Identify: Options to skip multiple results and single name performers (#3707)

Co-authored-by: WithoutPants <53250216+WithoutPants@users.noreply.github.com>

graphql/documents/data/config.graphql

@@ -124,6 +124,10 @@ fragment IdentifyMetadataOptionsData on IdentifyMetadataOptions {
   setCoverImage
   setOrganized
   includeMalePerformers
+  skipMultipleMatches
+  skipMultipleMatchTag
+  skipSingleNamePerformers
+  skipSingleNamePerformerTag
 }
 
 fragment ScraperSourceData on ScraperSource {

graphql/schema/types/metadata.graphql

@@ -185,6 +185,14 @@ input IdentifyMetadataOptionsInput {
   setOrganized: Boolean
   """defaults to true if not provided"""
   includeMalePerformers: Boolean
+  """defaults to true if not provided"""
+  skipMultipleMatches: Boolean
+  """tag to tag skipped multiple matches with"""
+  skipMultipleMatchTag: String
+  """defaults to true if not provided"""
+  skipSingleNamePerformers: Boolean
+  """tag to tag skipped single name performers with"""
+  skipSingleNamePerformerTag: String
 }
 
 input IdentifySourceInput {
@@ -222,6 +230,14 @@ type IdentifyMetadataOptions {
   setOrganized: Boolean
   """defaults to true if not provided"""
   includeMalePerformers: Boolean
+  """defaults to true if not provided"""
+  skipMultipleMatches: Boolean
+  """tag to tag skipped multiple matches with"""
+  skipMultipleMatchTag: String
+  """defaults to true if not provided"""
+  skipSingleNamePerformers: Boolean
+  """tag to tag skipped single name performers with"""
+  skipSingleNamePerformerTag: String
 }
 
 type IdentifySource {

internal/identify/identify.go

@@ -2,18 +2,33 @@ package identify
 
 import (
 	"context"
+	"errors"
 	"fmt"
+	"strconv"
 
 	"github.com/stashapp/stash/pkg/logger"
 	"github.com/stashapp/stash/pkg/models"
 	"github.com/stashapp/stash/pkg/scene"
 	"github.com/stashapp/stash/pkg/scraper"
+	"github.com/stashapp/stash/pkg/sliceutil/intslice"
 	"github.com/stashapp/stash/pkg/txn"
 	"github.com/stashapp/stash/pkg/utils"
 )
 
+var (
+	ErrSkipSingleNamePerformer = errors.New("a performer was skipped because they only had a single name and no disambiguation")
+)
+
+type MultipleMatchesFoundError struct {
+	Source ScraperSource
+}
+
+func (e *MultipleMatchesFoundError) Error() string {
+	return fmt.Sprintf("multiple matches found for %s", e.Source.Name)
+}
+
 type SceneScraper interface {
-	ScrapeScene(ctx context.Context, sceneID int) (*scraper.ScrapedScene, error)
+	ScrapeScenes(ctx context.Context, sceneID int) ([]*scraper.ScrapedScene, error)
 }
 
 type SceneUpdatePostHookExecutor interface {
@@ -31,7 +46,7 @@ type SceneIdentifier struct {
 	SceneReaderUpdater SceneReaderUpdater
 	StudioCreator      StudioCreator
 	PerformerCreator   PerformerCreator
-	TagCreator         TagCreator
+	TagCreatorFinder   TagCreatorFinder
 
 	DefaultOptions              *MetadataOptions
 	Sources                     []ScraperSource
@@ -39,13 +54,31 @@ type SceneIdentifier struct {
 }
 
 func (t *SceneIdentifier) Identify(ctx context.Context, txnManager txn.Manager, scene *models.Scene) error {
-	result, err := t.scrapeScene(ctx, scene)
+	result, err := t.scrapeScene(ctx, txnManager, scene)
+	var multipleMatchErr *MultipleMatchesFoundError
 	if err != nil {
+		if !errors.As(err, &multipleMatchErr) {
 			return err
 		}
+	}
 
 	if result == nil {
+		if multipleMatchErr != nil {
+			logger.Debugf("Identify skipped because multiple results returned for %s", scene.Path)
+
+			// find if the scene should be tagged for multiple results
+			options := t.getOptions(multipleMatchErr.Source)
+			if options.SkipMultipleMatchTag != nil && len(*options.SkipMultipleMatchTag) > 0 {
+				// Tag it with the multiple results tag
+				err := t.addTagToScene(ctx, txnManager, scene, *options.SkipMultipleMatchTag)
+				if err != nil {
+					return err
+				}
+				return nil
+			}
+		} else {
 			logger.Debugf("Unable to identify %s", scene.Path)
+		}
 		return nil
 	}
 
@@ -62,50 +95,80 @@ type scrapeResult struct {
 	source ScraperSource
 }
 
-func (t *SceneIdentifier) scrapeScene(ctx context.Context, scene *models.Scene) (*scrapeResult, error) {
+func (t *SceneIdentifier) scrapeScene(ctx context.Context, txnManager txn.Manager, scene *models.Scene) (*scrapeResult, error) {
 	// iterate through the input sources
 	for _, source := range t.Sources {
 		// scrape using the source
-		scraped, err := source.Scraper.ScrapeScene(ctx, scene.ID)
+		results, err := source.Scraper.ScrapeScenes(ctx, scene.ID)
 		if err != nil {
 			logger.Errorf("error scraping from %v: %v", source.Scraper, err)
 			continue
 		}
 
