dbbackup/internal/restore/engine.go

package restore

import (
	"archive/tar"
	"compress/gzip"
	"context"
	"database/sql"
	"fmt"
	"io"
	"os"
	"os/exec"
	"path/filepath"
	"strconv"
	"strings"
	"sync"
	"sync/atomic"
	"time"

	"dbbackup/internal/checks"
	"dbbackup/internal/config"
	"dbbackup/internal/database"
	"dbbackup/internal/logger"
	"dbbackup/internal/progress"
	"dbbackup/internal/security"

	"github.com/hashicorp/go-multierror"
	_ "github.com/jackc/pgx/v5/stdlib" // PostgreSQL driver
)

// ProgressCallback is called with progress updates during long operations
// Parameters: current bytes/items done, total bytes/items, description
type ProgressCallback func(current, total int64, description string)

// DatabaseProgressCallback is called with database count progress during cluster restore
type DatabaseProgressCallback func(done, total int, dbName string)

// DatabaseProgressWithTimingCallback is called with database progress including timing info
// Parameters: done count, total count, database name, elapsed time for current restore phase, avg duration per DB
type DatabaseProgressWithTimingCallback func(done, total int, dbName string, phaseElapsed, avgPerDB time.Duration)

// Engine handles database restore operations
type Engine struct {
	cfg              *config.Config
	log              logger.Logger
	db               database.Database
	progress         progress.Indicator
	detailedReporter *progress.DetailedReporter
	dryRun           bool
	debugLogPath     string // Path to save debug log on error

	// TUI progress callback for detailed progress reporting
	progressCallback         ProgressCallback
	dbProgressCallback       DatabaseProgressCallback
	dbProgressTimingCallback DatabaseProgressWithTimingCallback
}

// New creates a new restore engine
func New(cfg *config.Config, log logger.Logger, db database.Database) *Engine {
	progressIndicator := progress.NewIndicator(true, "line")
	detailedReporter := progress.NewDetailedReporter(progressIndicator, &loggerAdapter{logger: log})

	return &Engine{
		cfg:              cfg,
		log:              log,
		db:               db,
		progress:         progressIndicator,
		detailedReporter: detailedReporter,
		dryRun:           false,
	}
}

// NewSilent creates a new restore engine with no stdout progress (for TUI mode)
func NewSilent(cfg *config.Config, log logger.Logger, db database.Database) *Engine {
	progressIndicator := progress.NewNullIndicator()
	detailedReporter := progress.NewDetailedReporter(progressIndicator, &loggerAdapter{logger: log})

	return &Engine{
		cfg:              cfg,
		log:              log,
		db:               db,
		progress:         progressIndicator,
		detailedReporter: detailedReporter,
		dryRun:           false,
	}
}

// NewWithProgress creates a restore engine with custom progress indicator
func NewWithProgress(cfg *config.Config, log logger.Logger, db database.Database, progressIndicator progress.Indicator, dryRun bool) *Engine {
	if progressIndicator == nil {
		progressIndicator = progress.NewNullIndicator()
	}

	detailedReporter := progress.NewDetailedReporter(progressIndicator, &loggerAdapter{logger: log})

	return &Engine{
		cfg:              cfg,
		log:              log,
		db:               db,
		progress:         progressIndicator,
		detailedReporter: detailedReporter,
		dryRun:           dryRun,
	}
}

// SetDebugLogPath enables saving detailed error reports on failure
func (e *Engine) SetDebugLogPath(path string) {
	e.debugLogPath = path
}

// SetProgressCallback sets a callback for detailed progress reporting (for TUI mode)
func (e *Engine) SetProgressCallback(cb ProgressCallback) {
	e.progressCallback = cb
}

// SetDatabaseProgressCallback sets a callback for database count progress during cluster restore
func (e *Engine) SetDatabaseProgressCallback(cb DatabaseProgressCallback) {
	e.dbProgressCallback = cb
}

// SetDatabaseProgressWithTimingCallback sets a callback for database progress with timing info
func (e *Engine) SetDatabaseProgressWithTimingCallback(cb DatabaseProgressWithTimingCallback) {
	e.dbProgressTimingCallback = cb
}

// reportProgress safely calls the progress callback if set
func (e *Engine) reportProgress(current, total int64, description string) {
	if e.progressCallback != nil {
		e.progressCallback(current, total, description)
	}
}

// reportDatabaseProgress safely calls the database progress callback if set
func (e *Engine) reportDatabaseProgress(done, total int, dbName string) {
	if e.dbProgressCallback != nil {
		e.dbProgressCallback(done, total, dbName)
	}
}

// reportDatabaseProgressWithTiming safely calls the timing-aware callback if set
func (e *Engine) reportDatabaseProgressWithTiming(done, total int, dbName string, phaseElapsed, avgPerDB time.Duration) {
	if e.dbProgressTimingCallback != nil {
		e.dbProgressTimingCallback(done, total, dbName, phaseElapsed, avgPerDB)
	}
}

// loggerAdapter adapts our logger to the progress.Logger interface
type loggerAdapter struct {
	logger logger.Logger
}

func (la *loggerAdapter) Info(msg string, args ...any) {
	la.logger.Info(msg, args...)
}

func (la *loggerAdapter) Warn(msg string, args ...any) {
	la.logger.Warn(msg, args...)
}

func (la *loggerAdapter) Error(msg string, args ...any) {
	la.logger.Error(msg, args...)
}

func (la *loggerAdapter) Debug(msg string, args ...any) {
	la.logger.Debug(msg, args...)
}

// RestoreSingle restores a single database from an archive
func (e *Engine) RestoreSingle(ctx context.Context, archivePath, targetDB string, cleanFirst, createIfMissing bool) error {
	operation := e.log.StartOperation("Single Database Restore")

	// Validate and sanitize archive path
	validArchivePath, pathErr := security.ValidateArchivePath(archivePath)
	if pathErr != nil {
		operation.Fail(fmt.Sprintf("Invalid archive path: %v", pathErr))
		return fmt.Errorf("invalid archive path: %w", pathErr)
	}
	archivePath = validArchivePath

	// Validate archive exists
	if _, err := os.Stat(archivePath); os.IsNotExist(err) {
		operation.Fail("Archive not found")
		return fmt.Errorf("archive not found: %s", archivePath)
	}

	// Verify checksum if .sha256 file exists
	if checksumErr := security.LoadAndVerifyChecksum(archivePath); checksumErr != nil {
		e.log.Warn("Checksum verification failed", "error", checksumErr)
		e.log.Warn("Continuing restore without checksum verification (use with caution)")
	} else {
		e.log.Info("[OK] Archive checksum verified successfully")
	}

	// Detect archive format
	format := DetectArchiveFormat(archivePath)
	e.log.Info("Detected archive format", "format", format, "path", archivePath)

	// Check version compatibility for PostgreSQL dumps
	if format == FormatPostgreSQLDump || format == FormatPostgreSQLDumpGz {
		if compatResult, err := e.CheckRestoreVersionCompatibility(ctx, archivePath); err == nil && compatResult != nil {
			e.log.Info(compatResult.Message,
				"source_version", compatResult.SourceVersion.Full,
				"target_version", compatResult.TargetVersion.Full,
				"compatibility", compatResult.Level.String())

			// Block unsupported downgrades
			if !compatResult.Compatible {
				operation.Fail(compatResult.Message)
				return fmt.Errorf("version compatibility error: %s", compatResult.Message)
			}

			// Show warnings for risky upgrades
			if compatResult.Level == CompatibilityLevelRisky || compatResult.Level == CompatibilityLevelWarning {
				for _, warning := range compatResult.Warnings {
					e.log.Warn(warning)
				}
			}
		}
	}

	if e.dryRun {
		e.log.Info("DRY RUN: Would restore single database", "archive", archivePath, "target", targetDB)
		return e.previewRestore(archivePath, targetDB, format)
	}

	// Start progress tracking
	e.progress.Start(fmt.Sprintf("Restoring database '%s' from %s", targetDB, filepath.Base(archivePath)))

	// Create database if requested and it doesn't exist
	if createIfMissing {
		e.log.Info("Checking if target database exists", "database", targetDB)
		if err := e.ensureDatabaseExists(ctx, targetDB); err != nil {
			operation.Fail(fmt.Sprintf("Failed to create database: %v", err))
			return fmt.Errorf("failed to create database '%s': %w", targetDB, err)
		}
	}

	// Handle different archive formats
	var err error
	switch format {
	case FormatPostgreSQLDump, FormatPostgreSQLDumpGz:
		err = e.restorePostgreSQLDump(ctx, archivePath, targetDB, format == FormatPostgreSQLDumpGz, cleanFirst)
	case FormatPostgreSQLSQL, FormatPostgreSQLSQLGz:
		err = e.restorePostgreSQLSQL(ctx, archivePath, targetDB, format == FormatPostgreSQLSQLGz)
	case FormatMySQLSQL, FormatMySQLSQLGz:
		err = e.restoreMySQLSQL(ctx, archivePath, targetDB, format == FormatMySQLSQLGz)
	default:
		operation.Fail("Unsupported archive format")
		return fmt.Errorf("unsupported archive format: %s", format)
	}

	if err != nil {
		e.progress.Fail(fmt.Sprintf("Restore failed: %v", err))
		operation.Fail(fmt.Sprintf("Restore failed: %v", err))
		return err
	}

	e.progress.Complete(fmt.Sprintf("Database '%s' restored successfully", targetDB))
	operation.Complete(fmt.Sprintf("Restored database '%s' from %s", targetDB, filepath.Base(archivePath)))
	return nil
}

// restorePostgreSQLDump restores from PostgreSQL custom dump format
func (e *Engine) restorePostgreSQLDump(ctx context.Context, archivePath, targetDB string, compressed bool, cleanFirst bool) error {
	// Build restore command
	opts := database.RestoreOptions{
		Parallel:          1,
		Clean:             cleanFirst,
		NoOwner:           true,
		NoPrivileges:      true,
		SingleTransaction: false, // CRITICAL: Disabled to prevent lock exhaustion with large objects
		Verbose:           true,  // Enable verbose for single database restores (not cluster)
	}

	cmd := e.db.BuildRestoreCommand(targetDB, archivePath, opts)

	if compressed {
		// For compressed dumps, decompress first
		return e.executeRestoreWithDecompression(ctx, archivePath, cmd)
	}

	return e.executeRestoreCommand(ctx, cmd)
}

// restorePostgreSQLDumpWithOwnership restores from PostgreSQL custom dump with ownership control
func (e *Engine) restorePostgreSQLDumpWithOwnership(ctx context.Context, archivePath, targetDB string, compressed bool, preserveOwnership bool) error {
	// Check if dump contains large objects (BLOBs) - if so, use phased restore
	// to prevent lock table exhaustion (max_locks_per_transaction OOM)
	hasLargeObjects := e.checkDumpHasLargeObjects(archivePath)

	if hasLargeObjects {
		e.log.Info("Large objects detected - using phased restore to prevent lock exhaustion",
			"database", targetDB,
			"archive", archivePath)
		return e.restorePostgreSQLDumpPhased(ctx, archivePath, targetDB, preserveOwnership)
	}

