refac: atomic merge, lost-update race, RunPurge cost, duplicated walks

internal/domain/contenttype/fieldSchema.go

@@ -115,6 +115,21 @@ func (f Field) validDefault() bool {
 	return !f.Unique && f.Violation(f.Default) == ""
 }
 
+type MediaRef struct {
+	Field Field
+	ID    uint32
+}
+
+func MediaRefs(fields []Field, data map[string]any) []MediaRef {
+	var refs []MediaRef
+	for _, field := range MediaFields(fields) {
+		if id, set := MediaID(data[field.Key]); set {
+			refs = append(refs, MediaRef{Field: field, ID: id})
+		}
+	}
+	return refs
+}
+
 func MediaFields(fields []Field) []Field {
 	var media []Field
 	for _, field := range fields {

internal/domain/entry/entryData.go

@@ -51,12 +51,12 @@ func validateField(field contenttype.Field, data map[string]any) error {
 	value := data[field.Key]
 	if value == nil {
 		if field.Required {
-			return invalidField(field.Key, RuleRequired)
+			return InvalidField(field.Key, RuleRequired)
 		}
 		return nil
 	}
 	if rule := field.Violation(value); rule != "" {
-		return invalidField(field.Key, rule)
+		return InvalidField(field.Key, rule)
 	}
 	return nil
 }
@@ -76,7 +76,7 @@ func firstUnknownField(fields []contenttype.Field, data map[string]any) error {
 		return nil
 	}
 	sort.Strings(unknown)
-	return invalidField(unknown[0], RuleUnknownField)
+	return InvalidField(unknown[0], RuleUnknownField)
 }
 
 func UniqueValues(fields []contenttype.Field, data map[string]any) []UniqueValue {
@@ -91,10 +91,15 @@ func UniqueValues(fields []contenttype.Field, data map[string]any) []UniqueValue
 
 func MediaIDs(fields []contenttype.Field, data map[string]any) map[string]uint32 {
 	ids := map[string]uint32{}
-	for _, field := range contenttype.MediaFields(fields) {
-		if id, ok := contenttype.MediaID(data[field.Key]); ok {
-			ids[field.Key] = id
-		}
+	for _, ref := range contenttype.MediaRefs(fields, data) {
+		ids[ref.Field.Key] = ref.ID
 	}
 	return ids
 }
+
+func MissingMedia(field contenttype.Field, unchanged bool) error {
+	if unchanged {
+		return InvalidField(field.Key, RuleRequired)
+	}
+	return InvalidField(field.Key, contenttype.RuleType)
+}

internal/domain/entry/entryDomain.go

@@ -55,12 +55,8 @@ func (e FieldError) Unwrap() error {
 	return e.Cause
 }
 
-func invalidField(key string, rule string) error {
-	return FieldError{Key: key, Rule: rule, Cause: ErrInvalidEntry}
-}
-
 func InvalidField(key string, rule string) error {
-	return invalidField(key, rule)
+	return FieldError{Key: key, Rule: rule, Cause: ErrInvalidEntry}
 }
 
 func UniqueViolation(key string) error {

internal/domain/image/imageDomain.go

@@ -246,3 +246,52 @@ type Image struct {
 	CreatedAt    time.Time `json:"created_at"`
 	UpdatedAt    time.Time `json:"updated_at"`
 }
+
+func isStoredNameByte(char byte) bool {
+	return char == '-' || char == '_' || char == '.' || char >= '0' && char <= '9' || char >= 'a' && char <= 'z' || char >= 'A' && char <= 'Z'
+}
+
+func continuesName(body string, end int) bool {
+	if end >= len(body) || !isStoredNameByte(body[end]) {
+		return false
+	}
+	return body[end] != '.' || end+1 < len(body) && isStoredNameByte(body[end+1])
+}
+
+func storedNameMatches(body string, name string) []int {
+	var starts []int
+	if name == "" {
+		return starts
+	}
+	for from := 0; from < len(body); {
+		found := strings.Index(body[from:], name)
+		if found < 0 {
+			break
+		}
+		start := from + found
+		end := start + len(name)
+		from = start + 1
+		if start > 0 && isStoredNameByte(body[start-1]) || continuesName(body, end) {
+			continue
+		}
+		starts = append(starts, start)
+		from = end
+	}
+	return starts
+}
+
+func ContainsStoredName(body string, name string) bool {
+	return len(storedNameMatches(body, name)) > 0
+}
+
+func ReplaceStoredName(body string, from string, to string) string {
+	var rewritten strings.Builder
+	last := 0
+	for _, start := range storedNameMatches(body, from) {
+		rewritten.WriteString(body[last:start])
+		rewritten.WriteString(to)
+		last = start + len(from)
+	}
+	rewritten.WriteString(body[last:])
+	return rewritten.String()
+}

internal/domain/transaction/transactionDomain.go

@@ -0,0 +1,22 @@
+package transaction
+
+import (
+	"RenewCMS/internal/domain/article"
+	"RenewCMS/internal/domain/author"
+	"RenewCMS/internal/domain/entry"
+	"RenewCMS/internal/domain/image"
+	"RenewCMS/internal/domain/site"
+	"context"
+)
+
+type Repositories struct {
+	Articles article.Repository
+	Sites    site.Repository
+	Authors  author.Repository
+	Entries  entry.Repository
+	Images   image.Repository
+}
+
+type Transactor interface {
+	Run(ctx context.Context, work func(Repositories) error) error
+}

