Merge pull request 'scan_rework' (#1) from scan_rework into main

Reviewed-on: #1

constants/constants.go

@@ -87,6 +87,7 @@ const (
 	FILE_STATUS_MISSING
 	FILE_STATUS_EXISTS
 	FILE_STATUS_CHANGED
+	FILE_STATUS_NEW
 )
 
 func (s FileStatus) String() string {
@@ -99,6 +100,8 @@ func (s FileStatus) String() string {
 		return "Exists"
 	case FILE_STATUS_CHANGED:
 		return "Changed"
+	case FILE_STATUS_NEW:
+		return "New"
 	default:
 		return fmt.Sprintf("%d", int(s))
 	}

go.mod

@@ -26,4 +26,5 @@ require (
 	golang.org/x/sync v0.7.0 // indirect
 	golang.org/x/sys v0.20.0 // indirect
 	golang.org/x/text v0.15.0 // indirect
+	gopkg.in/vansante/go-ffprobe.v2 v2.2.0 // indirect
 )

go.sum

@@ -68,6 +68,8 @@ golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 golang.org/x/text v0.15.0 h1:h1V/4gjBv8v9cjcR6+AR5+/cIYK5N/WAgiv4xlsEtAk=
 golang.org/x/text v0.15.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/vansante/go-ffprobe.v2 v2.2.0 h1:iuOqTsbfYuqIz4tAU9NWh22CmBGxlGHdgj4iqP+NUmY=
+gopkg.in/vansante/go-ffprobe.v2 v2.2.0/go.mod h1:qF0AlAjk7Nqzqf3y333Ly+KxN3cKF2JqA3JT5ZheUGE=
 gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 nhooyr.io/websocket v1.8.11 h1:f/qXNc2/3DpoSZkHt1DQu6rj4zGC8JmkkLkWss0MgN0=
 nhooyr.io/websocket v1.8.11/go.mod h1:rN9OFWIUwuxg4fR5tELlYC04bXYowCP9GX47ivo2l+c=

migrations/000017_alter_files_table_hash.down.sql

@@ -0,0 +1,3 @@
+ALTER TABLE files
+ALTER COLUMN md5 SET NOT NULL,
+DROP IF EXISTS mod_time;

migrations/000017_alter_files_table_hash.up.sql

@@ -0,0 +1,3 @@
+ALTER TABLE files
+ALTER COLUMN md5 DROP NOT NULL,
+ADD mod_time TIMESTAMP NOT NULL DEFAULT current_timestamp;

migrations/000018_alter_files_table_ffprobe_data.down.sql

@@ -0,0 +1,2 @@
+ALTER TABLE files
+DROP IF EXISTS ffprobe;

migrations/000018_alter_files_table_ffprobe_data.up.sql

@@ -0,0 +1,2 @@
+ALTER TABLE files
+ADD ffprobe_data JSONB;

server/hash.go

@@ -0,0 +1,62 @@
+package server
+
+import (
+	"context"
+	"crypto/md5"
+	"fmt"
+	"io"
+	"log/slog"
+	"os"
+	"path/filepath"
+	"slices"
+
+	"git.lastassault.de/speatzle/morffix/constants"
+)
+
+func hashFile(ctx context.Context, id uint) error {
+	slog.InfoContext(ctx, "Hashing File", "id", id)
+
+	var filePath string
+	var libraryPath string
+	var oldHash []byte
+	err := db.QueryRow(ctx, "SELECT f.path, l.path, md5 FROM files f INNER JOIN libraries l ON l.id = f.library_id WHERE f.id = $1", id).Scan(&filePath, &libraryPath, &oldHash)
+	if err != nil {
+		return fmt.Errorf("Get File: %w", err)
+	}
+
+	fullPath := filepath.Join(libraryPath, filePath)
+
+	file, err := os.Open(fullPath)
+	if err != nil {
+		return fmt.Errorf("Opening File: %w", err)
+	}
+
+	hash := md5.New()
+	if _, err := io.Copy(hash, file); err != nil {
+		return fmt.Errorf("Reading File: %w", err)
+	}
+	newHash := hash.Sum(nil)
+
+	if slices.Compare[[]byte](newHash, oldHash) != 0 {
+		// File has changed
+		// TODO Queue healthcheck / transcode if enabled
+	}
+
+	tx, err := db.Begin(ctx)
+	if err != nil {
+		return fmt.Errorf("Begin Transaction: %w", err)
+	}
+	defer tx.Rollback(ctx)
+
+	_, err = tx.Exec(ctx, "UPDATE files SET status = $2, md5 = $3 WHERE id = $1", id, constants.FILE_STATUS_EXISTS, newHash)
+	if err != nil {
+		return fmt.Errorf("Updating File in DB: %w", err)
+	}
+
+	err = tx.Commit(ctx)
+	if err != nil {
+		return fmt.Errorf("Committing Changes: %w", err)
+	}
+	slog.InfoContext(ctx, "Hashing File Done", "id", id)
+	return nil
+}