+		if len(results) > 0 {
+			options := t.getOptions(source)
+			if len(results) > 1 && utils.IsTrue(options.SkipMultipleMatches) {
+				return nil, &MultipleMatchesFoundError{
+					Source: source,
+				}
+			} else {
 				// if results were found then return
-		if scraped != nil {
 				return &scrapeResult{
-				result: scraped,
+					result: results[0],
 					source: source,
 				}, nil
 			}
 		}
+	}
 
 	return nil, nil
 }
 
+// Returns a MetadataOptions object with any default options overwritten by source specific options
+func (t *SceneIdentifier) getOptions(source ScraperSource) MetadataOptions {
+	options := *t.DefaultOptions
+	if source.Options == nil {
+		return options
+	}
+	if source.Options.SetCoverImage != nil {
+		options.SetCoverImage = source.Options.SetCoverImage
+	}
+	if source.Options.SetOrganized != nil {
+		options.SetOrganized = source.Options.SetOrganized
+	}
+	if source.Options.IncludeMalePerformers != nil {
+		options.IncludeMalePerformers = source.Options.IncludeMalePerformers
+	}
+	if source.Options.SkipMultipleMatches != nil {
+		options.SkipMultipleMatches = source.Options.SkipMultipleMatches
+	}
+	if source.Options.SkipMultipleMatchTag != nil && len(*source.Options.SkipMultipleMatchTag) > 0 {
+		options.SkipMultipleMatchTag = source.Options.SkipMultipleMatchTag
+	}
+	if source.Options.SkipSingleNamePerformers != nil {
+		options.SkipSingleNamePerformers = source.Options.SkipSingleNamePerformers
+	}
+	if source.Options.SkipSingleNamePerformerTag != nil && len(*source.Options.SkipSingleNamePerformerTag) > 0 {
+		options.SkipSingleNamePerformerTag = source.Options.SkipSingleNamePerformerTag
+	}
+	return options
+}
+
 func (t *SceneIdentifier) getSceneUpdater(ctx context.Context, s *models.Scene, result *scrapeResult) (*scene.UpdateSet, error) {
 	ret := &scene.UpdateSet{
 		ID: s.ID,
 	}
 
-	options := []MetadataOptions{}
+	allOptions := []MetadataOptions{}
 	if result.source.Options != nil {
-		options = append(options, *result.source.Options)
+		allOptions = append(allOptions, *result.source.Options)
 	}
 	if t.DefaultOptions != nil {
-		options = append(options, *t.DefaultOptions)
+		allOptions = append(allOptions, *t.DefaultOptions)
 	}
 
-	fieldOptions := getFieldOptions(options)
+	fieldOptions := getFieldOptions(allOptions)
-
+	options := t.getOptions(result.source)
-	setOrganized := false
-	for _, o := range options {
-		if o.SetOrganized != nil {
-			setOrganized = *o.SetOrganized
-			break
-		}
-	}
 
 	scraped := result.result
 
@@ -113,12 +176,17 @@ func (t *SceneIdentifier) getSceneUpdater(ctx context.Context, s *models.Scene,
 		sceneReader:              t.SceneReaderUpdater,
 		studioCreator:            t.StudioCreator,
 		performerCreator:         t.PerformerCreator,
-		tagCreator:       t.TagCreator,
+		tagCreatorFinder:         t.TagCreatorFinder,
 		scene:                    s,
 		result:                   result,
 		fieldOptions:             fieldOptions,
+		skipSingleNamePerformers: *options.SkipSingleNamePerformers,
 	}
 
+	setOrganized := false
+	if options.SetOrganized != nil {
+		setOrganized = *options.SetOrganized
+	}
 	ret.Partial = getScenePartial(s, scraped, fieldOptions, setOrganized)
 
 	studioID, err := rel.studio(ctx)
@@ -130,18 +198,20 @@ func (t *SceneIdentifier) getSceneUpdater(ctx context.Context, s *models.Scene,
 		ret.Partial.StudioID = models.NewOptionalInt(*studioID)
 	}
 
-	ignoreMale := false
+	includeMalePerformers := true
-	for _, o := range options {
+	if options.IncludeMalePerformers != nil {
-		if o.IncludeMalePerformers != nil {
+		includeMalePerformers = *options.IncludeMalePerformers
-			ignoreMale = !*o.IncludeMalePerformers
-			break
-		}
 	}
 
-	performerIDs, err := rel.performers(ctx, ignoreMale)
+	addSkipSingleNamePerformerTag := false
+	performerIDs, err := rel.performers(ctx, !includeMalePerformers)
 	if err != nil {
+		if errors.Is(err, ErrSkipSingleNamePerformer) {
+			addSkipSingleNamePerformerTag = true
+		} else {
 			return nil, err
 		}
+	}
 	if performerIDs != nil {
 		ret.Partial.PerformerIDs = &models.UpdateIDs{
 			IDs:  performerIDs,
@@ -153,6 +223,14 @@ func (t *SceneIdentifier) getSceneUpdater(ctx context.Context, s *models.Scene,
 	if err != nil {
 		return nil, err
 	}
+	if addSkipSingleNamePerformerTag {
+		tagID, err := strconv.ParseInt(*options.SkipSingleNamePerformerTag, 10, 64)
+		if err != nil {
+			return nil, fmt.Errorf("error converting tag ID %s: %w", *options.SkipSingleNamePerformerTag, err)
+		}
+
+		tagIDs = intslice.IntAppendUnique(tagIDs, int(tagID))
+	}
 	if tagIDs != nil {
 		ret.Partial.TagIDs = &models.UpdateIDs{
 			IDs:  tagIDs,
@@ -171,15 +249,7 @@ func (t *SceneIdentifier) getSceneUpdater(ctx context.Context, s *models.Scene,
 		}
 	}
 