	// Standard restore for dumps without large objects
	opts := database.RestoreOptions{
		Parallel:          1,
		Clean:             false,              // We already dropped the database
		NoOwner:           !preserveOwnership, // Preserve ownership if we're superuser
		NoPrivileges:      !preserveOwnership, // Preserve privileges if we're superuser
		SingleTransaction: false,              // CRITICAL: Disabled to prevent lock exhaustion with large objects
		Verbose:           false,              // CRITICAL: disable verbose to prevent OOM on large restores
	}

	e.log.Info("Restoring database",
		"database", targetDB,
		"preserveOwnership", preserveOwnership,
		"noOwner", opts.NoOwner,
		"noPrivileges", opts.NoPrivileges)

	cmd := e.db.BuildRestoreCommand(targetDB, archivePath, opts)

	if compressed {
		// For compressed dumps, decompress first
		return e.executeRestoreWithDecompression(ctx, archivePath, cmd)
	}

	return e.executeRestoreCommand(ctx, cmd)
}

// restorePostgreSQLDumpPhased performs a multi-phase restore to prevent lock table exhaustion
// Phase 1: pre-data (schema, types, functions)
// Phase 2: data (table data, excluding BLOBs)
// Phase 3: blobs (large objects in smaller batches)
// Phase 4: post-data (indexes, constraints, triggers)
//
// This approach prevents OOM errors by committing and releasing locks between phases.
func (e *Engine) restorePostgreSQLDumpPhased(ctx context.Context, archivePath, targetDB string, preserveOwnership bool) error {
	e.log.Info("Starting phased restore for database with large objects",
		"database", targetDB,
		"archive", archivePath)

	// Phase definitions with --section flag
	phases := []struct {
		name    string
		section string
		desc    string
	}{
		{"pre-data", "pre-data", "Schema, types, functions"},
		{"data", "data", "Table data"},
		{"post-data", "post-data", "Indexes, constraints, triggers"},
	}

	for i, phase := range phases {
		e.log.Info(fmt.Sprintf("Phase %d/%d: Restoring %s", i+1, len(phases), phase.name),
			"database", targetDB,
			"section", phase.section,
			"description", phase.desc)

		if err := e.restoreSection(ctx, archivePath, targetDB, phase.section, preserveOwnership); err != nil {
			// Check if it's an ignorable error
			if e.isIgnorableError(err.Error()) {
				e.log.Warn(fmt.Sprintf("Phase %d completed with ignorable errors", i+1),
					"section", phase.section,
					"error", err)
				continue
			}
			return fmt.Errorf("phase %d (%s) failed: %w", i+1, phase.name, err)
		}

		e.log.Info(fmt.Sprintf("Phase %d/%d completed successfully", i+1, len(phases)),
			"section", phase.section)
	}

	e.log.Info("Phased restore completed successfully", "database", targetDB)
	return nil
}

// restoreSection restores a specific section of a PostgreSQL dump
func (e *Engine) restoreSection(ctx context.Context, archivePath, targetDB, section string, preserveOwnership bool) error {
	// Build pg_restore command with --section flag
	args := []string{"pg_restore"}

	// Connection parameters
	if e.cfg.Host != "localhost" {
		args = append(args, "-h", e.cfg.Host)
		args = append(args, "-p", fmt.Sprintf("%d", e.cfg.Port))
		args = append(args, "--no-password")
	}
	args = append(args, "-U", e.cfg.User)

	// Section-specific restore
	args = append(args, "--section="+section)

	// Options
	if !preserveOwnership {
		args = append(args, "--no-owner", "--no-privileges")
	}

	// Skip data for failed tables (prevents cascading errors)
	args = append(args, "--no-data-for-failed-tables")

	// Database and input
	args = append(args, "--dbname="+targetDB)
	args = append(args, archivePath)

	return e.executeRestoreCommand(ctx, args)
}

// checkDumpHasLargeObjects checks if a PostgreSQL custom dump contains large objects (BLOBs)
func (e *Engine) checkDumpHasLargeObjects(archivePath string) bool {
	// Use pg_restore -l to list contents without restoring
	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	cmd := exec.CommandContext(ctx, "pg_restore", "-l", archivePath)
	output, err := cmd.Output()

	if err != nil {
		// If listing fails, assume no large objects (safer to use standard restore)
		e.log.Debug("Could not list dump contents, assuming no large objects", "error", err)
		return false
	}

	outputStr := string(output)

	// Check for BLOB/LARGE OBJECT indicators
	if strings.Contains(outputStr, "BLOB") ||
		strings.Contains(outputStr, "LARGE OBJECT") ||
		strings.Contains(outputStr, " BLOBS ") ||
		strings.Contains(outputStr, "lo_create") {
		return true
	}

	return false
}

// restorePostgreSQLSQL restores from PostgreSQL SQL script
func (e *Engine) restorePostgreSQLSQL(ctx context.Context, archivePath, targetDB string, compressed bool) error {
	// Pre-validate SQL dump to detect truncation BEFORE attempting restore
	// This saves time by catching corrupted files early (vs 49min failures)
	if err := e.quickValidateSQLDump(archivePath, compressed); err != nil {
		e.log.Error("Pre-restore validation failed - dump file appears corrupted",
			"file", archivePath,
			"error", err)
		return fmt.Errorf("dump validation failed: %w - the backup file may be truncated or corrupted", err)
	}

	// Use psql for SQL scripts
	var cmd []string

	// For localhost, omit -h to use Unix socket (avoids Ident auth issues)
	// But always include -p for port (in case of non-standard port)
	hostArg := ""
	portArg := fmt.Sprintf("-p %d", e.cfg.Port)
	if e.cfg.Host != "localhost" && e.cfg.Host != "" {
		hostArg = fmt.Sprintf("-h %s", e.cfg.Host)
	}

	if compressed {
		// Use ON_ERROR_STOP=1 to fail fast on first error (prevents millions of errors on truncated dumps)
		psqlCmd := fmt.Sprintf("psql %s -U %s -d %s -v ON_ERROR_STOP=1", portArg, e.cfg.User, targetDB)
		if hostArg != "" {
			psqlCmd = fmt.Sprintf("psql %s %s -U %s -d %s -v ON_ERROR_STOP=1", hostArg, portArg, e.cfg.User, targetDB)
		}
		// Set PGPASSWORD in the bash command for password-less auth
		cmd = []string{
			"bash", "-c",
			fmt.Sprintf("PGPASSWORD='%s' gunzip -c %s | %s", e.cfg.Password, archivePath, psqlCmd),
		}
	} else {
		if hostArg != "" {
			cmd = []string{
				"psql",
				"-h", e.cfg.Host,
				"-p", fmt.Sprintf("%d", e.cfg.Port),
				"-U", e.cfg.User,
				"-d", targetDB,
				"-v", "ON_ERROR_STOP=1",
				"-f", archivePath,
			}
		} else {
			cmd = []string{
				"psql",
				"-p", fmt.Sprintf("%d", e.cfg.Port),
				"-U", e.cfg.User,
				"-d", targetDB,
				"-v", "ON_ERROR_STOP=1",
				"-f", archivePath,
			}
		}
	}

	return e.executeRestoreCommand(ctx, cmd)
}

// restoreMySQLSQL restores from MySQL SQL script
func (e *Engine) restoreMySQLSQL(ctx context.Context, archivePath, targetDB string, compressed bool) error {
	options := database.RestoreOptions{}

	cmd := e.db.BuildRestoreCommand(targetDB, archivePath, options)

	if compressed {
		// For compressed SQL, decompress on the fly
		cmd = []string{
			"bash", "-c",
			fmt.Sprintf("gunzip -c %s | %s", archivePath, strings.Join(cmd, " ")),
		}
	}

	return e.executeRestoreCommand(ctx, cmd)
}

// executeRestoreCommand executes a restore command
func (e *Engine) executeRestoreCommand(ctx context.Context, cmdArgs []string) error {
	return e.executeRestoreCommandWithContext(ctx, cmdArgs, "", "", FormatUnknown)
}

// executeRestoreCommandWithContext executes a restore command with error collection context
func (e *Engine) executeRestoreCommandWithContext(ctx context.Context, cmdArgs []string, archivePath, targetDB string, format ArchiveFormat) error {
	e.log.Info("Executing restore command", "command", strings.Join(cmdArgs, " "))

	cmd := exec.CommandContext(ctx, cmdArgs[0], cmdArgs[1:]...)

	// Set environment variables
	cmd.Env = append(os.Environ(),
		fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password),
		fmt.Sprintf("MYSQL_PWD=%s", e.cfg.Password),
	)

	// Create error collector if debug log path is set
	var collector *ErrorCollector
	if e.debugLogPath != "" {
		collector = NewErrorCollector(e.cfg, e.log, archivePath, targetDB, format, true)
	}

	// Stream stderr to avoid memory issues with large output
	// Don't use CombinedOutput() as it loads everything into memory
	stderr, err := cmd.StderrPipe()
	if err != nil {
		return fmt.Errorf("failed to create stderr pipe: %w", err)
	}

	if err := cmd.Start(); err != nil {
		return fmt.Errorf("failed to start restore command: %w", err)
	}

	// Read stderr in goroutine to avoid blocking
	var lastError string
	var errorCount int
	stderrDone := make(chan struct{})
	go func() {
		defer close(stderrDone)
		buf := make([]byte, 4096)
		const maxErrors = 10 // Limit captured errors to prevent OOM
		for {
			n, err := stderr.Read(buf)
			if n > 0 {
				chunk := string(buf[:n])

				// Feed to error collector if enabled
				if collector != nil {
					collector.CaptureStderr(chunk)
				}

				// Only capture REAL errors, not verbose output
				if strings.Contains(chunk, "ERROR:") || strings.Contains(chunk, "FATAL:") || strings.Contains(chunk, "error:") {
					lastError = strings.TrimSpace(chunk)
					errorCount++
					if errorCount <= maxErrors {
						e.log.Warn("Restore stderr", "output", chunk)
					}
				}
				// Note: --verbose output is discarded to prevent OOM
			}
			if err != nil {
				break
			}
		}
	}()

	// Wait for command with proper context handling
	cmdDone := make(chan error, 1)
	go func() {
		cmdDone <- cmd.Wait()
	}()

	var cmdErr error
	select {
	case cmdErr = <-cmdDone:
		// Command completed (success or failure)
	case <-ctx.Done():
		// Context cancelled - kill process
		e.log.Warn("Restore cancelled - killing process")
		cmd.Process.Kill()
		<-cmdDone
		cmdErr = ctx.Err()
	}

	// Wait for stderr reader to finish
	<-stderrDone

	if cmdErr != nil {
		// Get exit code
		exitCode := 1
		if exitErr, ok := cmdErr.(*exec.ExitError); ok {
			exitCode = exitErr.ExitCode()
		}

