bmc_hub/app/services/customer_consistency.py

"""
Customer Data Consistency Service
Compares customer data across BMC Hub, vTiger Cloud, and e-conomic
"""
import logging
import asyncio
from typing import Dict, List, Optional, Tuple, Any
from app.core.database import execute_query, execute_query_single, execute_update, execute_insert
from app.services.vtiger_service import VTigerService
from app.services.economic_service import EconomicService
from app.core.config import settings

logger = logging.getLogger(__name__)


class CustomerConsistencyService:
    """Service for checking and syncing customer data across systems"""
    
    # Field mapping: hub_field -> (vtiger_field, economic_field)
    FIELD_MAP = {
        'name': ('accountname', 'name'),
        'cvr_number': ('cf_856', 'corporateIdentificationNumber'),
        'address': ('bill_street', 'address'),
        'city': ('bill_city', 'city'),
        'postal_code': ('bill_code', 'zip'),
        'country': ('bill_country', 'country'),
        'phone': ('phone', 'telephoneAndFaxNumber'),
        'mobile_phone': ('mobile', 'mobilePhone'),
        'email': ('email1', 'email'),
        'website': ('website', 'website'),
        'invoice_email': ('email2', 'email'),
    }
    
    def __init__(self):
        self.vtiger = VTigerService()
        self.economic = EconomicService()

    @staticmethod
    def _clean_contact_value(value: Any) -> Optional[str]:
        if value is None:
            return None
        text = str(value).strip()
        return text or None

    @classmethod
    def _normalize_contact(cls, source: str, row: Dict[str, Any]) -> Dict[str, Any]:
        first_name = cls._clean_contact_value(
            row.get('first_name') if source == 'hub' else row.get('firstname')
        )
        last_name = cls._clean_contact_value(
            row.get('last_name') if source == 'hub' else row.get('lastname')
        )
        email = cls._clean_contact_value(row.get('email'))
        phone = cls._clean_contact_value(row.get('phone'))
        mobile = cls._clean_contact_value(row.get('mobile'))
        title = cls._clean_contact_value(row.get('title'))
        department = cls._clean_contact_value(row.get('department'))
        vtiger_id = cls._clean_contact_value(
            row.get('vtiger_id') if source == 'hub' else row.get('id')
        )
        display_name = " ".join(part for part in [first_name, last_name] if part).strip()
        if not display_name:
            display_name = email or phone or mobile or vtiger_id or 'Ukendt kontakt'

        return {
            'id': row.get('id'),
            'vtiger_id': vtiger_id,
            'first_name': first_name,
            'last_name': last_name,
            'email': email,
            'phone': phone,
            'mobile': mobile,
            'title': title,
            'department': department,
            'display_name': display_name,
            'is_primary': bool(row.get('is_primary')) if source == 'hub' else False,
            'role': cls._clean_contact_value(row.get('role')) if source == 'hub' else None,
            'customer_count': row.get('customer_count', 0) or 0,
        }

    @staticmethod
    def _contact_match_key(contact: Dict[str, Any]) -> str:
        vtiger_id = str(contact.get('vtiger_id') or '').strip().lower()
        if vtiger_id:
            return f"vtiger:{vtiger_id}"

        email = str(contact.get('email') or '').strip().lower()
        if email:
            return f"email:{email}"

        first_name = str(contact.get('first_name') or '').strip().lower()
        last_name = str(contact.get('last_name') or '').strip().lower()
        phone = str(contact.get('phone') or '').strip().lower()
        mobile = str(contact.get('mobile') or '').strip().lower()
        return f"name:{first_name}|{last_name}|{phone}|{mobile}"

    async def fetch_hub_contacts(self, customer_id: int) -> List[Dict[str, Any]]:
        query = """
            SELECT
                c.*,
                cc.is_primary,
                cc.role,
                (
                    SELECT COUNT(*)
                    FROM contact_companies cc2
                    WHERE cc2.contact_id = c.id
                ) AS customer_count
            FROM contacts c
            JOIN contact_companies cc ON cc.contact_id = c.id
            WHERE cc.customer_id = %s
              AND c.is_active = TRUE
            ORDER BY cc.is_primary DESC, c.first_name, c.last_name, c.id
        """
        rows = await asyncio.to_thread(execute_query, query, (customer_id,))
        return [self._normalize_contact('hub', row) for row in (rows or [])]

    async def fetch_vtiger_contacts(self, vtiger_customer_id: Optional[str]) -> List[Dict[str, Any]]:
        if not vtiger_customer_id or not settings.VTIGER_URL:
            return []

        safe_customer_id = self.vtiger._sanitize_vtiger_id(vtiger_customer_id)
        if not safe_customer_id:
            return []

        rows: List[Dict[str, Any]] = []
        seen_ids = set()

