bmc_hub/app/system/backend/sync_router.py

"""
System Sync Router
API endpoints for syncing data between vTiger, e-conomic and Hub
"""

import logging
from fastapi import APIRouter, HTTPException
from typing import Dict, Any
from app.core.database import execute_query
from app.services.vtiger_service import get_vtiger_service
import re

logger = logging.getLogger(__name__)

router = APIRouter()


def normalize_name(name: str) -> str:
    """Normalize company name for matching"""
    if not name:
        return ""
    # Remove common suffixes and punctuation
    name = re.sub(r'\b(A/S|ApS|I/S|IVS|v/)\b', '', name, flags=re.IGNORECASE)
    name = re.sub(r'[^\w\s]', '', name)  # Remove punctuation
    return name.lower().strip()


@router.post("/sync/vtiger")
async def sync_from_vtiger() -> Dict[str, Any]:
    """
    Link vTiger accounts to existing Hub customers
    Matches by CVR or normalized name, updates vtiger_id
    """
    try:
        logger.info("🔄 Starting vTiger link sync...")
        
        vtiger = get_vtiger_service()
        
        # Fetch ALL accounts - vTiger query API doesn't support LIMIT/OFFSET
        # Instead, use recursive queries based on ID to get all records
        all_accounts = []
        last_id = None
        batch_size = 100  # vTiger typically returns ~100 records per query
        
        while True:
            # Build query with ID filter to paginate
            if last_id is None:
                query = f"SELECT id, accountname, email1, siccode, cf_accounts_cvr, website, bill_city, bill_code, bill_country FROM Accounts ORDER BY id LIMIT {batch_size};"
            else:
                # Get records with ID > last_id to continue pagination
                query = f"SELECT id, accountname, email1, siccode, cf_accounts_cvr, website, bill_city, bill_code, bill_country FROM Accounts WHERE id > '{last_id}' ORDER BY id LIMIT {batch_size};"
            
            batch = await vtiger.query(query)
            
            if not batch or len(batch) == 0:
                break
                
            all_accounts.extend(batch)
            last_id = batch[-1].get('id')  # Track last ID for next query
            logger.info(f"📥 Fetched batch: {len(batch)} accounts (last ID: {last_id}, total: {len(all_accounts)})")
            
            # If we got less than batch_size, we've reached the end
            if len(batch) < batch_size:
                break
        
        logger.info(f"📥 Fetched total of {len(all_accounts)} accounts from vTiger")
        accounts = all_accounts
        
        linked_count = 0
        updated_count = 0
        not_found_count = 0
        
        for account in accounts:
            vtiger_id = account.get('id')
            name = account.get('accountname', '').strip()
            cvr = account.get('cf_accounts_cvr') or account.get('siccode')
            
            if not name or not vtiger_id:
                not_found_count += 1
                continue
            
            # Clean CVR number
            if cvr:
                cvr = re.sub(r'\D', '', str(cvr))[:8]  # Remove non-digits, max 8 chars
                if len(cvr) != 8:
                    cvr = None
            
            # Find existing Hub customer by CVR or normalized name
            existing = None
            match_method = None
            
            if cvr:
                existing = execute_query(
                    "SELECT id, name, vtiger_id FROM customers WHERE cvr_number = %s",
                    (cvr,)
                )
                if existing:
                    match_method = "CVR"
            
            if not existing:
                # Match by normalized name
                normalized = normalize_name(name)
                all_customers = execute_query("SELECT id, name, vtiger_id FROM customers")
                for customer in all_customers:
                    if normalize_name(customer['name']) == normalized:
                        existing = [customer]
                        match_method = "navn"
                        break
            
            if existing:
                # Link vTiger ID to existing customer
                current_vtiger_id = existing[0].get('vtiger_id')
                
                # Check if this vtiger_id is already assigned to another customer
                vtiger_owner = execute_query(
                    "SELECT id, name FROM customers WHERE vtiger_id = %s",
                    (vtiger_id,)
                )
                
                if vtiger_owner and vtiger_owner[0]['id'] != existing[0]['id']:
                    # vTiger ID already belongs to another customer - skip
                    logger.warning(f"⚠️  vTiger #{vtiger_id} allerede tilknyttet '{vtiger_owner[0]['name']}' - springer over '{existing[0]['name']}'")
                    not_found_count += 1
                    continue
                