		// PostgreSQL pg_restore returns exit code 1 even for ignorable errors
		// Check if errors are ignorable (already exists, duplicate, etc.)
		if lastError != "" && e.isIgnorableError(lastError) {
			e.log.Warn("Restore completed with ignorable errors", "error_count", errorCount, "last_error", lastError)
			return nil // Success despite ignorable errors
		}

		// Classify error and provide helpful hints
		var classification *checks.ErrorClassification
		var errType, errHint string
		if lastError != "" {
			classification = checks.ClassifyError(lastError)
			errType = classification.Type
			errHint = classification.Hint
			e.log.Error("Restore command failed",
				"error", err,
				"last_stderr", lastError,
				"error_count", errorCount,
				"error_type", classification.Type,
				"hint", classification.Hint,
				"action", classification.Action)
		} else {
			e.log.Error("Restore command failed", "error", err, "error_count", errorCount)
		}

		// Generate and save error report if collector is enabled
		if collector != nil {
			collector.SetExitCode(exitCode)
			report := collector.GenerateReport(
				lastError,
				errType,
				errHint,
			)

			// Print report to console
			collector.PrintReport(report)

			// Save to file
			if e.debugLogPath != "" {
				if saveErr := collector.SaveReport(report, e.debugLogPath); saveErr != nil {
					e.log.Warn("Failed to save debug log", "error", saveErr)
				} else {
					e.log.Info("Debug log saved", "path", e.debugLogPath)
					fmt.Printf("\n[LOG] Detailed error report saved to: %s\n", e.debugLogPath)
				}
			}
		}

		if lastError != "" {
			return fmt.Errorf("restore failed: %w (last error: %s, total errors: %d) - %s",
				err, lastError, errorCount, errHint)
		}
		return fmt.Errorf("restore failed: %w", err)
	}

	e.log.Info("Restore command completed successfully")
	return nil
}

// executeRestoreWithDecompression handles decompression during restore
func (e *Engine) executeRestoreWithDecompression(ctx context.Context, archivePath string, restoreCmd []string) error {
	// Check if pigz is available for faster decompression
	decompressCmd := "gunzip"
	if _, err := exec.LookPath("pigz"); err == nil {
		decompressCmd = "pigz"
		e.log.Info("Using pigz for parallel decompression")
	}

	// Build pipeline: decompress | restore
	pipeline := fmt.Sprintf("%s -dc %s | %s", decompressCmd, archivePath, strings.Join(restoreCmd, " "))
	cmd := exec.CommandContext(ctx, "bash", "-c", pipeline)

	cmd.Env = append(os.Environ(),
		fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password),
		fmt.Sprintf("MYSQL_PWD=%s", e.cfg.Password),
	)

	// Stream stderr to avoid memory issues with large output
	stderr, err := cmd.StderrPipe()
	if err != nil {
		return fmt.Errorf("failed to create stderr pipe: %w", err)
	}

	if err := cmd.Start(); err != nil {
		return fmt.Errorf("failed to start restore command: %w", err)
	}

	// Read stderr in goroutine to avoid blocking
	var lastError string
	var errorCount int
	stderrDone := make(chan struct{})
	go func() {
		defer close(stderrDone)
		buf := make([]byte, 4096)
		const maxErrors = 10 // Limit captured errors to prevent OOM
		for {
			n, err := stderr.Read(buf)
			if n > 0 {
				chunk := string(buf[:n])
				// Only capture REAL errors, not verbose output
				if strings.Contains(chunk, "ERROR:") || strings.Contains(chunk, "FATAL:") || strings.Contains(chunk, "error:") {
					lastError = strings.TrimSpace(chunk)
					errorCount++
					if errorCount <= maxErrors {
						e.log.Warn("Restore stderr", "output", chunk)
					}
				}
				// Note: --verbose output is discarded to prevent OOM
			}
			if err != nil {
				break
			}
		}
	}()

	// Wait for command with proper context handling
	cmdDone := make(chan error, 1)
	go func() {
		cmdDone <- cmd.Wait()
	}()

	var cmdErr error
	select {
	case cmdErr = <-cmdDone:
		// Command completed (success or failure)
	case <-ctx.Done():
		// Context cancelled - kill process
		e.log.Warn("Restore with decompression cancelled - killing process")
		cmd.Process.Kill()
		<-cmdDone
		cmdErr = ctx.Err()
	}

	// Wait for stderr reader to finish
	<-stderrDone

	if cmdErr != nil {
		// PostgreSQL pg_restore returns exit code 1 even for ignorable errors
		// Check if errors are ignorable (already exists, duplicate, etc.)
		if lastError != "" && e.isIgnorableError(lastError) {
			e.log.Warn("Restore with decompression completed with ignorable errors", "error_count", errorCount, "last_error", lastError)
			return nil // Success despite ignorable errors
		}

		// Classify error and provide helpful hints
		if lastError != "" {
			classification := checks.ClassifyError(lastError)
			e.log.Error("Restore with decompression failed",
				"error", cmdErr,
				"last_stderr", lastError,
				"error_count", errorCount,
				"error_type", classification.Type,
				"hint", classification.Hint,
				"action", classification.Action)
			return fmt.Errorf("restore failed: %w (last error: %s, total errors: %d) - %s",
				cmdErr, lastError, errorCount, classification.Hint)
		}

		e.log.Error("Restore with decompression failed", "error", cmdErr, "last_stderr", lastError, "error_count", errorCount)
		return fmt.Errorf("restore failed: %w", cmdErr)
	}

	return nil
}

// previewRestore shows what would be done without executing
func (e *Engine) previewRestore(archivePath, targetDB string, format ArchiveFormat) error {
	fmt.Println("\n" + strings.Repeat("=", 60))
	fmt.Println(" RESTORE PREVIEW (DRY RUN)")
	fmt.Println(strings.Repeat("=", 60))

	stat, _ := os.Stat(archivePath)
	fmt.Printf("\nArchive: %s\n", filepath.Base(archivePath))
	fmt.Printf("Format: %s\n", format)
	if stat != nil {
		fmt.Printf("Size: %s\n", FormatBytes(stat.Size()))
		fmt.Printf("Modified: %s\n", stat.ModTime().Format("2006-01-02 15:04:05"))
	}
	fmt.Printf("Target Database: %s\n", targetDB)
	fmt.Printf("Target Host: %s:%d\n", e.cfg.Host, e.cfg.Port)

	fmt.Println("\nOperations that would be performed:")
	switch format {
	case FormatPostgreSQLDump:
		fmt.Printf("  1. Execute: pg_restore -d %s %s\n", targetDB, archivePath)
	case FormatPostgreSQLDumpGz:
		fmt.Printf("  1. Decompress: %s\n", archivePath)
		fmt.Printf("  2. Execute: pg_restore -d %s\n", targetDB)
	case FormatPostgreSQLSQL, FormatPostgreSQLSQLGz:
		fmt.Printf("  1. Execute: psql -d %s -f %s\n", targetDB, archivePath)
	case FormatMySQLSQL, FormatMySQLSQLGz:
		fmt.Printf("  1. Execute: mysql %s < %s\n", targetDB, archivePath)
	}

	fmt.Println("\n[WARN]  WARNING: This will restore data to the target database.")
	fmt.Println("   Existing data may be overwritten or merged.")
	fmt.Println("\nTo execute this restore, add the --confirm flag.")
	fmt.Println(strings.Repeat("=", 60) + "\n")

	return nil
}

// RestoreCluster restores a full cluster from a tar.gz archive
func (e *Engine) RestoreCluster(ctx context.Context, archivePath string) error {
	operation := e.log.StartOperation("Cluster Restore")

	// Validate and sanitize archive path
	validArchivePath, pathErr := security.ValidateArchivePath(archivePath)
	if pathErr != nil {
		operation.Fail(fmt.Sprintf("Invalid archive path: %v", pathErr))
		return fmt.Errorf("invalid archive path: %w", pathErr)
	}
	archivePath = validArchivePath

	// Validate archive exists
	if _, err := os.Stat(archivePath); os.IsNotExist(err) {
		operation.Fail("Archive not found")
		return fmt.Errorf("archive not found: %s", archivePath)
	}

	// Verify checksum if .sha256 file exists
	if checksumErr := security.LoadAndVerifyChecksum(archivePath); checksumErr != nil {
		e.log.Warn("Checksum verification failed", "error", checksumErr)
		e.log.Warn("Continuing restore without checksum verification (use with caution)")
	} else {
		e.log.Info("[OK] Cluster archive checksum verified successfully")
	}

	format := DetectArchiveFormat(archivePath)
	if format != FormatClusterTarGz {
		operation.Fail("Invalid cluster archive format")
		return fmt.Errorf("not a cluster archive: %s (detected format: %s)", archivePath, format)
	}

	// Check disk space before starting restore
	e.log.Info("Checking disk space for restore")
	archiveInfo, err := os.Stat(archivePath)
	if err == nil {
		spaceCheck := checks.CheckDiskSpaceForRestore(e.cfg.BackupDir, archiveInfo.Size())

		if spaceCheck.Critical {
			operation.Fail("Insufficient disk space")
			return fmt.Errorf("insufficient disk space for restore: %.1f%% used - need at least 4x archive size", spaceCheck.UsedPercent)
		}

		if spaceCheck.Warning {
			e.log.Warn("Low disk space - restore may fail",
				"available_gb", float64(spaceCheck.AvailableBytes)/(1024*1024*1024),
				"used_percent", spaceCheck.UsedPercent)
		}
	}

	if e.dryRun {
		e.log.Info("DRY RUN: Would restore cluster", "archive", archivePath)
		return e.previewClusterRestore(archivePath)
	}

	e.progress.Start(fmt.Sprintf("Restoring cluster from %s", filepath.Base(archivePath)))

	// Create temporary extraction directory in configured WorkDir
	workDir := e.cfg.GetEffectiveWorkDir()
	tempDir := filepath.Join(workDir, fmt.Sprintf(".restore_%d", time.Now().Unix()))
	if err := os.MkdirAll(tempDir, 0755); err != nil {
		operation.Fail("Failed to create temporary directory")
		return fmt.Errorf("failed to create temp directory in %s: %w", workDir, err)
	}
	defer os.RemoveAll(tempDir)

	// Extract archive
	e.log.Info("Extracting cluster archive", "archive", archivePath, "tempDir", tempDir)
	if err := e.extractArchive(ctx, archivePath, tempDir); err != nil {
		operation.Fail("Archive extraction failed")
		return fmt.Errorf("failed to extract archive: %w", err)
	}

	// Check if user has superuser privileges (required for ownership restoration)
	e.progress.Update("Checking privileges...")
	isSuperuser, err := e.checkSuperuser(ctx)
	if err != nil {
		e.log.Warn("Could not verify superuser status", "error", err)
		isSuperuser = false // Assume not superuser if check fails
	}

	if !isSuperuser {
		e.log.Warn("Current user is not a superuser - database ownership may not be fully restored")
		e.progress.Update("[WARN]  Warning: Non-superuser - ownership restoration limited")
		time.Sleep(2 * time.Second) // Give user time to see warning
	} else {
		e.log.Info("Superuser privileges confirmed - full ownership restoration enabled")
	}

