update merged performer upon batch update (#5664)

* update merged performer upon batch update
* Handle aliases and name for merged performer
* Refactor merge performer code

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

graphql/stash-box/query.graphql

@@ -49,6 +49,8 @@ fragment PerformerFragment on Performer {
   aliases
   gender
   merged_ids
+  deleted
+  merged_into_id
   urls {
     ...URLFragment
   }

internal/manager/task_stash_box_tag.go

@@ -9,6 +9,7 @@ import (
 	"github.com/stashapp/stash/pkg/models"
 	"github.com/stashapp/stash/pkg/performer"
 	"github.com/stashapp/stash/pkg/scraper/stashbox"
+	"github.com/stashapp/stash/pkg/sliceutil"
 	"github.com/stashapp/stash/pkg/studio"
 )
 
@@ -119,6 +120,18 @@ func (t *StashBoxBatchTagTask) findStashBoxPerformer(ctx context.Context) (*mode
 		}
 		if remoteID != "" {
 			performer, err = client.FindStashBoxPerformerByID(ctx, remoteID)
+
+			if performer != nil && performer.RemoteMergedIntoId != nil {
+				mergedPerformer, err := t.handleMergedPerformer(ctx, performer, client)
+				if err != nil {
+					return nil, err
+				}
+
+				if mergedPerformer != nil {
+					logger.Infof("Performer id %s merged into %s, updating local performer", remoteID, *performer.RemoteMergedIntoId)
+					performer = mergedPerformer
+				}
+			}
 		}
 	} else {
 		var name string
@@ -133,6 +146,21 @@ func (t *StashBoxBatchTagTask) findStashBoxPerformer(ctx context.Context) (*mode
 	return performer, err
 }
 
+func (t *StashBoxBatchTagTask) handleMergedPerformer(ctx context.Context, performer *models.ScrapedPerformer, client *stashbox.Client) (mergedPerformer *models.ScrapedPerformer, err error) {
+	mergedPerformer, err = client.FindStashBoxPerformerByID(ctx, *performer.RemoteMergedIntoId)
+	if err != nil {
+		return nil, fmt.Errorf("loading merged performer %s from stashbox", *performer.RemoteMergedIntoId)
+	}
+
+	if mergedPerformer.StoredID != nil && *mergedPerformer.StoredID != *performer.StoredID {
+		logger.Warnf("Performer %s merged into %s, but both exist locally, not merging", *performer.StoredID, *mergedPerformer.StoredID)
+		return nil, nil
+	}
+
+	mergedPerformer.StoredID = performer.StoredID
+	return mergedPerformer, nil
+}
+
 func (t *StashBoxBatchTagTask) processMatchedPerformer(ctx context.Context, p *models.ScrapedPerformer, excluded map[string]bool) {
 	// Refreshing an existing performer
 	if t.performer != nil {
@@ -156,6 +184,19 @@ func (t *StashBoxBatchTagTask) processMatchedPerformer(ctx context.Context, p *m
 
 			partial := p.ToPartial(t.box.Endpoint, excluded, existingStashIDs)
 
+			// if we're setting the performer's aliases, and not the name, then filter out the name
+			// from the aliases to avoid duplicates
+			// add the name to the aliases if it's not already there
+			if partial.Aliases != nil && !partial.Name.Set {
+				partial.Aliases.Values = sliceutil.Filter(partial.Aliases.Values, func(s string) bool {
+					return s != t.performer.Name
+				})
+
+				if p.Name != nil && t.performer.Name != *p.Name {
+					partial.Aliases.Values = sliceutil.AppendUnique(partial.Aliases.Values, *p.Name)
+				}
+			}
+
 			if err := performer.ValidateUpdate(ctx, t.performer.ID, partial, qb); err != nil {
 				return err
 			}

