#app.py
import gradio as gr
import pandas as pd
import os
import logging
from collections import defaultdict
import matplotlib
matplotlib.use('Agg')

# --- Module Imports ---
from utils.gradio_utils import get_url_user_token
from config import (
    PLOT_ID_TO_FORMULA_KEY_MAP,
    LINKEDIN_CLIENT_ID_ENV_VAR,
    BUBBLE_APP_NAME_ENV_VAR,
    BUBBLE_API_KEY_PRIVATE_ENV_VAR,
    BUBBLE_API_ENDPOINT_ENV_VAR
)

# REMOVED: from services.analytics_tab_module import AnalyticsTab
from services.state_manager import load_data_from_bubble
from ui.ui_generators import (
    # REMOVED: build_analytics_tab_plot_area,
    build_dynamic_home_tab_ui,
    create_enhanced_report_tab,
    BOMB_ICON, EXPLORE_ICON, FORMULA_ICON, ACTIVE_ICON
)

from services.home_tab_module import refresh_home_tab_ui
from ui.config import custom_title_css
from ui.okr_ui_generator import create_enhanced_okr_tab, format_okrs_for_enhanced_display, get_initial_okr_display
# REMOVED: from ui.analytics_plot_generator import update_analytics_plots_figures, create_placeholder_plot
# REMOVED: from formulas import PLOT_FORMULAS

from config import GRAPH_GROUPS, GRAPH_GROUP_TITLES, GRAPH_GROUP_DESCRIPTIONS
from services.sidebar_graphs_module import generate_sidebar_graphs_for_group, _create_empty_plotly_figure
from data_processing.analytics_data_processing import prepare_filtered_analytics_data

# REMOVED: from features.chatbot.chatbot_prompts import get_initial_insight_prompt_and_suggestions
# REMOVED: from features.chatbot.chatbot_handler import generate_llm_response

try:
    from run_agentic_pipeline import load_and_display_agentic_results
    from services.report_data_handler import fetch_and_reconstruct_data_from_bubble
    from ui.insights_ui_generator import format_report_for_display
    AGENTIC_MODULES_LOADED = True
except ImportError as e:
    logging.error(f"Could not import agentic modules: {e}")
    AGENTIC_MODULES_LOADED = False
    
    def load_and_display_agentic_results(*args, **kwargs):
        empty_header_html = """
        <div class="report-title">📝 Comprehensive Analysis Report</div>
        <div class="report-subtitle">AI-Generated Insights</div>
        """
        empty_body_markdown = """
        <div class="empty-state">
            <div class="empty-state-icon">📄</div>
            <div class="empty-state-title">No Report Selected</div>
        </div>
        """
        return (
            gr.update(value="Modules not loaded."),
            gr.update(choices=[], value=None),
            gr.update(choices=[], value=[]),
            gr.update(value="Modules not loaded."),
            None,
            [],
            [],
            gr.update(value=empty_header_html),
            gr.update(value=empty_body_markdown),
            {},
            gr.update(value=get_initial_okr_display()),
            gr.update(value={})
        )
    
    def fetch_and_reconstruct_data_from_bubble(*args, **kwargs):
        return None, {}
    
    def format_report_for_display(report_data):
        return {'header_html': '<h1>Modules not loaded.</h1>', 'body_markdown': 'Unavailable.'}


with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue", secondary_hue="sky"),
                     title="LinkedIn Organization Dashboard") as app:
    # --- STATE MANAGEMENT ---
    token_state = gr.State(value={
        "token": None, "client_id": None, "org_urn": None,
        "bubble_posts_df": pd.DataFrame(), "bubble_post_stats_df": pd.DataFrame(),
        "bubble_mentions_df": pd.DataFrame(), "bubble_follower_stats_df": pd.DataFrame(),
        "bubble_agentic_analysis_data": pd.DataFrame(),
        "url_user_token_temp_storage": None,
        "config_date_col_posts": "published_at", "config_date_col_mentions": "date",
        "config_date_col_followers": "date", "config_media_type_col": "media_type",
        "config_eb_labels_col": "li_eb_label"
    })

    # REMOVED: chat_histories_st = gr.State({})
    # REMOVED: current_chat_plot_id_st = gr.State(None)
    # REMOVED: plot_data_for_chatbot_st = gr.State({})

    orchestration_raw_results_st = gr.State(None)
    key_results_for_selection_st = gr.State([])
    selected_key_result_ids_st = gr.State([])
    reconstruction_cache_st = gr.State({})
    actionable_okrs_data_st = gr.State({})
    active_graph_group_st = gr.State(None)