	// Restore global objects FIRST (roles, tablespaces) - CRITICAL for ownership
	globalsFile := filepath.Join(tempDir, "globals.sql")
	if _, err := os.Stat(globalsFile); err == nil {
		e.log.Info("Restoring global objects (roles, tablespaces)")
		e.progress.Update("Restoring global objects (roles, tablespaces)...")
		if err := e.restoreGlobals(ctx, globalsFile); err != nil {
			e.log.Error("Failed to restore global objects", "error", err)
			if isSuperuser {
				// If we're superuser and can't restore globals, this is a problem
				e.progress.Fail("Failed to restore global objects")
				operation.Fail("Global objects restoration failed")
				return fmt.Errorf("failed to restore global objects: %w", err)
			} else {
				e.log.Warn("Continuing without global objects (may cause ownership issues)")
			}
		} else {
			e.log.Info("Successfully restored global objects")
		}
	} else {
		e.log.Warn("No globals.sql file found in backup - roles and tablespaces will not be restored")
	}

	// Restore individual databases
	dumpsDir := filepath.Join(tempDir, "dumps")
	if _, err := os.Stat(dumpsDir); err != nil {
		operation.Fail("No database dumps found in archive")
		return fmt.Errorf("no database dumps found in archive")
	}

	entries, err := os.ReadDir(dumpsDir)
	if err != nil {
		operation.Fail("Failed to read dumps directory")
		return fmt.Errorf("failed to read dumps directory: %w", err)
	}

	// PRE-VALIDATE all SQL dumps BEFORE starting restore
	// This catches truncated files early instead of failing after hours of work
	e.log.Info("Pre-validating dump files before restore...")
	e.progress.Update("Pre-validating dump files...")
	var corruptedDumps []string
	diagnoser := NewDiagnoser(e.log, false)
	for _, entry := range entries {
		if entry.IsDir() {
			continue
		}
		dumpFile := filepath.Join(dumpsDir, entry.Name())
		if strings.HasSuffix(dumpFile, ".sql.gz") {
			result, err := diagnoser.DiagnoseFile(dumpFile)
			if err != nil {
				e.log.Warn("Could not validate dump file", "file", entry.Name(), "error", err)
				continue
			}
			if result.IsTruncated || result.IsCorrupted || !result.IsValid {
				dbName := strings.TrimSuffix(entry.Name(), ".sql.gz")
				errDetail := "unknown issue"
				if len(result.Errors) > 0 {
					errDetail = result.Errors[0]
				}
				corruptedDumps = append(corruptedDumps, fmt.Sprintf("%s: %s", dbName, errDetail))
				e.log.Error("CORRUPTED dump file detected",
					"database", dbName,
					"file", entry.Name(),
					"truncated", result.IsTruncated,
					"errors", result.Errors)
			}
		} else if strings.HasSuffix(dumpFile, ".dump") {
			// Validate custom format dumps using pg_restore --list
			cmd := exec.CommandContext(ctx, "pg_restore", "--list", dumpFile)
			output, err := cmd.CombinedOutput()
			if err != nil {
				dbName := strings.TrimSuffix(entry.Name(), ".dump")
				errDetail := strings.TrimSpace(string(output))
				if len(errDetail) > 100 {
					errDetail = errDetail[:100] + "..."
				}
				// Check for truncation indicators
				if strings.Contains(errDetail, "unexpected end") || strings.Contains(errDetail, "invalid") {
					corruptedDumps = append(corruptedDumps, fmt.Sprintf("%s: %s", dbName, errDetail))
					e.log.Error("CORRUPTED custom dump file detected",
						"database", dbName,
						"file", entry.Name(),
						"error", errDetail)
				} else {
					e.log.Warn("pg_restore --list warning (may be recoverable)",
						"file", entry.Name(),
						"error", errDetail)
				}
			}
		}
	}
	if len(corruptedDumps) > 0 {
		operation.Fail("Corrupted dump files detected")
		e.progress.Fail(fmt.Sprintf("Found %d corrupted dump files - restore aborted", len(corruptedDumps)))
		return fmt.Errorf("pre-validation failed: %d corrupted dump files detected: %s - the backup archive appears to be damaged, restore from a different backup",
			len(corruptedDumps), strings.Join(corruptedDumps, ", "))
	}
	e.log.Info("All dump files passed validation")

	// Run comprehensive preflight checks (Linux system + PostgreSQL + Archive analysis)
	preflight, preflightErr := e.RunPreflightChecks(ctx, dumpsDir, entries)
	if preflightErr != nil {
		e.log.Warn("Preflight checks failed", "error", preflightErr)
	}

	// Calculate optimal lock boost based on BLOB count
	lockBoostValue := 2048 // Default
	if preflight != nil && preflight.Archive.RecommendedLockBoost > 0 {
		lockBoostValue = preflight.Archive.RecommendedLockBoost
	}

	// AUTO-TUNE: Boost PostgreSQL settings for large restores
	e.progress.Update("Tuning PostgreSQL for large restore...")
	originalSettings, tuneErr := e.boostPostgreSQLSettings(ctx, lockBoostValue)
	if tuneErr != nil {
		e.log.Warn("Could not boost PostgreSQL settings - restore may fail on BLOB-heavy databases",
			"error", tuneErr)
	} else {
		e.log.Info("Boosted PostgreSQL settings for restore",
			"max_locks_per_transaction", fmt.Sprintf("%d → %d", originalSettings.MaxLocks, lockBoostValue),
			"maintenance_work_mem", fmt.Sprintf("%s → 2GB", originalSettings.MaintenanceWorkMem))
		// Ensure we reset settings when done (even on failure)
		defer func() {
			if resetErr := e.resetPostgreSQLSettings(ctx, originalSettings); resetErr != nil {
				e.log.Warn("Could not reset PostgreSQL settings", "error", resetErr)
			} else {
				e.log.Info("Reset PostgreSQL settings to original values")
			}
		}()
	}

	var restoreErrors *multierror.Error
	var restoreErrorsMu sync.Mutex
	totalDBs := 0

	// Count total databases
	for _, entry := range entries {
		if !entry.IsDir() {
			totalDBs++
		}
	}

	// Create ETA estimator for database restores
	estimator := progress.NewETAEstimator("Restoring cluster", totalDBs)
	e.progress.SetEstimator(estimator)

	// Check for large objects in dump files and adjust parallelism
	hasLargeObjects := e.detectLargeObjectsInDumps(dumpsDir, entries)

	// Use worker pool for parallel restore
	parallelism := e.cfg.ClusterParallelism
	if parallelism < 1 {
		parallelism = 1 // Ensure at least sequential
	}

	// Automatically reduce parallelism if large objects detected
	if hasLargeObjects && parallelism > 1 {
		e.log.Warn("Large objects detected in dump files - reducing parallelism to avoid lock contention",
			"original_parallelism", parallelism,
			"adjusted_parallelism", 1)
		e.progress.Update("[WARN]  Large objects detected - using sequential restore to avoid lock conflicts")
		time.Sleep(2 * time.Second) // Give user time to see warning
		parallelism = 1
	}

	var successCount, failCount int32
	var mu sync.Mutex // Protect shared resources (progress, logger)

	// Timing tracking for restore phase progress
	restorePhaseStart := time.Now()
	var completedDBTimes []time.Duration // Track duration for each completed DB restore
	var completedDBTimesMu sync.Mutex

	// Create semaphore to limit concurrency
	semaphore := make(chan struct{}, parallelism)
	var wg sync.WaitGroup

	dbIndex := 0
	for _, entry := range entries {
		if entry.IsDir() {
			continue
		}

		wg.Add(1)
		semaphore <- struct{}{} // Acquire

		go func(idx int, filename string) {
			defer wg.Done()
			defer func() { <-semaphore }() // Release

			// Panic recovery - prevent one database failure from crashing entire cluster restore
			defer func() {
				if r := recover(); r != nil {
					e.log.Error("Panic in database restore goroutine", "file", filename, "panic", r)
					atomic.AddInt32(&failCount, 1)
				}
			}()

			// Check for context cancellation before starting
			if ctx.Err() != nil {
				e.log.Warn("Context cancelled - skipping database restore", "file", filename)
				atomic.AddInt32(&failCount, 1)
				restoreErrorsMu.Lock()
				restoreErrors = multierror.Append(restoreErrors, fmt.Errorf("%s: restore skipped (context cancelled)", strings.TrimSuffix(strings.TrimSuffix(filename, ".dump"), ".sql.gz")))
				restoreErrorsMu.Unlock()
				return
			}

			// Track timing for this database restore
			dbRestoreStart := time.Now()

			// Update estimator progress (thread-safe)
			mu.Lock()
			estimator.UpdateProgress(idx)
			mu.Unlock()

			dumpFile := filepath.Join(dumpsDir, filename)
			dbName := filename
			dbName = strings.TrimSuffix(dbName, ".dump")
			dbName = strings.TrimSuffix(dbName, ".sql.gz")

			dbProgress := 15 + int(float64(idx)/float64(totalDBs)*85.0)

			// Calculate average time per DB and report progress with timing
			completedDBTimesMu.Lock()
			var avgPerDB time.Duration
			if len(completedDBTimes) > 0 {
				var totalDuration time.Duration
				for _, d := range completedDBTimes {
					totalDuration += d
				}
				avgPerDB = totalDuration / time.Duration(len(completedDBTimes))
			}
			phaseElapsed := time.Since(restorePhaseStart)
			completedDBTimesMu.Unlock()

			mu.Lock()
			statusMsg := fmt.Sprintf("Restoring database %s (%d/%d)", dbName, idx+1, totalDBs)
			e.progress.Update(statusMsg)
			e.log.Info("Restoring database", "name", dbName, "file", dumpFile, "progress", dbProgress)
			// Report database progress for TUI (both callbacks)
			e.reportDatabaseProgress(idx, totalDBs, dbName)
			e.reportDatabaseProgressWithTiming(idx, totalDBs, dbName, phaseElapsed, avgPerDB)
			mu.Unlock()

			// STEP 1: Drop existing database completely (clean slate)
			e.log.Info("Dropping existing database for clean restore", "name", dbName)
			if err := e.dropDatabaseIfExists(ctx, dbName); err != nil {
				e.log.Warn("Could not drop existing database", "name", dbName, "error", err)
			}

			// STEP 2: Create fresh database
			if err := e.ensureDatabaseExists(ctx, dbName); err != nil {
				e.log.Error("Failed to create database", "name", dbName, "error", err)
				restoreErrorsMu.Lock()
				restoreErrors = multierror.Append(restoreErrors, fmt.Errorf("%s: failed to create database: %w", dbName, err))
				restoreErrorsMu.Unlock()
				atomic.AddInt32(&failCount, 1)
				return
			}

