Indexer: Merge identical photos (stack files) #576

internal/api/index.go

@@ -48,7 +48,8 @@ func StartIndexing(router *gin.RouterGroup) {
 
 		indOpt := photoprism.IndexOptions{
 			Rescan:  f.Rescan,
-			Convert: f.Convert && conf.SidecarWritable(),
+			Convert: conf.Settings().Index.Convert && conf.SidecarWritable(),
+			Stack:   conf.Settings().Index.Stack,
 			Path:    filepath.Clean(f.Path),
 		}
 

internal/api/photo_unstack.go

@@ -131,7 +131,7 @@ func PhotoUnstack(router *gin.RouterGroup) {
 		ind := service.Index()
 
 		// Index new, unstacked file.
-		if res := ind.File(mediaFile.FileName()); res.Failed() {
+		if res := ind.SingleFile(mediaFile.FileName()); res.Failed() {
 			log.Errorf("photo: %s (unstack %s)", res.Err, txt.Quote(baseName))
 			AbortSaveFailed(c)
 			return
@@ -148,7 +148,7 @@ func PhotoUnstack(router *gin.RouterGroup) {
 		oldPrimaryName := photoprism.FileName(oldPrimary.FileRoot, oldPrimary.FileName)
 
 		// Re-index old, existing primary file.
-		if res := ind.File(oldPrimaryName); res.Failed() {
+		if res := ind.SingleFile(oldPrimaryName); res.Failed() {
 			log.Errorf("photo: %s (unstack %s)", res.Err, txt.Quote(baseName))
 			AbortSaveFailed(c)
 			return

internal/commands/index.go

@@ -63,6 +63,7 @@ func indexAction(ctx *cli.Context) error {
 		Path:    subPath,
 		Rescan:  ctx.Bool("all"),
 		Convert: conf.Settings().Index.Convert && conf.SidecarWritable(),
+		Stack:   conf.Settings().Index.Stack,
 	}
 
 	indexed := ind.Start(indOpt)

internal/config/settings.go

@@ -33,6 +33,7 @@ type IndexSettings struct {
 	Convert   bool   `json:"convert" yaml:"convert"`
 	Rescan    bool   `json:"rescan" yaml:"rescan"`
 	Sequences bool   `json:"sequences" yaml:"sequences"`
+	Stack     bool   `json:"stack" yaml:"stack"`
 }
 
 // ImportSettings represents import settings.
@@ -107,6 +108,7 @@ func NewSettings() *Settings {
 			Rescan:    false,
 			Convert:   true,
 			Sequences: true,
+			Stack:     true,
 		},
 	}
 }

internal/config/testdata/configEmpty.yml

@@ -28,3 +28,4 @@ index:
   convert: true
   rescan: false
   sequences: true
+  stack: true

internal/entity/photo.go

@@ -4,6 +4,7 @@ import (
 	"errors"
 	"fmt"
 	"path"
+	"strconv"
 	"strings"
 	"time"
 