    # --- UI LAYOUT ---
    url_user_token_display = gr.Textbox(label="User Token", interactive=False, visible=False)
    org_urn_display = gr.Textbox(label="Org URN", interactive=False, visible=False)
    status_box = gr.Textbox(label="Status", interactive=False, value="", visible=False)

    app.load(fn=get_url_user_token, inputs=None, outputs=[url_user_token_display, org_urn_display], 
             api_name="get_url_params", show_progress=False)
    app.load(
        fn=None,
        js="""
        const processedButtons = new Set();
        function setAllTooltips() {
            const bombBtns = document.querySelectorAll('.analytics-bomb-btn button');
            bombBtns.forEach(btn => {
                if (!processedButtons.has(btn)) {
                    btn.setAttribute('title', 'Analizza il grafico con l\\'intelligenza artificiale.');
                    processedButtons.add(btn);
                }
            });
            const formulaBtns = document.querySelectorAll('.analytics-formula-btn button');
            formulaBtns.forEach(btn => {
                if (!processedButtons.has(btn)) {
                    btn.setAttribute('title', 'Scopri come viene calcolato il dato.');
                    processedButtons.add(btn);
                }
            });
            const exploreBtns = document.querySelectorAll('.analytics-explore-btn button');
            exploreBtns.forEach(btn => {
                if (!processedButtons.has(btn)) {
                    btn.setAttribute('title', 'Ingrandisci il grafico per un\\'analisi dettagliata.');
                    processedButtons.add(btn);
                }
            });
        }
        setInterval(setAllTooltips, 250);
        """
    )

    def initial_data_load_sequence(url_token, org_urn_val, current_state):
        """Handles initial data loading from Bubble."""
        status_msg, new_state = load_data_from_bubble(url_token, org_urn_val, current_state)

        show_status_box = False
        if new_state.get("token") is None:
            show_status_box = True
        elif "Error" in status_msg or "Warning" in status_msg:
            show_status_box = True
        
        return gr.update(value=status_msg, visible=show_status_box), new_state

    # REMOVED: analytics_icons and analytics_tab_instance initialization

    def update_report_display(selected_report_id: str, current_token_state: dict):
        """Updates report display when a new report is selected."""
        empty_header_html = """
        <div class="report-title">📝 Comprehensive Analysis Report</div>
        <div class="report-subtitle">AI-Generated Insights</div>
        """
        empty_body_markdown_no_selection = """
        <div class="empty-state">
            <div class="empty-state-icon">📋</div>
            <div class="empty-state-title">Select a Report</div>
        </div>
        """

        if not selected_report_id:
            return gr.update(value=empty_header_html), gr.update(value=empty_body_markdown_no_selection)

        agentic_df = current_token_state.get("bubble_agentic_analysis_data")
        if agentic_df is None or agentic_df.empty:
            return gr.update(value=empty_header_html), gr.update(value="<div class='empty-state'>No data</div>")

        selected_report_series_df = agentic_df[agentic_df['_id'] == selected_report_id]
        if selected_report_series_df.empty:
            return gr.update(value=empty_header_html), gr.update(value="<div class='empty-state'>Not found</div>")

        selected_report_series = selected_report_series_df.iloc[0]
        formatted_content_parts = format_report_for_display(selected_report_series)

        return (
            gr.update(value=formatted_content_parts['header_html']),
            gr.update(value=formatted_content_parts['body_markdown'])
        )

    def handle_home_filter_change(token_state_value, date_filter_option, custom_start_date, custom_end_date):
        """Combined function to handle home date filter changes."""
        if date_filter_option == "Personalizza":
            custom_row_update = gr.update(visible=True)
        else:
            custom_row_update = gr.update(visible=False)
        
        kpi_updates = refresh_home_tab_ui(
            token_state_value, 
            date_filter_option, 
            custom_start_date, 
            custom_end_date
        )
        
        return (custom_row_update,) + kpi_updates

    def prepare_table_for_group(graph_group: str, token_state_value: dict):
        """
        Prepara la tabella sidebar con i dati base del gruppo.
        """
        try:
            logging.info(f"Preparing sidebar table for group: {graph_group}")
            
