"""
Complaints Celery tasks

This module contains tasks for:
- Checking overdue complaints
- Sending SLA reminders
- Triggering resolution satisfaction surveys
- Creating PX actions from complaints
- AI-powered complaint analysis
"""
import logging
from typing import Optional, Dict, Any, Tuple

from celery import shared_task
from django.db import transaction
from django.db.models import Q
from django.utils import timezone

logger = logging.getLogger(__name__)


def match_staff_from_name(staff_name: str, hospital_id: str, department_name: Optional[str] = None) -> Tuple[Optional[str], float, str]:
    """
    Match staff member from extracted name using multiple matching strategies.

    Args:
        staff_name: Name extracted from complaint (without titles)
        hospital_id: Hospital ID to search within
        department_name: Optional department name to prioritize matching

    Returns:
        Tuple of (staff_id, confidence_score, matching_method)
        - staff_id: UUID of matched staff or None
        - confidence_score: Float from 0.0 to 1.0
        - matching_method: Description of how staff was matched
    """
    from apps.organizations.models import Staff, Department

    if not staff_name or not staff_name.strip():
        return None, 0.0, "No staff name provided"

    staff_name = staff_name.strip()

    # Build base query - staff from this hospital, active status
    base_query = Staff.objects.filter(
        hospital_id=hospital_id,
        status='active'
    )

    # If department is specified, prioritize staff in that department
    dept_id = None
    if department_name:
        department = Department.objects.filter(
            hospital_id=hospital_id,
            name__iexact=department_name,
            status='active'
        ).first()
        if department:
            dept_id = department.id

    # Layer 1: Exact English match (full name)
    # Try "first_name last_name" format
    words = staff_name.split()
    if len(words) >= 2:
        first_name = words[0]
        last_name = ' '.join(words[1:])

        # Exact match in correct department
        exact_query = base_query.filter(
            Q(first_name__iexact=first_name) & Q(last_name__iexact=last_name)
        )
        if dept_id:
            exact_query = exact_query.filter(department_id=dept_id)

        staff = exact_query.first()
        if staff:
            confidence = 0.95 if dept_id else 0.90
            method = f"Exact English match in {'correct' if dept_id else 'any'} department"
            logger.info(f"Matched staff using exact English match: {staff.first_name} {staff.last_name} (confidence: {confidence})")
            return str(staff.id), confidence, method

    # Layer 2: Exact Arabic match
    arabic_query = base_query.filter(
        Q(first_name_ar__iexact=staff_name) | Q(last_name_ar__iexact=staff_name)
    )
    if dept_id:
        arabic_query = arabic_query.filter(department_id=dept_id)

    # Try full Arabic name match
    for staff in arabic_query:
        full_arabic_name = f"{staff.first_name_ar} {staff.last_name_ar}".strip()
        if full_arabic_name == staff_name:
            confidence = 0.95 if dept_id else 0.90
            method = f"Exact Arabic match in {'correct' if dept_id else 'any'} department"
            logger.info(f"Matched staff using exact Arabic match: {staff.first_name_ar} {staff.last_name_ar} (confidence: {confidence})")
            return str(staff.id), confidence, method

    # Layer 3: Partial match (first name or last name)
    partial_query = base_query.filter(
        Q(first_name__icontains=staff_name) |
        Q(last_name__icontains=staff_name) |
        Q(first_name_ar__icontains=staff_name) |
        Q(last_name_ar__icontains=staff_name)
    )
    if dept_id:
        partial_query = partial_query.filter(department_id=dept_id)

    staff = partial_query.first()
    if staff:
        confidence = 0.70 if dept_id else 0.60
        method = f"Partial match in {'correct' if dept_id else 'any'} department"
        logger.info(f"Matched staff using partial match: {staff.first_name} {staff.last_name} (confidence: {confidence})")
        return str(staff.id), confidence, method

