MokoSuiteBackup/source/packages/com_mokosuitebackup/src/Engine/DatabaseDumper.php

<?php

/**
 * @package     MokoSuiteBackup
 * @subpackage  com_mokosuitebackup
 * @author      Moko Consulting <hello@mokoconsulting.tech>
 * @copyright   Copyright (C) 2026 Moko Consulting. All rights reserved.
 * @license     GNU General Public License version 3 or later; see LICENSE
 */

namespace Joomla\Component\MokoSuiteBackup\Administrator\Engine;

defined('_JEXEC') or die;

use Joomla\CMS\Factory;

class DatabaseDumper
{
	/** @var array Tables to exclude entirely (both structure and data) */
	private array $excludeBoth = [];

	/** @var array Tables to exclude data only (structure is kept) */
	private array $excludeDataOnly = [];

	/** @var array Tables to exclude structure only (data is kept — unusual) */
	private array $excludeStructureOnly = [];

	private int $tablesCount = 0;

	/** @var bool Whether to sanitize user passwords */
	private bool $sanitizePasswords = false;

	/** @var bool Whether to preserve super admin password when sanitizing */
	private bool $preserveSuperAdmin = false;

	/** @var bool Whether to sanitize user emails */
	private bool $sanitizeEmails = false;

	/** @var bool Whether to clear session data */
	private bool $sanitizeSessions = false;

	/** Known invalid bcrypt hash used for sanitized passwords */
	private const SANITIZED_HASH = '$2y$10$SANITIZED.MOKOSUITEBACKUP.INVALID.HASH.DO.NOT.USE.000000';

	/**
	 * @param   array  $excludeTables       Table names to exclude (with #__ prefix).
	 * @param   bool   $sanitizePasswords   Replace user password hashes with invalid value
	 * @param   bool   $preserveSuperAdmin  Keep super admin password when sanitizing
	 * @param   bool   $sanitizeEmails      Replace user emails with sanitized placeholders
	 * @param   bool   $sanitizeSessions    Skip session table data entirely
	 */
	public function __construct(
		array $excludeTables = [],
		bool $sanitizePasswords = false,
		bool $preserveSuperAdmin = false,
		bool $sanitizeEmails = false,
		bool $sanitizeSessions = false
	)
	{
		foreach ($excludeTables as $entry) {
			if (str_ends_with($entry, ':data-only')) {
				$this->excludeDataOnly[] = substr($entry, 0, -10);
			} elseif (str_ends_with($entry, ':structure-only')) {
				$this->excludeStructureOnly[] = substr($entry, 0, -15);
			} else {
				$this->excludeBoth[] = $entry;
			}
		}

		$this->sanitizePasswords  = $sanitizePasswords;
		$this->preserveSuperAdmin = $preserveSuperAdmin;
		$this->sanitizeEmails     = $sanitizeEmails;
		$this->sanitizeSessions   = $sanitizeSessions;

		/* If session sanitization is on, auto-exclude session table data */
		if ($sanitizeSessions) {
			$this->excludeDataOnly[] = '#__session';
		}
	}

	/**
	 * Dump all database tables to SQL.
	 *
	 * @return  string  The SQL dump
	 */
	public function dump(): string
	{
		$db     = Factory::getDbo();
		$prefix = $db->getPrefix();
		$output = [];

		$output[] = '-- MokoSuiteBackup Database Dump';
		$output[] = '-- Generated: ' . date('Y-m-d H:i:s');
		$output[] = '-- Server: ' . $db->getServerType();
		$output[] = '-- Database: ' . $db->getName();
		$output[] = '-- Original Prefix: ' . $prefix;
		$output[] = '-- Abstract Prefix: #__';
		$output[] = '-- Note: Table names use #__ placeholder. Replace with your prefix on restore.';
		$output[] = '';
		$output[] = 'SET SQL_MODE = "NO_AUTO_VALUE_ON_ZERO";';
		$output[] = 'SET time_zone = "+00:00";';
		$output[] = '';

		// Get all tables with the site prefix
		$tables = $db->getTableList();
		$siteTables = [];

		foreach ($tables as $table) {
			if (str_starts_with($table, $prefix)) {
				$siteTables[] = $table;
			}
		}

		foreach ($siteTables as $table) {
			// Check if excluded
			$abstractName = '#__' . substr($table, strlen($prefix));

			if ($this->isExcludedBoth($abstractName, $table)) {
				continue;
			}

			$skipData      = $this->isExcludedDataOnly($abstractName, $table);
			$skipStructure = $this->isExcludedStructureOnly($abstractName, $table);

			$this->tablesCount++;

			$output[] = '-- --------------------------------------------------------';
			$output[] = '-- Table: ' . $abstractName;

			if ($skipData) {
				$output[] = '-- (data excluded)';
			}

			if ($skipStructure) {
				$output[] = '-- (structure excluded)';
			}

			$output[] = '-- --------------------------------------------------------';
			$output[] = '';