-	setCoverImage := false
+	if options.SetCoverImage != nil && *options.SetCoverImage {
-	for _, o := range options {
-		if o.SetCoverImage != nil {
-			setCoverImage = *o.SetCoverImage
-			break
-		}
-	}
-
-	if setCoverImage {
 		ret.CoverImage, err = rel.cover(ctx)
 		if err != nil {
 			return nil, err
@@ -241,6 +311,41 @@ func (t *SceneIdentifier) modifyScene(ctx context.Context, txnManager txn.Manage
 	return nil
 }
 
+func (t *SceneIdentifier) addTagToScene(ctx context.Context, txnManager txn.Manager, s *models.Scene, tagToAdd string) error {
+	if err := txn.WithTxn(ctx, txnManager, func(ctx context.Context) error {
+		tagID, err := strconv.Atoi(tagToAdd)
+		if err != nil {
+			return fmt.Errorf("error converting tag ID %s: %w", tagToAdd, err)
+		}
+
+		if err := s.LoadTagIDs(ctx, t.SceneReaderUpdater); err != nil {
+			return err
+		}
+		existing := s.TagIDs.List()
+
+		if intslice.IntInclude(existing, tagID) {
+			// skip if the scene was already tagged
+			return nil
+		}
+
+		if err := scene.AddTag(ctx, t.SceneReaderUpdater, s, tagID); err != nil {
+			return err
+		}
+
+		ret, err := t.TagCreatorFinder.Find(ctx, tagID)
+		if err != nil {
+			logger.Infof("Added tag id %s to skipped scene %s", tagToAdd, s.Path)
+		} else {
+			logger.Infof("Added tag %s to skipped scene %s", ret.Name, s.Path)
+		}
+
+		return nil
+	}); err != nil {
+		return err
+	}
+	return nil
+}
+
 func getFieldOptions(options []MetadataOptions) map[string]*FieldOptions {
 	// prefer source-specific field strategies, then the defaults
 	ret := make(map[string]*FieldOptions)
@@ -291,8 +396,7 @@ func getScenePartial(scene *models.Scene, scraped *scraper.ScrapedScene, fieldOp
 	}
 
 	if setOrganized && !scene.Organized {
-		// just reuse the boolean since we know it's true
+		partial.Organized = models.NewOptionalBool(true)
-		partial.Organized = models.NewOptionalBool(setOrganized)
 	}
 
 	return partial

internal/identify/identify_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"reflect"
+	"strconv"
 	"testing"
 
 	"github.com/stashapp/stash/pkg/models"
@@ -17,10 +18,10 @@ var testCtx = context.Background()
 
 type mockSceneScraper struct {
 	errIDs  []int
-	results map[int]*scraper.ScrapedScene
+	results map[int][]*scraper.ScrapedScene
 }
 
-func (s mockSceneScraper) ScrapeScene(ctx context.Context, sceneID int) (*scraper.ScrapedScene, error) {
+func (s mockSceneScraper) ScrapeScenes(ctx context.Context, sceneID int) ([]*scraper.ScrapedScene, error) {
 	if intslice.IntInclude(s.errIDs, sceneID) {
 		return nil, errors.New("scrape scene error")
 	}
@@ -40,39 +41,74 @@ func TestSceneIdentifier_Identify(t *testing.T) {
 		missingID
 		found1ID
 		found2ID
+		multiFoundID
+		multiFound2ID
 		errUpdateID
 	)
 
-	var scrapedTitle = "scrapedTitle"
+	var (
+		skipMultipleTagID    = 1
+		skipMultipleTagIDStr = strconv.Itoa(skipMultipleTagID)
+	)
 
-	defaultOptions := &MetadataOptions{}
+	var (
+		scrapedTitle  = "scrapedTitle"
+		scrapedTitle2 = "scrapedTitle2"
+
+		boolFalse = false
+		boolTrue  = true
+	)
+
+	defaultOptions := &MetadataOptions{
+		SetOrganized:             &boolFalse,
+		SetCoverImage:            &boolFalse,
+		IncludeMalePerformers:    &boolFalse,
+		SkipSingleNamePerformers: &boolFalse,
+	}
 	sources := []ScraperSource{
 		{
 			Scraper: mockSceneScraper{
 				errIDs: []int{errID1},
-				results: map[int]*scraper.ScrapedScene{
+				results: map[int][]*scraper.ScrapedScene{
-					found1ID: {
+					found1ID: {{
 						Title: &scrapedTitle,
-					},
+					}},
 				},
 			},
 		},
 		{
 			Scraper: mockSceneScraper{
 				errIDs: []int{errID2},
-				results: map[int]*scraper.ScrapedScene{
+				results: map[int][]*scraper.ScrapedScene{
-					found2ID: {
+					found2ID: {{
+						Title: &scrapedTitle,
+					}},
+					errUpdateID: {{
+						Title: &scrapedTitle,
+					}},
+					multiFoundID: {
+						{
 							Title: &scrapedTitle,
 						},
-					errUpdateID: {
+						{
+							Title: &scrapedTitle2,
+						},
+					},
+					multiFound2ID: {
+						{
 							Title: &scrapedTitle,
 						},
+						{
+							Title: &scrapedTitle2,
+						},
+					},
 				},
 			},
 		},
 	}
 
 	mockSceneReaderWriter := &mocks.SceneReaderWriter{}
+	mockTagFinderCreator := &mocks.TagReaderWriter{}
 
 	mockSceneReaderWriter.On("UpdatePartial", mock.Anything, mock.MatchedBy(func(id int) bool {
 		return id == errUpdateID
@@ -81,52 +117,84 @@ func TestSceneIdentifier_Identify(t *testing.T) {
 		return id != errUpdateID
 	}), mock.Anything).Return(nil, nil)
 