        # vTiger installations are inconsistent about which relation field is exposed on Contacts.
        relation_fields = ("account_id", "accountid", "parent_id", "account")
        for field_name in relation_fields:
            query = f"SELECT * FROM Contacts WHERE {field_name}='{safe_customer_id}';"
            result = await self.vtiger.query(query)
            for row in (result or []):
                row_id = str(row.get("id") or "").strip()
                dedupe_key = row_id or f"{row.get('email')}|{row.get('firstname')}|{row.get('lastname')}"
                if dedupe_key in seen_ids:
                    continue
                seen_ids.add(dedupe_key)
                rows.append(row)

        return [self._normalize_contact('vtiger', row) for row in rows]
    
    @staticmethod
    def normalize_value(value: Any) -> Optional[str]:
        """
        Normalize value for comparison
        - Convert to string
        - Strip whitespace
        - Lowercase
        - Convert empty strings to None
        """
        if value is None:
            return None
        
        # Convert to string
        str_value = str(value).strip()
        
        # Empty string to None
        if not str_value:
            return None
        
        # Lowercase for case-insensitive comparison
        return str_value.lower()
    
    async def fetch_all_data(self, customer_id: int) -> Dict[str, Any]:
        """
        Fetch customer data from all three systems in parallel
        
        Args:
            customer_id: Hub customer ID
            
        Returns:
            Dict with keys 'hub', 'vtiger', 'economic' containing raw data (or None)
        """
        logger.info(f"🔍 Fetching customer data from all systems for customer {customer_id}")
        
        # Fetch Hub data first to get mapping IDs
        hub_query = """
            SELECT * FROM customers WHERE id = %s
        """
        hub_data = await asyncio.to_thread(execute_query_single, hub_query, (customer_id,))
        
        if not hub_data:
            raise ValueError(f"Customer {customer_id} not found in Hub")
        
        # Prepare async tasks for vTiger and e-conomic
        vtiger_task = None
        economic_task = None
        hub_contacts_task = self.fetch_hub_contacts(customer_id)
        vtiger_contacts_task = None
        
        # Fetch vTiger data if we have an ID and vTiger is configured
        if hub_data.get('vtiger_id') and settings.VTIGER_URL:
            vtiger_task = self.vtiger.get_account_by_id(hub_data['vtiger_id'])
            vtiger_contacts_task = self.fetch_vtiger_contacts(hub_data['vtiger_id'])
        
        # Fetch e-conomic data if we have a customer number and e-conomic is configured
        if hub_data.get('economic_customer_number') and settings.ECONOMIC_APP_SECRET_TOKEN:
            economic_task = self.economic.get_customer(hub_data['economic_customer_number'])
        
        # Parallel fetch with error handling
        tasks = {}
        if vtiger_task:
            tasks['vtiger'] = vtiger_task
        if economic_task:
            tasks['economic'] = economic_task
        tasks['hub_contacts'] = hub_contacts_task
        if vtiger_contacts_task:
            tasks['vtiger_contacts'] = vtiger_contacts_task
        
        results = {}
        if tasks:
            task_results = await asyncio.gather(
                *tasks.values(),
                return_exceptions=True
            )
            
            # Map results back
            for key, result in zip(tasks.keys(), task_results):
                if isinstance(result, Exception):
                    logger.error(f"❌ Error fetching {key} data: {result}")
                    results[key] = None
                else:
                    results[key] = result
        
        return {
            'hub': hub_data,
            'vtiger': results.get('vtiger'),
            'economic': results.get('economic'),
            'hub_contacts': results.get('hub_contacts') or [],
            'vtiger_contacts': results.get('vtiger_contacts') or [],
        }
    
    @classmethod
    def compare_data(cls, all_data: Dict[str, Optional[Dict[str, Any]]]) -> Dict[str, Dict[str, Any]]:
        """
        Compare data across systems and identify discrepancies
        
        Args:
            all_data: Dict with 'hub', 'vtiger', 'economic' data (values may be None)
            
        Returns:
            Dict of discrepancies: {
                field_name: {
                    'hub': value,
                    'vtiger': value,
                    'economic': value,
                    'discrepancy': True/False
                }
            }
        """
        discrepancies = {}
        hub_data = all_data.get('hub', {})
        vtiger_data = all_data.get('vtiger', {})
        economic_data = all_data.get('economic', {})
        
        for hub_field, (vtiger_field, economic_field) in cls.FIELD_MAP.items():
            # Get raw values
            hub_value = hub_data.get(hub_field)
            vtiger_value = vtiger_data.get(vtiger_field) if vtiger_data else None
            economic_value = economic_data.get(economic_field) if economic_data else None
            