                if current_vtiger_id is None:
                    execute_query(
                        "UPDATE customers SET vtiger_id = %s, last_synced_at = NOW() WHERE id = %s",
                        (vtiger_id, existing[0]['id'])
                    )
                    linked_count += 1
                    logger.info(f"🔗 Linket: {existing[0]['name']} → vTiger #{vtiger_id} (match: {match_method}, CVR: {cvr or 'ingen'})")
                elif current_vtiger_id != vtiger_id:
                    # Update if different vTiger ID
                    execute_query(
                        "UPDATE customers SET vtiger_id = %s, last_synced_at = NOW() WHERE id = %s",
                        (vtiger_id, existing[0]['id'])
                    )
                    updated_count += 1
                    logger.info(f"✏️  Opdateret vTiger ID: {existing[0]['name']} → {vtiger_id} (var: {current_vtiger_id})")
                else:
                    # Already linked, just update timestamp
                    execute_query("UPDATE customers SET last_synced_at = NOW() WHERE id = %s", (existing[0]['id'],))
            else:
                not_found_count += 1
                logger.debug(f"⏭️  Ikke fundet i Hub: '{name}' (CVR: {cvr or 'ingen'}, normalized: '{normalize_name(name)}')")
        
        logger.info(f"✅ vTiger link sync fuldført: {linked_count} nye linket, {updated_count} opdateret, {not_found_count} ikke fundet af {len(accounts)} totalt")
        
        return {
            "status": "success",
            "linked": linked_count,
            "updated": updated_count,
            "not_found": not_found_count,
            "total_processed": len(accounts)
        }
        
    except Exception as e:
        logger.error(f"❌ vTiger sync error: {e}")
        raise HTTPException(status_code=500, detail=str(e))


@router.post("/sync/vtiger-contacts")
async def sync_vtiger_contacts() -> Dict[str, Any]:
    """
    SIMPEL TILGANG - Sync contacts from vTiger and link to customers
    Step 1: Fetch all contacts from vTiger
    Step 2: Create/update contact in Hub
    Step 3: Link to customer IF account_id matches customer.vtiger_id
    """
    try:
        logger.info("🔄 Starting vTiger contacts sync (SIMPEL VERSION)...")
        
        vtiger = get_vtiger_service()
        
        # ===== STEP 1: FETCH ALL CONTACTS =====
        logger.info("📥 STEP 1: Fetching contacts from vTiger...")
        all_contacts = []
        last_id = None
        batch_size = 100
        
        while True:
            if last_id is None:
                query = f"SELECT id, firstname, lastname, email, phone, mobile, title, department, account_id FROM Contacts ORDER BY id LIMIT {batch_size};"
            else:
                query = f"SELECT id, firstname, lastname, email, phone, mobile, title, department, account_id FROM Contacts WHERE id > '{last_id}' ORDER BY id LIMIT {batch_size};"
            
            batch = await vtiger.query(query)
            if not batch or len(batch) == 0:
                break
                
            all_contacts.extend(batch)
            last_id = batch[-1].get('id')
            logger.info(f"  → Batch: {len(batch)} contacts (ID: {last_id}, total: {len(all_contacts)})")
            
            if len(batch) < batch_size:
                break
        
        logger.info(f"✅ STEP 1 DONE: {len(all_contacts)} contacts fetched")
        
        # ===== STEP 2 & 3: PROCESS EACH CONTACT =====
        logger.info("🔄 STEP 2 & 3: Creating/updating contacts and linking to customers...")
        
        created_count = 0
        updated_count = 0
        skipped_count = 0
        linked_count = 0
        debug_count = 0
        
        for i, contact in enumerate(all_contacts, 1):
            vtiger_id = contact.get('id')
            first_name = (contact.get('firstname') or '').strip()
            last_name = (contact.get('lastname') or '').strip()
            email = contact.get('email')
            account_id = contact.get('account_id')  # This is the vTiger Account ID
            
            # Show progress every 100 contacts
            if i % 100 == 0:
                logger.info(f"  → Progress: {i}/{len(all_contacts)} contacts processed...")
            
            # Must have name
            if not first_name and not last_name:
                skipped_count += 1
                continue
            
            # --- STEP 2A: Check if contact exists ---
            existing = execute_query(
                "SELECT id FROM contacts WHERE vtiger_id = %s",
                (vtiger_id,)
            ) if vtiger_id else None
            
