ferrero-opentext/Python-Version/scripts/a2_to_a3_upload_polling.py

#!/usr/bin/env python3
"""
A2→A3 Upload Handler - Box Folder Polling Version
Polls Box folder for new files with V2 naming, uploads to DAM
Updates status to A3 only when ALL assets for campaign uploaded
Supports --A3update flag to force status update for testing
Compatible with Python 3.6+
"""

import sys
import os
import time
import logging
import argparse

# Add shared library to path
sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))

from shared.config_loader import load_config, load_field_mappings
from shared.dam_client import DAMClient
from shared.box_client import BoxClient
from shared.database import Database
from shared.notifier import Notifier
from shared.filename_parser import FilenameParser
from shared.metadata_extractor_mvp import MetadataExtractorMVP

# Load configuration
config = load_config('config/config.yaml')
field_mappings = load_field_mappings(config)

# Setup logging with rotation
from logging.handlers import RotatingFileHandler

# Create logs directory if it doesn't exist
os.makedirs('logs', exist_ok=True)
os.makedirs('logs/backup', exist_ok=True)

# Configure logging with rotation
# Keep 1 week of active logs (7 days * 10MB = 70MB)
# Backup rotates keep 4 weeks (28 backups * 10MB = 280MB total)
log_handler = RotatingFileHandler(
    'logs/a2_to_a3.log',
    maxBytes=10*1024*1024,  # 10MB per file
    backupCount=28  # Keep 28 rotated files (approximately 1 month)
)
log_handler.setLevel(logging.INFO)
log_handler.setFormatter(logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s'))

console_handler = logging.StreamHandler()
console_handler.setLevel(logging.INFO)
console_handler.setFormatter(logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s'))

logging.basicConfig(
    level=logging.INFO,
    handlers=[log_handler, console_handler]
)

logger = logging.getLogger('A2toA3')

def process_box_file(file_info, dam, box, db, parser, mvp_extractor, config, keep_files=False, dryrun=False):
    """
    Process a single file from Box folder

    Args:
        keep_files: If True, don't delete file from Box after upload (for testing)
        dryrun: If True, build metadata but don't upload to DAM (shows full JSON)

    Returns:
        dict with success, asset_id, tracking_id
    """
    file_id = file_info['id']
    filename = file_info['name']

    logger.info("Processing: {}".format(filename))

    try:
        # 1. Parse V2 filename
        parsed = parser.parse_filename(filename)

        if not parsed['is_valid']:
            raise ValueError("Invalid V2 filename: {} - {}".format(
                filename, ', '.join(parsed['validation_errors'])
            ))

        tracking_id = parsed['tracking_id']
        tracking_mode = parsed.get('tracking_mode', 'full')
        subfolder_path = file_info.get('subfolder_path')

        if not tracking_id:
            raise ValueError("No tracking ID in filename")

        logger.info("Tracking ID: {} (mode: {})".format(tracking_id, tracking_mode))
        if subfolder_path:
            logger.info("From Box subfolder: {} -> will create in DAM".format(subfolder_path))

        # 2. Load master metadata from database
        master_asset = db.get_master_asset(tracking_id)

        if not master_asset:
            raise ValueError("No master asset for tracking ID: {}".format(tracking_id))

        # 3. Get CreativeX score from database (lookup by original Box filename)
        # The PDF contains the filename field with the full name (job + tracking ID)
        # So we lookup using the original filename from Box, not the stripped version
        creativex_data = db.get_creativex_score_by_filename(filename)

        # Build box_metadata dict (for compatibility with existing code)
        if creativex_data:
            box_metadata = {
                'score': creativex_data['quality_score'],
                'url': creativex_data['creativex_url']
            }
            logger.info("CreativeX score found in database: Score={}, URL={}".format(
                creativex_data['quality_score'], creativex_data['creativex_url']
            ))
            creativex_found = True
        else:
            # Use default values when no CreativeX score found
            box_metadata = {
                'score': '0',
                'url': 'https://app.creativex.com/preflight/pretests'
            }
            logger.warning("No CreativeX score found for: {} - Using default values (Score: 0, Placeholder URL)".format(
                filename
            ))
            creativex_found = False

