services/agentic_handlers.py

# handlers/agentic_handlers.py
import gradio as gr
import logging
from collections import defaultdict
import json # Added for JSON serialization/deserialization

# Attempt to import agentic pipeline functions and UI formatters
try:
    from run_agentic_pipeline import run_full_analytics_orchestration
    from ui.insights_ui_generator import (
        format_report_to_markdown,
        extract_key_results_for_selection,
        format_single_okr_for_display
    )
    AGENTIC_MODULES_LOADED = True
except ImportError as e:
    logging.error(f"Could not import agentic pipeline modules for AgenticHandlers: {e}.")
    AGENTIC_MODULES_LOADED = False
    # Define placeholder functions if modules are not loaded to avoid NameErrors during class definition
    async def run_full_analytics_orchestration(*args, **kwargs): return None
    def format_report_to_markdown(report_string): return "Agentic modules not loaded. Report unavailable."
    def extract_key_results_for_selection(okrs_dict): return []
    def format_single_okr_for_display(okr_data, **kwargs): return "Agentic modules not loaded. OKR display unavailable."


class AgenticHandlers:
    def __init__(self, agentic_report_components, agentic_okrs_components,
                 token_state_ref, orchestration_raw_results_st_ref, 
                 key_results_for_selection_st_ref, selected_key_result_ids_st_ref):
        
        self.report_components = agentic_report_components
        self.okrs_components = agentic_okrs_components
        
        # References to global states
        self.token_state = token_state_ref
        self.orchestration_raw_results_st = orchestration_raw_results_st_ref
        self.key_results_for_selection_st = key_results_for_selection_st_ref
        self.selected_key_result_ids_st = selected_key_result_ids_st_ref
        
        self.agentic_modules_really_loaded = AGENTIC_MODULES_LOADED
        logging.info(f"AgenticHandlers initialized. Modules loaded: {self.agentic_modules_really_loaded}")

    async def run_agentic_pipeline_autonomously_on_update(self, current_token_state_val):
        """
        This function is intended to be triggered by changes in token_state.
        It yields updates for the agentic report and OKR tabs.
        State values (5th, 6th, 7th) are serialized to JSON strings.
        Updates for key_results_cbg are now simplified without choices/interactive.
        """
        logging.info(f"Agentic pipeline auto-trigger. Token: {'Set' if current_token_state_val.get('token') else 'Not Set'}")

        initial_report_status = "Pipeline AI: In attesa dei dati necessari..."
        initial_okr_details = "Pipeline AI: In attesa dei dati necessari..."
        
        initial_orchestration_results = self.orchestration_raw_results_st.value 
        initial_selected_krs = self.selected_key_result_ids_st.value
        initial_krs_for_selection = self.key_results_for_selection_st.value
        
        report_status_md_update = gr.update(value=initial_report_status) if self.report_components.get("agentic_pipeline_status_md") else gr.update()
        report_display_md_update = gr.update() 
        
        # Simple update for checkbox component - just reset value
        okrs_cbg_update = gr.update(value=[]) if self.okrs_components.get("key_results_cbg") else gr.update()
        
        okrs_detail_md_update = gr.update(value=initial_okr_details) if self.okrs_components.get("okr_detail_display_md") else gr.update()

        if not current_token_state_val or not current_token_state_val.get("token"):
            logging.info("Agentic pipeline: Token not available in token_state. Skipping actual run.")
            yield (
                report_status_md_update, 
                report_display_md_update, 
                okrs_cbg_update,          
                okrs_detail_md_update,    
                json.dumps(initial_orchestration_results), # Serialize to JSON
                json.dumps(initial_selected_krs),          # Serialize to JSON
                json.dumps(initial_krs_for_selection)      # Serialize to JSON
            )
            return

        in_progress_status = "Analisi AI (Sempre) in corso..."
        if self.report_components.get("agentic_pipeline_status_md"):
            report_status_md_update = gr.update(value=in_progress_status)
        if self.okrs_components.get("okr_detail_display_md"): 
             okrs_detail_md_update = gr.update(value="Dettagli OKR (Sempre) in corso di generazione...")
        