    # Layer 4: Fuzzy match using individual words
    # Handle cases like "Dr. Ahmed" or "Nurse Sarah"
    word_query = base_query.filter(
        Q(first_name__in=words) | Q(first_name_ar__in=words)
    )
    if dept_id:
        word_query = word_query.filter(department_id=dept_id)

    staff = word_query.first()
    if staff:
        confidence = 0.50 if dept_id else 0.45
        method = f"Word match in {'correct' if dept_id else 'any'} department"
        logger.info(f"Matched staff using word match: {staff.first_name} {staff.last_name} (confidence: {confidence})")
        return str(staff.id), confidence, method

    # No match found
    logger.warning(f"No staff match found for name: {staff_name}")
    return None, 0.0, "No match found"


@shared_task
def check_overdue_complaints():
    """
    Periodic task to check for overdue complaints.

    Runs every 15 minutes (configured in config/celery.py).
    Updates is_overdue flag for complaints past their SLA deadline.
    Triggers automatic escalation based on escalation rules.
    """
    from apps.complaints.models import Complaint, ComplaintStatus

    # Get active complaints (not closed or cancelled)
    active_complaints = Complaint.objects.filter(
        status__in=[ComplaintStatus.OPEN, ComplaintStatus.IN_PROGRESS, ComplaintStatus.RESOLVED]
    ).select_related('hospital', 'patient', 'department')

    overdue_count = 0
    escalated_count = 0

    for complaint in active_complaints:
        if complaint.check_overdue():
            overdue_count += 1
            logger.warning(
                f"Complaint {complaint.id} is overdue: {complaint.title} "
                f"(due: {complaint.due_at})"
            )

            # Trigger automatic escalation
            result = escalate_complaint_auto.delay(str(complaint.id))
            if result:
                escalated_count += 1

    if overdue_count > 0:
        logger.info(f"Found {overdue_count} overdue complaints, triggered {escalated_count} escalations")

    return {
        'overdue_count': overdue_count,
        'escalated_count': escalated_count
    }


@shared_task
def send_complaint_resolution_survey(complaint_id):
    """
    Send resolution satisfaction survey when complaint is closed.

    This task is triggered when a complaint status changes to CLOSED.

    Args:
        complaint_id: UUID of the Complaint

    Returns:
        dict: Result with survey_instance_id
    """
    from apps.complaints.models import Complaint
    from apps.core.services import create_audit_log
    from apps.surveys.models import SurveyInstance, SurveyTemplate

    try:
        complaint = Complaint.objects.select_related(
            'patient', 'hospital'
        ).get(id=complaint_id)

        # Check if survey already sent
        if complaint.resolution_survey:
            logger.info(f"Resolution survey already sent for complaint {complaint_id}")
            return {'status': 'skipped', 'reason': 'already_sent'}

        # Get resolution satisfaction survey template
        try:
            survey_template = SurveyTemplate.objects.get(
                hospital=complaint.hospital,
                survey_type='complaint_resolution',
                is_active=True
            )
        except SurveyTemplate.DoesNotExist:
            logger.warning(
                f"No resolution satisfaction survey template found for hospital {complaint.hospital.name}"
            )
            return {'status': 'skipped', 'reason': 'no_template'}

        # Create survey instance
        with transaction.atomic():
            survey_instance = SurveyInstance.objects.create(
                survey_template=survey_template,
                patient=complaint.patient,
                encounter_id=complaint.encounter_id,
                delivery_channel='sms',  # Default
                recipient_phone=complaint.patient.phone,
                recipient_email=complaint.patient.email,
                metadata={
                    'complaint_id': str(complaint.id),
                    'complaint_title': complaint.title
                }
            )

            # Link survey to complaint
            complaint.resolution_survey = survey_instance
            complaint.resolution_survey_sent_at = timezone.now()
            complaint.save(update_fields=['resolution_survey', 'resolution_survey_sent_at'])