@@ -30,6 +31,11 @@ func (m Photos) UIDs() []string {
 	return result
 }
 
+// MapKey returns a key referencing time and location for indexing.
+func MapKey(takenAt time.Time, cellId string) string {
+	return path.Join(strconv.FormatInt(takenAt.Unix(), 36), cellId)
+}
+
 // Photo represents a photo, all its properties, and link to all its images and sidecar files.
 type Photo struct {
 	ID               uint         `gorm:"primary_key" yaml:"-"`
@@ -48,6 +54,7 @@ type Photo struct {
 	PhotoName        string       `gorm:"type:VARBINARY(255);" json:"Name" yaml:"-"`
 	OriginalName     string       `gorm:"type:VARBINARY(768);" json:"OriginalName" yaml:"OriginalName,omitempty"`
 	PhotoFavorite    bool         `json:"Favorite" yaml:"Favorite,omitempty"`
+	PhotoSingle      bool         `json:"Single" yaml:"Single,omitempty"`
 	PhotoPrivate     bool         `json:"Private" yaml:"Private,omitempty"`
 	PhotoScan        bool         `json:"Scan" yaml:"Scan,omitempty"`
 	PhotoPanorama    bool         `json:"Panorama" yaml:"Panorama,omitempty"`
@@ -998,3 +1005,39 @@ func (m *Photo) Links() Links {
 func (m *Photo) PrimaryFile() (File, error) {
 	return PrimaryFile(m.PhotoUID)
 }
+
+// MapKey returns a key referencing time and location for indexing.
+func (m *Photo) MapKey() string {
+	return MapKey(m.TakenAt, m.CellID)
+}
+
+// Stack merges the photo with identical ones.
+func (m *Photo) Stack() (identical Photos, err error) {
+	if err := Db().
+		Where("id <> ?", m.ID).
+		Where("taken_at = ?", m.TakenAt).
+		Where("cell_id = ?", m.CellID).
+		Where("camera_serial = ?", m.CameraSerial).
+		Where("photo_single = 0").
+		Find(&identical).Error; err != nil {
+		return identical, err
+	}
+
+	for _, photo := range identical {
+		if err := UnscopedDb().Model(File{}).Where("photo_id = ?", photo.ID).Updates(File{PhotoID: m.ID, PhotoUID: m.PhotoUID}).Error; err != nil {
+			return identical, err
+		}
+
+		UnscopedDb().Model(PhotoKeyword{}).Where("photo_id = ?", photo.ID).Updates(PhotoKeyword{PhotoID: m.ID})
+		UnscopedDb().Model(PhotoLabel{}).Where("photo_id = ?", photo.ID).Updates(PhotoLabel{PhotoID: m.ID})
+		UnscopedDb().Model(PhotoAlbum{}).Where("photo_uid = ?", photo.PhotoUID).Updates(PhotoAlbum{PhotoUID: m.PhotoUID})
+
+		if err := photo.Updates(map[string]interface{}{"DeletedAt": Timestamp(), "PhotoQuality": -1}); err != nil {
+			return identical, err
+		}
+	}
+
+	_, err = m.Optimize()
+
+	return identical, err
+}

internal/form/index_options.go

@@ -1,7 +1,6 @@
 package form
 
 type IndexOptions struct {
-	Path    string `json:"path"`
-	Convert bool   `json:"convert"`
-	Rescan  bool   `json:"rescan"`
+	Path   string `json:"path"`
+	Rescan bool   `json:"rescan"`
 }

internal/meta/data.go

@@ -4,6 +4,8 @@ import (
 	"math"
 	"time"
 
+	"github.com/photoprism/photoprism/pkg/s2"
+
 	"github.com/photoprism/photoprism/pkg/rnd"
 )
 
@@ -115,3 +117,8 @@ func (data Data) ActualHeight() int {
 
 	return data.Height
 }
+
+// CellID returns the S2 cell ID.
+func (data Data) CellID() string {
+	return s2.PrefixedToken(float64(data.Lat), float64(data.Lng))
+}

internal/photoprism/files.go

@@ -16,7 +16,7 @@ type Files struct {
 	mutex sync.RWMutex
 }
 
-// NewFiles returns a new Files instance pointer.
+// NewFiles returns a new Files instance.
 func NewFiles() *Files {
 	m := &Files{
 		files: make(query.FileMap),
@@ -42,7 +42,7 @@ func (m *Files) Init() error {
 	files, err := query.IndexedFiles()
 
 	if err != nil {
-		return fmt.Errorf("%s (query indexed files)", err.Error())
+		return fmt.Errorf("%s (find indexed files)", err.Error())
 	} else {
 		m.files = files
 		m.count = len(files)

internal/photoprism/import_test.go

@@ -16,7 +16,7 @@ func TestNewImport(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 	imp := NewImport(conf, ind, convert)
 
 	assert.IsType(t, &Import{}, imp)
@@ -31,7 +31,7 @@ func TestImport_DestinationFilename(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 
 	imp := NewImport(conf, ind, convert)
 
@@ -63,7 +63,7 @@ func TestImport_Start(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 
 	imp := NewImport(conf, ind, convert)
 

internal/photoprism/index.go

@@ -9,6 +9,8 @@ import (
 	"strings"
 	"sync"
 
+	"github.com/photoprism/photoprism/internal/query"
+
 	"github.com/karrick/godirwalk"
 	"github.com/photoprism/photoprism/internal/classify"
 	"github.com/photoprism/photoprism/internal/config"
@@ -27,16 +29,18 @@ type Index struct {
 	nsfwDetector *nsfw.Detector
 	convert      *Convert
 	files        *Files
+	photos       *Photos
 }
 
 // NewIndex returns a new indexer and expects its dependencies as arguments.
-func NewIndex(conf *config.Config, tensorFlow *classify.TensorFlow, nsfwDetector *nsfw.Detector, convert *Convert, files *Files) *Index {
+func NewIndex(conf *config.Config, tensorFlow *classify.TensorFlow, nsfwDetector *nsfw.Detector, convert *Convert, files *Files, photos *Photos) *Index {
 	i := &Index{
 		conf:         conf,
 		tensorFlow:   tensorFlow,
 		nsfwDetector: nsfwDetector,
 		convert:      convert,
 		files:        files,
+		photos:       photos,
 	}
 
 	return i
@@ -217,6 +221,12 @@ func (ind *Index) Start(opt IndexOptions) fs.Done {
 		log.Error(err.Error())
 	}
 
