src/utils/s3_logger.py

"""
Structured JSON Logging for S3 Backup/Restore Operations

Provides consistent logging format for observability and debugging.
All logs are emitted as JSON for easy parsing by log aggregation systems.
"""

import json
import logging
from datetime import datetime
from typing import Optional, Dict, Any


logger = logging.getLogger(__name__)


def _log_event(event: str, **kwargs) -> None:
    """
    Internal helper to emit structured JSON log messages.
    
    Args:
        event: Event name (e.g., "backup_started", "restore_completed")
        **kwargs: Additional event-specific fields
    """
    log_data = {
        'event': event,
        'timestamp': datetime.utcnow().isoformat() + 'Z',
        **kwargs
    }
    
    # Mask sensitive data
    if 's3_key' in log_data:
        # S3 keys are safe to log
        pass
    if 'error' in log_data and isinstance(log_data['error'], Exception):
        # Convert exception to string
        log_data['error'] = str(log_data['error'])
    
    logger.info(json.dumps(log_data))


def backup_started(db_path: str, db_size_bytes: int) -> None:
    """
    Log that a backup operation has started.
    
    Args:
        db_path: Path to the database file being backed up
        db_size_bytes: Size of the database in bytes
    """
    _log_event(
        'backup_started',
        db_path=db_path,
        db_size_bytes=db_size_bytes
    )


def backup_completed(
    duration_seconds: float,
    s3_key: str,
    upload_size_bytes: int
) -> None:
    """
    Log that a backup operation completed successfully.
    
    Args:
        duration_seconds: Time taken for the backup operation
        s3_key: S3 object key where backup was stored
        upload_size_bytes: Size of uploaded file in bytes
    """
    _log_event(
        'backup_completed',
        duration_seconds=round(duration_seconds, 2),
        s3_key=s3_key,
        upload_size_bytes=upload_size_bytes
    )


def backup_failed(
    error: str,
    retry_attempt: Optional[int] = None,
    max_retries: Optional[int] = None
) -> None:
    """
    Log that a backup operation failed.
    
    Args:
        error: Error message or exception
        retry_attempt: Current retry attempt number (if retrying)
        max_retries: Maximum number of retries (if retrying)
    """
    log_data = {
        'error': str(error)
    }
    
    if retry_attempt is not None:
        log_data['retry_attempt'] = retry_attempt
    if max_retries is not None:
        log_data['max_retries'] = max_retries
    
    _log_event('backup_failed', **log_data)


def restore_started() -> None:
    """Log that a restore operation has started."""
    _log_event('restore_started')


def restore_completed(
    duration_seconds: float,
    s3_key: Optional[str],
    download_size_bytes: Optional[int],
    result: str
) -> None:
    """
    Log that a restore operation completed.
    
    Args:
        duration_seconds: Time taken for the restore operation
        s3_key: S3 object key that was restored (None if no restore needed)
        download_size_bytes: Size of downloaded file (None if no download)
        result: Restore result enum value (e.g., "restored_from_s3", "local_newer")
    """
    log_data = {
        'duration_seconds': round(duration_seconds, 2),
        'result': result
    }
    
    if s3_key:
        log_data['s3_key'] = s3_key
    if download_size_bytes:
        log_data['download_size_bytes'] = download_size_bytes
    
    _log_event('restore_completed', **log_data)


def restore_fallback(reason: str, fallback_action: str) -> None:
    """
    Log that a restore operation fell back to an alternative.
    
    Args:
        reason: Reason for fallback (e.g., "validation_failed", "network_error")
        fallback_action: Action taken (e.g., "using_local_database", "initializing_empty")
    """
    _log_event(
        'restore_fallback',
        reason=reason,
        fallback_action=fallback_action
    )


def backup_debounced(pending_requests: int, debounce_seconds: int) -> None:
    """
    Log that backup requests are being debounced.
    
    Args:
        pending_requests: Number of backup requests collapsed into one
        debounce_seconds: Debounce period in seconds
    """
    _log_event(
        'backup_debounced',
        pending_requests=pending_requests,
        debounce_seconds=debounce_seconds
    )


def s3_credentials_validated(bucket: str, endpoint: Optional[str]) -> None:
    """
    Log that S3 credentials were successfully validated.
    
    Args:
        bucket: S3 bucket name
        endpoint: S3 endpoint URL (None for AWS S3)
    """
    _log_event(
        's3_credentials_validated',
        bucket=bucket,
        endpoint=endpoint or 'AWS S3'
    )


def s3_credentials_invalid(error: str) -> None:
    """
    Log that S3 credentials validation failed.
    
    Args:
        error: Error message
    """
    _log_event(
        's3_credentials_invalid',
        error=str(error)
    )


def backup_skip_reason(reason: str, operation: str) -> None:
    """
    Log why a backup was skipped.
    
    Args:
        reason: Reason for skipping (e.g., "s3_not_enabled", "non_critical_update")
        operation: Operation that would have triggered backup
    """
    _log_event(
        'backup_skip',
        reason=reason,
        operation=operation
    )