			// STEP 3: Restore with ownership preservation if superuser
			preserveOwnership := isSuperuser
			isCompressedSQL := strings.HasSuffix(dumpFile, ".sql.gz")

			var restoreErr error
			if isCompressedSQL {
				mu.Lock()
				e.log.Info("Detected compressed SQL format, using psql + gunzip", "file", dumpFile, "database", dbName)
				mu.Unlock()
				restoreErr = e.restorePostgreSQLSQL(ctx, dumpFile, dbName, true)
			} else {
				mu.Lock()
				e.log.Info("Detected custom dump format, using pg_restore", "file", dumpFile, "database", dbName)
				mu.Unlock()
				restoreErr = e.restorePostgreSQLDumpWithOwnership(ctx, dumpFile, dbName, false, preserveOwnership)
			}

			if restoreErr != nil {
				mu.Lock()
				e.log.Error("Failed to restore database", "name", dbName, "file", dumpFile, "error", restoreErr)
				mu.Unlock()

				// Check for specific recoverable errors
				errMsg := restoreErr.Error()
				if strings.Contains(errMsg, "max_locks_per_transaction") {
					mu.Lock()
					e.log.Warn("Database restore failed due to insufficient locks - this is a PostgreSQL configuration issue",
						"database", dbName,
						"solution", "increase max_locks_per_transaction in postgresql.conf")
					mu.Unlock()
				} else if strings.Contains(errMsg, "total errors:") && strings.Contains(errMsg, "2562426") {
					mu.Lock()
					e.log.Warn("Database has massive error count - likely data corruption or incompatible dump format",
						"database", dbName,
						"errors", "2562426")
					mu.Unlock()
				}

				restoreErrorsMu.Lock()
				// Include more context in the error message
				restoreErrors = multierror.Append(restoreErrors, fmt.Errorf("%s: restore failed: %w", dbName, restoreErr))
				restoreErrorsMu.Unlock()
				atomic.AddInt32(&failCount, 1)
				return
			}

			// Track completed database restore duration for ETA calculation
			dbRestoreDuration := time.Since(dbRestoreStart)
			completedDBTimesMu.Lock()
			completedDBTimes = append(completedDBTimes, dbRestoreDuration)
			completedDBTimesMu.Unlock()

			atomic.AddInt32(&successCount, 1)
			
			// Small delay to ensure PostgreSQL fully closes connections before next restore
			time.Sleep(100 * time.Millisecond)
		}(dbIndex, entry.Name())

		dbIndex++
	}

	// Wait for all restores to complete
	wg.Wait()

	successCountFinal := int(atomic.LoadInt32(&successCount))
	failCountFinal := int(atomic.LoadInt32(&failCount))

	// SANITY CHECK: Verify all databases were accounted for
	// This catches any goroutine that exited without updating counters
	accountedFor := successCountFinal + failCountFinal
	if accountedFor != totalDBs {
		missingCount := totalDBs - accountedFor
		e.log.Error("INTERNAL ERROR: Some database restore goroutines did not report status",
			"expected", totalDBs,
			"success", successCountFinal,
			"failed", failCountFinal,
			"unaccounted", missingCount)

		// Treat unaccounted databases as failures
		failCountFinal += missingCount
		restoreErrorsMu.Lock()
		restoreErrors = multierror.Append(restoreErrors, fmt.Errorf("%d database(s) did not complete (possible goroutine crash or deadlock)", missingCount))
		restoreErrorsMu.Unlock()
	}

	// CRITICAL: Check if no databases were restored at all
	if successCountFinal == 0 {
		e.progress.Fail(fmt.Sprintf("Cluster restore FAILED: 0 of %d databases restored", totalDBs))
		operation.Fail("No databases were restored")

		if failCountFinal > 0 && restoreErrors != nil {
			return fmt.Errorf("cluster restore failed: all %d database(s) failed:\n%s", failCountFinal, restoreErrors.Error())
		}
		return fmt.Errorf("cluster restore failed: no databases were restored (0 of %d total). Check PostgreSQL logs for details", totalDBs)
	}

	if failCountFinal > 0 {
		// Format multi-error with detailed output
		restoreErrors.ErrorFormat = func(errs []error) string {
			if len(errs) == 1 {
				return errs[0].Error()
			}
			points := make([]string, len(errs))
			for i, err := range errs {
				points[i] = fmt.Sprintf("  • %s", err.Error())
			}
			return fmt.Sprintf("%d database(s) failed:\n%s", len(errs), strings.Join(points, "\n"))
		}

		// Log summary
		e.log.Info("Cluster restore completed with failures",
			"succeeded", successCountFinal,
			"failed", failCountFinal,
			"total", totalDBs)

		e.progress.Fail(fmt.Sprintf("Cluster restore: %d succeeded, %d failed out of %d total", successCountFinal, failCountFinal, totalDBs))
		operation.Complete(fmt.Sprintf("Partial restore: %d/%d databases succeeded", successCountFinal, totalDBs))

		return fmt.Errorf("cluster restore completed with %d failures:\n%s", failCountFinal, restoreErrors.Error())
	}

	e.progress.Complete(fmt.Sprintf("Cluster restored successfully: %d databases", successCountFinal))
	operation.Complete(fmt.Sprintf("Restored %d databases from cluster archive", successCountFinal))
	return nil
}

// extractArchive extracts a tar.gz archive with progress reporting
func (e *Engine) extractArchive(ctx context.Context, archivePath, destDir string) error {
	// If progress callback is set, use Go's archive/tar for progress tracking
	if e.progressCallback != nil {
		return e.extractArchiveWithProgress(ctx, archivePath, destDir)
	}

	// Otherwise use fast shell tar (no progress)
	return e.extractArchiveShell(ctx, archivePath, destDir)
}

// extractArchiveWithProgress extracts using Go's archive/tar with detailed progress reporting
func (e *Engine) extractArchiveWithProgress(ctx context.Context, archivePath, destDir string) error {
	// Get archive size for progress calculation
	archiveInfo, err := os.Stat(archivePath)
	if err != nil {
		return fmt.Errorf("failed to stat archive: %w", err)
	}
	totalSize := archiveInfo.Size()

	// Open the archive file
	file, err := os.Open(archivePath)
	if err != nil {
		return fmt.Errorf("failed to open archive: %w", err)
	}
	defer file.Close()

	// Wrap with progress reader
	progressReader := &progressReader{
		reader:    file,
		totalSize: totalSize,
		callback:  e.progressCallback,
		desc:      "Extracting archive",
	}

	// Create gzip reader
	gzReader, err := gzip.NewReader(progressReader)
	if err != nil {
		return fmt.Errorf("failed to create gzip reader: %w", err)
	}
	defer gzReader.Close()

	// Create tar reader
	tarReader := tar.NewReader(gzReader)

	// Extract files
	for {
		select {
		case <-ctx.Done():
			return ctx.Err()
		default:
		}

		header, err := tarReader.Next()
		if err == io.EOF {
			break // End of archive
		}
		if err != nil {
			return fmt.Errorf("failed to read tar header: %w", err)
		}

		// Sanitize and validate path
		targetPath := filepath.Join(destDir, header.Name)

		// Security check: ensure path is within destDir (prevent path traversal)
		if !strings.HasPrefix(filepath.Clean(targetPath), filepath.Clean(destDir)) {
			e.log.Warn("Skipping potentially malicious path in archive", "path", header.Name)
			continue
		}

		switch header.Typeflag {
		case tar.TypeDir:
			if err := os.MkdirAll(targetPath, 0755); err != nil {
				return fmt.Errorf("failed to create directory %s: %w", targetPath, err)
			}
		case tar.TypeReg:
			// Ensure parent directory exists
			if err := os.MkdirAll(filepath.Dir(targetPath), 0755); err != nil {
				return fmt.Errorf("failed to create parent directory: %w", err)
			}

			// Create the file
			outFile, err := os.OpenFile(targetPath, os.O_CREATE|os.O_WRONLY|os.O_TRUNC, os.FileMode(header.Mode))
			if err != nil {
				return fmt.Errorf("failed to create file %s: %w", targetPath, err)
			}

			// Copy file contents
			if _, err := io.Copy(outFile, tarReader); err != nil {
				outFile.Close()
				return fmt.Errorf("failed to write file %s: %w", targetPath, err)
			}
			outFile.Close()
		case tar.TypeSymlink:
			// Handle symlinks (common in some archives)
			if err := os.Symlink(header.Linkname, targetPath); err != nil {
				// Ignore symlink errors (may already exist or not supported)
				e.log.Debug("Could not create symlink", "path", targetPath, "target", header.Linkname)
			}
		}
	}

	// Final progress update
	e.reportProgress(totalSize, totalSize, "Extraction complete")
	return nil
}

// progressReader wraps an io.Reader to report read progress
type progressReader struct {
	reader      io.Reader
	totalSize   int64
	bytesRead   int64
	callback    ProgressCallback
	desc        string
	lastReport  time.Time
	reportEvery time.Duration
}

func (pr *progressReader) Read(p []byte) (n int, err error) {
	n, err = pr.reader.Read(p)
	pr.bytesRead += int64(n)

	// Throttle progress reporting to every 100ms
	if pr.reportEvery == 0 {
		pr.reportEvery = 100 * time.Millisecond
	}
	if time.Since(pr.lastReport) > pr.reportEvery {
		if pr.callback != nil {
			pr.callback(pr.bytesRead, pr.totalSize, pr.desc)
		}
		pr.lastReport = time.Now()
	}

	return n, err
}

// extractArchiveShell extracts using shell tar command (faster but no progress)
func (e *Engine) extractArchiveShell(ctx context.Context, archivePath, destDir string) error {
	cmd := exec.CommandContext(ctx, "tar", "-xzf", archivePath, "-C", destDir)

	// Stream stderr to avoid memory issues - tar can produce lots of output for large archives
	stderr, err := cmd.StderrPipe()
	if err != nil {
		return fmt.Errorf("failed to create stderr pipe: %w", err)
	}

	if err := cmd.Start(); err != nil {
		return fmt.Errorf("failed to start tar: %w", err)
	}

	// Discard stderr output in chunks to prevent memory buildup
	stderrDone := make(chan struct{})
	go func() {
		defer close(stderrDone)
		buf := make([]byte, 4096)
		for {
			_, err := stderr.Read(buf)
			if err != nil {
				break
			}
		}
	}()

	// Wait for command with proper context handling
	cmdDone := make(chan error, 1)
	go func() {
		cmdDone <- cmd.Wait()
	}()

	var cmdErr error
	select {
	case cmdErr = <-cmdDone:
		// Command completed
	case <-ctx.Done():
		e.log.Warn("Archive extraction cancelled - killing process")
		cmd.Process.Kill()
		<-cmdDone
		cmdErr = ctx.Err()
	}

	<-stderrDone

	if cmdErr != nil {
		return fmt.Errorf("tar extraction failed: %w", cmdErr)
	}
	return nil
}