        yield (
            report_status_md_update, report_display_md_update, okrs_cbg_update, okrs_detail_md_update,
            json.dumps(initial_orchestration_results), # Serialize to JSON
            json.dumps(initial_selected_krs),          # Serialize to JSON
            json.dumps(initial_krs_for_selection)      # Serialize to JSON
        )

        if not self.agentic_modules_really_loaded:
            logging.warning("Agentic modules not loaded. Skipping autonomous pipeline actual run.")
            error_status = "Moduli AI non caricati. Operazione non disponibile."
            if self.report_components.get("agentic_pipeline_status_md"):
                report_status_md_update = gr.update(value=error_status)
            if self.report_components.get("agentic_report_display_md"):
                report_display_md_update = gr.update(value=error_status) 
            
            # Simple update for checkbox in error case
            if self.okrs_components.get("key_results_cbg"):
                okrs_cbg_update = gr.update(value=[])
            
            if self.okrs_components.get("okr_detail_display_md"):
                 okrs_detail_md_update = gr.update(value=error_status) 

            yield (
                report_status_md_update, report_display_md_update, okrs_cbg_update, okrs_detail_md_update,
                json.dumps(None), json.dumps([]), json.dumps([]) # Serialize to JSON
            )
            return

        try:
            date_filter_val_agentic = "Sempre"
            custom_start_val_agentic = None
            custom_end_val_agentic = None
            
            logging.info("Agentic pipeline: Calling run_full_analytics_orchestration...")
            orchestration_output = await run_full_analytics_orchestration(
                current_token_state_val, 
                date_filter_val_agentic, 
                custom_start_val_agentic, 
                custom_end_val_agentic
            )
            
            final_status_text = "Pipeline AI (Sempre) completata."
            logging.info(f"Autonomous agentic pipeline finished. Output keys: {orchestration_output.keys() if orchestration_output else 'None'}")

            orchestration_results_update_val = None
            selected_krs_update_val = [] # This will be the value for the CheckboxGroup, initially empty
            krs_for_selection_update_val = []

            if orchestration_output:
                orchestration_results_update_val = orchestration_output 
                
                report_str = orchestration_output.get('comprehensive_analysis_report', "Nessun report testuale fornito.")
                if self.report_components.get("agentic_report_display_md"):
                    formatted_report = format_report_to_markdown(report_str)
                    # Ensure we have a string, not a list
                    if isinstance(formatted_report, list):
                        formatted_report = "\n".join(formatted_report)
                    elif not isinstance(formatted_report, str):
                        formatted_report = str(formatted_report)
                    report_display_md_update = gr.update(value=formatted_report)
                
                actionable_okrs = orchestration_output.get('actionable_okrs_and_tasks')
                krs_for_ui_selection_list = extract_key_results_for_selection(actionable_okrs)
                krs_for_selection_update_val = krs_for_ui_selection_list # This is the list of dicts
                
                # For checkbox, just reset the value - don't update choices programmatically
                if self.okrs_components.get("key_results_cbg"):
                    okrs_cbg_update = gr.update(value=[])
                
                all_okrs_md_parts = []
                if actionable_okrs and isinstance(actionable_okrs.get("okrs"), list):
                    for okr_idx, okr_item in enumerate(actionable_okrs["okrs"]):
                        all_okrs_md_parts.append(format_single_okr_for_display(okr_item, accepted_kr_indices=None, okr_main_index=okr_idx))
                
                if not all_okrs_md_parts:
                    if self.okrs_components.get("okr_detail_display_md"):
                        okrs_detail_md_update = gr.update(value="Nessun OKR generato o trovato (Sempre).")
                else:
                    if self.okrs_components.get("okr_detail_display_md"):
                        okrs_detail_md_update = gr.update(value="\n\n---\n\n".join(all_okrs_md_parts))
                
                selected_krs_update_val = [] # Reset CheckboxGroup selection
            else: 
                final_status_text = "Pipeline AI (Sempre): Nessun risultato prodotto."
                if self.report_components.get("agentic_report_display_md"):
                    report_display_md_update = gr.update(value="Nessun report generato dalla pipeline AI (Sempre).")
                