server/library.go

@@ -63,10 +63,6 @@ func handleLibrary(w http.ResponseWriter, r *http.Request) {
 		Enable: enabled,
 	}
 
-	if r.Method == "PUT" {
-		// TODO
-	}
-
 	rows, err := db.Query(r.Context(), "SELECT id, path, size, status, health, transcode, md5, updated_at FROM files where library_id = $1", id)
 	if err != nil {
 		slog.ErrorContext(r.Context(), "Query Files", "err", err)

server/scan.go

@@ -3,18 +3,20 @@ package server
 import (
 	"bytes"
 	"context"
-	"crypto/md5"
 	"errors"
 	"fmt"
-	"io"
 	"log/slog"
 	"net/http"
 	"os"
 	"path/filepath"
 	"slices"
+	"strconv"
+	"sync"
+	"time"
 
 	"git.lastassault.de/speatzle/morffix/constants"
 	"github.com/jackc/pgx/v5"
+	"gopkg.in/vansante/go-ffprobe.v2"
 )
 
 var videoFileExtensions = []string{".mkv", ".mp4", ".webm", ".flv", ".avi"}
@@ -25,18 +27,16 @@ func handleScan(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	id := r.PathValue("id")
+	id, err := strconv.Atoi(r.FormValue("id"))
-	if id == "" {
+	if err != nil {
-		http.Error(w, "No ID Set", http.StatusBadRequest)
+		http.Error(w, "Parsing ID", http.StatusBadRequest)
 		return
 	}
 
-	full := r.FormValue("full") == "on"
-
 	var name string
 	var path string
 	var enabled bool
-	err := db.QueryRow(r.Context(), "SELECT name, path, enable FROM libraries WHERE id = $1", id).Scan(&name, &path, &enabled)
+	err = db.QueryRow(r.Context(), "SELECT name, path, enable FROM libraries WHERE id = $1", id).Scan(&name, &path, &enabled)
 	if err != nil {
 		slog.ErrorContext(r.Context(), "Get Library", "err", err)
 		http.Error(w, "Error Get Library: "+err.Error(), http.StatusInternalServerError)
@@ -44,7 +44,7 @@ func handleScan(w http.ResponseWriter, r *http.Request) {
 	}
 
 	scanCtx := context.Background()
-	go scan(scanCtx, id, full)
+	go scanLibrary(scanCtx, id)
 
 	message := "Scan Started"
 
@@ -79,17 +79,81 @@ func scanStatus(w http.ResponseWriter, r *http.Request) {
 	}
 }
 
-func scan(ctx context.Context, id string, full bool) {
+func manageScan(stop chan bool) {
-	slog.InfoContext(ctx, "Starting Scan", "id", id)
+	scanTicker := time.NewTicker(time.Minute)
+	hashTicker := time.NewTicker(time.Second)
+	scanRunning := false
+	hashRunning := false
+	ctx, cancel := context.WithCancel(context.Background())
 