        # Send survey
        from apps.notifications.services import NotificationService
        notification_log = NotificationService.send_survey_invitation(
            survey_instance=survey_instance,
            language='en'  # TODO: Get from patient preference
        )

        # Update survey status
        survey_instance.status = 'active'
        survey_instance.sent_at = timezone.now()
        survey_instance.save(update_fields=['status', 'sent_at'])

        # Log audit event
        create_audit_log(
            event_type='survey_sent',
            description=f"Resolution satisfaction survey sent for complaint: {complaint.title}",
            content_object=survey_instance,
            metadata={
                'complaint_id': str(complaint.id),
                'survey_template': survey_template.name
            }
        )

        logger.info(
            f"Resolution satisfaction survey sent for complaint {complaint.id}"
        )

        return {
            'status': 'sent',
            'survey_instance_id': str(survey_instance.id),
            'notification_log_id': str(notification_log.id)
        }

    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}

    except Exception as e:
        error_msg = f"Error sending resolution survey: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}


@shared_task
def check_resolution_survey_threshold(survey_instance_id, complaint_id):
    """
    Check if resolution survey score breaches threshold and create PX Action if needed.

    This task is triggered when a complaint resolution survey is completed.

    Args:
        survey_instance_id: UUID of the SurveyInstance
        complaint_id: UUID of the Complaint

    Returns:
        dict: Result with action status
    """
    from apps.complaints.models import Complaint, ComplaintThreshold
    from apps.surveys.models import SurveyInstance
    from apps.px_action_center.models import PXAction
    from django.contrib.contenttypes.models import ContentType

    try:
        survey = SurveyInstance.objects.get(id=survey_instance_id)
        complaint = Complaint.objects.select_related('hospital', 'patient').get(id=complaint_id)

        # Get threshold for this hospital
        try:
            threshold = ComplaintThreshold.objects.get(
                hospital=complaint.hospital,
                threshold_type='resolution_survey_score',
                is_active=True
            )
        except ComplaintThreshold.DoesNotExist:
            logger.info(f"No resolution survey threshold configured for hospital {complaint.hospital.name_en}")
            return {'status': 'no_threshold'}

        # Check if threshold is breached
        if threshold.check_threshold(survey.score):
            logger.warning(
                f"Resolution survey score {survey.score} breaches threshold {threshold.threshold_value} "
                f"for complaint {complaint_id}"
            )

            # Create PX Action
            complaint_ct = ContentType.objects.get_for_model(Complaint)

            action = PXAction.objects.create(
                title=f"Low Resolution Satisfaction: {complaint.title[:100]}",
                description=(
                    f"Complaint resolution survey scored {survey.score}% "
                    f"(threshold: {threshold.threshold_value}%). "
                    f"Original complaint: {complaint.description[:200]}"
                ),
                source='complaint_resolution_survey',
                priority='high' if survey.score < 30 else 'medium',
                hospital=complaint.hospital,
                department=complaint.department,
                patient=complaint.patient,
                content_type=complaint_ct,
                object_id=complaint.id,
                metadata={
                    'complaint_id': str(complaint.id),
                    'survey_id': str(survey.id),
                    'survey_score': survey.score,
                    'threshold_value': threshold.threshold_value,
                }
            )

            # Log audit
            from apps.core.services import create_audit_log
            create_audit_log(
                event_type='px_action_created',
                description=f"PX Action created from low resolution survey score",
                content_object=action,
                metadata={
                    'complaint_id': str(complaint.id),
                    'survey_score': survey.score,
                    'trigger': 'resolution_survey_threshold'
                }
            )

            logger.info(f"Created PX Action {action.id} from low resolution survey score")

            return {
                'status': 'action_created',
                'action_id': str(action.id),
                'survey_score': survey.score,
                'threshold': threshold.threshold_value
            }
        else:
            logger.info(f"Resolution survey score {survey.score} is above threshold {threshold.threshold_value}")
            return {'status': 'threshold_not_breached', 'survey_score': survey.score}

    except SurveyInstance.DoesNotExist:
        error_msg = f"SurveyInstance {survey_instance_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}
    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}
    except Exception as e:
        error_msg = f"Error checking resolution survey threshold: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}


@shared_task
def create_action_from_complaint(complaint_id):
    """
    Create PX Action from complaint (if configured).