+	if opt.Stack {
+		if err := ind.StackIdenticalPhotos(); err != nil {
+			log.Errorf("index: %s", err)
+		}
+	}
+
 	if filesIndexed > 0 {
 		if err := entity.UpdatePhotoCounts(); err != nil {
 			log.Errorf("index: %s", err)
@@ -230,8 +240,29 @@ func (ind *Index) Start(opt IndexOptions) fs.Done {
 	return done
 }
 
+// StackIdenticalPhotos stacks files that belong to the same photo.
+func (ind *Index) StackIdenticalPhotos() error {
+	photos, err := query.IdenticalPhotos()
+
+	if err != nil {
+		return err
+	}
+
+	for _, photo := range photos {
+		if merged, err := photo.Stack(); err != nil {
+			log.Errorf("index: %s", err)
+		} else {
+			log.Infof("index: merged photo uid %s with %s", photo.PhotoUID, merged.UIDs())
+			event.EntitiesUpdated("photos", []entity.Photo{photo})
+			event.EntitiesDeleted("photos", merged.UIDs())
+		}
+	}
+
+	return nil
+}
+
 // File indexes a single file and returns the result.
-func (ind *Index) File(name string) (result IndexResult) {
+func (ind *Index) SingleFile(name string) (result IndexResult) {
 	file, err := NewMediaFile(name)
 
 	if err != nil {
@@ -250,5 +281,5 @@ func (ind *Index) File(name string) (result IndexResult) {
 		return result
 	}
 
-	return IndexRelated(related, ind, IndexOptionsAll())
+	return IndexRelated(related, ind, IndexOptionsSingle())
 }

internal/photoprism/index_mediafile.go

@@ -160,22 +160,25 @@ func (ind *Index) MediaFile(m *MediaFile, o IndexOptions, originalName string) (
 	if !fileExists {
 		photoQuery = entity.UnscopedDb().First(&photo, "photo_path = ? AND photo_name = ?", filePath, fileBase)
 
-		// Try to find existing photo by exact time and location.
-		if photoQuery.Error != nil && m.MetaData().HasTimeAndPlace() {
-			metaData = m.MetaData()
-			photoQuery = entity.UnscopedDb().First(&photo, "photo_lat = ? AND photo_lng = ? AND taken_at = ? AND camera_serial = ?", metaData.Lat, metaData.Lng, metaData.TakenAt, metaData.CameraSerial)
+		// Add file to existing photo (file stack)?
+		if o.Stack {
+			// Try to find existing photo by exact time and location.
+			if photoQuery.Error != nil && m.MetaData().HasTimeAndPlace() {
+				metaData = m.MetaData()
+				photoQuery = entity.UnscopedDb().First(&photo, "photo_lat = ? AND photo_lng = ? AND taken_at = ? AND camera_serial = ?", metaData.Lat, metaData.Lng, metaData.TakenAt, metaData.CameraSerial)
 
-			if photoQuery.Error == nil {
-				fileStacked = true
+				if photoQuery.Error == nil {
+					fileStacked = true
+				}
 			}
-		}
 
-		// Try to find existing photo by unique image id.
-		if photoQuery.Error != nil && m.MetaData().HasDocumentID() {
-			photoQuery = entity.UnscopedDb().First(&photo, "uuid = ?", m.MetaData().DocumentID)
+			// Try to find existing photo by unique image id.
+			if photoQuery.Error != nil && m.MetaData().HasDocumentID() {
+				photoQuery = entity.UnscopedDb().First(&photo, "uuid = ?", m.MetaData().DocumentID)
 
-			if photoQuery.Error == nil {
-				fileStacked = true
+				if photoQuery.Error == nil {
+					fileStacked = true
+				}
 			}
 		}
 	} else {
@@ -210,6 +213,7 @@ func (ind *Index) MediaFile(m *MediaFile, o IndexOptions, originalName string) (
 	// Try to recover photo metadata from backup if not exists.
 	if !photoExists {
 		photo.PhotoQuality = -1
+		photo.PhotoSingle = !o.Stack
 
 		if yamlName := fs.TypeYaml.FindFirst(m.FileName(), []string{Config().SidecarPath(), fs.HiddenPath}, Config().OriginalsPath(), stripSequence); yamlName != "" {
 			if err := photo.LoadFromYaml(yamlName); err != nil {

internal/photoprism/index_mediafile_test.go

@@ -23,7 +23,7 @@ func TestIndex_MediaFile(t *testing.T) {
 		nd := nsfw.New(conf.NSFWModelPath())
 		convert := NewConvert(conf)
 