pkg/models/model_scraped_item.go

@@ -128,13 +128,15 @@ type ScrapedPerformer struct {
 	Aliases        *string       `json:"aliases"`
 	Tags           []*ScrapedTag `json:"tags"`
 	// This should be a base64 encoded data URL
-	Image        *string  `json:"image"` // deprecated: use Images
-	Images       []string `json:"images"`
-	Details      *string  `json:"details"`
-	DeathDate    *string  `json:"death_date"`
-	HairColor    *string  `json:"hair_color"`
-	Weight       *string  `json:"weight"`
-	RemoteSiteID *string  `json:"remote_site_id"`
+	Image              *string  `json:"image"` // deprecated: use Images
+	Images             []string `json:"images"`
+	Details            *string  `json:"details"`
+	DeathDate          *string  `json:"death_date"`
+	HairColor          *string  `json:"hair_color"`
+	Weight             *string  `json:"weight"`
+	RemoteSiteID       *string  `json:"remote_site_id"`
+	RemoteDeleted      bool     `json:"remote_deleted"`
+	RemoteMergedIntoId *string  `json:"remote_merged_into_id"`
 }
 
 func (ScrapedPerformer) IsScrapedContent() {}

pkg/scraper/stashbox/graphql/generated_client.go

@@ -196,6 +196,8 @@ type PerformerFragment struct {
 	Aliases         []string                    "json:\"aliases\" graphql:\"aliases\""
 	Gender          *GenderEnum                 "json:\"gender,omitempty\" graphql:\"gender\""
 	MergedIds       []string                    "json:\"merged_ids\" graphql:\"merged_ids\""
+	Deleted         bool                        "json:\"deleted\" graphql:\"deleted\""
+	MergedIntoID    *string                     "json:\"merged_into_id,omitempty\" graphql:\"merged_into_id\""
 	Urls            []*URLFragment              "json:\"urls\" graphql:\"urls\""
 	Images          []*ImageFragment            "json:\"images\" graphql:\"images\""
 	BirthDate       *string                     "json:\"birth_date,omitempty\" graphql:\"birth_date\""
@@ -249,6 +251,18 @@ func (t *PerformerFragment) GetMergedIds() []string {
 	}
 	return t.MergedIds
 }
+func (t *PerformerFragment) GetDeleted() bool {
+	if t == nil {
+		t = &PerformerFragment{}
+	}
+	return t.Deleted
+}
+func (t *PerformerFragment) GetMergedIntoID() *string {
+	if t == nil {
+		t = &PerformerFragment{}
+	}
+	return t.MergedIntoID
+}
 func (t *PerformerFragment) GetUrls() []*URLFragment {
 	if t == nil {
 		t = &PerformerFragment{}
@@ -860,6 +874,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -993,6 +1009,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -1126,6 +1144,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -1259,6 +1279,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -1331,6 +1353,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -1408,6 +1432,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}
@@ -1546,6 +1572,8 @@ fragment PerformerFragment on Performer {
 	aliases
 	gender
 	merged_ids
+	deleted
+	merged_into_id
 	urls {
 		... URLFragment
 	}

pkg/scraper/stashbox/performer.go

@@ -297,16 +297,18 @@ func performerFragmentToScrapedPerformer(p graphql.PerformerFragment) *models.Sc
 	}
 
 	sp := &models.ScrapedPerformer{
-		Name:           &p.Name,
-		Disambiguation: p.Disambiguation,
-		Country:        p.Country,
-		Measurements:   formatMeasurements(*p.Measurements),
-		CareerLength:   formatCareerLength(p.CareerStartYear, p.CareerEndYear),
-		Tattoos:        formatBodyModifications(p.Tattoos),
-		Piercings:      formatBodyModifications(p.Piercings),
-		Twitter:        findURL(p.Urls, "TWITTER"),
-		RemoteSiteID:   &p.ID,
-		Images:         images,
+		Name:               &p.Name,
+		Disambiguation:     p.Disambiguation,
+		Country:            p.Country,
+		Measurements:       formatMeasurements(*p.Measurements),
+		CareerLength:       formatCareerLength(p.CareerStartYear, p.CareerEndYear),
+		Tattoos:            formatBodyModifications(p.Tattoos),
+		Piercings:          formatBodyModifications(p.Piercings),
+		Twitter:            findURL(p.Urls, "TWITTER"),
+		RemoteSiteID:       &p.ID,
+		RemoteDeleted:      p.Deleted,
+		RemoteMergedIntoId: p.MergedIntoID,
+		Images:             images,
 		// TODO - tags not currently supported
 		// graphql schema change to accommodate this. Leave off for now.
 	}