            # Normalize for comparison
            hub_norm = cls.normalize_value(hub_value)
            vtiger_norm = cls.normalize_value(vtiger_value)
            economic_norm = cls.normalize_value(economic_value)
            
            # Check if all values are the same
            # Only compare systems that are available
            available_values = []
            if hub_data:
                available_values.append(hub_norm)
            if vtiger_data:
                available_values.append(vtiger_norm)
            if economic_data:
                available_values.append(economic_norm)
            
            # Has discrepancy if there are different non-None values
            has_discrepancy = len(set(available_values)) > 1 if len(available_values) > 1 else False
            
            discrepancies[hub_field] = {
                'hub': hub_value,
                'vtiger': vtiger_value,
                'economic': economic_value,
                'discrepancy': has_discrepancy
            }
        
        return discrepancies

    @classmethod
    def compare_contacts(cls, all_data: Dict[str, Any]) -> List[Dict[str, Any]]:
        hub_contacts = all_data.get('hub_contacts') or []
        vtiger_contacts = all_data.get('vtiger_contacts') or []

        hub_by_key = {cls._contact_match_key(contact): contact for contact in hub_contacts}
        discrepancies: List[Dict[str, Any]] = []

        for vtiger_contact in vtiger_contacts:
            match_key = cls._contact_match_key(vtiger_contact)
            hub_contact = hub_by_key.get(match_key)
            if not hub_contact:
                discrepancies.append({
                    'match_key': match_key,
                    'action': 'create_or_link',
                    'reason': 'Kontakten findes i vTiger men ikke på denne kunde i Hub',
                    'hub': None,
                    'vtiger': vtiger_contact,
                    'selectable': True,
                })
                continue

            changed_fields = []
            for field in ('first_name', 'last_name', 'email', 'phone', 'mobile', 'title', 'department'):
                if cls.normalize_value(hub_contact.get(field)) != cls.normalize_value(vtiger_contact.get(field)):
                    changed_fields.append(field)

            if changed_fields:
                discrepancies.append({
                    'match_key': match_key,
                    'action': 'update_hub',
                    'reason': 'Kontakt findes begge steder men har feltforskelle',
                    'hub': hub_contact,
                    'vtiger': vtiger_contact,
                    'changed_fields': changed_fields,
                    'selectable': True,
                })
            else:
                discrepancies.append({
                    'match_key': match_key,
                    'action': 'matched',
                    'reason': 'Kontakt findes allerede i Hub og matcher vTiger',
                    'hub': hub_contact,
                    'vtiger': vtiger_contact,
                    'changed_fields': [],
                    'selectable': False,
                })

        return discrepancies
    
    async def sync_field(
        self, 
        customer_id: int, 
        field_name: str, 
        source_system: str, 
        source_value: Any
    ) -> Dict[str, bool]:
        """
        Sync a field value to all enabled systems
        
        Args:
            customer_id: Hub customer ID
            field_name: Hub field name (from FIELD_MAP keys)
            source_system: 'hub', 'vtiger', or 'economic'
            source_value: The correct value to sync
            
        Returns:
            Dict with sync status: {'hub': True/False, 'vtiger': True/False, 'economic': True/False}
        """
        logger.info(f"🔄 Syncing {field_name} from {source_system} with value: {source_value}")
        
        if field_name not in self.FIELD_MAP:
            raise ValueError(f"Unknown field: {field_name}")
        
        _, economic_field = self.FIELD_MAP[field_name]
        
        # Fetch Hub data to get mapping IDs
        hub_query = "SELECT * FROM customers WHERE id = %s"
        hub_data = await asyncio.to_thread(execute_query_single, hub_query, (customer_id,))
        
        if not hub_data:
            raise ValueError(f"Customer {customer_id} not found")
        
        results = {}
        
        # Update Hub if not the source
        if source_system != 'hub':
            try:
                update_query = f"UPDATE customers SET {field_name} = %s WHERE id = %s"
                await asyncio.to_thread(execute_update, update_query, (source_value, customer_id))
                results['hub'] = True
                logger.info(f"✅ Hub {field_name} updated")
            except Exception as e:
                logger.error(f"❌ Failed to update Hub: {e}")
                results['hub'] = False
        else:
            results['hub'] = True  # Already correct
        
        # vTiger is read-only for this workflow.
        results['vtiger'] = True
        
        # Update e-conomic if enabled and not the source
        if settings.ECONOMIC_SYNC_ENABLED and source_system != 'economic' and hub_data.get('economic_customer_number'):
            try:
                # e-conomic update requires different handling based on field
                update_data = {economic_field: source_value}
                
