Changelog for CLI

Improved

• Database Download Strategy: Reversed the database download logic to attempt a fast, direct download first. If the direct download fails (e.g., due to hosting restrictions), it now automatically falls back to using the PHP streaming API. This logic is consolidated in a new download_db_file function and replaces the v2.2.0 "stream-first" approach.
• Database Backup Efficiency: Refactored the database backup logic for both the backup and sync commands into a new, centralized processDatabaseBackup function.
• Hybrid Batching: The new processDatabaseBackup function introduces a "hybrid batching" strategy, which groups numerous small tables into batches to be exported together while processing very large tables individually in parts. This improves reliability for complex databases.
• Sync Reliability: The sync command's file download step for subsequent syncs is now more robust. It processes new or changed files in chunks (of 2,500 files) and includes a retry mechanism to handle intermittent network or server errors during the download and extraction of a chunk.
• Help Command: Reorganized the help output (disembark or disembark --help) for better readability, grouping backup and sync as "Primary Commands".

Changed

• Removed Backup Confirmation: Removed the "Do you want to proceed? (yes/no)" prompt from the backup command to allow for non-interactive scripting and automation.

Added

• New sync Command: A new disembark sync <site-url> [<folder>] command has been added to create and update a local mirror of a remote site.
  • On the first run, it performs an "initial sync" to download all database tables and file chunks.
  • On subsequent runs, it intelligently compares a local manifest with the remote one, downloading only new or changed files and removing deleted ones.
  • The sync command supports --debug for saving manifest files and --session-id to reuse a manifest from the web UI.
• Session Reuse for backup: The backup command now supports a --session-id=<id> argument. This allows you to reuse an existing backup session from the plugin's web UI, skipping the analysis step and downloading files from the pre-generated manifest.

Improved

• Hosting Download Compatibility: The method for downloading database (.sql) files has been re-architected. This change specifically supports restrictive hosting environments (like Flywheel) that block direct URL access to .sql files.
  • The tool now attempts to download database files using a new download_file_via_stream function, which calls a /wp-json/disembark/v1/stream-file API endpoint to stream the file content via PHP.
  • If this streaming method fails, it automatically falls back to the original direct download method.
• Removed curl Dependency: The download_file_direct function (used for file chunks and as a fallback) was rewritten to use the internal rmccue/requests PHP library. This removes the dependency on the curl command-line utility, increasing portability and reliability.
• Code Refactoring: The file analysis and backup summary logic, which was previously part of the runPreview function, was refactored into new, separate functions: runManifestGeneration and displayBackupSummary. This allows the logic to be reused by the backup, backup --preview, and new sync commands.
• Build Process: The build-phar.php script was improved to manually add only the required disembark script and vendor directory, which creates a smaller and cleaner final .phar file.

Improved

• Backup Efficiency: The backup process has been optimized to run the file analysis and manifest generation only once. The analysis now runs during the initial preview step, and the generated file manifest is reused for the actual backup. This avoids the previous behavior of running the analysis twice (once for preview, once for the backup), significantly speeding up the operation.
• Preview & Cleanup Logic: The runPreview function was refactored to accept a backup_token and return the generated file manifest. The main backup function now correctly calls cleanupTemporaryFiles if the user exits after a --preview or cancels the backup. Previously, runPreview generated its own temporary token and cleaned up after itself.

Changed

• Moved the changelog from the bottom of readme.md to its own dedicated changelog.md file.

Changed

• Re-architected Backup Process: The backup process is now "local-first." The CLI downloads all database tables and file chunks individually and assembles the final .zip file on the user's local machine.
• Removed the dependency on the remote disembark.host/generate-zip script for backup finalization.

Added

• The backup command now creates a local temporary snapshot-<timestamp> directory to build the backup locally.
• Added new local helper functions for zipping (zip_directory), unzipping (unzip_file), and deleting directories (delete_directory) to support the new local backup process.
• Added a download_file_direct function using curl for more reliable downloading of backup parts.
• The backup process now calls a /cleanup-file endpoint to delete temporary backup chunks from the server as they are downloaded.

Improved

• File Manifest: The file manifest generation is now a more robust, multi-step process (initiate, scan, chunkify, process, finalize) to better handle large sites and complex exclusion rules.
• Backup Preview: The backup --preview command was rewritten to use the new multi-step manifest generation, providing a much more accurate and reliable preview of the backup content before the backup runs.
• Configuration: The connect command is now more resilient and can read legacy ~/.disembark configuration files that were stored as a single object instead of an array.
• Updated the readme.md installation URL to point to the official GitHub Releases page instead of the main branch.

Added

• A new list command has been added to display all connected sites.
• A new upgrade command to update Disembark CLI to latest version.
• The backup command now includes an --exclude-tables option, which allows you to exclude specific database tables from the backup. This feature supports the use of wildcards for more flexible table selection.
• You can now use the -x argument with the backup command to exclude certain files or directories from the backup. You can use this option multiple times to exclude several paths.
• A new --preview argument has been added to the backup command to display a list of files and tables that will be included in the backup without actually running the backup process.

v1.0.0 - June 8th 2024

Initial Release

• Provides a connect command to securely save a site URL and token.
• Credentials are saved to a .disembark file in the user's home directory.
• The tool can add new site credentials or update the token for an existing site.
• Verifies that the site URL format is valid before attempting to connect.
• Includes a backup command to initiate the backup process for a specified site.
• Reads the appropriate token for the site from the configuration file.
• Performs backups by making a series of requests to the Disembark WordPress plugin API.
• The backup process includes exporting individual database tables, zipping the full database, and processing site files in batches.
• Finalizes the backup by executing a remote script via curl to generate a complete zip archive.
• The script is designed to be run from the command line.
• A version command is included to display the current version of the tool.
• A showHelp function displays usage instructions for available commands.
• Provides real-time progress and status messages during the backup process.
• Includes a utility to convert bytes into a human-readable file size format.