    This task is triggered when a complaint is created,
    if the hospital configuration requires automatic action creation.

    Args:
        complaint_id: UUID of the Complaint

    Returns:
        dict: Result with action_id
    """
    from apps.complaints.models import Complaint
    from apps.organizations.models import Hospital
    from apps.px_action_center.models import PXAction
    from django.contrib.contenttypes.models import ContentType

    try:
        complaint = Complaint.objects.select_related('hospital', 'patient', 'department').get(id=complaint_id)

        # Check if hospital has auto-create enabled
        # For now, we'll check metadata on hospital or use a simple rule
        # In production, you'd have a HospitalComplaintConfig model
        # Handle case where metadata field might not exist (legacy data)
        hospital_metadata = getattr(complaint.hospital, 'metadata', None)
        if hospital_metadata is None:
            hospital_metadata = {}
        auto_create = hospital_metadata.get('auto_create_action_on_complaint', False)

        if not auto_create:
            logger.info(f"Auto-create PX Action disabled for hospital {complaint.hospital.name}")
            return {'status': 'disabled'}

        # Use JSON-serializable values instead of model objects
        category_name = complaint.category.name_en if complaint.category else None
        category_id = str(complaint.category.id) if complaint.category else None

        # Create PX Action
        complaint_ct = ContentType.objects.get_for_model(Complaint)

        action = PXAction.objects.create(
            title=f"New Complaint: {complaint.title[:100]}",
            description=complaint.description[:500],
            source='complaint',
            priority=complaint.priority,
            hospital=complaint.hospital,
            department=complaint.department,
            patient=complaint.patient,
            content_type=complaint_ct,
            object_id=complaint.id,
            metadata={
                'complaint_id': str(complaint.id),
                'complaint_category': category_name,
                'complaint_category_id': category_id,
                'complaint_severity': complaint.severity,
            }
        )

        # Log audit
        from apps.core.services import create_audit_log
        create_audit_log(
            event_type='px_action_created',
            description=f"PX Action created from complaint",
            content_object=action,
            metadata={
                'complaint_id': str(complaint.id),
                'trigger': 'complaint_creation'
            }
        )

        logger.info(f"Created PX Action {action.id} from complaint {complaint_id}")

        return {
            'status': 'action_created',
            'action_id': str(action.id)
        }

    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}
    except Exception as e:
        error_msg = f"Error creating action from complaint: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}


@shared_task
def escalate_complaint_auto(complaint_id):
    """
    Automatically escalate complaint based on escalation rules.

    This task is triggered when a complaint becomes overdue.
    It finds matching escalation rules and reassigns the complaint.

    Args:
        complaint_id: UUID of the Complaint

    Returns:
        dict: Result with escalation status
    """
    from apps.complaints.models import Complaint, ComplaintUpdate, EscalationRule
    from apps.accounts.models import User

    try:
        complaint = Complaint.objects.select_related(
            'hospital', 'department', 'assigned_to'
        ).get(id=complaint_id)

        # Calculate hours overdue
        hours_overdue = (timezone.now() - complaint.due_at).total_seconds() / 3600

        # Get applicable escalation rules for this hospital
        rules = EscalationRule.objects.filter(
            hospital=complaint.hospital,
            is_active=True,
            trigger_on_overdue=True
        ).order_by('order')

        # Filter rules by severity and priority if specified
        if complaint.severity:
            rules = rules.filter(
                Q(severity_filter='') | Q(severity_filter=complaint.severity)
            )

        if complaint.priority:
            rules = rules.filter(
                Q(priority_filter='') | Q(priority_filter=complaint.priority)
            )