            # --- STEP 2B: Create or update contact ---
            contact_id = None
            
            if existing:
                # UPDATE existing
                execute_query("""
                    UPDATE contacts 
                    SET first_name = %s, last_name = %s, email = %s, 
                        phone = %s, mobile = %s, title = %s, department = %s, 
                        updated_at = NOW()
                    WHERE id = %s
                """, (
                    first_name, last_name, email,
                    contact.get('phone'), contact.get('mobile'),
                    contact.get('title'), contact.get('department'),
                    existing[0]['id']
                ))
                contact_id = existing[0]['id']
                updated_count += 1
            else:
                # CREATE new
                result = execute_query("""
                    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
                """, (
                    first_name, last_name, email,
                    contact.get('phone'), contact.get('mobile'),
                    contact.get('title'), contact.get('department'),
                    vtiger_id
                ))
                
                if result and len(result) > 0:
                    contact_id = result[0]['id']
                    created_count += 1
                else:
                    skipped_count += 1
                    continue
            
            # --- DEBUG LOGGING FOR FIRST 20 CONTACTS ---
            if debug_count < 20:
                logger.warning(f"🔍 DEBUG {debug_count+1}: Contact '{first_name} {last_name}' (vtiger_id={vtiger_id}, contact_id={contact_id}, account_id={account_id})")
                debug_count += 1
            
            # --- STEP 3: LINK TO CUSTOMER ---
            if not account_id:
                # No account_id means contact is not linked to any account in vTiger
                if debug_count <= 20:
                    logger.warning(f"  ↳ No account_id - cannot link")
                continue
            
            # Find Hub customer with matching vtiger_id
            customer_result = execute_query(
                "SELECT id, name FROM customers WHERE vtiger_id = %s",
                (account_id,)
            )
            
            if not customer_result or len(customer_result) == 0:
                # No matching customer found
                if debug_count <= 20:
                    logger.warning(f"  ↳ No customer found with vtiger_id={account_id}")
                continue
            
            customer_id = customer_result[0]['id']
            customer_name = customer_result[0]['name']
            
            # Check if link already exists
            existing_link = execute_query(
                "SELECT id FROM contact_companies WHERE contact_id = %s AND customer_id = %s",
                (contact_id, customer_id)
            )
            
            if existing_link:
                # Already linked
                if debug_count <= 20:
                    logger.warning(f"  ↳ Already linked to '{customer_name}'")
                continue
            
            # CREATE LINK
            execute_query(
                "INSERT INTO contact_companies (contact_id, customer_id, is_primary) VALUES (%s, %s, false)",
                (contact_id, customer_id)
            )
            linked_count += 1
            
            if linked_count <= 10:  # Log first 10 successful links
                logger.info(f"🔗 LINKED: {first_name} {last_name} → {customer_name}")
        
        logger.info(f"✅ SYNC COMPLETE: created={created_count}, updated={updated_count}, linked={linked_count}, skipped={skipped_count}, total={len(all_contacts)}")
        
        return {
            "status": "success",
            "created": created_count,
            "updated": updated_count,
            "linked": linked_count,
            "skipped": skipped_count,
            "total_processed": len(all_contacts)
        }
        
    except Exception as e:
        logger.error(f"❌ vTiger contacts sync error: {e}", exc_info=True)
        raise HTTPException(status_code=500, detail=str(e))


@router.post("/sync/economic")
async def sync_from_economic() -> Dict[str, Any]:
    """
    Sync customers from e-conomic (PRIMARY SOURCE)
    Creates/updates Hub customers with e-conomic data
    """
    try:
        logger.info("🔄 Starting e-conomic customer sync (PRIMARY SOURCE)...")
        
        from app.services.economic_service import EconomicService
        economic = EconomicService()
        
        # Get all customers from e-conomic (max 1000 per page)
        all_customers = []
        page = 0
        while True:
            customers = await economic.get_customers(page=page, page_size=1000)
            if not customers:
                break
            all_customers.extend(customers)
            page += 1
            if len(customers) < 1000:  # Last page
                break
        
        economic_customers = all_customers
        logger.info(f"📥 Fetched {len(economic_customers)} customers from e-conomic ({page} pages)")
        
        created_count = 0
        updated_count = 0
        skipped_count = 0
        
        for eco_customer in economic_customers:
            customer_number = eco_customer.get('customerNumber')
            cvr = eco_customer.get('corporateIdentificationNumber')
            name = eco_customer.get('name', '').strip()
            address = eco_customer.get('address', '')
            city = eco_customer.get('city', '')
            zip_code = eco_customer.get('zip', '')
            country = eco_customer.get('country', 'DK')
            email = eco_customer.get('email', '')
            website = eco_customer.get('website', '')
            
            if not customer_number or not name:
                skipped_count += 1
                continue
            
            # Clean CVR
            if cvr:
                cvr = re.sub(r'\D', '', str(cvr))[:8]
                if len(cvr) != 8:
                    cvr = None
            