			// Get CREATE TABLE statement (unless structure is excluded)
			if (!$skipStructure) {
				$db->setQuery('SHOW CREATE TABLE ' . $db->quoteName($table));
				$createRow = $db->loadRow();

				if (!$createRow || empty($createRow[1])) {
					continue;
				}

				// Replace all occurrences of the live prefix with #__ in CREATE TABLE
				// output — covers the table itself and FK REFERENCES to other tables
				$createSql = str_replace('`' . $prefix, '`#__', $createRow[1]);
				$output[] = 'DROP TABLE IF EXISTS `' . $abstractName . '`;';
				$output[] = $createSql . ';';
				$output[] = '';
			}

			// Dump data (unless data is excluded)
			if ($skipData) {
				$output[] = '';
				continue;
			}

			$db->setQuery('SELECT COUNT(*) FROM ' . $db->quoteName($table));
			$rowCount = (int) $db->loadResult();

			if ($rowCount === 0) {
				$output[] = '-- (empty table)';
				$output[] = '';
				continue;
			}

			$chunkSize = 500;

			for ($offset = 0; $offset < $rowCount; $offset += $chunkSize) {
				$db->setQuery(
					$db->getQuery(true)
						->select('*')
						->from($db->quoteName($table)),
					$offset,
					$chunkSize
				);
				$rows = $db->loadAssocList();

				if (empty($rows)) {
					break;
				}

				foreach ($rows as $row) {
					$this->sanitizeRow($row, $abstractName, $db);
					$values = [];

					foreach ($row as $value) {
						if ($value === null) {
							$values[] = 'NULL';
						} else {
							$values[] = $db->quote($value);
						}
					}

					$columns = array_map([$db, 'quoteName'], array_keys($row));
					$output[] = 'INSERT INTO `' . $abstractName . '`'
						. ' (' . implode(', ', $columns) . ')'
						. ' VALUES (' . implode(', ', $values) . ');';
				}
			}

			$output[] = '';
		}

		return implode("\n", $output);
	}

	/**
	 * Check if a table is fully excluded (both data and structure).
	 */
	private function isExcludedBoth(string $abstractName, string $realName): bool
	{
		foreach ($this->excludeBoth as $pattern) {
			if ($pattern === $abstractName || $pattern === $realName) {
				return true;
			}
		}

		return false;
	}

	/**
	 * Check if a table's data is excluded (structure only).
	 */
	private function isExcludedDataOnly(string $abstractName, string $realName): bool
	{
		foreach ($this->excludeDataOnly as $pattern) {
			if ($pattern === $abstractName || $pattern === $realName) {
				return true;
			}
		}

		return false;
	}

	/**
	 * Check if a table's structure is excluded (data only).
	 */
	private function isExcludedStructureOnly(string $abstractName, string $realName): bool
	{
		foreach ($this->excludeStructureOnly as $pattern) {
			if ($pattern === $abstractName || $pattern === $realName) {
				return true;
			}
		}

		return false;
	}

	/**
	 * Dump all database tables directly to a file, streaming row by row.
	 * Avoids loading the entire dump into RAM.
	 *
	 * @param   string  $filePath  Absolute path to write the SQL file
	 *
	 * @return  int  Size of the dump file in bytes
	 */
	public function dumpToFile(string $filePath): int
	{
		$db     = Factory::getDbo();
		$prefix = $db->getPrefix();

		$fp = fopen($filePath, 'w');

		if ($fp === false) {
			throw new \RuntimeException('Cannot open dump file for writing: ' . $filePath);
		}

		fwrite($fp, "-- MokoSuiteBackup Database Dump\n");
		fwrite($fp, "-- Generated: " . date('Y-m-d H:i:s') . "\n");
		fwrite($fp, "-- Server: " . $db->getServerType() . "\n");
		fwrite($fp, "-- Database: " . $db->getName() . "\n");
		fwrite($fp, "-- Original Prefix: " . $prefix . "\n");
		fwrite($fp, "-- Abstract Prefix: #__\n");
		fwrite($fp, "-- Note: Table names use #__ placeholder. Replace with your prefix on restore.\n\n");
		fwrite($fp, "SET SQL_MODE = \"NO_AUTO_VALUE_ON_ZERO\";\n");
		fwrite($fp, "SET time_zone = \"+00:00\";\n\n");

		// Get all tables with the site prefix
		$tables = $db->getTableList();
		$siteTables = [];

		foreach ($tables as $table) {
			if (str_starts_with($table, $prefix)) {
				$siteTables[] = $table;
			}
		}

		foreach ($siteTables as $table) {
			$abstractName = '#__' . substr($table, strlen($prefix));

			if ($this->isExcludedBoth($abstractName, $table)) {
				continue;
			}

			$skipData      = $this->isExcludedDataOnly($abstractName, $table);
			$skipStructure = $this->isExcludedStructureOnly($abstractName, $table);

