
#     Output("source-filter", "value"),
#     Input("news-data", "data")
# )
# def update_source_dropdown(data):

#     df = pd.DataFrame(data)
#     unique_sources = df["Fonte"].unique().tolist()

#     options = [{"label": "Select All", "value": "ALL"}] + [
#         {"label": src, "value": src} for src in unique_sources
#     ]

#     return options, unique_sources

# To update the source filter
@app.callback(
    Output("source-filter", "value", allow_duplicate=True),
    [Input("source-filter", "value")],
    State("source-filter", "options"),
    prevent_initial_call=True
)
def update_source_selection(selected_sources, options):
    all_sources = data["Fonte"].unique().tolist()

    if "ALL" in selected_sources:
        # If "Select All" is clicked, return all sources
        if "Utilizador" in [src["value"] for src in options]:
            return all_sources + ["Utilizador"]
        return all_sources

    else:
        # else return selected sources normally
        return selected_sources

# To updated the graph size based on chart type
@app.callback(
    Output("news-plot", "style"),
    Input("chart-type", "value")
)
def update_graph_height(chart_type):
    if chart_type == "1D":
        return {"height": "400px"}
    elif chart_type == "2D":
        return {"height": "800px"}
    elif chart_type == "3D":
        return {"height": "1000px"}
    return {"height": "800px"}

# To disable Y-axis and Z-axis dropdowns based on chart type
@app.callback(
    [Output("y-axis", "disabled"),
     Output("z-axis", "disabled")],
    [Input("chart-type", "value")]
)
def update_dropdown_states(chart_type):
    if chart_type == "1D":
        return True, True
    elif chart_type == "2D":
        return False, True
    else:
        return False, False

# To hide Y-axis and Z-axis dropdowns based on chart type
@app.callback(
    [Output("y-axis-container", "style"),
     Output("z-axis-container", "style")],
    [Input("chart-type", "value")]
)
def update_dropdown_visibility(chart_type):
    base_style = {'marginBottom': '10px'}
    disabled_style = {'marginBottom': '10px', 'opacity': '0.5'}
    
    if chart_type == "1D":
        return disabled_style, disabled_style
    
    elif chart_type == "2D":
        return base_style, disabled_style
    
    else:
        return base_style, base_style
    
# show/hide filters based on chart type
@app.callback(
    [Output("x-filter", "disabled"),
     Output("y-filter", "disabled"),
     Output("z-filter", "disabled")],
    [Input("chart-type", "value"),
     Input("x-axis", "value"),
     Input("y-axis", "value"),
     Input("z-axis", "value")]
)
def update_filter_availability(chart_type, x_axis, y_axis, z_axis):

    x_disabled = False
    y_disabled = chart_type == "1D"
    z_disabled = chart_type != "3D"
    
    return x_disabled, y_disabled, z_disabled

# To update the range sliders based on selected axis
@app.callback(
    [Output("x-filter", "min"), Output("x-filter", "max"), 
     Output("x-filter", "marks"), Output("x-filter", "value"),
     Output("y-filter", "min"), Output("y-filter", "max"), 
     Output("y-filter", "marks"), Output("y-filter", "value"),
     Output("z-filter", "min"), Output("z-filter", "max"), 
     Output("z-filter", "marks"), Output("z-filter", "value")],
    [Input("x-axis", "value"), Input("y-axis", "value"), Input("z-axis", "value")]
)
def update_range_sliders(x_axis, y_axis, z_axis):
    axis_data = {
        "Viés Político": [-100, 100, {-100: "-100", -50: "-50", 0: "0", 50: "50", 100: "100"}],
        "Fiabilidade": [-100, 100, {-100: "-100", -50: "-50", 0: "0", 50: "50", 100: "100"}],
        "Objetividade": [-100, 100, {-100: "-100", -50: "-50", 0: "0", 50: "50", 100: "100"}],
        "Legibilidade": [0, 100, {0: "0", 25: "25", 50: "50", 75: "75", 100: "100"}]
    }
    
    x_min, x_max, x_marks = axis_data[x_axis]
    y_min, y_max, y_marks = axis_data[y_axis]
    z_min, z_max, z_marks = axis_data[z_axis]
    
    return (x_min, x_max, x_marks, [x_min, x_max],
            y_min, y_max, y_marks, [y_min, y_max],
            z_min, z_max, z_marks, [z_min, z_max])