        # Find first matching rule based on hours overdue
        matching_rule = None
        for rule in rules:
            if hours_overdue >= rule.trigger_hours_overdue:
                matching_rule = rule
                break

        if not matching_rule:
            logger.info(f"No matching escalation rule found for complaint {complaint_id}")
            return {'status': 'no_matching_rule'}

        # Determine escalation target
        escalation_target = None

        if matching_rule.escalate_to_role == 'department_manager':
            if complaint.department and complaint.department.manager:
                escalation_target = complaint.department.manager

        elif matching_rule.escalate_to_role == 'hospital_admin':
            # Find hospital admin for this hospital
            escalation_target = User.objects.filter(
                hospital=complaint.hospital,
                groups__name='Hospital Admin',
                is_active=True
            ).first()

        elif matching_rule.escalate_to_role == 'px_admin':
            # Find PX admin
            escalation_target = User.objects.filter(
                groups__name='PX Admin',
                is_active=True
            ).first()

        elif matching_rule.escalate_to_role == 'specific_user':
            escalation_target = matching_rule.escalate_to_user

        if not escalation_target:
            logger.warning(
                f"Could not find escalation target for rule {matching_rule.name} "
                f"on complaint {complaint_id}"
            )
            return {'status': 'no_target_found', 'rule': matching_rule.name}

        # Perform escalation
        old_assignee = complaint.assigned_to
        complaint.assigned_to = escalation_target
        complaint.escalated_at = timezone.now()
        complaint.save(update_fields=['assigned_to', 'escalated_at'])

        # Create update
        ComplaintUpdate.objects.create(
            complaint=complaint,
            update_type='escalation',
            message=(
                f"Automatically escalated to {escalation_target.get_full_name()} "
                f"(Rule: {matching_rule.name}). "
                f"Complaint is {hours_overdue:.1f} hours overdue."
            ),
            created_by=None,  # System action
            metadata={
                'rule_id': str(matching_rule.id),
                'rule_name': matching_rule.name,
                'hours_overdue': hours_overdue,
                'old_assignee_id': str(old_assignee.id) if old_assignee else None,
                'new_assignee_id': str(escalation_target.id)
            }
        )

        # Send notifications
        send_complaint_notification.delay(
            complaint_id=str(complaint.id),
            event_type='escalated'
        )

        # Log audit
        from apps.core.services import create_audit_log
        create_audit_log(
            event_type='complaint_escalated',
            description=f"Complaint automatically escalated to {escalation_target.get_full_name()}",
            content_object=complaint,
            metadata={
                'rule': matching_rule.name,
                'hours_overdue': hours_overdue,
                'escalated_to': escalation_target.get_full_name()
            }
        )

        logger.info(
            f"Escalated complaint {complaint_id} to {escalation_target.get_full_name()} "
            f"using rule '{matching_rule.name}'"
        )

        return {
            'status': 'escalated',
            'rule': matching_rule.name,
            'escalated_to': escalation_target.get_full_name(),
            'hours_overdue': round(hours_overdue, 2)
        }

    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}
    except Exception as e:
        error_msg = f"Error escalating complaint: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}


@shared_task
def analyze_complaint_with_ai(complaint_id):
    """
    Analyze a complaint using AI to determine severity and priority and category.

    This task is triggered when a complaint is created.
    It uses the AI service to analyze the complaint content and classify it.