            (filtered_posts_df, 
             filtered_mentions_df, 
             filtered_comments_df, 
             _, _, _, _) = prepare_filtered_analytics_data(
                token_state_value, "Sempre", None, None
            )
    
            if graph_group == "posts":
                if filtered_posts_df.empty:
                    return gr.update(
                        value=pd.DataFrame({"Info": ["Nessun post disponibile"]}),
                        visible=True,
                        headers=["Data", "Post", "Likes", "Comments", "Shares", "Impressions"]
                    )
                    
                df = filtered_posts_df.copy()
                df['date'] = pd.to_datetime(df['published_at'], errors='coerce')
                df = df.dropna(subset=['date']).sort_values('date', ascending=False).head(10)
                
                if 'text' not in df.columns:
                    df['text'] = "N/A"
                    
                output_df = pd.DataFrame({
                    "Data": df['date'].dt.strftime('%Y-%m-%d'),
                    "Post link": df['id'].apply(lambda post_id: f'<a href="https://www.linkedin.com/feed/update/{post_id}" target="_blank" style="color: #3b82f6; text-decoration: none;">Post Link</a>'),
                    "Post": df['text'].astype(str).str[:80] + "...",
                    "Likes": pd.to_numeric(df.get('likeCount', 0), errors='coerce').fillna(0).astype(int),
                    "Comments": pd.to_numeric(df.get('commentCount', 0), errors='coerce').fillna(0).astype(int),
                    "Shares": pd.to_numeric(df.get('shareCount', 0), errors='coerce').fillna(0).astype(int),
                    "Impressions": pd.to_numeric(df.get('impressionCount', 0), errors='coerce').fillna(0).astype(int)
                })
                
                logging.info(f"✅ Table populated with {len(output_df)} posts")
                return gr.update(
                    value=output_df,
                    visible=True,
                    headers=["Data", "Post link", "Post", "Likes", "Comments", "Shares", "Impressions"],
                    datatype=["date", "markdown", "str", "number", "number", "number", "number"],
                    column_widths=["12%", "12%", "30%", "7%", "7%", "7%", "7%"]
                )
                
            elif graph_group == "sentiment":
                mentions_list = []
                if not filtered_mentions_df.empty and 'sentiment_label' in filtered_mentions_df.columns:
                    mentions_df = filtered_mentions_df.copy()
                    mentions_df['date'] = pd.to_datetime(mentions_df['date'], errors='coerce')
                    mentions_df = mentions_df.dropna(subset=['date']).sort_values('date', ascending=False).head(25)
                    
                    mentions_list = [{
                        "Data": row['date'].strftime('%Y-%m-%d'),
                        "Contenuto": str(row.get('mention_text', 'N/A'))[:80] + "...",
                        "Tipo": "Menzione",
                        "Sentiment": row.get('sentiment_label', 'N/A')
                    } for _, row in mentions_df.iterrows()]
                
                comments_list = []
                if not filtered_comments_df.empty and not filtered_posts_df.empty:
                    comments_df = filtered_comments_df.copy()
                    
                    posts_dates_df = filtered_posts_df[['id', 'published_at']].copy()
                    posts_dates_df['date'] = pd.to_datetime(posts_dates_df['published_at'], errors='coerce')
                    
                    comments_with_dates = pd.merge(
                        comments_df,
                        posts_dates_df[['id', 'date']],
                        left_on='post_id',
                        right_on='id',
                        how='left'
                    )
                    
                    comments_df_sorted = comments_with_dates.dropna(subset=['date']).sort_values('date', ascending=False).head(25)
    
                    sentiment_col = 'sentiment_label' if 'sentiment_label' in comments_df_sorted.columns else 'sentiment'
                    
                    for _, row in comments_df_sorted.iterrows():
                        content_text = str(row.get('comment_text', 'N/A'))[:80] + "..."
                        
                        comments_list.append({
                            "Data": row['date'].strftime('%Y-%m-%d'),
                            "Contenuto": content_text,
                            "Tipo": "Comment",
                            "Sentiment": row.get(sentiment_col, 'N/A')
                        })
                
                combined = mentions_list + comments_list
                if not combined:
                    return gr.update(
                        value=pd.DataFrame({"Info": ["Nessun dato sentiment disponibile"]}),
                        visible=True
                    )
                
                combined_sorted = sorted(combined, key=lambda x: x['Data'], reverse=True)
                
                output_df = pd.DataFrame(combined_sorted).head(50)
                logging.info(f"✅ Table populated with {len(output_df)} sentiment records")
                return gr.update(
                    value=output_df,
                    visible=True,
                    headers=["Data", "Contenuto", "Tipo", "Sentiment"],
                    datatype=["str", "str", "str", "str"],
                    column_widths=["20%", "40%", "15%", "15%"]
                )
                
            else:
                logging.info(f"Table hidden for group: {graph_group}")
                return gr.update(visible=False)
                
        except Exception as e:
            logging.error(f"Error preparing table: {e}", exc_info=True)
            return gr.update(
                value=pd.DataFrame({"Errore": [f"Errore: {str(e)}"]}),
                visible=True
            )
        
    def open_analysis_page(graph_group: str, current_token_state):
        """
        Apre la pagina dettagliata rendendo visibile il Column overlay
        e nascondendo il contenuto principale della Home.
        """
        try:
            date_filter = "Sempre"
            logging.info(f"Opening analysis page for group '{graph_group}'")
            