@app.callback(
    Output('upload-feedback', 'children', allow_duplicate=True),
    Output("news-plot", "figure", allow_duplicate=True),
    Output("source-filter", "options", allow_duplicate=True),
    Output("source-filter", "value", allow_duplicate=True),
    Output("data-store", "data", allow_duplicate=True), 
    Input('upload-article', 'contents'),
    State('upload-article', 'filename'),
    State("source-filter", "options"),
    State("source-filter", "value"),
    State("data-store", "data"),
    prevent_initial_call=True
)
def classify_article(contents, filename, options, selected_sources, data_records):
    data = pd.DataFrame(data_records)
    if contents is None:
        raise PreventUpdate

    # Decode .txt content
    content_type, content_string = contents.split(',')
    decoded = u.decode_text_file(content_string)

    openai = OpenAI(
        api_key=os.environ.get("API_KEY"),
        base_url="https://api.deepinfra.com/v1/openai",
    )

    try:
        # political_bias_db, reliability_db, objectivity_db = u.setup_db()

        descriptor_political_bias = u.get_descriptor(decoded, "political_bias", u.prompts, openai)
        descriptor_reliability = u.get_descriptor(decoded, "reliability", u.prompts, openai)
        descriptor_objectivity = u.get_descriptor(decoded, "objectivity", u.prompts, openai)

        political_bias_score = u.get_score(descriptor_political_bias, embedding_model, 0.5, political_bias_db, 50)
        reliability_score = u.get_score(descriptor_reliability, embedding_model, 0.2, reliability_db, 7505)
        objectivity_score = u.get_score(descriptor_objectivity, embedding_model, 0.2, objectivity_db, 9000)
        reliability_score = u.classify_readability(decoded)

        new_point = {
            "Título": filename,
            "Texto": decoded,
            "Fonte": "Utilizador",
            "Descritor de Viés Político": descriptor_political_bias,
            "Descritor de Fiabilidade": descriptor_reliability,
            "Descritor de Objetividade": descriptor_objectivity,
            "Viés Político": political_bias_score,
            "Fiabilidade": reliability_score,
            "Objetividade": objectivity_score,
            "Legibilidade": reliability_score
        }

        data = pd.concat([pd.DataFrame(data), pd.DataFrame([new_point])], ignore_index=True)

        unique_sources = selected_sources + [new_point["Fonte"]]
        options = options + [{"label": new_point["Fonte"], "value": new_point["Fonte"]}]

        return f"✅ Classificado e adicionado '{filename}'", dash.no_update, options, unique_sources, data.to_dict("records")


    except Exception as e:

        return f"❌ Erro a classificar o artigo: {e}", dash.no_update, dash.no_update, dash.no_update, dash.no_update

@app.callback(
    Output('upload-feedback', 'children'),
    Output("news-plot", "figure"),
    Output("source-filter", "options"),
    Output("source-filter", "value"),
    Output("data-store", "data"), 
    Input('submit-url-button', 'n_clicks'),
    State('url-input', 'value'),
    State("source-filter", "options"),
    State("source-filter", "value"),
    State("data-store", "data")
)
def classify_url(n_clicks, url, options, selected_sources, data_records):
    data = pd.DataFrame(data_records)
    if n_clicks > 0 and url:

        res = requests.get(url)
        soup = bs4.BeautifulSoup(res.text, 'lxml')

        if "expresso.pt" in url:
            outlet = "Expresso"
            title = soup.select('h1')[0].text
            text = soup.select('.full-article-fragment.full-article-body.article-content.first')[0].getText()
        