                # Simple update for checkbox if no output
                if self.okrs_components.get("key_results_cbg"):
                    okrs_cbg_update = gr.update(value=[])
                
                if self.okrs_components.get("okr_detail_display_md"):
                    okrs_detail_md_update = gr.update(value="Nessun OKR generato o errore nella pipeline AI (Sempre).")
            
            if self.report_components.get("agentic_pipeline_status_md"):
                 report_status_md_update = gr.update(value=final_status_text)

            yield (
                report_status_md_update, report_display_md_update, okrs_cbg_update, okrs_detail_md_update,
                json.dumps(orchestration_results_update_val), # Serialize to JSON
                json.dumps(selected_krs_update_val),          # Serialize to JSON (value for selected_key_result_ids_st)
                json.dumps(krs_for_selection_update_val)      # Serialize to JSON (value for key_results_for_selection_st)
            )

        except Exception as e:
            logging.error(f"Error during autonomous agentic pipeline execution: {e}", exc_info=True)
            error_status_text = f"Errore pipeline AI (Sempre): {str(e)}"
            if self.report_components.get("agentic_pipeline_status_md"):
                report_status_md_update = gr.update(value=error_status_text)
            if self.report_components.get("agentic_report_display_md"):
                 report_display_md_update = gr.update(value=f"Errore generazione report AI (Sempre): {str(e)}")
            
            # Simple update for checkbox in case of exception
            if self.okrs_components.get("key_results_cbg"):
                okrs_cbg_update = gr.update(value=[])

            if self.okrs_components.get("okr_detail_display_md"):
                okrs_detail_md_update = gr.update(value=f"Errore generazione OKR AI (Sempre): {str(e)}")

            yield (
                report_status_md_update, report_display_md_update, okrs_cbg_update, okrs_detail_md_update,
                json.dumps(None), json.dumps([]), json.dumps([]) # Serialize to JSON
            )

    def update_okr_display_on_kr_selection(self, selected_kr_unique_ids: list, 
                                           raw_orchestration_results_json: str, 
                                           all_krs_for_selection_list_json: str):
        """
        Updates the OKR detail display when Key Results are selected in the CheckboxGroup.
        raw_orchestration_results_json and all_krs_for_selection_list_json are expected
        to be JSON strings from state.
        """
        if not self.agentic_modules_really_loaded:
            return gr.update(value="Moduli AI non caricati. Impossibile visualizzare i dettagli OKR.")
        
        # Handle case where selected_kr_unique_ids might be None or not a list
        if not isinstance(selected_kr_unique_ids, list):
            selected_kr_unique_ids = []
        
        parsed_orchestration_results = None
        try:
            if raw_orchestration_results_json: # Check if the string is not empty
                parsed_orchestration_results = json.loads(raw_orchestration_results_json)
        except (json.JSONDecodeError, TypeError) as e:
            logging.error(f"Failed to parse raw_orchestration_results_json: {raw_orchestration_results_json}. Error: {e}")
            return gr.update(value="Errore: Dati interni corrotti (orchestration results).")

        if not parsed_orchestration_results: # This covers None or empty after parsing
            return gr.update(value="Nessun dato dalla pipeline AI (orchestration results).")

        parsed_krs_for_selection_list = []
        try:
            if all_krs_for_selection_list_json: # Check if the string is not empty
                 parsed_krs_for_selection_list = json.loads(all_krs_for_selection_list_json)
        except (json.JSONDecodeError, TypeError) as e:
            logging.error(f"Failed to parse all_krs_for_selection_list_json: {all_krs_for_selection_list_json}. Error: {e}")
            return gr.update(value="Errore: Dati interni corrotti (krs for selection).")
        