-		ind := NewIndex(conf, tf, nd, convert, NewFiles())
+		ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 		indexOpt := IndexOptionsAll()
 		mediaFile, err := NewMediaFile(conf.ExamplesPath() + "/blue-go-video.mp4")
 		if err != nil {
@@ -44,7 +44,7 @@ func TestIndex_MediaFile(t *testing.T) {
 		nd := nsfw.New(conf.NSFWModelPath())
 		convert := NewConvert(conf)
 
-		ind := NewIndex(conf, tf, nd, convert, NewFiles())
+		ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 		indexOpt := IndexOptionsAll()
 
 		result := ind.MediaFile(nil, indexOpt, "blue-go-video.mp4")

internal/photoprism/index_options.go

@@ -4,6 +4,7 @@ type IndexOptions struct {
 	Path    string
 	Rescan  bool
 	Convert bool
+	Stack   bool
 }
 
 func (o *IndexOptions) SkipUnchanged() bool {
@@ -16,6 +17,19 @@ func IndexOptionsAll() IndexOptions {
 		Path:    "/",
 		Rescan:  true,
 		Convert: true,
+		Stack:   true,
+	}
+
+	return result
+}
+
+// IndexOptionsSingle returns new index options for unstacked, single files.
+func IndexOptionsSingle() IndexOptions {
+	result := IndexOptions{
+		Path:    "/",
+		Rescan:  true,
+		Convert: true,
+		Stack:   false,
 	}
 
 	return result

internal/photoprism/index_related_test.go

@@ -54,7 +54,7 @@ func TestIndexRelated(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 	opt := IndexOptionsAll()
 
 	result := IndexRelated(related, ind, opt)

internal/photoprism/index_test.go

@@ -23,7 +23,7 @@ func TestIndex_Start(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 	imp := NewImport(conf, ind, convert)
 	opt := ImportOptionsMove(conf.ImportPath())
 
@@ -48,8 +48,8 @@ func TestIndex_File(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 
-	err := ind.File("xxx")
+	err := ind.SingleFile("xxx")
 	assert.Equal(t, IndexFailed, err.Status)
 }

internal/photoprism/photos.go

@@ -0,0 +1,68 @@
+package photoprism
+
+import (
+	"fmt"
+	"sync"
+	"time"
+
+	"github.com/photoprism/photoprism/internal/entity"
+
+	"github.com/photoprism/photoprism/internal/query"
+)
+
+// Photos represents photo id lookup table, sorted by date and S2 cell id.
+type Photos struct {
+	count  int
+	photos query.PhotoMap
+	mutex  sync.RWMutex
+}
+
+// NewPhotos returns a new Photos instance.
+func NewPhotos() *Photos {
+	m := &Photos{
+		photos: make(query.PhotoMap),
+	}
+
+	return m
+}
+
+// Init fetches the list from the database once.
+func (m *Photos) Init() error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	if len(m.photos) > 0 {
+		m.count = len(m.photos)
+		return nil
+	}
+
+	photos, err := query.IndexedPhotos()
+
+	if err != nil {
+		return fmt.Errorf("%s (find indexed photos)", err.Error())
+	} else {
+		m.photos = photos
+		m.count = len(photos)
+		return nil
+	}
+}
+
+// Remove a photo from the lookup table.
+func (m *Photos) Remove(takenAt time.Time, cellId string) {
+	key := entity.MapKey(takenAt, cellId)
+
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	delete(m.photos, key)
+}
+
+// Find returns the photo ID for a time and cell id.
+func (m *Photos) Find(takenAt time.Time, cellId string) uint {
+	key := entity.MapKey(takenAt, cellId)
+
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	return m.photos[key]
+}

internal/photoprism/photos_test.go

@@ -0,0 +1,13 @@
+package photoprism
+
+import (
+	"testing"
+)
+
+func TestPhotos_Init(t *testing.T) {
+	photos := NewPhotos()
+
+	if err := photos.Init(); err != nil {
+		t.Fatal(err)
+	}
+}

internal/photoprism/resample_test.go

@@ -32,7 +32,7 @@ func TestResample_Start(t *testing.T) {
 	nd := nsfw.New(conf.NSFWModelPath())
 	convert := NewConvert(conf)
 