        elif "publico.pt" in url:
            outlet = "Público"
            title = soup.select('.headline.story__headline')[0].getText()
            match = re.search(r'\s*(.*?)\s*$', title.strip())
            if match:
                title = match.group(1)
            text = soup.select('.story__body')[0].getText()
        
        elif "eco.sapo.pt" in url:
            outlet = "Eco Sapo"
            title = soup.select('.title')[0].get_text()
            title = re.sub(r'\s+', ' ', title).strip()
            text = soup.select('.entry__content')[0].get_text()
        
        elif "dn.pt" in url:
            outlet = "Diário de Notícias"
            title = soup.select('.arrow-component.arr--story-headline.story-headline-m_wrapper__1Wey6')[0].getText()
            text = soup.select('.arr--story-page-card-wrapper')[0].getText() 

        openai = OpenAI(
            api_key=os.environ.get("API_KEY"),
            base_url="https://api.deepinfra.com/v1/openai",
        )

        try:
            # political_bias_db, reliability_db, objectivity_db = u.setup_db()

            descriptor_political_bias = u.get_descriptor(text, "political_bias", u.prompts, openai)
            descriptor_reliability = u.get_descriptor(text, "reliability", u.prompts, openai)
            descriptor_objectivity = u.get_descriptor(text, "objectivity", u.prompts, openai)

            political_bias_score = u.get_score(descriptor_political_bias, embedding_model, 0.5, political_bias_db, 50)
            reliability_score = u.get_score(descriptor_reliability, embedding_model, 0.2, reliability_db, 7505)
            objectivity_score = u.get_score(descriptor_objectivity, embedding_model, 0.2, objectivity_db, 9000)
            reliability_score = u.classify_readability(text)
            
            new_point = {
                "Título": f"{outlet}: {title}",
                "Texto": text,
                "Fonte": "Utilizador",
                "Descritor de Viés Político": descriptor_political_bias,
                "Descritor de Fiabilidade": descriptor_reliability,
                "Descritor de Objetividade": descriptor_objectivity,
                "Viés Político": political_bias_score,
                "Fiabilidade": reliability_score,
                "Objetividade": objectivity_score,
                "Legibilidade": reliability_score
            }

            data = pd.concat([pd.DataFrame(data), pd.DataFrame([new_point])], ignore_index=True)

            unique_sources = selected_sources + [new_point["Fonte"]]
            options = options + [{"label": new_point["Fonte"], "value": new_point["Fonte"]}]

            return f"✅ O artigo pedido foi classificao e adicionado.", dash.no_update, options, unique_sources, data.to_dict("records")


        except Exception as e:
            
            return f"❌ Erro a classificar o artigo: {e}", dash.no_update, dash.no_update, dash.no_update, dash.no_update
    
    else:
        raise PreventUpdate

# to open window with article text
@app.callback(
    Output("modal", "is_open"),
    Output("modal-body", "children"),
    Output("modal-title", "children"),
    Input("news-plot", "clickData"),
    Input("close", "n_clicks"),
    State("modal", "is_open"),
    State("data-store", "data")
)
def display_modal(clickData, close_clicks, is_open, data_records):
    data = pd.DataFrame(data_records)
    ctx = dash.callback_context

    if ctx.triggered_id == "news-plot" and clickData:
        title = clickData['points'][0]['hovertext']
        data_index = data[data["Título"] == title].index
        news_text = data.loc[data_index, 'Texto']
        return True, news_text, title
    elif ctx.triggered_id == "close" and is_open:
        return False, None, None
    return is_open, None, None

# to open informational window
@app.callback(
    Output("modal-info", "is_open"),
    [Input("open-info", "n_clicks"), Input("close-info", "n_clicks")],
    [State("modal-info", "is_open")],
)
def toggle_modal(n1, n2, is_open):
    if n1 or n2:
        return not is_open
    return is_open

if __name__ == '__main__':
    app.run(host="0.0.0.0", port=7860)