            (filtered_posts_df, filtered_mentions_df, filtered_comments_df,
             filtered_follower_df, raw_follower_df, _, _) = prepare_filtered_analytics_data(
                current_token_state, date_filter, None, None
            )
            
            figures = generate_sidebar_graphs_for_group(
                graph_group,
                filtered_posts_df,
                filtered_mentions_df,
                filtered_comments_df,
                filtered_follower_df,
                raw_follower_df,
                current_token_state
            )
            
            logging.info(f"Generated {len(figures)} figures for group '{graph_group}'")
            
            figure_updates = []
            for i in range(6):
                if i < len(figures):
                    figure_updates.append(gr.update(value=figures[i], visible=True))
                else:
                    figure_updates.append(gr.update(value=None, visible=False))
            
            is_posts_group = (graph_group == "posts")
            col2_visibility_update = gr.update(visible=not is_posts_group)
            
            table_update = prepare_table_for_group(graph_group, current_token_state)
            
            analysis_page_visibility = gr.update(visible=True)
            home_content_visibility = gr.update(visible=False)
            
            logging.info(f"✅ Analysis page opened successfully for '{graph_group}'")
            return (
                analysis_page_visibility,
                *figure_updates,
                graph_group,
                col2_visibility_update,
                table_update,
                home_content_visibility
            )
            
        except Exception as e:
            logging.error(f"Error opening analysis page: {e}", exc_info=True)
            empty_updates = [gr.update(value=None, visible=False) for _ in range(6)]
            return (
                gr.update(visible=True),
                *empty_updates,
                None,
                gr.update(visible=True),
                gr.update(visible=False),
                gr.update(visible=False)
            )

    def close_analysis_page():
        """
        Chiude la pagina dettagliata nascondendo il Column overlay
        e ripristinando la visibilità del contenuto Home.
        """
        return (
            gr.update(visible=False),
            gr.update(visible=True)
        )

    with gr.Tabs() as tabs:
        with gr.TabItem("🏠 Home", id="tab_home"):
            (
                home_content_col,
                home_date_filter_selector, home_custom_dates_row, home_custom_start_date, home_custom_end_date, home_apply_filter_btn,
                home_kpi_new_followers, home_kpi_growth_rate, home_kpi_brand_sentiment,
                home_perf_engagement_plot, home_perf_detail_btn,
                home_kpi_top_topics, home_kpi_top_formats,
                home_kpi_follower_persona,
                analysis_page_container,
                home_sidebar_graph1, home_sidebar_graph2, home_sidebar_graph3,
                home_sidebar_graph4, home_sidebar_graph5, home_sidebar_graph6,
                home_sidebar_column_2,
                home_sidebar_close_btn,
                home_follower_detail_btn, home_sentiment_detail_btn,
                home_sidebar_selection_table
            ) = build_dynamic_home_tab_ui()
            
            home_refresh_outputs_list = [
                home_kpi_new_followers, home_kpi_growth_rate, home_kpi_brand_sentiment,
                home_perf_engagement_plot, home_kpi_top_topics, home_kpi_top_formats,
                home_kpi_follower_persona
            ]

            home_change_handler_outputs = [home_custom_dates_row] + home_refresh_outputs_list

            sidebar_outputs = [
                analysis_page_container,
                home_sidebar_graph1, home_sidebar_graph2, 
                home_sidebar_graph3, home_sidebar_graph4,
                home_sidebar_graph5, home_sidebar_graph6,
                active_graph_group_st,
                home_sidebar_column_2,
                home_sidebar_selection_table,
                home_content_col
            ]
            
            home_perf_detail_btn.click(
                fn=lambda ts: open_analysis_page("posts", ts),
                inputs=[token_state],
                outputs=sidebar_outputs
            )
            
            home_follower_detail_btn.click(
                fn=lambda ts: open_analysis_page("followers", ts),
                inputs=[token_state],
                outputs=sidebar_outputs
            )
            