+	mockTagFinderCreator.On("Find", mock.Anything, skipMultipleTagID).Return(&models.Tag{
+		ID:   skipMultipleTagID,
+		Name: skipMultipleTagIDStr,
+	}, nil)
+
 	tests := []struct {
 		name    string
 		sceneID int
+		options *MetadataOptions
 		wantErr bool
 	}{
 		{
 			"error scraping",
 			errID1,
+			nil,
 			false,
 		},
 		{
 			"error scraping from second",
 			errID2,
+			nil,
 			false,
 		},
 		{
 			"found in first scraper",
 			found1ID,
+			nil,
 			false,
 		},
 		{
 			"found in second scraper",
 			found2ID,
+			nil,
 			false,
 		},
 		{
 			"not found",
 			missingID,
+			nil,
 			false,
 		},
 		{
 			"error modifying",
 			errUpdateID,
+			nil,
 			true,
 		},
+		{
+			"multiple found",
+			multiFoundID,
+			nil,
+			false,
+		},
+		{
+			"multiple found - set tag",
+			multiFound2ID,
+			&MetadataOptions{
+				SkipMultipleMatches:  &boolTrue,
+				SkipMultipleMatchTag: &skipMultipleTagIDStr,
+			},
+			false,
+		},
 	}
 
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
 			identifier := SceneIdentifier{
 				SceneReaderUpdater:          mockSceneReaderWriter,
+				TagCreatorFinder:            mockTagFinderCreator,
 				DefaultOptions:              defaultOptions,
 				Sources:                     sources,
 				SceneUpdatePostHookExecutor: mockHookExecutor{},
 			}
 
-	for _, tt := range tests {
+			if tt.options != nil {
-		t.Run(tt.name, func(t *testing.T) {
+				identifier.DefaultOptions = tt.options
+			}
+
 			scene := &models.Scene{
 				ID:           tt.sceneID,
 				PerformerIDs: models.NewRelatedIDs([]int{}),
@@ -144,7 +212,16 @@ func TestSceneIdentifier_modifyScene(t *testing.T) {
 	repo := models.Repository{
 		TxnManager: &mocks.TxnManager{},
 	}
-	tr := &SceneIdentifier{}
+	boolFalse := false
+	defaultOptions := &MetadataOptions{
+		SetOrganized:             &boolFalse,
+		SetCoverImage:            &boolFalse,
+		IncludeMalePerformers:    &boolFalse,
+		SkipSingleNamePerformers: &boolFalse,
+	}
+	tr := &SceneIdentifier{
+		DefaultOptions: defaultOptions,
+	}
 
 	type args struct {
 		scene  *models.Scene
@@ -165,6 +242,9 @@ func TestSceneIdentifier_modifyScene(t *testing.T) {
 				},
 				&scrapeResult{
 					result: &scraper.ScrapedScene{},
+					source: ScraperSource{
+						Options: defaultOptions,
+					},
 				},
 			},
 			false,

internal/identify/options.go

@@ -33,6 +33,14 @@ type MetadataOptions struct {
 	SetOrganized  *bool `json:"setOrganized"`
 	// defaults to true if not provided
 	IncludeMalePerformers *bool `json:"includeMalePerformers"`
+	// defaults to true if not provided
+	SkipMultipleMatches *bool `json:"skipMultipleMatches"`
+	// ID of tag to tag skipped multiple matches with
+	SkipMultipleMatchTag *string `json:"skipMultipleMatchTag"`
+	// defaults to true if not provided
+	SkipSingleNamePerformers *bool `json:"skipSingleNamePerformers"`
+	// ID of tag to tag skipped single name performers with
+	SkipSingleNamePerformerTag *string `json:"skipSingleNamePerformerTag"`
 }
 
 type FieldOptions struct {

internal/identify/performer.go

@@ -4,17 +4,20 @@ import (
 	"context"
 	"fmt"
 	"strconv"
+	"strings"
 	"time"
 
 	"github.com/stashapp/stash/pkg/models"
 	"github.com/stashapp/stash/pkg/sliceutil/stringslice"
+	"github.com/stashapp/stash/pkg/utils"
 )
 
 type PerformerCreator interface {
 	Create(ctx context.Context, newPerformer *models.Performer) error
+	UpdateImage(ctx context.Context, performerID int, image []byte) error
 }
 
-func getPerformerID(ctx context.Context, endpoint string, w PerformerCreator, p *models.ScrapedPerformer, createMissing bool) (*int, error) {
+func getPerformerID(ctx context.Context, endpoint string, w PerformerCreator, p *models.ScrapedPerformer, createMissing bool, skipSingleNamePerformers bool) (*int, error) {
 	if p.StoredID != nil {
 		// existing performer, just add it
 		performerID, err := strconv.Atoi(*p.StoredID)
@@ -24,6 +27,10 @@ func getPerformerID(ctx context.Context, endpoint string, w PerformerCreator, p
 
 		return &performerID, nil
 	} else if createMissing && p.Name != nil { // name is mandatory
+		// skip single name performers with no disambiguation
+		if skipSingleNamePerformers && !strings.Contains(*p.Name, " ") && (p.Disambiguation == nil || len(*p.Disambiguation) == 0) {
+			return nil, ErrSkipSingleNamePerformer
+		}
 		return createMissingPerformer(ctx, endpoint, w, p)
 	}
 
@@ -46,6 +53,19 @@ func createMissingPerformer(ctx context.Context, endpoint string, w PerformerCre
 		return nil, fmt.Errorf("error creating performer: %w", err)
 	}
 
+	// update image table
+	if p.Image != nil && len(*p.Image) > 0 {
+		imageData, err := utils.ReadImageFromURL(ctx, *p.Image)
+		if err != nil {
+			return nil, err
+		}
+
+		err = w.UpdateImage(ctx, performerInput.ID, imageData)
+		if err != nil {
+			return nil, err
+		}
+	}
+
 	return &performerInput.ID, nil
 }
 