// restoreGlobals restores global objects (roles, tablespaces)
// Note: psql returns 0 even when some statements fail (e.g., role already exists)
// We track errors but only fail on FATAL errors that would prevent restore
func (e *Engine) restoreGlobals(ctx context.Context, globalsFile string) error {
	args := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-f", globalsFile,
	}

	// Only add -h flag if host is not localhost (to use Unix socket for peer auth)
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		args = append([]string{"-h", e.cfg.Host}, args...)
	}

	cmd := exec.CommandContext(ctx, "psql", args...)

	cmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

	// Stream output to avoid memory issues with large globals.sql files
	stderr, err := cmd.StderrPipe()
	if err != nil {
		return fmt.Errorf("failed to create stderr pipe: %w", err)
	}

	if err := cmd.Start(); err != nil {
		return fmt.Errorf("failed to start psql: %w", err)
	}

	// Read stderr in chunks in goroutine
	var lastError string
	var errorCount int
	var fatalError bool
	stderrDone := make(chan struct{})
	go func() {
		defer close(stderrDone)
		buf := make([]byte, 4096)
		for {
			n, err := stderr.Read(buf)
			if n > 0 {
				chunk := string(buf[:n])
				// Track different error types
				if strings.Contains(chunk, "FATAL") {
					fatalError = true
					lastError = chunk
					e.log.Error("Globals restore FATAL error", "output", chunk)
				} else if strings.Contains(chunk, "ERROR") {
					errorCount++
					lastError = chunk
					// Only log first few errors to avoid spam
					if errorCount <= 5 {
						// Check if it's an ignorable "already exists" error
						if strings.Contains(chunk, "already exists") {
							e.log.Debug("Globals restore: object already exists (expected)", "output", chunk)
						} else {
							e.log.Warn("Globals restore error", "output", chunk)
						}
					}
				}
			}
			if err != nil {
				break
			}
		}
	}()

	// Wait for command with proper context handling
	cmdDone := make(chan error, 1)
	go func() {
		cmdDone <- cmd.Wait()
	}()

	var cmdErr error
	select {
	case cmdErr = <-cmdDone:
		// Command completed
	case <-ctx.Done():
		e.log.Warn("Globals restore cancelled - killing process")
		cmd.Process.Kill()
		<-cmdDone
		cmdErr = ctx.Err()
	}

	<-stderrDone

	// Only fail on actual command errors or FATAL PostgreSQL errors
	// Regular ERROR messages (like "role already exists") are expected
	if cmdErr != nil {
		return fmt.Errorf("failed to restore globals: %w (last error: %s)", cmdErr, lastError)
	}

	// If we had FATAL errors, those are real problems
	if fatalError {
		return fmt.Errorf("globals restore had FATAL error: %s", lastError)
	}

	// Log summary if there were errors (but don't fail)
	if errorCount > 0 {
		e.log.Info("Globals restore completed with some errors (usually 'already exists' - expected)",
			"error_count", errorCount)
	}

	return nil
}

// checkSuperuser verifies if the current user has superuser privileges
func (e *Engine) checkSuperuser(ctx context.Context) (bool, error) {
	args := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-tAc", "SELECT usesuper FROM pg_user WHERE usename = current_user",
	}

	// Only add -h flag if host is not localhost (to use Unix socket for peer auth)
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		args = append([]string{"-h", e.cfg.Host}, args...)
	}

	cmd := exec.CommandContext(ctx, "psql", args...)

	// Always set PGPASSWORD (empty string is fine for peer/ident auth)
	cmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

	output, err := cmd.CombinedOutput()
	if err != nil {
		return false, fmt.Errorf("failed to check superuser status: %w", err)
	}

	isSuperuser := strings.TrimSpace(string(output)) == "t"
	return isSuperuser, nil
}

// terminateConnections kills all active connections to a database
func (e *Engine) terminateConnections(ctx context.Context, dbName string) error {
	query := fmt.Sprintf(`
		SELECT pg_terminate_backend(pid)
		FROM pg_stat_activity
		WHERE datname = '%s'
		AND pid <> pg_backend_pid()
	`, dbName)

	args := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-tAc", query,
	}

	// Only add -h flag if host is not localhost (to use Unix socket for peer auth)
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		args = append([]string{"-h", e.cfg.Host}, args...)
	}

	cmd := exec.CommandContext(ctx, "psql", args...)

	// Always set PGPASSWORD (empty string is fine for peer/ident auth)
	cmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

	output, err := cmd.CombinedOutput()
	if err != nil {
		e.log.Warn("Failed to terminate connections", "database", dbName, "error", err, "output", string(output))
		// Don't fail - database might not exist or have no connections
	}

	return nil
}

// dropDatabaseIfExists drops a database completely (clean slate)
// Uses PostgreSQL 13+ WITH (FORCE) option to forcefully drop even with active connections
func (e *Engine) dropDatabaseIfExists(ctx context.Context, dbName string) error {
	// First terminate all connections
	if err := e.terminateConnections(ctx, dbName); err != nil {
		e.log.Warn("Could not terminate connections", "database", dbName, "error", err)
	}

	// Wait a moment for connections to terminate
	time.Sleep(500 * time.Millisecond)

	// Try to revoke new connections (prevents race condition)
	// This only works if we have the privilege to do so
	revokeArgs := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-c", fmt.Sprintf("REVOKE CONNECT ON DATABASE \"%s\" FROM PUBLIC", dbName),
	}
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		revokeArgs = append([]string{"-h", e.cfg.Host}, revokeArgs...)
	}
	revokeCmd := exec.CommandContext(ctx, "psql", revokeArgs...)
	revokeCmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))
	revokeCmd.Run() // Ignore errors - database might not exist

	// Terminate connections again after revoking connect privilege
	e.terminateConnections(ctx, dbName)
	time.Sleep(200 * time.Millisecond)

	// Try DROP DATABASE WITH (FORCE) first (PostgreSQL 13+)
	// This forcefully terminates connections and drops the database atomically
	forceArgs := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-c", fmt.Sprintf("DROP DATABASE IF EXISTS \"%s\" WITH (FORCE)", dbName),
	}
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		forceArgs = append([]string{"-h", e.cfg.Host}, forceArgs...)
	}
	forceCmd := exec.CommandContext(ctx, "psql", forceArgs...)
	forceCmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

	output, err := forceCmd.CombinedOutput()
	if err == nil {
		e.log.Info("Dropped existing database (with FORCE)", "name", dbName)
		return nil
	}

	// If FORCE option failed (PostgreSQL < 13), try regular drop
	if strings.Contains(string(output), "syntax error") || strings.Contains(string(output), "WITH (FORCE)") {
		e.log.Debug("WITH (FORCE) not supported, using standard DROP", "name", dbName)

		args := []string{
			"-p", fmt.Sprintf("%d", e.cfg.Port),
			"-U", e.cfg.User,
			"-d", "postgres",
			"-c", fmt.Sprintf("DROP DATABASE IF EXISTS \"%s\"", dbName),
		}
		if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
			args = append([]string{"-h", e.cfg.Host}, args...)
		}

		cmd := exec.CommandContext(ctx, "psql", args...)
		cmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

		output, err = cmd.CombinedOutput()
		if err != nil {
			return fmt.Errorf("failed to drop database '%s': %w\nOutput: %s", dbName, err, string(output))
		}
	} else if err != nil {
		return fmt.Errorf("failed to drop database '%s': %w\nOutput: %s", dbName, err, string(output))
	}

	e.log.Info("Dropped existing database", "name", dbName)
	return nil
}

// ensureDatabaseExists checks if a database exists and creates it if not
func (e *Engine) ensureDatabaseExists(ctx context.Context, dbName string) error {
	// Route to appropriate implementation based on database type
	if e.cfg.DatabaseType == "mysql" || e.cfg.DatabaseType == "mariadb" {
		return e.ensureMySQLDatabaseExists(ctx, dbName)
	}
	return e.ensurePostgresDatabaseExists(ctx, dbName)
}

// ensureMySQLDatabaseExists checks if a MySQL database exists and creates it if not
func (e *Engine) ensureMySQLDatabaseExists(ctx context.Context, dbName string) error {
	// Build mysql command
	args := []string{
		"-h", e.cfg.Host,
		"-P", fmt.Sprintf("%d", e.cfg.Port),
		"-u", e.cfg.User,
		"-e", fmt.Sprintf("CREATE DATABASE IF NOT EXISTS `%s`", dbName),
	}

	if e.cfg.Password != "" {
		args = append(args, fmt.Sprintf("-p%s", e.cfg.Password))
	}

	cmd := exec.CommandContext(ctx, "mysql", args...)
	output, err := cmd.CombinedOutput()
	if err != nil {
		e.log.Warn("MySQL database creation failed", "name", dbName, "error", err, "output", string(output))
		return fmt.Errorf("failed to create database '%s': %w (output: %s)", dbName, err, strings.TrimSpace(string(output)))
	}

	e.log.Info("Successfully ensured MySQL database exists", "name", dbName)
	return nil
}

// ensurePostgresDatabaseExists checks if a PostgreSQL database exists and creates it if not
// It attempts to extract encoding/locale from the dump file to preserve original settings
func (e *Engine) ensurePostgresDatabaseExists(ctx context.Context, dbName string) error {
	// Skip creation for postgres and template databases - they should already exist
	if dbName == "postgres" || dbName == "template0" || dbName == "template1" {
		e.log.Info("Skipping create for system database (assume exists)", "name", dbName)
		return nil
	}

	// Build psql command with authentication
	buildPsqlCmd := func(ctx context.Context, database, query string) *exec.Cmd {
		args := []string{
			"-p", fmt.Sprintf("%d", e.cfg.Port),
			"-U", e.cfg.User,
			"-d", database,
			"-tAc", query,
		}

		// Only add -h flag if host is not localhost (to use Unix socket for peer auth)
		if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
			args = append([]string{"-h", e.cfg.Host}, args...)
		}

		cmd := exec.CommandContext(ctx, "psql", args...)