    Args:
        complaint_id: UUID of the Complaint

    Returns:
        dict: Result with severity, priority, category, and reasoning
    """
    from apps.complaints.models import Complaint
    from apps.core.ai_service import AIService, AIServiceError

    try:
        complaint = Complaint.objects.select_related('hospital').get(id=complaint_id)

        logger.info(f"Starting AI analysis for complaint {complaint_id}")

        # Get category name if category exists
        category_name = None
        if complaint.category:
            category_name = complaint.category.name_en

        # Analyze complaint using AI service
        try:
            analysis = AIService.analyze_complaint(
                title=complaint.title,
                description=complaint.description,
                category=category_name,
                hospital_id=complaint.hospital.id
            )

            # Analyze emotion using AI service
            emotion_analysis = AIService.analyze_emotion(
                text=complaint.description
            )

            # Update complaint with AI-determined values
            old_severity = complaint.severity
            old_priority = complaint.priority
            old_category = complaint.category
            old_department = complaint.department

            complaint.severity = analysis['severity']
            complaint.priority = analysis['priority']

            from apps.complaints.models import ComplaintCategory
            if category := ComplaintCategory.objects.filter(name_en=analysis['category']).first():
                complaint.category = category

            # Update department from AI analysis
            department_name = analysis.get('department', '')
            if department_name:
                from apps.organizations.models import Department
                # Try exact match first (case-insensitive)
                if department := Department.objects.filter(
                    hospital_id=complaint.hospital.id,
                    name__iexact=department_name,
                    status='active'
                ).first():
                    complaint.department = department
                    logger.info(f"Matched department exactly: {department.name}")
                # If no exact match, try partial match
                elif department := Department.objects.filter(
                    hospital_id=complaint.hospital.id,
                    name__icontains=department_name,
                    status='active'
                ).first():
                    complaint.department = department
                    logger.info(f"Matched department partially: {department.name} from '{department_name}'")
                else:
                    logger.warning(f"AI suggested department '{department_name}' but no match found in hospital '{complaint.hospital.name}'")

            # Update title from AI analysis (use English version)
            if analysis.get('title_en'):
                complaint.title = analysis['title_en']
            elif analysis.get('title'):
                complaint.title = analysis['title']

            # Get staff_name from analyze_complaint result (already extracted by AI)
            staff_name = analysis.get('staff_name', '').strip()

            matched_staff_id = None
            staff_confidence = 0.0
            staff_matching_method = None

            # Capture old staff before matching
            old_staff = complaint.staff

            if staff_name:
                logger.info(f"AI extracted staff name: {staff_name}")

                # Try matching WITH department filter first (higher confidence if match found)
                matched_staff_id, staff_confidence, staff_matching_method = match_staff_from_name(
                    staff_name=staff_name,
                    hospital_id=str(complaint.hospital.id),
                    department_name=department_name
                )

                # If no match found with department, try WITHOUT department filter
                if not matched_staff_id:
                    logger.info(f"No match found with department filter '{department_name}', trying without department filter...")
                    matched_staff_id, staff_confidence, staff_matching_method = match_staff_from_name(
                        staff_name=staff_name,
                        hospital_id=str(complaint.hospital.id),
                        department_name=None  # Search all departments
                    )

                # Only assign staff if confidence is above threshold (0.6)
                if matched_staff_id and staff_confidence >= 0.6:
                    from apps.organizations.models import Staff
                    try:
                        staff = Staff.objects.get(id=matched_staff_id)
                        complaint.staff = staff
                        logger.info(
                            f"Assigned staff {staff.first_name} {staff.last_name} "
                            f"to complaint {complaint_id} "
                            f"(confidence: {staff_confidence:.2f}, method: {staff_matching_method})"
                        )
                    except Staff.DoesNotExist:
                        logger.warning(f"Staff {matched_staff_id} not found in database")
                else:
                    logger.info(
                        f"Staff match confidence {staff_confidence:.2f} below threshold 0.6, "
                        f"or no match found. Not assigning staff."
                    )

            # Save reasoning in metadata
            # Use JSON-serializable values instead of model objects
            old_category_name = old_category.name_en if old_category else None
            old_category_id = str(old_category.id) if old_category else None
            old_department_name = old_department.name if old_department else None
            old_department_id = str(old_department.id) if old_department else None
            old_staff_name = f"{old_staff.first_name} {old_staff.last_name}" if old_staff else None
            old_staff_id = str(old_staff.id) if old_staff else None