			$this->tablesCount++;

			fwrite($fp, "-- --------------------------------------------------------\n");
			fwrite($fp, "-- Table: " . $abstractName . "\n");

			if ($skipData) {
				fwrite($fp, "-- (data excluded)\n");
			}

			if ($skipStructure) {
				fwrite($fp, "-- (structure excluded)\n");
			}

			fwrite($fp, "-- --------------------------------------------------------\n\n");

			if (!$skipStructure) {
				$db->setQuery('SHOW CREATE TABLE ' . $db->quoteName($table));
				$createRow = $db->loadRow();

				if (!$createRow || empty($createRow[1])) {
					continue;
				}

				$createSql = str_replace('`' . $prefix, '`#__', $createRow[1]);
				fwrite($fp, 'DROP TABLE IF EXISTS `' . $abstractName . "`;\\n");
				fwrite($fp, $createSql . ";\n\n");
			}

			if ($skipData) {
				fwrite($fp, "\n");
				continue;
			}

			$db->setQuery('SELECT COUNT(*) FROM ' . $db->quoteName($table));
			$rowCount = (int) $db->loadResult();

			if ($rowCount === 0) {
				fwrite($fp, "-- (empty table)\n\n");
				continue;
			}

			$chunkSize = 500;

			for ($offset = 0; $offset < $rowCount; $offset += $chunkSize) {
				$db->setQuery(
					$db->getQuery(true)
						->select('*')
						->from($db->quoteName($table)),
					$offset,
					$chunkSize
				);
				$rows = $db->loadAssocList();

				if (empty($rows)) {
					break;
				}

				foreach ($rows as $row) {
					$this->sanitizeRow($row, $abstractName, $db);
					$values = [];

					foreach ($row as $value) {
						if ($value === null) {
							$values[] = 'NULL';
						} else {
							$values[] = $db->quote($value);
						}
					}

					$columns = array_map([$db, 'quoteName'], array_keys($row));
					fwrite($fp, 'INSERT INTO `' . $abstractName . '`'
						. ' (' . implode(', ', $columns) . ')'
						. ' VALUES (' . implode(', ', $values) . ");\n");
				}
			}

			fwrite($fp, "\n");
		}

		fclose($fp);

		return filesize($filePath) ?: 0;
	}

	/**
	 * Sanitize a row if it belongs to the users table and sanitization is enabled.
	 *
	 * Replaces the password column with an invalid hash so the backup
	 * cannot be used to extract user credentials.
	 */
	private function sanitizeRow(array &$row, string $abstractTable, object $db): void
	{
		if ($abstractTable !== '#__users') {
			return;
		}

		if (!$this->sanitizePasswords && !$this->sanitizeEmails) {
			return;
		}

		if ($this->sanitizeEmails && isset($row['email']) && isset($row['id'])) {
			$userId = (int) $row['id'];

			/* Preserve super admin emails if preserving super admin */
			if (!$this->preserveSuperAdmin || !$this->isSuperAdmin($userId, $db)) {
				$row['email'] = 'user' . $userId . '@sanitized.example.com';
			}
		}

		if (!$this->sanitizePasswords || !isset($row['password'])) {
			return;
		}

		if ($this->preserveSuperAdmin && isset($row['id'])) {
			if ($this->isSuperAdmin((int) $row['id'], $db)) {
				return;
			}
		}

		$row['password'] = self::SANITIZED_HASH;
	}

	/**
	 * Check if a user ID belongs to the Super Users group (group_id = 8).
	 */
	private function isSuperAdmin(int $userId, object $db): bool
	{
		static $superAdminIds = null;

		if ($superAdminIds === null) {
			$prefix = $db->getPrefix();

			try {
				$db->setQuery(
					$db->getQuery(true)
						->select('DISTINCT ' . $db->quoteName('user_id'))
						->from($db->quoteName($prefix . 'user_usergroup_map'))
						->where($db->quoteName('group_id') . ' = 8')
				);
				$superAdminIds = array_map('intval', $db->loadColumn() ?: []);
			} catch (\Throwable $e) {
				$superAdminIds = [];
			}
		}

		return in_array($userId, $superAdminIds, true);
	}

	/**
	 * Check if passwords were sanitized (for use by callers to log the action).
	 */
	public function isPasswordSanitizationEnabled(): bool
	{
		return $this->sanitizePasswords;
	}

	/**
	 * Get the sentinel hash used for sanitized passwords.
	 */
	public static function getSanitizedHash(): string
	{
		return self::SANITIZED_HASH;
	}

	public function getTablesCount(): int
	{
		return $this->tablesCount;
	}
}