		// Always set PGPASSWORD (empty string is fine for peer/ident auth)
		cmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

		return cmd
	}

	// Check if database exists
	checkCmd := buildPsqlCmd(ctx, "postgres", fmt.Sprintf("SELECT 1 FROM pg_database WHERE datname = '%s'", dbName))

	output, err := checkCmd.CombinedOutput()
	if err != nil {
		e.log.Warn("Database existence check failed", "name", dbName, "error", err, "output", string(output))
		// Continue anyway - maybe we can create it
	}

	// If database exists, we're done
	if strings.TrimSpace(string(output)) == "1" {
		e.log.Info("Database already exists", "name", dbName)
		return nil
	}

	// Database doesn't exist, create it
	// IMPORTANT: Use template0 to avoid duplicate definition errors from local additions to template1
	// Also use UTF8 encoding explicitly as it's the most common and safest choice
	// See PostgreSQL docs: https://www.postgresql.org/docs/current/app-pgrestore.html#APP-PGRESTORE-NOTES
	e.log.Info("Creating database from template0 with UTF8 encoding", "name", dbName)

	// Get server's default locale for LC_COLLATE and LC_CTYPE
	// This ensures compatibility while using the correct encoding
	localeCmd := buildPsqlCmd(ctx, "postgres", "SHOW lc_collate")
	localeOutput, _ := localeCmd.CombinedOutput()
	serverLocale := strings.TrimSpace(string(localeOutput))
	if serverLocale == "" {
		serverLocale = "en_US.UTF-8" // Fallback to common default
	}

	// Build CREATE DATABASE command with encoding and locale
	// Using ENCODING 'UTF8' explicitly ensures the dump can be restored
	createSQL := fmt.Sprintf(
		"CREATE DATABASE \"%s\" WITH TEMPLATE template0 ENCODING 'UTF8' LC_COLLATE '%s' LC_CTYPE '%s'",
		dbName, serverLocale, serverLocale,
	)

	createArgs := []string{
		"-p", fmt.Sprintf("%d", e.cfg.Port),
		"-U", e.cfg.User,
		"-d", "postgres",
		"-c", createSQL,
	}

	// Only add -h flag if host is not localhost (to use Unix socket for peer auth)
	if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
		createArgs = append([]string{"-h", e.cfg.Host}, createArgs...)
	}

	createCmd := exec.CommandContext(ctx, "psql", createArgs...)

	// Always set PGPASSWORD (empty string is fine for peer/ident auth)
	createCmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

	output, err = createCmd.CombinedOutput()
	if err != nil {
		// If encoding/locale fails, try simpler CREATE DATABASE
		e.log.Warn("Database creation with encoding failed, trying simple create", "name", dbName, "error", err)

		simpleArgs := []string{
			"-p", fmt.Sprintf("%d", e.cfg.Port),
			"-U", e.cfg.User,
			"-d", "postgres",
			"-c", fmt.Sprintf("CREATE DATABASE \"%s\" WITH TEMPLATE template0", dbName),
		}
		if e.cfg.Host != "localhost" && e.cfg.Host != "127.0.0.1" && e.cfg.Host != "" {
			simpleArgs = append([]string{"-h", e.cfg.Host}, simpleArgs...)
		}

		simpleCmd := exec.CommandContext(ctx, "psql", simpleArgs...)
		simpleCmd.Env = append(os.Environ(), fmt.Sprintf("PGPASSWORD=%s", e.cfg.Password))

		output, err = simpleCmd.CombinedOutput()
		if err != nil {
			e.log.Warn("Database creation failed", "name", dbName, "error", err, "output", string(output))
			return fmt.Errorf("failed to create database '%s': %w (output: %s)", dbName, err, strings.TrimSpace(string(output)))
		}
	}

	e.log.Info("Successfully created database from template0", "name", dbName)
	return nil
}

// previewClusterRestore shows cluster restore preview
func (e *Engine) previewClusterRestore(archivePath string) error {
	fmt.Println("\n" + strings.Repeat("=", 60))
	fmt.Println(" CLUSTER RESTORE PREVIEW (DRY RUN)")
	fmt.Println(strings.Repeat("=", 60))

	stat, _ := os.Stat(archivePath)
	fmt.Printf("\nArchive: %s\n", filepath.Base(archivePath))
	if stat != nil {
		fmt.Printf("Size: %s\n", FormatBytes(stat.Size()))
		fmt.Printf("Modified: %s\n", stat.ModTime().Format("2006-01-02 15:04:05"))
	}
	fmt.Printf("Target Host: %s:%d\n", e.cfg.Host, e.cfg.Port)

	fmt.Println("\nOperations that would be performed:")
	fmt.Println("  1. Extract cluster archive to temporary directory")
	fmt.Println("  2. Restore global objects (roles, tablespaces)")
	fmt.Println("  3. Restore all databases found in archive")
	fmt.Println("  4. Cleanup temporary files")

	fmt.Println("\n[WARN]  WARNING: This will restore multiple databases.")
	fmt.Println("   Existing databases may be overwritten or merged.")
	fmt.Println("\nTo execute this restore, add the --confirm flag.")
	fmt.Println(strings.Repeat("=", 60) + "\n")

	return nil
}

// detectLargeObjectsInDumps checks if any dump files contain large objects
func (e *Engine) detectLargeObjectsInDumps(dumpsDir string, entries []os.DirEntry) bool {
	hasLargeObjects := false
	checkedCount := 0
	maxChecks := 5 // Only check first 5 dumps to avoid slowdown

	for _, entry := range entries {
		if entry.IsDir() || checkedCount >= maxChecks {
			continue
		}

		dumpFile := filepath.Join(dumpsDir, entry.Name())

		// Skip compressed SQL files (can't easily check without decompressing)
		if strings.HasSuffix(dumpFile, ".sql.gz") {
			continue
		}

		// Use pg_restore -l to list contents (fast, doesn't restore data)
		// 2 minutes for large dumps with many objects
		ctx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
		defer cancel()

		cmd := exec.CommandContext(ctx, "pg_restore", "-l", dumpFile)
		output, err := cmd.Output()

		if err != nil {
			// If pg_restore -l fails, it might not be custom format - skip
			continue
		}

		checkedCount++

		// Check if output contains "BLOB" or "LARGE OBJECT" entries
		outputStr := string(output)
		if strings.Contains(outputStr, "BLOB") ||
			strings.Contains(outputStr, "LARGE OBJECT") ||
			strings.Contains(outputStr, " BLOBS ") {
			e.log.Info("Large objects detected in dump file", "file", entry.Name())
			hasLargeObjects = true
			// Don't break - log all files with large objects
		}
	}

	if hasLargeObjects {
		e.log.Warn("Cluster contains databases with large objects - parallel restore may cause lock contention")
	}

	return hasLargeObjects
}

// isIgnorableError checks if an error message represents an ignorable PostgreSQL restore error
func (e *Engine) isIgnorableError(errorMsg string) bool {
	// Convert to lowercase for case-insensitive matching
	lowerMsg := strings.ToLower(errorMsg)

	// CRITICAL: Syntax errors are NOT ignorable - indicates corrupted dump
	if strings.Contains(lowerMsg, "syntax error") {
		e.log.Error("CRITICAL: Syntax error in dump file - dump may be corrupted", "error", errorMsg)
		return false
	}

	// CRITICAL: If error count is extremely high (>100k), dump is likely corrupted
	if strings.Contains(errorMsg, "total errors:") {
		// Extract error count if present in message
		parts := strings.Split(errorMsg, "total errors:")
		if len(parts) > 1 {
			errorCountStr := strings.TrimSpace(strings.Split(parts[1], ")")[0])
			// Try to parse as number
			var count int
			if _, err := fmt.Sscanf(errorCountStr, "%d", &count); err == nil && count > 100000 {
				e.log.Error("CRITICAL: Excessive errors indicate corrupted dump", "error_count", count)
				return false
			}
		}
	}

	// List of ignorable error patterns (objects that already exist)
	ignorablePatterns := []string{
		"already exists",
		"duplicate key",
		"does not exist, skipping", // For DROP IF EXISTS
		"no pg_hba.conf entry",     // Permission warnings (not fatal)
	}

	for _, pattern := range ignorablePatterns {
		if strings.Contains(lowerMsg, pattern) {
			return true
		}
	}

	return false
}

// FormatBytes formats bytes to human readable format
func FormatBytes(bytes int64) string {
	const unit = 1024
	if bytes < unit {
		return fmt.Sprintf("%d B", bytes)
	}
	div, exp := int64(unit), 0
	for n := bytes / unit; n >= unit; n /= unit {
		div *= unit
		exp++
	}
	return fmt.Sprintf("%.1f %cB", float64(bytes)/float64(div), "KMGTPE"[exp])
}

// quickValidateSQLDump performs a fast validation of SQL dump files
// by checking for truncated COPY blocks. This catches corrupted dumps
// BEFORE attempting a full restore (which could waste 49+ minutes).
func (e *Engine) quickValidateSQLDump(archivePath string, compressed bool) error {
	e.log.Debug("Pre-validating SQL dump file", "path", archivePath, "compressed", compressed)

	diagnoser := NewDiagnoser(e.log, false) // non-verbose for speed
	result, err := diagnoser.DiagnoseFile(archivePath)
	if err != nil {
		return fmt.Errorf("diagnosis error: %w", err)
	}

	// Check for critical issues that would cause restore failure
	if result.IsTruncated {
		errMsg := "SQL dump file is TRUNCATED"
		if result.Details != nil && result.Details.UnterminatedCopy {
			errMsg = fmt.Sprintf("%s - unterminated COPY block for table '%s' at line %d",
				errMsg, result.Details.LastCopyTable, result.Details.LastCopyLineNumber)
			if len(result.Details.SampleCopyData) > 0 {
				errMsg = fmt.Sprintf("%s (sample orphaned data: %s)", errMsg, result.Details.SampleCopyData[0])
			}
		}
		return fmt.Errorf("%s", errMsg)
	}

	if result.IsCorrupted {
		return fmt.Errorf("SQL dump file is corrupted: %v", result.Errors)
	}

	if !result.IsValid {
		if len(result.Errors) > 0 {
			return fmt.Errorf("dump validation failed: %s", result.Errors[0])
		}
		return fmt.Errorf("dump file is invalid (unknown reason)")
	}

	// Log any warnings but don't fail
	for _, warning := range result.Warnings {
		e.log.Warn("Dump validation warning", "warning", warning)
	}

	e.log.Debug("SQL dump validation passed", "path", archivePath)
	return nil
}

// boostLockCapacity temporarily increases max_locks_per_transaction to prevent OOM
// during large restores with many BLOBs. Returns the original value for later reset.
// Uses ALTER SYSTEM + pg_reload_conf() so no restart is needed.
func (e *Engine) boostLockCapacity(ctx context.Context) (int, error) {
	// Connect to PostgreSQL to run system commands
	connStr := fmt.Sprintf("host=%s port=%d user=%s password=%s dbname=postgres sslmode=disable",
		e.cfg.Host, e.cfg.Port, e.cfg.User, e.cfg.Password)

	// For localhost, use Unix socket
	if e.cfg.Host == "localhost" || e.cfg.Host == "" {
		connStr = fmt.Sprintf("user=%s password=%s dbname=postgres sslmode=disable",
			e.cfg.User, e.cfg.Password)
	}

	db, err := sql.Open("pgx", connStr)
	if err != nil {
		return 0, fmt.Errorf("failed to connect: %w", err)
	}
	defer db.Close()

	// Get current value
	var currentValue int
	err = db.QueryRowContext(ctx, "SHOW max_locks_per_transaction").Scan(&currentValue)
	if err != nil {
		// Try parsing as string (some versions return string)
		var currentValueStr string
		err = db.QueryRowContext(ctx, "SHOW max_locks_per_transaction").Scan(&currentValueStr)
		if err != nil {
			return 0, fmt.Errorf("failed to get current max_locks_per_transaction: %w", err)
		}
		fmt.Sscanf(currentValueStr, "%d", &currentValue)
	}