            home_sentiment_detail_btn.click(
                fn=lambda ts: open_analysis_page("sentiment", ts),
                inputs=[token_state],
                outputs=sidebar_outputs
            )
            
            home_sidebar_close_btn.click(
                fn=close_analysis_page,
                outputs=[analysis_page_container, home_content_col]
            )
            
            home_date_filter_selector.change(
                fn=handle_home_filter_change,
                inputs=[token_state, home_date_filter_selector, home_custom_start_date, home_custom_end_date],
                outputs=home_change_handler_outputs,
                show_progress="full"
            )
            
            home_apply_filter_btn.click(
                fn=refresh_home_tab_ui,
                inputs=[token_state, home_date_filter_selector, home_custom_start_date, home_custom_end_date],
                outputs=home_refresh_outputs_list,
                show_progress="full"
            )

        # REMOVED: analytics_tab_instance.create_tab_ui()
        
        with gr.TabItem("📝 AI Analysis Reports", id="tab_agentic_report", visible=AGENTIC_MODULES_LOADED):
            agentic_pipeline_status_md, report_selector_dd, report_header_html_display, report_body_markdown_display = \
                create_enhanced_report_tab(AGENTIC_MODULES_LOADED)

        with gr.TabItem("🎯 OKRs & Action Items", id="tab_agentic_okrs", visible=AGENTIC_MODULES_LOADED):
            gr.Markdown("## 🎯 AI Generated OKRs")
            
            if not AGENTIC_MODULES_LOADED:
                gr.Markdown("🔴 **Error:** Agentic modules not loaded.")

            with gr.Column(visible=False):
                key_results_cbg = gr.CheckboxGroup(label="Select Key Results", choices=[], value=[])
                okr_detail_display_md = gr.Markdown("Details will appear here.")

            enhanced_okr_display_html = create_enhanced_okr_tab()
            
    if AGENTIC_MODULES_LOADED:
        report_selector_dd.change(
            fn=update_report_display,
            inputs=[report_selector_dd, token_state],
            outputs=[report_header_html_display, report_body_markdown_display],
            show_progress="minimal"
        )

    agentic_display_outputs = [
        agentic_pipeline_status_md,
        report_selector_dd,
        key_results_cbg,
        okr_detail_display_md,
        orchestration_raw_results_st,
        selected_key_result_ids_st,
        key_results_for_selection_st,
        report_header_html_display,
        report_body_markdown_display,
        reconstruction_cache_st,
        enhanced_okr_display_html,
        actionable_okrs_data_st
    ]

    # REFACTORED: Simplified event chain without analytics_load_event
    initial_load_event = org_urn_display.change(
        fn=lambda: gr.update(value="Loading data...", visible=True),
        inputs=[],
        outputs=[status_box],
        show_progress="full"
    ).then(
        fn=initial_data_load_sequence,
        inputs=[url_user_token_display, org_urn_display, token_state],
        outputs=[status_box, token_state],
        show_progress="full"
    )
    
    # REFACTORED: home_load_event now follows directly after initial_load_event
    home_load_event = initial_load_event.then(
        fn=refresh_home_tab_ui,
        inputs=[token_state, home_date_filter_selector, home_custom_start_date, home_custom_end_date],
        outputs=home_refresh_outputs_list,
        show_progress="minimal"
    )
    
    home_load_event.then(
        fn=load_and_display_agentic_results,
        inputs=[token_state, reconstruction_cache_st],
        outputs=agentic_display_outputs,
        show_progress="minimal"
    ).then( 
        fn=format_okrs_for_enhanced_display,
        inputs=[reconstruction_cache_st],
        outputs=[enhanced_okr_display_html],
        show_progress="minimal"
    )


if __name__ == "__main__":
    if not os.environ.get(LINKEDIN_CLIENT_ID_ENV_VAR):
        logging.warning(f"WARNING: '{LINKEDIN_CLIENT_ID_ENV_VAR}' not set.")
    if not all(os.environ.get(var) for var in [BUBBLE_APP_NAME_ENV_VAR, BUBBLE_API_KEY_PRIVATE_ENV_VAR, BUBBLE_API_ENDPOINT_ENV_VAR]):
        logging.warning("WARNING: Bubble environment variables not set.")
    if not AGENTIC_MODULES_LOADED:
        logging.warning("CRITICAL: Agentic modules failed to load.")
    if not os.environ.get("GEMINI_API_KEY"):
        logging.warning("WARNING: 'GEMINI_API_KEY' not set.")

    app.launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 7860)), debug=True)