@@ -56,6 +76,9 @@ func scrapedToPerformerInput(performer *models.ScrapedPerformer) models.Performe
 		CreatedAt: currentTime,
 		UpdatedAt: currentTime,
 	}
+	if performer.Disambiguation != nil {
+		ret.Disambiguation = *performer.Disambiguation
+	}
 	if performer.Birthdate != nil {
 		d := models.NewDate(*performer.Birthdate)
 		ret.Birthdate = &d
@@ -126,6 +149,12 @@ func scrapedToPerformerInput(performer *models.ScrapedPerformer) models.Performe
 	if performer.Instagram != nil {
 		ret.Instagram = *performer.Instagram
 	}
+	if performer.URL != nil {
+		ret.URL = *performer.URL
+	}
+	if performer.Details != nil {
+		ret.Details = *performer.Details
+	}
 
 	return ret
 }

internal/identify/performer_test.go

@@ -34,6 +34,7 @@ func Test_getPerformerID(t *testing.T) {
 		endpoint       string
 		p              *models.ScrapedPerformer
 		createMissing  bool
+		skipSingleName bool
 	}
 	tests := []struct {
 		name    string
@@ -47,6 +48,7 @@ func Test_getPerformerID(t *testing.T) {
 				emptyEndpoint,
 				&models.ScrapedPerformer{},
 				false,
+				false,
 			},
 			nil,
 			false,
@@ -59,6 +61,7 @@ func Test_getPerformerID(t *testing.T) {
 					StoredID: &invalidStoredID,
 				},
 				false,
+				false,
 			},
 			nil,
 			true,
@@ -71,6 +74,7 @@ func Test_getPerformerID(t *testing.T) {
 					StoredID: &validStoredIDStr,
 				},
 				false,
+				false,
 			},
 			&validStoredID,
 			false,
@@ -83,6 +87,7 @@ func Test_getPerformerID(t *testing.T) {
 					Name: &name,
 				},
 				false,
+				false,
 			},
 			nil,
 			false,
@@ -93,10 +98,24 @@ func Test_getPerformerID(t *testing.T) {
 				emptyEndpoint,
 				&models.ScrapedPerformer{},
 				true,
+				false,
 			},
 			nil,
 			false,
 		},
+		{
+			"single name no disambig creating",
+			args{
+				emptyEndpoint,
+				&models.ScrapedPerformer{
+					Name: &name,
+				},
+				true,
+				true,
+			},
+			nil,
+			true,
+		},
 		{
 			"valid name creating",
 			args{
@@ -105,6 +124,7 @@ func Test_getPerformerID(t *testing.T) {
 					Name: &name,
 				},
 				true,
+				false,
 			},
 			&validStoredID,
 			false,
@@ -112,7 +132,7 @@ func Test_getPerformerID(t *testing.T) {
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			got, err := getPerformerID(testCtx, tt.args.endpoint, &mockPerformerReaderWriter, tt.args.p, tt.args.createMissing)
+			got, err := getPerformerID(testCtx, tt.args.endpoint, &mockPerformerReaderWriter, tt.args.p, tt.args.createMissing, tt.args.skipSingleName)
 			if (err != nil) != tt.wantErr {
 				t.Errorf("getPerformerID() error = %v, wantErr %v", err, tt.wantErr)
 				return
@@ -207,7 +227,7 @@ func Test_scrapedToPerformerInput(t *testing.T) {
 	name := "name"
 
 	var stringValues []string
-	for i := 0; i < 17; i++ {
+	for i := 0; i < 20; i++ {
 		stringValues = append(stringValues, strconv.Itoa(i))
 	}
 
@@ -241,6 +261,7 @@ func Test_scrapedToPerformerInput(t *testing.T) {
 			"set all",
 			&models.ScrapedPerformer{
 				Name:           &name,
+				Disambiguation: nextVal(),
 				Birthdate:      nextVal(),
 				DeathDate:      nextVal(),
 				Gender:         nextVal(),
@@ -258,9 +279,12 @@ func Test_scrapedToPerformerInput(t *testing.T) {
 				Aliases:        nextVal(),
 				Twitter:        nextVal(),
 				Instagram:      nextVal(),
+				URL:            nextVal(),
+				Details:        nextVal(),
 			},
 			models.Performer{
 				Name:           name,
+				Disambiguation: *nextVal(),
 				Birthdate:      dateToDatePtr(models.NewDate(*nextVal())),
 				DeathDate:      dateToDatePtr(models.NewDate(*nextVal())),
 				Gender:         genderPtr(models.GenderEnum(*nextVal())),
@@ -278,6 +302,8 @@ func Test_scrapedToPerformerInput(t *testing.T) {
 				Aliases:        models.NewRelatedStrings([]string{*nextVal()}),
 				Twitter:        *nextVal(),
 				Instagram:      *nextVal(),
+				URL:            *nextVal(),
+				Details:        *nextVal(),
 			},
 		},
 		{

internal/identify/scene.go

@@ -3,6 +3,7 @@ package identify
 import (
 	"bytes"
 	"context"
+	"errors"
 	"fmt"
 	"strconv"
 	"strings"
@@ -13,6 +14,7 @@ import (
 	"github.com/stashapp/stash/pkg/scene"
 	"github.com/stashapp/stash/pkg/sliceutil"
 	"github.com/stashapp/stash/pkg/sliceutil/intslice"
+	"github.com/stashapp/stash/pkg/tag"
 	"github.com/stashapp/stash/pkg/utils"
 )
 
@@ -24,18 +26,20 @@ type SceneReaderUpdater interface {
 	models.StashIDLoader
 }
 