	// Skip if already high enough
	if currentValue >= 2048 {
		e.log.Info("max_locks_per_transaction already sufficient", "value", currentValue)
		return currentValue, nil
	}

	// Boost to 2048 (enough for most BLOB-heavy databases)
	_, err = db.ExecContext(ctx, "ALTER SYSTEM SET max_locks_per_transaction = 2048")
	if err != nil {
		return currentValue, fmt.Errorf("failed to set max_locks_per_transaction: %w", err)
	}

	// Reload config without restart
	_, err = db.ExecContext(ctx, "SELECT pg_reload_conf()")
	if err != nil {
		return currentValue, fmt.Errorf("failed to reload config: %w", err)
	}

	return currentValue, nil
}

// resetLockCapacity restores the original max_locks_per_transaction value
func (e *Engine) resetLockCapacity(ctx context.Context, originalValue int) error {
	connStr := fmt.Sprintf("host=%s port=%d user=%s password=%s dbname=postgres sslmode=disable",
		e.cfg.Host, e.cfg.Port, e.cfg.User, e.cfg.Password)

	if e.cfg.Host == "localhost" || e.cfg.Host == "" {
		connStr = fmt.Sprintf("user=%s password=%s dbname=postgres sslmode=disable",
			e.cfg.User, e.cfg.Password)
	}

	db, err := sql.Open("pgx", connStr)
	if err != nil {
		return fmt.Errorf("failed to connect: %w", err)
	}
	defer db.Close()

	// Reset to original value (or use RESET to go back to default)
	if originalValue == 64 { // Default value
		_, err = db.ExecContext(ctx, "ALTER SYSTEM RESET max_locks_per_transaction")
	} else {
		_, err = db.ExecContext(ctx, fmt.Sprintf("ALTER SYSTEM SET max_locks_per_transaction = %d", originalValue))
	}
	if err != nil {
		return fmt.Errorf("failed to reset max_locks_per_transaction: %w", err)
	}

	// Reload config
	_, err = db.ExecContext(ctx, "SELECT pg_reload_conf()")
	if err != nil {
		return fmt.Errorf("failed to reload config: %w", err)
	}

	return nil
}

// OriginalSettings stores PostgreSQL settings to restore after operation
type OriginalSettings struct {
	MaxLocks           int
	MaintenanceWorkMem string
}

// boostPostgreSQLSettings boosts multiple PostgreSQL settings for large restores
// NOTE: max_locks_per_transaction requires a PostgreSQL RESTART to take effect!
// maintenance_work_mem can be changed with pg_reload_conf().
func (e *Engine) boostPostgreSQLSettings(ctx context.Context, lockBoostValue int) (*OriginalSettings, error) {
	connStr := e.buildConnString()
	db, err := sql.Open("pgx", connStr)
	if err != nil {
		return nil, fmt.Errorf("failed to connect: %w", err)
	}
	defer db.Close()

	original := &OriginalSettings{}

	// Get current max_locks_per_transaction
	var maxLocksStr string
	if err := db.QueryRowContext(ctx, "SHOW max_locks_per_transaction").Scan(&maxLocksStr); err == nil {
		original.MaxLocks, _ = strconv.Atoi(maxLocksStr)
	}

	// Get current maintenance_work_mem
	db.QueryRowContext(ctx, "SHOW maintenance_work_mem").Scan(&original.MaintenanceWorkMem)

	// CRITICAL: max_locks_per_transaction requires a PostgreSQL RESTART!
	// pg_reload_conf() is NOT sufficient for this parameter.
	needsRestart := false
	if original.MaxLocks < lockBoostValue {
		_, err = db.ExecContext(ctx, fmt.Sprintf("ALTER SYSTEM SET max_locks_per_transaction = %d", lockBoostValue))
		if err != nil {
			e.log.Warn("Could not set max_locks_per_transaction", "error", err)
		} else {
			needsRestart = true
			e.log.Warn("max_locks_per_transaction requires PostgreSQL restart to take effect",
				"current", original.MaxLocks,
				"target", lockBoostValue)
		}
	}

	// Boost maintenance_work_mem to 2GB for faster index creation
	// (this one CAN be applied via pg_reload_conf)
	_, err = db.ExecContext(ctx, "ALTER SYSTEM SET maintenance_work_mem = '2GB'")
	if err != nil {
		e.log.Warn("Could not boost maintenance_work_mem", "error", err)
	}

	// Reload config to apply maintenance_work_mem
	_, err = db.ExecContext(ctx, "SELECT pg_reload_conf()")
	if err != nil {
		return original, fmt.Errorf("failed to reload config: %w", err)
	}

	// If max_locks_per_transaction needs a restart, try to do it
	if needsRestart {
		if restarted := e.tryRestartPostgreSQL(ctx); restarted {
			e.log.Info("PostgreSQL restarted successfully - max_locks_per_transaction now active")
			// Wait for PostgreSQL to be ready
			time.Sleep(3 * time.Second)
		} else {
			// Cannot restart - warn user but continue
			// The setting is written to postgresql.auto.conf and will take effect on next restart
			e.log.Warn("=" + strings.Repeat("=", 70))
			e.log.Warn("NOTE: max_locks_per_transaction change requires PostgreSQL restart")
			e.log.Warn("Current value: " + strconv.Itoa(original.MaxLocks) + ", target: " + strconv.Itoa(lockBoostValue))
			e.log.Warn("")
			e.log.Warn("The setting has been saved to postgresql.auto.conf and will take")
			e.log.Warn("effect on the next PostgreSQL restart. If restore fails with")
			e.log.Warn("'out of shared memory' errors, ask your DBA to restart PostgreSQL.")
			e.log.Warn("")
			e.log.Warn("Continuing with restore - this may succeed if your databases")
			e.log.Warn("don't have many large objects (BLOBs).")
			e.log.Warn("=" + strings.Repeat("=", 70))
			// Continue anyway - might work for small restores or DBs without BLOBs
		}
	}

	return original, nil
}

// canRestartPostgreSQL checks if we have the ability to restart PostgreSQL
// Returns false if running in a restricted environment (e.g., su postgres on enterprise systems)
func (e *Engine) canRestartPostgreSQL() bool {
	// Check if we're running as postgres user - if so, we likely can't restart
	// because PostgreSQL is managed by init/systemd, not directly by pg_ctl
	currentUser := os.Getenv("USER")
	if currentUser == "" {
		currentUser = os.Getenv("LOGNAME")
	}

	// If we're the postgres user, check if we have sudo access
	if currentUser == "postgres" {
		// Try a quick sudo check - if this fails, we can't restart
		ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
		defer cancel()
		cmd := exec.CommandContext(ctx, "sudo", "-n", "true")
		cmd.Stdin = nil
		if err := cmd.Run(); err != nil {
			e.log.Info("Running as postgres user without sudo access - cannot restart PostgreSQL",
				"user", currentUser,
				"hint", "Ask system administrator to restart PostgreSQL if needed")
			return false
		}
	}

	return true
}

// tryRestartPostgreSQL attempts to restart PostgreSQL using various methods
// Returns true if restart was successful
// IMPORTANT: Uses short timeouts and non-interactive sudo to avoid blocking on password prompts
// NOTE: This function will return false immediately if running as postgres without sudo
func (e *Engine) tryRestartPostgreSQL(ctx context.Context) bool {
	// First check if we can even attempt a restart
	if !e.canRestartPostgreSQL() {
		e.log.Info("Skipping PostgreSQL restart attempt (no privileges)")
		return false
	}

	e.progress.Update("Attempting PostgreSQL restart for lock settings...")

	// Use short timeout for each restart attempt (don't block on sudo password prompts)
	runWithTimeout := func(args ...string) bool {
		cmdCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
		defer cancel()
		cmd := exec.CommandContext(cmdCtx, args[0], args[1:]...)
		// Set stdin to /dev/null to prevent sudo from waiting for password
		cmd.Stdin = nil
		return cmd.Run() == nil
	}

	// Method 1: systemctl (most common on modern Linux) - use sudo -n for non-interactive
	if runWithTimeout("sudo", "-n", "systemctl", "restart", "postgresql") {
		return true
	}

	// Method 2: systemctl with version suffix (e.g., postgresql-15)
	for _, ver := range []string{"17", "16", "15", "14", "13", "12"} {
		if runWithTimeout("sudo", "-n", "systemctl", "restart", "postgresql-"+ver) {
			return true
		}
	}

	// Method 3: service command (older systems)
	if runWithTimeout("sudo", "-n", "service", "postgresql", "restart") {
		return true
	}

	// Method 4: pg_ctl as postgres user (if we ARE postgres user, no sudo needed)
	if runWithTimeout("pg_ctl", "restart", "-D", "/var/lib/postgresql/data", "-m", "fast") {
		return true
	}

	// Method 5: Try common PGDATA paths with pg_ctl directly (for postgres user)
	pgdataPaths := []string{
		"/var/lib/pgsql/data",
		"/var/lib/pgsql/17/data",
		"/var/lib/pgsql/16/data",
		"/var/lib/pgsql/15/data",
		"/var/lib/postgresql/17/main",
		"/var/lib/postgresql/16/main",
		"/var/lib/postgresql/15/main",
	}
	for _, pgdata := range pgdataPaths {
		if runWithTimeout("pg_ctl", "restart", "-D", pgdata, "-m", "fast") {
			return true
		}
	}

	return false
}

// resetPostgreSQLSettings restores original PostgreSQL settings
// NOTE: max_locks_per_transaction changes are written but require restart to take effect.
// We don't restart here since we're done with the restore.
func (e *Engine) resetPostgreSQLSettings(ctx context.Context, original *OriginalSettings) error {
	connStr := e.buildConnString()
	db, err := sql.Open("pgx", connStr)
	if err != nil {
		return fmt.Errorf("failed to connect: %w", err)
	}
	defer db.Close()

	// Reset max_locks_per_transaction (will take effect on next restart)
	if original.MaxLocks == 64 { // Default
		db.ExecContext(ctx, "ALTER SYSTEM RESET max_locks_per_transaction")
	} else if original.MaxLocks > 0 {
		db.ExecContext(ctx, fmt.Sprintf("ALTER SYSTEM SET max_locks_per_transaction = %d", original.MaxLocks))
	}

	// Reset maintenance_work_mem (takes effect immediately with reload)
	if original.MaintenanceWorkMem == "64MB" { // Default
		db.ExecContext(ctx, "ALTER SYSTEM RESET maintenance_work_mem")
	} else if original.MaintenanceWorkMem != "" {
		db.ExecContext(ctx, fmt.Sprintf("ALTER SYSTEM SET maintenance_work_mem = '%s'", original.MaintenanceWorkMem))
	}

	// Reload config (only maintenance_work_mem will take effect immediately)
	_, err = db.ExecContext(ctx, "SELECT pg_reload_conf()")
	if err != nil {
		return fmt.Errorf("failed to reload config: %w", err)
	}

	e.log.Info("PostgreSQL settings reset queued",
		"note", "max_locks_per_transaction will revert on next PostgreSQL restart")

	return nil
}