        # Ensure parsed_krs_for_selection_list is a list, even if JSON was 'null' or other non-list type
        if not isinstance(parsed_krs_for_selection_list, list):
            logging.warning(f"Parsed all_krs_for_selection_list is not a list: {type(parsed_krs_for_selection_list)}. Defaulting to empty list.")
            parsed_krs_for_selection_list = []

        actionable_okrs_dict = parsed_orchestration_results.get("actionable_okrs_and_tasks") if isinstance(parsed_orchestration_results, dict) else None
        
        if not actionable_okrs_dict or not isinstance(actionable_okrs_dict.get("okrs"), list):
            return gr.update(value="Nessun OKR trovato nei risultati della pipeline (o dati in formato imprevisto).")

        okrs_list = actionable_okrs_dict["okrs"]
        if not okrs_list: 
            return gr.update(value="Nessun OKR generato.")

        kr_id_to_indices = {}
        if isinstance(parsed_krs_for_selection_list, list): # Ensure it's a list before iterating
            for kr_info in parsed_krs_for_selection_list:
                if isinstance(kr_info, dict) and 'unique_kr_id' in kr_info and 'okr_index' in kr_info and 'kr_index' in kr_info:
                    kr_id_to_indices[kr_info['unique_kr_id']] = (kr_info['okr_index'], kr_info['kr_index'])
                else:
                    logging.warning(f"Skipping invalid kr_info item: {kr_info}")

        selected_krs_by_okr_idx = defaultdict(list)
        # selected_kr_unique_ids comes directly from CheckboxGroup, should be a list of strings/values
        if isinstance(selected_kr_unique_ids, list): 
            for kr_unique_id in selected_kr_unique_ids:
                if kr_unique_id in kr_id_to_indices:
                    okr_idx, kr_idx_in_okr = kr_id_to_indices[kr_unique_id]
                    selected_krs_by_okr_idx[okr_idx].append(kr_idx_in_okr)
        
        output_md_parts = []
        for okr_idx, okr_data in enumerate(okrs_list):
            accepted_indices_for_this_okr = selected_krs_by_okr_idx.get(okr_idx)

            if selected_kr_unique_ids: 
                if accepted_indices_for_this_okr is not None: 
                    formatted_okr_md = format_single_okr_for_display(
                        okr_data, 
                        accepted_kr_indices=accepted_indices_for_this_okr, 
                        okr_main_index=okr_idx
                    )
                    output_md_parts.append(formatted_okr_md)
            else: 
                formatted_okr_md = format_single_okr_for_display(
                    okr_data, 
                    accepted_kr_indices=None, 
                    okr_main_index=okr_idx
                )
                output_md_parts.append(formatted_okr_md)

        if not output_md_parts and selected_kr_unique_ids:
            final_md = "Nessun OKR corrisponde alla selezione corrente o i KR selezionati non hanno task dettagliati."
        elif not output_md_parts and not selected_kr_unique_ids: 
            final_md = "Nessun OKR generato."
        else:
            final_md = "\n\n---\n\n".join(output_md_parts)
            
        return gr.update(value=final_md)

    def setup_event_handlers(self):
        """Sets up event handlers for the agentic OKRs tab."""
        if not self.agentic_modules_really_loaded:
            logging.warning("Agentic modules not loaded. Skipping agentic event handler setup.")
            return

        if self.okrs_components.get("key_results_cbg"):
            self.okrs_components['key_results_cbg'].change(
                fn=self.update_okr_display_on_kr_selection,
                inputs=[
                    self.okrs_components['key_results_cbg'], 
                    self.orchestration_raw_results_st,     
                    self.key_results_for_selection_st      
                ],
                outputs=[self.okrs_components['okr_detail_display_md']],
                api_name="update_okr_display_on_kr_selection" # Keep api_name for Gradio
            )
            logging.info("Agentic OKR selection handler setup complete.")
        else:
            logging.warning("key_results_cbg component not found for agentic OKR handler setup.")