        # 4. Download from Box
        temp_file = os.path.join('temp/downloads', filename)
        box.download_file(file_id, temp_file)

        # 5. Get clean filename
        clean_filename = parser.strip_upload_components(filename)

        # 6. Build MVP asset representation with CreativeX data from database
        asset_rep = mvp_extractor.build_mvp_asset_representation(
            master_metadata=master_asset['full_metadata'],
            clean_filename=clean_filename,
            parsed_filename=parsed,
            box_metadata=box_metadata,  # Pass CreativeX data from database
            tracking_mode=tracking_mode  # Pass tracking mode for folder-only handling
        )

        # DRYRUN MODE: Display full asset representation and exit
        if dryrun:
            import json
            logger.info("")
            logger.info("=" * 80)
            logger.info("DRYRUN MODE - Asset Representation (will NOT upload to DAM)")
            logger.info("=" * 80)
            logger.info("")
            logger.info("FULL ASSET REPRESENTATION (JSON):")
            logger.info("")
            logger.info(json.dumps(asset_rep, indent=2, ensure_ascii=False))
            logger.info("")
            logger.info("=" * 80)
            logger.info("Field Count: {} fields".format(len(asset_rep)))
            logger.info("=" * 80)
            logger.info("")
            logger.info("CreativeX Status:")
            logger.info("  Found in database: {}".format(creativex_found))
            logger.info("  Score: {}".format(box_metadata.get('score')))
            logger.info("  URL: {}".format(box_metadata.get('url')))
            logger.info("")
            logger.info("DRYRUN: No upload performed, file kept in Box")
            logger.info("=" * 80)

            return {
                'success': True,
                'asset_id': 'DRYRUN_NO_UPLOAD',
                'tracking_id': tracking_id,
                'filename': filename,
                'clean_filename': clean_filename,
                'creativex_found': creativex_found,
                'creativex_score': box_metadata.get('score', '0'),
                'creativex_url': box_metadata.get('url', 'https://app.creativex.com/preflight/pretests'),
                'dryrun': True
            }

        # 7. Rename to clean filename
        clean_temp_file = os.path.join('temp/downloads', clean_filename)
        if os.path.exists(clean_temp_file):
            os.remove(clean_temp_file)
        os.rename(temp_file, clean_temp_file)

        # 7. Upload to DAM (with subfolder structure if present)
        upload_folder_id = master_asset['upload_directory']  # Base "01. Final Assets" folder

        # If file was in a Box subfolder, create/use matching DAM subfolder
        if subfolder_path:
            logger.info("Creating DAM subfolder path: {}".format(subfolder_path))
            upload_folder_id = dam.get_or_create_subfolder_path(
                base_folder_id=upload_folder_id,
                subfolder_path=subfolder_path
            )
            logger.info("Will upload to: 01. Final Assets/{}".format(subfolder_path))

        upload_result = dam.upload_asset(
            file_path=clean_temp_file,
            folder_id=upload_folder_id,
            asset_representation=asset_rep
        )

        if not upload_result['success']:
            raise Exception("Upload failed: {}".format(upload_result.get('error')))

        # 8. Store derivative record
        db.store_derivative_asset(
            tracking_id=tracking_id,
            master_asset_id=None,
            dam_asset_id=upload_result['asset_id'],
            filename=clean_filename
        )

        # 9. Delete file from Box after successful upload (unless --keep-files flag set)
        if keep_files:
            logger.info("--keep-files flag set - File kept in Box: {}".format(filename))
        else:
            try:
                box_file = box.client.file(file_id)
                box_file.delete()
                logger.info("Deleted file from Box: {}".format(filename))
            except Exception as e:
                logger.warning("Could not delete file from Box: {}".format(str(e)))