-type TagCreator interface {
+type TagCreatorFinder interface {
 	Create(ctx context.Context, newTag *models.Tag) error
+	tag.Finder
 }
 
 type sceneRelationships struct {
 	sceneReader              SceneReaderUpdater
 	studioCreator            StudioCreator
 	performerCreator         PerformerCreator
-	tagCreator       TagCreator
+	tagCreatorFinder         TagCreatorFinder
 	scene                    *models.Scene
 	result                   *scrapeResult
 	fieldOptions             map[string]*FieldOptions
+	skipSingleNamePerformers bool
 }
 
 func (g sceneRelationships) studio(ctx context.Context) (*int, error) {
@@ -93,13 +97,19 @@ func (g sceneRelationships) performers(ctx context.Context, ignoreMale bool) ([]
 		performerIDs = originalPerformerIDs
 	}
 
+	singleNamePerformerSkipped := false
+
 	for _, p := range scraped {
 		if ignoreMale && p.Gender != nil && strings.EqualFold(*p.Gender, models.GenderEnumMale.String()) {
 			continue
 		}
 
-		performerID, err := getPerformerID(ctx, endpoint, g.performerCreator, p, createMissing)
+		performerID, err := getPerformerID(ctx, endpoint, g.performerCreator, p, createMissing, g.skipSingleNamePerformers)
 		if err != nil {
+			if errors.Is(err, ErrSkipSingleNamePerformer) {
+				singleNamePerformerSkipped = true
+				continue
+			}
 			return nil, err
 		}
 
@@ -110,9 +120,15 @@ func (g sceneRelationships) performers(ctx context.Context, ignoreMale bool) ([]
 
 	// don't return if nothing was added
 	if sliceutil.SliceSame(originalPerformerIDs, performerIDs) {
+		if singleNamePerformerSkipped {
+			return nil, ErrSkipSingleNamePerformer
+		}
 		return nil, nil
 	}
 
+	if singleNamePerformerSkipped {
+		return performerIDs, ErrSkipSingleNamePerformer
+	}
 	return performerIDs, nil
 }
 
@@ -156,7 +172,7 @@ func (g sceneRelationships) tags(ctx context.Context) ([]int, error) {
 				CreatedAt: now,
 				UpdatedAt: now,
 			}
-			err := g.tagCreator.Create(ctx, &newTag)
+			err := g.tagCreatorFinder.Create(ctx, &newTag)
 			if err != nil {
 				return nil, fmt.Errorf("error creating tag: %w", err)
 			}

internal/identify/scene_test.go

@@ -375,7 +375,7 @@ func Test_sceneRelationships_tags(t *testing.T) {
 
 	tr := sceneRelationships{
 		sceneReader:      mockSceneReaderWriter,
-		tagCreator:   mockTagReaderWriter,
+		tagCreatorFinder: mockTagReaderWriter,
 		fieldOptions:     make(map[string]*FieldOptions),
 	}
 

internal/identify/studio.go

@@ -7,11 +7,13 @@ import (
 
 	"github.com/stashapp/stash/pkg/hash/md5"
 	"github.com/stashapp/stash/pkg/models"
+	"github.com/stashapp/stash/pkg/utils"
 )
 
 type StudioCreator interface {
 	Create(ctx context.Context, newStudio *models.Studio) error
 	UpdateStashIDs(ctx context.Context, studioID int, stashIDs []models.StashID) error
+	UpdateImage(ctx context.Context, studioID int, image []byte) error
 }
 
 func createMissingStudio(ctx context.Context, endpoint string, w StudioCreator, studio *models.ScrapedStudio) (*int, error) {
@@ -21,6 +23,19 @@ func createMissingStudio(ctx context.Context, endpoint string, w StudioCreator,
 		return nil, fmt.Errorf("error creating studio: %w", err)
 	}
 
+	// update image table
+	if studio.Image != nil && len(*studio.Image) > 0 {
+		imageData, err := utils.ReadImageFromURL(ctx, *studio.Image)
+		if err != nil {
+			return nil, err
+		}
+
+		err = w.UpdateImage(ctx, studioInput.ID, imageData)
+		if err != nil {
+			return nil, err
+		}
+	}
+
 	if endpoint != "" && studio.RemoteSiteID != nil {
 		if err := w.UpdateStashIDs(ctx, studioInput.ID, []models.StashID{
 			{

internal/manager/task_identify.go

@@ -136,7 +136,7 @@ func (j *IdentifyJob) identifyScene(ctx context.Context, s *models.Scene, source
 			SceneReaderUpdater: instance.Repository.Scene,
 			StudioCreator:      instance.Repository.Studio,
 			PerformerCreator:   instance.Repository.Performer,
-			TagCreator:         instance.Repository.Tag,
+			TagCreatorFinder:   instance.Repository.Tag,
 
 			DefaultOptions:              j.input.Options,
 			Sources:                     sources,
@@ -248,14 +248,14 @@ type stashboxSource struct {
 	endpoint string
 }
 
-func (s stashboxSource) ScrapeScene(ctx context.Context, sceneID int) (*scraper.ScrapedScene, error) {
+func (s stashboxSource) ScrapeScenes(ctx context.Context, sceneID int) ([]*scraper.ScrapedScene, error) {
 	results, err := s.FindStashBoxSceneByFingerprints(ctx, sceneID)
 	if err != nil {
 		return nil, fmt.Errorf("error querying stash-box using scene ID %d: %w", sceneID, err)
 	}
 
 	if len(results) > 0 {
-		return results[0], nil
+		return results, nil
 	}
 
 	return nil, nil
@@ -270,7 +270,7 @@ type scraperSource struct {
 	scraperID string
 }
 