                # Check safety flags
                if settings.ECONOMIC_READ_ONLY or settings.ECONOMIC_DRY_RUN:
                    logger.warning(f"⚠️ e-conomic update blocked by safety flags (READ_ONLY={settings.ECONOMIC_READ_ONLY}, DRY_RUN={settings.ECONOMIC_DRY_RUN})")
                    results['economic'] = False
                else:
                    await self.economic.update_customer(hub_data['economic_customer_number'], update_data)
                    results['economic'] = True
                    logger.info(f"✅ e-conomic {economic_field} updated")
            except Exception as e:
                logger.error(f"❌ Failed to update e-conomic: {e}")
                results['economic'] = False
        else:
            results['economic'] = True  # Not applicable or already correct

        return results

    async def sync_vtiger_contacts_to_hub(
        self,
        customer_id: int,
        selected_match_keys: List[str],
    ) -> Dict[str, Any]:
        if not selected_match_keys:
            return {"selected": 0, "created": 0, "updated": 0, "linked": 0, "skipped": 0}

        all_data = await self.fetch_all_data(customer_id)
        discrepancies = self.compare_contacts(all_data)
        selected = {str(item or '').strip() for item in selected_match_keys if str(item or '').strip()}
        selected_rows = [row for row in discrepancies if row.get('match_key') in selected]

        stats = {"selected": len(selected_rows), "created": 0, "updated": 0, "linked": 0, "skipped": 0}

        for row in selected_rows:
            vtiger_contact = row.get('vtiger') or {}
            hub_contact = row.get('hub')
            if not vtiger_contact:
                stats["skipped"] += 1
                continue

            contact_id = hub_contact.get('id') if hub_contact else None
            if not contact_id:
                vtiger_id = vtiger_contact.get("vtiger_id")
                email = vtiger_contact.get("email")
                if vtiger_id:
                    existing_global = await asyncio.to_thread(
                        execute_query_single,
                        "SELECT id FROM contacts WHERE vtiger_id = %s LIMIT 1",
                        (vtiger_id,),
                    )
                    contact_id = existing_global.get("id") if existing_global else None
                if not contact_id and email:
                    existing_global = await asyncio.to_thread(
                        execute_query_single,
                        "SELECT id FROM contacts WHERE LOWER(COALESCE(email, '')) = %s LIMIT 1",
                        (str(email).strip().lower(),),
                    )
                    contact_id = existing_global.get("id") if existing_global else None

            if contact_id:
                update_fields = {
                    "first_name": vtiger_contact.get("first_name"),
                    "last_name": vtiger_contact.get("last_name"),
                    "email": vtiger_contact.get("email"),
                    "phone": vtiger_contact.get("phone"),
                    "mobile": vtiger_contact.get("mobile"),
                    "title": vtiger_contact.get("title"),
                    "department": vtiger_contact.get("department"),
                    "vtiger_id": vtiger_contact.get("vtiger_id"),
                }
                await asyncio.to_thread(
                    execute_update,
                    """
                    UPDATE contacts
                    SET first_name = %s,
                        last_name = %s,
                        email = %s,
                        phone = %s,
                        mobile = %s,
                        title = %s,
                        department = %s,
                        vtiger_id = COALESCE(%s, vtiger_id)
                    WHERE id = %s
                    """,
                    (
                        update_fields["first_name"],
                        update_fields["last_name"],
                        update_fields["email"],
                        update_fields["phone"],
                        update_fields["mobile"],
                        update_fields["title"],
                        update_fields["department"],
                        update_fields["vtiger_id"],
                        contact_id,
                    ),
                )
                stats["updated"] += 1
            else:
                contact_id = await asyncio.to_thread(
                    execute_insert,
                    """
                    INSERT INTO contacts (
                        first_name, last_name, email, phone, mobile, title, department, vtiger_id
                    )
                    VALUES (%s, %s, %s, %s, %s, %s, %s, %s)
                    RETURNING id
                    """,
                    (
                        vtiger_contact.get("first_name"),
                        vtiger_contact.get("last_name"),
                        vtiger_contact.get("email"),
                        vtiger_contact.get("phone"),
                        vtiger_contact.get("mobile"),
                        vtiger_contact.get("title"),
                        vtiger_contact.get("department"),
                        vtiger_contact.get("vtiger_id"),
                    ),
                )
                stats["created"] += 1

            link_exists = await asyncio.to_thread(
                execute_query_single,
                """
                SELECT id
                FROM contact_companies
                WHERE contact_id = %s AND customer_id = %s
                """,
                (contact_id, customer_id),
            )
            if not link_exists:
                await asyncio.to_thread(
                    execute_update,
                    """
                    INSERT INTO contact_companies (contact_id, customer_id, is_primary, role)
                    VALUES (%s, %s, %s, %s)
                    ON CONFLICT (contact_id, customer_id) DO NOTHING
                    """,
                    (contact_id, customer_id, False, None),
                )
                stats["linked"] += 1

        return stats