        # 10. Clean up local temp file
        os.remove(clean_temp_file)

        logger.info("✓ Success: {} → Asset ID: {}".format(filename, upload_result['asset_id']))

        return {
            'success': True,
            'asset_id': upload_result['asset_id'],
            'tracking_id': tracking_id,
            'filename': filename,
            'clean_filename': clean_filename,
            'creativex_found': creativex_found,
            'creativex_score': box_metadata.get('score', '0'),
            'creativex_url': box_metadata.get('url', 'https://app.creativex.com/preflight/pretests')
        }

    except Exception as e:
        logger.error("✗ Failed: {} - {}".format(filename, str(e)))
        return {
            'success': False,
            'error': str(e),
            'filename': filename,
            'tracking_id': tracking_id if 'tracking_id' in locals() else None
        }

def main():
    """Main entry point - single run mode"""
    # Parse command-line arguments
    parser_args = argparse.ArgumentParser(description='Ferrero A2→A3 Upload Handler')
    parser_args.add_argument('--auth-pfx', action='store_true',
                       help='Use mTLS certificate authentication instead of OAuth2')
    parser_args.add_argument('--A3update', action='store_true',
                       help='Force update campaign status A2→A3 after upload (for testing)')
    parser_args.add_argument('--keep-files', action='store_true',
                       help='Keep files in Box after upload (don\'t delete, for testing)')
    parser_args.add_argument('--dryrun', action='store_true',
                       help='Build metadata but don\'t upload to DAM (shows full JSON for debugging)')
    args = parser_args.parse_args()

    logger.info("=" * 60)
    logger.info("Ferrero A2→A3 Upload Handler Starting (Polling Mode)")
    if args.auth_pfx:
        logger.info("Authentication: mTLS Certificate (--auth-pfx)")
    else:
        logger.info("Authentication: OAuth2 (default)")
    if args.A3update:
        logger.info("Mode: Auto-update campaign status A2→A3 (--A3update)")
    if args.keep_files:
        logger.info("Mode: Keep files in Box after upload (--keep-files)")
    if args.dryrun:
        logger.info("Mode: DRYRUN - Build metadata but DON'T upload (--dryrun)")
    logger.info("=" * 60)

    # Initialize clients
    dam = DAMClient(config, use_mtls=args.auth_pfx)
    # Use A2→A3 Box folder for polling
    box = BoxClient(config, root_folder_id=config['box'].get('root_folder_a2_a3'))
    db = Database(config)
    notifier = Notifier(config)
    parser = FilenameParser()
    mvp_extractor = MetadataExtractorMVP(field_mappings)

    # Test connections
    logger.info("Testing connections...")
    if not dam.test_connection():
        logger.error("DAM connection failed")
        sys.exit(1)

    if not box.test_connection():
        logger.error("Box connection failed")
        sys.exit(1)

    if not db.test_connection():
        logger.error("Database connection failed")
        sys.exit(1)

    logger.info("All connections OK")
    logger.info("")

    try:
        # Get Box folder ID to poll
        box_folder_id = config['box'].get('root_folder_a2_a3', config['box'].get('root_folder_id'))

        logger.info("Polling Box folder: {}".format(box_folder_id))

        # List files recursively in Box folder (skips 1st level job folders, preserves 2nd+ levels)
        files = box.list_folder_files_recursive(box_folder_id)
        logger.info("Recursive scan complete")

        if not files:
            logger.info("No files found in Box folder - exiting")
            db.close()
            sys.exit(0)

        logger.info("Found {} files in Box folder".format(len(files)))

        # Show subfolder distribution
        subfolders = set([f.get('subfolder_path') for f in files if f.get('subfolder_path')])
        if subfolders:
            logger.info("Files in {} subfolder(s): {}".format(
                len(subfolders), ', '.join(sorted(subfolders))
            ))

        files_at_root = len([f for f in files if not f.get('subfolder_path')])
        if files_at_root:
            logger.info("Files at job level (will go to DAM root): {}".format(files_at_root))