            # Clean country code (max 2 chars for ISO codes)
            if country:
                country = str(country).strip().upper()[:2]
                if not country:
                    country = 'DK'
            else:
                country = 'DK'
            
            # Extract email domain
            email_domain = email.split('@')[-1] if '@' in email else None
            
            # Check if customer exists by economic_customer_number OR CVR
            existing = execute_query(
                "SELECT id FROM customers WHERE economic_customer_number = %s",
                (customer_number,)
            )
            
            # If not found by customer number, try CVR (to avoid duplicates)
            if not existing and cvr:
                existing = execute_query(
                    "SELECT id FROM customers WHERE cvr_number = %s",
                    (cvr,)
                )
            
            if existing:
                # Update existing customer
                update_query = """
                    UPDATE customers SET 
                        name = %s,
                        cvr_number = %s,
                        email_domain = %s,
                        city = %s,
                        postal_code = %s,
                        country = %s,
                        website = %s,
                        last_synced_at = NOW()
                    WHERE id = %s
                """
                execute_query(update_query, (
                    name, cvr, email_domain, city, zip_code, country, website, existing[0]['id']
                ))
                updated_count += 1
                logger.info(f"✏️  Opdateret: {name} (e-conomic #{customer_number}, CVR: {cvr or 'ingen'})")
            else:
                # Create new customer from e-conomic
                insert_query = """
                    INSERT INTO customers 
                    (name, economic_customer_number, cvr_number, email_domain, 
                     city, postal_code, country, website, last_synced_at)
                    VALUES (%s, %s, %s, %s, %s, %s, %s, %s, NOW())
                    RETURNING id
                """
                result = execute_query(insert_query, (
                    name, customer_number, cvr, email_domain, city, zip_code, country, website
                ))
                
                if result:
                    created_count += 1
                    logger.info(f"✨ Oprettet: {name} (e-conomic #{customer_number}, CVR: {cvr or 'ingen'})")
                else:
                    skipped_count += 1
        
        logger.info(f"✅ e-conomic sync fuldført: {created_count} oprettet, {updated_count} opdateret, {skipped_count} sprunget over af {len(economic_customers)} totalt")
        
        return {
            "status": "success",
            "created": created_count,
            "updated": updated_count,
            "skipped": skipped_count,
            "total_processed": len(economic_customers)
        }
        
    except Exception as e:
        logger.error(f"❌ e-conomic sync error: {e}")
        raise HTTPException(status_code=500, detail=str(e))


@router.post("/sync/cvr-to-economic")
async def sync_cvr_to_economic() -> Dict[str, Any]:
    """
    Find customers in Hub with CVR but without e-conomic customer number
    Search e-conomic for matching CVR and update Hub
    """
    try:
        logger.info("🔄 Starting CVR to e-conomic sync...")
        
        from app.services.economic_service import EconomicService
        economic = EconomicService()
        
        # Find customers with CVR but no economic_customer_number
        customers = execute_query("""
            SELECT id, name, cvr_number 
            FROM customers 
            WHERE cvr_number IS NOT NULL 
              AND cvr_number != '' 
              AND economic_customer_number IS NULL
            LIMIT 100
        """)
        
        logger.info(f"📥 Found {len(customers)} customers with CVR but no e-conomic number")
        
        found_count = 0
        linked_count = 0
        
        for customer in customers:
            cvr = customer['cvr_number']
            
            # Search e-conomic for this CVR
            eco_customer = await economic.search_customer_by_cvr(cvr)
            
            if eco_customer:
                customer_number = eco_customer.get('customerNumber')
                if customer_number:
                    # Update Hub customer
                    execute_query(
                        "UPDATE customers SET economic_customer_number = %s, last_synced_at = NOW() WHERE id = %s",
                        (customer_number, customer['id'])
                    )
                    found_count += 1
                    linked_count += 1
                    logger.info(f"✅ Fundet og linket: {customer['name']} (CVR: {cvr}) → e-conomic kunde #{customer_number}")
                else:
                    found_count += 1
                    logger.warning(f"⚠️ Fundet men mangler kundenummer: {customer['name']} (CVR: {cvr})")
        
        logger.info(f"✅ CVR søgning fuldført: {found_count} fundet, {linked_count} linket af {len(customers)} kontrolleret")
        
        return {
            "status": "not_implemented",
            "message": "Requires e-conomic API search by CVR functionality",
            "found": 0,
            "candidates": len(customers)
        }
        
    except Exception as e:
        logger.error(f"❌ CVR to e-conomic sync error: {e}")
        raise HTTPException(status_code=500, detail=str(e))