-	// TODO Scan settings:
+	for {
-	// - Auto Queue Healthcheck for Changed Files
+		select {
-	// - Auto Queue Healthcheck for New Files
+		case <-scanTicker.C:
-	// - Auto Queue Transcode for New Files
+			if scanRunning {
-	// - Auto Queue Transcode for Changed Files (? Instead have library setting to queue transcode for changed files on healthcheck success)
+				continue
-	// - Auto Queue Health/Transcode for Unkown Status ? (might result in requeue loop)
+			}
-	// - Schedule Scans Periodically
+			scanRunning = true
-	// - Add File Monitoring for Setting Changed status and triggering tasks
+			go func() {
+				defer func() {
+					scanRunning = false
+				}()
+
+				rows, err := db.Query(ctx, "SELECT id FROM libraries WHERE enable = true")
+				if err != nil {
+					slog.ErrorContext(ctx, "Error getting Enabled Libraries for Scan", "err", err)
+					return
+				}
+
+				libraries, err := pgx.CollectRows[int](rows, pgx.RowTo[int])
+				if err != nil {
+					slog.ErrorContext(ctx, "Error Collecting Enabled Libraries for Scan", "err", err)
+					return
+				}
+
+				for _, l := range libraries {
+					scanLibrary(ctx, l)
+				}
+			}()
+
+		case <-hashTicker.C:
+			if hashRunning {
+				continue
+			}
+
+			hashRunning = true
+			go func() {
+				defer func() {
+					hashRunning = false
+				}()
+
+				var fileID uint
+				err := db.QueryRow(ctx, "SELECT id FROM files WHERE status = $1 OR status = $2 LIMIT 1", constants.FILE_STATUS_CHANGED, constants.FILE_STATUS_NEW).Scan(&fileID)
+				if err != nil {
+					if !errors.Is(err, pgx.ErrNoRows) {
+						slog.ErrorContext(ctx, "Error Getting Files for Hashing", "err", err)
+					}
+					return
+				}
+
+				err = hashFile(ctx, fileID)
+				if err != nil {
+					slog.ErrorContext(ctx, "Error Hashing File", "err", err)
+				}
+			}()
+		case <-stop:
+			cancel()
+			return
+		}
+	}
+}
+
+var scanLock sync.Mutex
+
+func scanLibrary(ctx context.Context, id int) {
+	slog.InfoContext(ctx, "Acquiring Scan Lock...")
+	scanLock.Lock()
+	defer scanLock.Unlock()
+	slog.InfoContext(ctx, "Starting Scan", "id", id)
 
 	var name string
 	var lpath string
@@ -150,20 +214,6 @@ func scan(ctx context.Context, id string, full bool) {
 				slog.InfoContext(ctx, "Skipping non video file", "path", fullPath)
 				return nil
 			}
-			slog.InfoContext(ctx, "Hashing File", "path", fullPath, "size", info.Size())
-
-			file, err := os.Open(fullPath)
-			if err != nil {
-				return fmt.Errorf("Opening File: %w", err)
-			}
-
-			hash := md5.New()
-			if _, err := io.Copy(hash, file); err != nil {
-				return fmt.Errorf("Reading File: %w", err)
-			}
-			newMD5 := hash.Sum(nil)
-
-			slog.InfoContext(ctx, "File MD5", "path", fullPath, "size", info.Size(), "md5", newMD5)
 
 			fPath, err := filepath.Rel(lpath, fullPath)
 			if err != nil {
@@ -171,14 +221,21 @@ func scan(ctx context.Context, id string, full bool) {
 			}
 
 			var fileID int
-			var oldMD5 []byte
+			var oldSize uint
-			var health constants.FileHealth
+			var oldModTime time.Time
-			err = tx.QueryRow(ctx, "SELECT id, md5, health FROM files WHERE library_id = $1 AND path = $2", id, fPath).Scan(&fileID, &oldMD5, &health)
+			err = tx.QueryRow(ctx, "SELECT id, size, mod_time FROM files WHERE library_id = $1 AND path = $2", id, fPath).Scan(&fileID, &oldSize, &oldModTime)
 			if errors.Is(err, pgx.ErrNoRows) {
 				// File Does not Exist Yet
 
-				slog.InfoContext(ctx, "File is New", "path", fullPath)
+				slog.InfoContext(ctx, "File is New, Running FFProbe...", "path", fullPath)
-				_, err = tx.Exec(ctx, "INSERT INTO files (library_id, path, size, status, health, md5) VALUES ($1, $2, $3, $4, $5, $6)", id, fPath, info.Size(), constants.FILE_STATUS_EXISTS, constants.FILE_HEALTH_UNKNOWN, newMD5)
+
+				ffprobeData, err := ffprobe.ProbeURL(ctx, fullPath)
+				if err != nil {
+					return fmt.Errorf("ffprobe New File: %w", err)
+				}
+				slog.InfoContext(ctx, "ffprobe Done", "path", fullPath)
+
+				_, err = tx.Exec(ctx, "INSERT INTO files (library_id, path, size, status, health, ffprobe_data) VALUES ($1, $2, $3, $4, $5, $6)", id, fPath, info.Size(), constants.FILE_STATUS_NEW, constants.FILE_HEALTH_UNKNOWN, ffprobeData)
 				if err != nil {
 					return fmt.Errorf("Add New File to DB: %w", err)
 				}
@@ -187,15 +244,31 @@ func scan(ctx context.Context, id string, full bool) {
 				return fmt.Errorf("Getting File: %w", err)
 			}
 