-func (s scraperSource) ScrapeScene(ctx context.Context, sceneID int) (*scraper.ScrapedScene, error) {
+func (s scraperSource) ScrapeScenes(ctx context.Context, sceneID int) ([]*scraper.ScrapedScene, error) {
 	content, err := s.cache.ScrapeID(ctx, s.scraperID, sceneID, scraper.ScrapeContentTypeScene)
 	if err != nil {
 		return nil, err
@@ -282,7 +282,7 @@ func (s scraperSource) ScrapeScene(ctx context.Context, sceneID int) (*scraper.S
 	}
 
 	if scene, ok := content.(scraper.ScrapedScene); ok {
-		return &scene, nil
+		return []*scraper.ScrapedScene{&scene}, nil
 	}
 
 	return nil, errors.New("could not convert content to scene")

pkg/scraper/stashbox/stash_box.go

@@ -722,6 +722,9 @@ func (c Client) sceneFragmentToScrapedScene(ctx context.Context, s *graphql.Scen
 				URL:          findURL(s.Studio.Urls, "HOME"),
 				RemoteSiteID: &studioID,
 			}
+			if s.Studio.Images != nil && len(s.Studio.Images) > 0 {
+				ss.Studio.Image = &s.Studio.Images[0].URL
+			}
 
 			err := match.ScrapedStudio(ctx, c.repository.Studio, ss.Studio, &c.box.Endpoint)
 			if err != nil {

ui/v2.5/src/components/Dialogs/IdentifyDialog/FieldOptions.tsx

@@ -311,7 +311,7 @@ export const FieldOptionsList: React.FC<IFieldOptionsList> = ({
   }
 
   return (
-    <Form.Group className="scraper-sources">
+    <Form.Group className="scraper-sources mt-3">
       <h5>
         <FormattedMessage id="config.tasks.identify.field_options" />
       </h5>

ui/v2.5/src/components/Dialogs/IdentifyDialog/IdentifyDialog.tsx

@@ -50,6 +50,10 @@ export const IdentifyDialog: React.FC<IIdentifyDialogProps> = ({
       includeMalePerformers: true,
       setCoverImage: true,
       setOrganized: false,
+      skipMultipleMatches: true,
+      skipMultipleMatchTag: undefined,
+      skipSingleNamePerformers: true,
+      skipSingleNamePerformerTag: undefined,
     };
   }
 
@@ -240,6 +244,8 @@ export const IdentifyDialog: React.FC<IIdentifyDialogProps> = ({
         const autoTagCopy = { ...autoTag };
         autoTagCopy.options = {
           setOrganized: false,
+          skipMultipleMatches: true,
+          skipSingleNamePerformers: true,
         };
         newSources.push(autoTagCopy);
       }

ui/v2.5/src/components/Dialogs/IdentifyDialog/Options.tsx

@@ -1,10 +1,11 @@
 import React from "react";
-import { Form } from "react-bootstrap";
+import { Col, Form, Row } from "react-bootstrap";
 import * as GQL from "src/core/generated-graphql";
 import { FormattedMessage, useIntl } from "react-intl";
 import { IScraperSource } from "./constants";
 import { FieldOptionsList } from "./FieldOptions";
 import { ThreeStateBoolean } from "./ThreeStateBoolean";
+import { TagSelect } from "src/components/Shared/Select";
 
 interface IOptionsEditor {
   options: GQL.IdentifyMetadataOptionsInput;
@@ -35,8 +36,76 @@ export const OptionsEditor: React.FC<IOptionsEditor> = ({
     indeterminateClassname: "text-muted",
   };
 
+  function maybeRenderMultipleMatchesTag() {
+    if (!options.skipMultipleMatches) {
+      return;
+    }
+
     return (
-    <Form.Group>
+      <Form.Group controlId="match_tags" className="ml-3 mt-1 mb-0" as={Row}>
+        <Form.Label
+          column
+          sm={{ span: 4, offset: 1 }}
+          title={intl.formatMessage({
+            id: "config.tasks.identify.tag_skipped_matches_tooltip",
+          })}
+        >
+          <FormattedMessage id="config.tasks.identify.tag_skipped_matches" />
+        </Form.Label>
+        <Col sm>
+          <TagSelect
+            onSelect={(tags) =>
+              setOptions({
+                skipMultipleMatchTag: tags[0]?.id,
+              })
+            }
+            ids={
+              options.skipMultipleMatchTag ? [options.skipMultipleMatchTag] : []
+            }
+            noSelectionString="Select/create tag..."
+          />
+        </Col>
+      </Form.Group>
+    );
+  }
+
+  function maybeRenderPerformersTag() {
+    if (!options.skipSingleNamePerformers) {
+      return;
+    }
+
+    return (
+      <Form.Group controlId="match_tags" className="ml-3 mt-1 mb-0" as={Row}>
+        <Form.Label
+          column
+          sm={{ span: 4, offset: 1 }}
+          title={intl.formatMessage({
+            id: "config.tasks.identify.tag_skipped_performer_tooltip",
+          })}
+        >
+          <FormattedMessage id="config.tasks.identify.tag_skipped_performers" />
+        </Form.Label>
+        <Col sm>
+          <TagSelect
+            onSelect={(tags) =>
+              setOptions({
+                skipSingleNamePerformerTag: tags[0]?.id,
+              })
+            }
+            ids={
+              options.skipSingleNamePerformerTag
+                ? [options.skipSingleNamePerformerTag]
+                : []
+            }
+            noSelectionString="Select/create tag..."
+          />
+        </Col>
+      </Form.Group>
+    );
+  }
+
+  return (
+    <Form.Group className="mb-0">
       <Form.Group>
         <h5>
           <FormattedMessage
@@ -52,7 +121,7 @@ export const OptionsEditor: React.FC<IOptionsEditor> = ({
           </Form.Text>
         )}
       </Form.Group>
-      <Form.Group>
+      <Form.Group className="mb-0">
         <ThreeStateBoolean
           id="include-male-performers"
           value={
@@ -104,6 +173,50 @@ export const OptionsEditor: React.FC<IOptionsEditor> = ({
           {...checkboxProps}
         />
       </Form.Group>
+      <ThreeStateBoolean
+        id="skip-multiple-match"
+        value={
+          options.skipMultipleMatches === null
+            ? undefined
+            : options.skipMultipleMatches
+        }
+        setValue={(v) =>
+          setOptions({
+            skipMultipleMatches: v,
+          })
+        }
+        label={intl.formatMessage({
+          id: "config.tasks.identify.skip_multiple_matches",
+        })}
+        defaultValue={defaultOptions?.skipMultipleMatches ?? undefined}
+        tooltip={intl.formatMessage({
+          id: "config.tasks.identify.skip_multiple_matches_tooltip",
+        })}
+        {...checkboxProps}
+      />
+      {maybeRenderMultipleMatchesTag()}
+      <ThreeStateBoolean
+        id="skip-single-name-performers"
+        value={
+          options.skipSingleNamePerformers === null
+            ? undefined
+            : options.skipSingleNamePerformers
+        }
+        setValue={(v) =>
+          setOptions({
+            skipSingleNamePerformers: v,
+          })
+        }
+        label={intl.formatMessage({
+          id: "config.tasks.identify.skip_single_name_performers",
+        })}
+        defaultValue={defaultOptions?.skipSingleNamePerformers ?? undefined}
+        tooltip={intl.formatMessage({
+          id: "config.tasks.identify.skip_single_name_performers_tooltip",
+        })}
+        {...checkboxProps}
+      />
+      {maybeRenderPerformersTag()}
 