            # Initialize metadata if needed
            if not complaint.metadata:
                complaint.metadata = {}

            # Update or create ai_analysis in metadata with bilingual support and emotion
            complaint.metadata['ai_analysis'] = {
                'title_en': analysis.get('title_en', ''),
                'title_ar': analysis.get('title_ar', ''),
                'short_description_en': analysis.get('short_description_en', ''),
                'short_description_ar': analysis.get('short_description_ar', ''),
                'suggested_action_en': analysis.get('suggested_action_en', ''),
                'suggested_action_ar': analysis.get('suggested_action_ar', ''),
                'reasoning_en': analysis.get('reasoning_en', ''),
                'reasoning_ar': analysis.get('reasoning_ar', ''),
                'emotion': emotion_analysis.get('emotion', 'neutral'),
                'emotion_intensity': emotion_analysis.get('intensity', 0.0),
                'emotion_confidence': emotion_analysis.get('confidence', 0.0),
                'analyzed_at': timezone.now().isoformat(),
                'old_severity': old_severity,
                'old_priority': old_priority,
                'old_category': old_category_name,
                'old_category_id': old_category_id,
                'old_department': old_department_name,
                'old_department_id': old_department_id,
                'old_staff': old_staff_name,
                'old_staff_id': old_staff_id,
                'extracted_staff_name': staff_name,
                'matched_staff_id': matched_staff_id,
                'staff_confidence': staff_confidence,
                'staff_matching_method': staff_matching_method
            }

            complaint.save(update_fields=['severity', 'priority', 'category', 'department', 'staff', 'title', 'metadata'])

            # Re-calculate SLA due date based on new severity
            complaint.due_at = complaint.calculate_sla_due_date()
            complaint.save(update_fields=['due_at'])

            # Create timeline update for AI completion
            from apps.complaints.models import ComplaintUpdate

            # Build bilingual message
            emotion_display = emotion_analysis.get('emotion', 'neutral')
            emotion_intensity = emotion_analysis.get('intensity', 0.0)

            # Build English message
            message_en = f"AI analysis complete: Severity={analysis['severity']}, Priority={analysis['priority']}, Category={analysis.get('category', 'N/A')}, Department={department_name or 'N/A'}"
            if matched_staff_id:
                message_en += f", Staff={f'{complaint.staff.first_name} {complaint.staff.last_name}' if complaint.staff else 'N/A'} (confidence: {staff_confidence:.2f})"
            message_en += f", Emotion={emotion_display} (Intensity: {emotion_intensity:.2f})"

            # Build Arabic message
            message_ar = f"اكتمل تحليل الذكاء الاصطناعي: الشدة={analysis['severity']}, الأولوية={analysis['priority']}, الفئة={analysis.get('category', 'N/A')}, القسم={department_name or 'N/A'}"
            if matched_staff_id and complaint.staff:
                staff_name_ar = complaint.staff.first_name_ar if complaint.staff.first_name_ar else complaint.staff.first_name
                message_ar += f", الموظف={staff_name_ar} {complaint.staff.last_name_ar if complaint.staff.last_name_ar else complaint.staff.last_name} (الثقة: {staff_confidence:.2f})"
            message_ar += f", العاطفة={emotion_display} (الشدة: {emotion_intensity:.2f})"

            message = f"{message_en}\n\n{message_ar}"