-	ind := NewIndex(conf, tf, nd, convert, NewFiles())
+	ind := NewIndex(conf, tf, nd, convert, NewFiles(), NewPhotos())
 
 	imp := NewImport(conf, ind, convert)
 	opt := ImportOptionsMove(conf.ImportPath())

internal/query/photo.go

@@ -109,3 +109,22 @@ func PhotosCheck(limit int, offset int) (entities entity.Photos, err error) {
 
 	return entities, err
 }
+
+// IdenticalPhotos returns photos sharing the same exact time, location and camera serial.
+func IdenticalPhotos() (entities entity.Photos, err error) {
+	err = Db().Table("photos").
+		Select("photos.*").
+		Joins(`JOIN photos dup ON photos.id < dup.id 
+				AND photos.photo_lat = dup.photo_lat 
+				AND photos.photo_lng = dup.photo_lng 
+				AND photos.taken_at = dup.taken_at 
+				AND photos.camera_serial = dup.camera_serial`).
+		Where(`photos.photo_single = 0 
+				AND dup.photo_single = 0 
+				AND photos.deleted_at IS NULL 
+				AND dup.deleted_at IS NULL`).
+		Group("photos.id").
+		Find(&entities).Error
+
+	return entities, err
+}

internal/query/photo_map.go

@@ -0,0 +1,32 @@
+package query
+
+import (
+	"path"
+	"strconv"
+	"time"
+)
+
+type PhotoMap map[string]uint
+
+// IndexedPhotos returns a map of already indexed files with their mod time unix timestamp as value.
+func IndexedPhotos() (result PhotoMap, err error) {
+	result = make(PhotoMap)
+
+	type Photo struct {
+		ID      uint
+		TakenAt time.Time
+		CellID  string
+	}
+
+	var rows []Photo
+
+	if err := UnscopedDb().Raw("SELECT id, taken_at, cell_id FROM photos WHERE deleted_at IS NULL").Scan(&rows).Error; err != nil {
+		return result, err
+	}
+
+	for _, row := range rows {
+		result[path.Join(strconv.FormatInt(row.TakenAt.Unix(), 36), row.CellID)] = row.ID
+	}
+
+	return result, err
+}

internal/query/photo_map_test.go

@@ -0,0 +1,15 @@
+package query
+
+import (
+	"testing"
+)
+
+func TestIndexedPhotos(t *testing.T) {
+	result, err := IndexedPhotos()
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	t.Logf("INDEXED Photos: %#v", result)
+}

internal/query/photo_results.go

@@ -32,6 +32,7 @@ type PhotoResult struct {
 	PhotoDay         int           `json:"Day"`
 	PhotoCountry     string        `json:"Country"`
 	PhotoFavorite    bool          `json:"Favorite"`
+	PhotoSingle      bool          `json:"Single"`
 	PhotoPrivate     bool          `json:"Private"`
 	PhotoIso         int           `json:"Iso"`
 	PhotoFocalLength int           `json:"FocalLength"`

internal/query/photo_test.go

@@ -78,3 +78,14 @@ func TestPhotosCheck(t *testing.T) {
 	}
 	assert.IsType(t, entity.Photos{}, result)
 }
+
+func TestIdenticalPhotos(t *testing.T) {
+	result, err := IdenticalPhotos()
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	assert.IsType(t, entity.Photos{}, result)
+	// t.Logf("%+v", result)
+}

internal/service/index.go

@@ -9,7 +9,7 @@ import (
 var onceIndex sync.Once
 
 func initIndex() {
-	services.Index = photoprism.NewIndex(Config(), Classify(), NsfwDetector(), Convert(), Files())
+	services.Index = photoprism.NewIndex(Config(), Classify(), NsfwDetector(), Convert(), Files(), Photos())
 }
 
 func Index() *photoprism.Index {

internal/service/photos.go

@@ -0,0 +1,19 @@
+package service
+
+import (
+	"sync"
+
+	"github.com/photoprism/photoprism/internal/photoprism"
+)
+
+var oncePhotos sync.Once
+
+func initPhotos() {
+	services.Photos = photoprism.NewPhotos()
+}
+
+func Photos() *photoprism.Photos {
+	oncePhotos.Do(initPhotos)
+
+	return services.Photos
+}

internal/service/service.go

@@ -19,6 +19,7 @@ var services struct {
 	Classify *classify.TensorFlow
 	Convert  *photoprism.Convert
 	Files    *photoprism.Files
+	Photos   *photoprism.Photos
 	Import   *photoprism.Import
 	Index    *photoprism.Index
 	Moments  *photoprism.Moments