verstak/internal/core/watcher/scanner.go

package watcher

import (
	"crypto/sha256"
	"encoding/hex"
	"fmt"
	"io"
	"io/fs"
	"log"
	"os"
	"path/filepath"
	"strings"
	"time"

	"verstak/internal/core/activity"
	"verstak/internal/core/files"
	"verstak/internal/core/nodes"
)

// SnapshotResult summarises what the snapshot scanner found.
type SnapshotResult struct {
	MissingFiles   int `json:"missing_files"`
	RestoredFiles  int `json:"restored_files"`
	ModifiedFiles  int `json:"modified_files"`
	NewFiles       int `json:"new_files"`
	NodesScanned   int `json:"nodes_scanned"`
}

// Scanner performs a one-shot scan of the vault filesystem.
type Scanner struct {
	vaultRoot string
	nodes     *nodes.Repository
	files     *files.Service
	activity  *activity.Service
}

// NewScanner creates a snapshot scanner.
func NewScanner(vaultRoot string, nr *nodes.Repository, fs *files.Service, as *activity.Service) *Scanner {
	return &Scanner{
		vaultRoot: vaultRoot,
		nodes:     nr,
		files:     fs,
		activity:  as,
	}
}

// Run performs one full scan and returns a summary.
func (s *Scanner) Run() (*SnapshotResult, error) {
	result := &SnapshotResult{}

	// 1. Collect all file records in the DB (including missing).
	dbFiles, err := s.files.ListAllVaultWithTrashed()
	if err != nil {
		return nil, fmt.Errorf("list vault files: %w", err)
	}

	// Index by path for O(1) lookup.
	byPath := make(map[string]*files.Record, len(dbFiles))
	for i := range dbFiles {
		rec := &dbFiles[i]
		byPath[rec.Path] = rec
	}

	// 2. Scan all nodes with FsPath to discover files on disk.
	allNodes, err := s.nodes.ListAllWithFsPath()
	if err != nil {
		return nil, fmt.Errorf("list nodes: %w", err)
	}

	scannedPaths := make(map[string]bool)

	for _, node := range allNodes {
		absDir := filepath.Join(s.vaultRoot, node.FsPath)
		info, err := os.Stat(absDir)
		if err != nil || !info.IsDir() {
			continue
		}

		result.NodesScanned++

		err = filepath.WalkDir(absDir, func(path string, d fs.DirEntry, err error) error {
			if err != nil {
				return nil // skip unreadable entries
			}
			if d.IsDir() {
				// Skip .verstak and hidden dirs.
				if strings.HasPrefix(d.Name(), ".") {
					return filepath.SkipDir
				}
				return nil
			}

			rel, err := filepath.Rel(s.vaultRoot, path)
			if err != nil {
				return nil
			}

			scannedPaths[rel] = true

			if rec, exists := byPath[rel]; exists {
				if rec.Missing {
					// File existed in DB as missing, now found on disk.
					_ = s.files.MarkMissing(rec.ID, false)
					s.logActivity(rec.NodeID, activity.TypeFileRestored, rec.Filename, rel)
					result.RestoredFiles++
				} else {
					// File exists in both places — check if content changed.
					fi, err := os.Stat(path)
					if err != nil {
						return nil
					}
					if fi.Size() != rec.Size || contentChanged(path, rec.SHA256) {
						sha, size := hashFile(path)
						// Update in DB. We don't have an Update method exposed,
						// but we can mark and re-add via the watcher state.
						_ = s.updateFileRecord(rec.ID, rec.NodeID, rec.Filename, rel, size, sha)
						s.logActivity(rec.NodeID, activity.TypeFileModified, rec.Filename, rel)
						result.ModifiedFiles++
					}
				}
				delete(byPath, rel)
			} else {
				// File on disk but no record in DB — new file.
				result.NewFiles++
			}
			return nil
		})
		if err != nil {
			log.Printf("[watcher] scan node %s error: %v", node.ID, err)
		}
	}

	// 3. Any remaining byPath entries are files in DB but missing on disk.
	for _, rec := range byPath {
		if !rec.Missing {
			_ = s.files.MarkMissing(rec.ID, true)
			s.logActivity(rec.NodeID, activity.TypeFileDeleted, rec.Filename, rec.Path)
			result.MissingFiles++
		}
	}

	return result, nil
}

func (s *Scanner) logActivity(nodeID, eventType, title, path string) {
	_ = s.activity.Record(nodeID, activity.TargetFile, "", path, eventType, title, "")
}

func (s *Scanner) updateFileRecord(id, nodeID, filename, path string, size int64, sha string) error {
	// Direct SQL update since files.Service doesn't expose an update method.
	_, err := s.files.DB().Exec(
		`UPDATE files SET size=?, sha256=?, updated_at=?, missing=0 WHERE id=?`,
		size, sha, time.Now().UTC().Format(time.RFC3339), id)
	return err
}

// hashFile computes SHA256 and returns size and hex hash.
func hashFile(absPath string) (string, int64) {
	f, err := os.Open(absPath)
	if err != nil {
		return "", 0
	}
	defer f.Close()

	h := sha256.New()
	n, err := io.Copy(h, f)
	if err != nil {
		return "", 0
	}
	return hex.EncodeToString(h.Sum(nil)), n
}

// contentChanged returns true if the file's SHA256 differs from the stored hash.
func contentChanged(absPath, storedHash string) bool {
	if storedHash == "" {
		return true
	}
	h, _ := hashFile(absPath)
	return h != storedHash
}