        # Filter for V2 filenames only
        valid_files = []
        for file_info in files:
            parsed = parser.parse_filename(file_info['name'])
            if parsed['is_valid'] and parsed.get('tracking_id'):
                valid_files.append(file_info)
            else:
                logger.info("Skipping invalid V2 file: {} - Errors: {}".format(
                    file_info['name'], parsed.get('validation_errors', [])
                ))

        logger.info("Found {} valid V2 files to process".format(len(valid_files)))

        if not valid_files:
            logger.info("No valid V2 files to process - exiting")
            db.close()
            sys.exit(0)

        # Process files one at a time (process first file only)
        file_info = valid_files[0]
        logger.info("Processing first file only (more will be processed on next run)")
        logger.info("")

        result = process_box_file(file_info, dam, box, db, parser, mvp_extractor, config, keep_files=args.keep_files, dryrun=args.dryrun)

        if result['success']:
            logger.info("")
            logger.info("=" * 60)
            logger.info("✓ File processed successfully")
            logger.info("  Filename: {}".format(result['filename']))
            logger.info("  Clean filename: {}".format(result['clean_filename']))
            logger.info("  Asset ID: {}".format(result['asset_id']))
            logger.info("  Tracking ID: {}".format(result['tracking_id']))
            logger.info("=" * 60)

            # Send success email notification with details
            # Get master asset info for email
            master_asset = db.get_master_asset(result['tracking_id'])

            notifier.send_email(
                template_name='a2_to_a3_file_uploaded',
                recipients=config['notifications']['recipients']['success'],
                data={
                    'filename': result['filename'],
                    'clean_filename': result['clean_filename'],
                    'asset_id': result['asset_id'],
                    'tracking_id': result['tracking_id'],
                    'master_asset_name': master_asset.get('opentext_id', 'Unknown') if master_asset else 'Unknown',
                    'upload_folder': master_asset.get('upload_directory', 'Unknown') if master_asset else 'Unknown',
                    'box_folder': box_folder_id,
                    'creativex_found': result.get('creativex_found', False),
                    'creativex_score': result.get('creativex_score', '0'),
                    'creativex_url': result.get('creativex_url', 'https://app.creativex.com/preflight/pretests')
                }
            )

            # Update campaign status A2→A3 if --A3update flag is set
            if args.A3update and master_asset:
                logger.info("")
                logger.info("--A3update flag set - Attempting to update campaign status")

                # Get campaign ID from master asset metadata
                full_metadata = master_asset.get('full_metadata', {})

                # Extract campaign ID from inherited_metadata_collections
                campaign_id = None
                collections = full_metadata.get('inherited_metadata_collections', [])

                for collection in collections:
                    if collection.get('container_type_name') == 'L7+ - CAMPAIGN':
                        campaign_id = collection.get('container_id')
                        break

                if campaign_id:
                    logger.info("Found campaign ID: {}".format(campaign_id))
                    logger.info("Updating campaign status A2 → A3...")

                    status_result = dam.update_campaign_status(campaign_id, 'A3')

                    if status_result['success']:
                        logger.info("✓ Campaign status updated successfully: A2 → A3")
                    else:
                        logger.error("✗ Campaign status update failed: {}".format(status_result.get('error')))
                else:
                    logger.warning("⚠ Campaign ID not found in master asset metadata - cannot update status")

            db.close()
            sys.exit(0)
        else:
            logger.warning("")
            logger.warning("=" * 60)
            logger.warning("✗ File processing failed")
            logger.warning("  Filename: {}".format(result['filename']))
            logger.warning("  Error: {}".format(result['error']))
            logger.warning("=" * 60)

            db.close()
            sys.exit(1)

    except Exception as e:
        logger.critical("Script error: {}".format(str(e)))
        db.close()
        sys.exit(1)

if __name__ == '__main__':
    main()