-			if slices.Compare[[]byte](newMD5, oldMD5) != 0 {
+			// Remove Timezone and Round to nearest Second
-				// File has changed on disk so reset health
+			newModTime := info.ModTime().UTC().Round(time.Second)
-				health = constants.FILE_HEALTH_UNKNOWN
+
-			}
+			// File Already Exists, Check if it has been changed
-			// File Exists so update Size, status and hash
+			if newModTime.Equal(oldModTime) && uint(info.Size()) == oldSize {
-			_, err = tx.Exec(ctx, "UPDATE files SET size = $2, status = $3, health = $4, md5 = $5 WHERE id = $1", fileID, info.Size(), constants.FILE_STATUS_EXISTS, health, newMD5)
+				slog.Debug("File stayed the same", "id", fileID)
-			if err != nil {
+				_, err = tx.Exec(ctx, "UPDATE files SET status = $2 WHERE id = $1", fileID, constants.FILE_STATUS_EXISTS)
-				return fmt.Errorf("Updating File in DB: %w", err)
+				if err != nil {
+					return fmt.Errorf("Updating Non Changed File in DB: %w", err)
+				}
+			} else {
+				slog.InfoContext(ctx, "File Has Changed", "path", fullPath, "old_mod_time", oldModTime, "new_mod_time", newModTime, "old_size", oldSize, "new_size", info.Size())
+
+				ffprobeData, err := ffprobe.ProbeURL(ctx, fullPath)
+				if err != nil {
+					return fmt.Errorf("ffprobe Changed File: %w", err)
+				}
+				slog.InfoContext(ctx, "ffprobe Done", "path", fullPath)
+
+				_, err = tx.Exec(ctx, "UPDATE files SET size = $2, status = $3, health = $4, mod_time = $5, ffprobe_data = $6 WHERE id = $1", fileID, info.Size(), constants.FILE_STATUS_CHANGED, constants.FILE_HEALTH_UNKNOWN, newModTime, ffprobeData)
+				if err != nil {
+					return fmt.Errorf("Updating Changed File in DB: %w", err)
+				}
 			}
+
 			return nil
 		})
 	if err != nil {
@@ -208,7 +281,5 @@ func scan(ctx context.Context, id string, full bool) {
 		slog.ErrorContext(ctx, "Error Committing Changes", "err", err)
 		return
 	}
-
-	// TODO, create health and transcode tasks if requested
 	slog.InfoContext(ctx, "Scan Done", "id", id)
 }

server/server.go

@@ -126,8 +126,11 @@ func Start(_conf config.Config, tmplFS embed.FS, staticFS embed.FS, migrationsFS
 		serverClose <- true
 	}()
 
-	stopCleanup := make(chan bool, 1)
+	stopWorkerManagement := make(chan bool, 1)
-	go manageWorkers(stopCleanup)
+	go manageWorkers(stopWorkerManagement)
+
+	stopScanning := make(chan bool, 1)
+	go manageScan(stopScanning)
 
 	sigs := make(chan os.Signal, 1)
 	signal.Notify(sigs, os.Interrupt)
@@ -139,7 +142,10 @@ func Start(_conf config.Config, tmplFS embed.FS, staticFS embed.FS, migrationsFS
 		stopCtx, cancel := context.WithTimeout(context.Background(), time.Second*10)
 		server.Shutdown(stopCtx)
 		cancel()
-		stopCleanup <- true
+		slog.Info("Stopping Worker Management...")
+		stopWorkerManagement <- true
+		slog.Info("Stopping Scanning...")
+		stopScanning <- true
 		slog.Info("Done")
 	}
 }