       <FieldOptionsList
         fieldOptions={options.fieldOptions ?? undefined}

ui/v2.5/src/components/Dialogs/IdentifyDialog/ThreeStateBoolean.tsx

@@ -10,6 +10,7 @@ interface IThreeStateBoolean {
   label?: React.ReactNode;
   disabled?: boolean;
   defaultValue?: boolean;
+  tooltip?: string | undefined;
 }
 
 export const ThreeStateBoolean: React.FC<IThreeStateBoolean> = ({
@@ -20,6 +21,7 @@ export const ThreeStateBoolean: React.FC<IThreeStateBoolean> = ({
   label,
   disabled,
   defaultValue,
+  tooltip,
 }) => {
   const intl = useIntl();
 
@@ -31,6 +33,7 @@ export const ThreeStateBoolean: React.FC<IThreeStateBoolean> = ({
         checked={value}
         label={label}
         onChange={() => setValue(!value)}
+        title={tooltip}
       />
     );
   }
@@ -79,7 +82,7 @@ export const ThreeStateBoolean: React.FC<IThreeStateBoolean> = ({
 
   return (
     <Form.Group>
-      <h6>{label}</h6>
+      <h6 title={tooltip}>{label}</h6>
       <Form.Group>
         {renderModeButton(undefined)}
         {renderModeButton(false)}

ui/v2.5/src/components/Dialogs/styles.scss

@@ -6,3 +6,13 @@
     justify-content: space-between;
   }
 }
+
+.form-group {
+  h6,
+  label {
+    &[title]:not([title=""]) {
+      cursor: help;
+      text-decoration: underline dotted;
+    }
+  }
+}

ui/v2.5/src/docs/en/Manual/Identify.md

@@ -15,6 +15,10 @@ The following options can be set:
 | Include male performers | If false, then male performers will not be created or set on scenes. |
 | Set cover images | If false, then scene cover images will not be modified. |
 | Set organised flag | If true, the organised flag is set to true when a scene is organised. |
+| Skip matches that have more than one result | If this is not enabled and more than one result is returned, one will be randomly chosen to match |
+| Tag skipped matches with | If the above option is set and a scene is skipped, this will add the tag so that you can filter for it in the Scene Tagger view and choose the correct match by hand |
+| Skip single name performers with no disambiguation | If this is not enabled, performers that are often generic like Samantha or Olga will be matched |
+| Tag skipped performers with | If the above options is set and a performer is skipped, this will add the tag so that you can filter for in it the Scene Tagger view and choose how you want to handle those performers |
 
 Field specific options may be set as well. Each field may have a Strategy. The behaviour for each strategy value is as follows:
 

ui/v2.5/src/locales/en-GB.json

@@ -455,10 +455,18 @@
         "include_male_performers": "Include male performers",
         "set_cover_images": "Set cover images",
         "set_organized": "Set organised flag",
+        "skip_multiple_matches": "Skip matches that have more than one result",
+        "skip_multiple_matches_tooltip": "If this is not enabled and more than one result is returned, one will be randomly chosen to match",
+        "skip_single_name_performers": "Skip single name performers with no disambiguation",
+        "skip_single_name_performers_tooltip": "If this is not enabled, performers that are often generic like Samantha or Olga will be matched",
         "source": "Source",
         "source_options": "{source} Options",
         "sources": "Sources",
-        "strategy": "Strategy"
+        "strategy": "Strategy",
+        "tag_skipped_matches": "Tag skipped matches with",
+        "tag_skipped_matches_tooltip": "Create a tag like 'Identify: Multiple Matches' that you can filter for in the Scene Tagger view and choose the correct match by hand",
+        "tag_skipped_performers": "Tag skipped performers with",
+        "tag_skipped_performer_tooltip": "Create a tag like 'Identify: Single Name Performer' that you can filter for in the Scene Tagger view and choose how you want to handle these performers"
       },
       "import_from_exported_json": "Import from exported JSON in the metadata directory. Wipes the existing database.",
       "incremental_import": "Incremental import from a supplied export zip file.",