            ComplaintUpdate.objects.create(
                complaint=complaint,
                update_type='note',
                message=message
            )

            logger.info(
                f"AI analysis complete for complaint {complaint_id}: "
                f"severity={old_severity}->{analysis['severity']}, "
                f"priority={old_priority}->{analysis['priority']}, "
                f"category={old_category_name}->{analysis['category']}, "
                f"department={old_department_name}->{department_name}, "
                f"title_en={analysis.get('title_en', '')}"
            )

            return {
                'status': 'success',
                'complaint_id': str(complaint_id),
                'severity': analysis['severity'],
                'priority': analysis['priority'],
                'category': analysis['category'],
                'department': department_name,
                'title_en': analysis.get('title_en', ''),
                'title_ar': analysis.get('title_ar', ''),
                'short_description_en': analysis.get('short_description_en', ''),
                'short_description_ar': analysis.get('short_description_ar', ''),
                'suggested_action_en': analysis.get('suggested_action_en', ''),
                'suggested_action_ar': analysis.get('suggested_action_ar', ''),
                'reasoning_en': analysis.get('reasoning_en', ''),
                'reasoning_ar': analysis.get('reasoning_ar', ''),
                'emotion': emotion_analysis.get('emotion', 'neutral'),
                'emotion_intensity': emotion_analysis.get('intensity', 0.0),
                'emotion_confidence': emotion_analysis.get('confidence', 0.0),
                'old_severity': old_severity,
                'old_priority': old_priority
            }

        except AIServiceError as e:
            logger.error(f"AI service error for complaint {complaint_id}: {str(e)}")
            # Keep default values (medium/medium) and log the error
            return {
                'status': 'ai_error',
                'complaint_id': str(complaint_id),
                'reason': str(e)
            }

    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}

    except Exception as e:
        error_msg = f"Error analyzing complaint {complaint_id} with AI: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}


@shared_task
def send_complaint_notification(complaint_id, event_type):
    """
    Send notification for complaint events.

    Args:
        complaint_id: UUID of the Complaint
        event_type: Type of event (created, assigned, overdue, escalated, resolved, closed)

    Returns:
        dict: Result with notification status
    """
    from apps.complaints.models import Complaint
    from apps.notifications.services import NotificationService

    try:
        complaint = Complaint.objects.select_related(
            'hospital', 'patient', 'assigned_to', 'department'
        ).get(id=complaint_id)

        # Determine recipients based on event type
        recipients = []

        if event_type == 'created':
            # Notify assigned user or department manager
            if complaint.assigned_to:
                recipients.append(complaint.assigned_to)
            elif complaint.department and complaint.department.manager:
                recipients.append(complaint.department.manager)

        elif event_type == 'assigned':
            # Notify the assignee
            if complaint.assigned_to:
                recipients.append(complaint.assigned_to)

        elif event_type in ['overdue', 'escalated']:
            # Notify assignee and their manager
            if complaint.assigned_to:
                recipients.append(complaint.assigned_to)
            if complaint.department and complaint.department.manager:
                recipients.append(complaint.department.manager)

        elif event_type == 'resolved':
            # Notify patient
            recipients.append(complaint.patient)

        elif event_type == 'closed':
            # Notify patient
            recipients.append(complaint.patient)

        # Send notifications
        notification_count = 0
        for recipient in recipients:
            try:
                # Check if NotificationService has send_notification method
                if hasattr(NotificationService, 'send_notification'):
                    NotificationService.send_notification(
                        recipient=recipient,
                        title=f"Complaint {event_type.title()}: {complaint.title[:50]}",
                        message=f"Complaint #{str(complaint.id)[:8]} has been {event_type}.",
                        notification_type='complaint',
                        related_object=complaint
                    )
                    notification_count += 1
                else:
                    logger.warning(f"NotificationService.send_notification method not available")
            except Exception as e:
                logger.error(f"Failed to send notification to {recipient}: {str(e)}")

        logger.info(f"Sent {notification_count} notifications for complaint {complaint_id} event: {event_type}")

        return {
            'status': 'sent',
            'notification_count': notification_count,
            'event_type': event_type
        }

    except Complaint.DoesNotExist:
        error_msg = f"Complaint {complaint_id} not found"
        logger.error(error_msg)
        return {'status': 'error', 'reason': error_msg}
    except Exception as e:
        error_msg = f"Error sending complaint notification: {str(e)}"
        logger.error(error_msg, exc_info=True)
        return {'status': 'error', 'reason': error_msg}