internal/infrastructure/persistence/repositories/transactor.go

@@ -0,0 +1,30 @@
+package persistence
+
+import (
+	domain "RenewCMS/internal/domain/transaction"
+	"context"
+
+	"gorm.io/gorm"
+)
+
+var _ domain.Transactor = &Transactor{}
+
+type Transactor struct {
+	db *gorm.DB
+}
+
+func NewTransactor(db *gorm.DB) *Transactor {
+	return &Transactor{db}
+}
+
+func (t *Transactor) Run(ctx context.Context, work func(domain.Repositories) error) error {
+	return t.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+		return work(domain.Repositories{
+			Articles: NewArticleRepository(tx),
+			Sites:    NewSiteRepository(tx),
+			Authors:  NewAuthorRepository(tx),
+			Entries:  NewEntryRepository(tx),
+			Images:   NewImageRepository(tx),
+		})
+	})
+}

internal/infrastructure/scheduler/runner.go

@@ -16,7 +16,7 @@ const (
 )
 
 type Job interface {
-	Tick(now time.Time) (purge.RunResult, bool, error)
+	Tick(ctx context.Context, now time.Time) (purge.RunResult, bool, error)
 }
 
 type CleanupJob interface {
@@ -24,15 +24,15 @@ type CleanupJob interface {
 }
 
 type task interface {
-	tick(now time.Time) (summary string, ran bool, err error)
+	tick(ctx context.Context, now time.Time) (summary string, ran bool, err error)
 }
 
 type purgeTask struct {
 	job Job
 }
 
-func (p purgeTask) tick(now time.Time) (string, bool, error) {
-	result, ran, err := p.job.Tick(now)
+func (p purgeTask) tick(ctx context.Context, now time.Time) (string, bool, error) {
+	result, ran, err := p.job.Tick(ctx, now)
 	if errors.Is(err, purge.ErrPurgeBusy) {
 		return "", false, nil
 	}
@@ -44,7 +44,7 @@ type cleanupTask struct {
 	job CleanupJob
 }
 
-func (c cleanupTask) tick(now time.Time) (string, bool, error) {
+func (c cleanupTask) tick(_ context.Context, now time.Time) (string, bool, error) {
 	result, ran, err := c.job.Tick(now)
 	return fmt.Sprintf("removed=%d", result.Removed), ran, err
 }
@@ -119,7 +119,7 @@ func (r *Runner) tick(ctx context.Context) {
 	if ctx.Err() != nil {
 		return
 	}
-	summary, ran, err := r.task.tick(r.now())
+	summary, ran, err := r.task.tick(ctx, r.now())
 	if err != nil {
 		r.logger.Printf("scheduler: %s failed: %v", r.label, err)
 		return

internal/infrastructure/useCases/GetMediaUsage.go

@@ -8,7 +8,6 @@ import (
 	"RenewCMS/internal/domain/image"
 	"RenewCMS/internal/domain/site"
 	"slices"
-	"strings"
 )
 
 const (
@@ -111,7 +110,7 @@ func (u *GetMediaUsageUseCase) articleUsages(img image.Image) ([]MediaUsage, err
 	}
 	byID := map[uint32]*MediaUsage{}
 	for _, found := range inBody {
-		if !strings.Contains(found.Body, storedName) {
+		if !image.ContainsStoredName(found.Body, storedName) {
 			continue
 		}
 		entry := usageFor(byID, found)
@@ -172,5 +171,5 @@ func (u *GetMediaUsageUseCase) siteUsages(img image.Image) ([]MediaUsage, error)
 }
 
 func articleMentionsImage(a article.Article, img image.Image) bool {
-	return strings.Contains(a.Body, image.StoredName(img.Path)) || a.UsesImage(img.ID)
+	return image.ContainsStoredName(a.Body, image.StoredName(img.Path)) || a.UsesImage(img.ID)
 }

internal/infrastructure/useCases/MergeMedia.go

@@ -3,17 +3,14 @@ package useCases
 import (
 	"RenewCMS/internal/domain/article"
 	"RenewCMS/internal/domain/audit"
-	"RenewCMS/internal/domain/author"
 	"RenewCMS/internal/domain/contenttype"
 	"RenewCMS/internal/domain/entry"
 	"RenewCMS/internal/domain/image"
-	"RenewCMS/internal/domain/site"
-	"cmp"
+	"RenewCMS/internal/domain/transaction"
 	"context"
 	"errors"
+	"log"
 	"slices"
-	"strconv"
-	"strings"
 )
 
 type MergeMediaResult struct {
@@ -30,13 +27,13 @@ type MergeMediaUseCase struct {
 	articleRepository article.Repository
 	remover           mediaRemover
 	recorder          audit.Recorder
-	sites             site.Repository
-	authors           author.Repository
 	entries           *entryMediaScanner
+	transactor        transaction.Transactor
+	logger            *log.Logger
 }
 
-func NewMergeMediaUseCase(imageRepository image.Repository, articleRepository article.Repository, storage image.Storage, variantFiles image.VariantFileRepository, recorder audit.Recorder, sites site.Repository, authors author.Repository, contentTypes contenttype.Repository, entries entry.Repository) *MergeMediaUseCase {
-	return &MergeMediaUseCase{imageRepository: imageRepository, articleRepository: articleRepository, remover: newMediaRemover(imageRepository, storage, variantFiles), recorder: recorder, sites: sites, authors: authors, entries: newEntryMediaScanner(contentTypes, entries)}
+func NewMergeMediaUseCase(imageRepository image.Repository, articleRepository article.Repository, storage image.Storage, variantFiles image.VariantFileRepository, recorder audit.Recorder, contentTypes contenttype.Repository, entries entry.Repository, transactor transaction.Transactor, logger *log.Logger) *MergeMediaUseCase {
+	return &MergeMediaUseCase{imageRepository: imageRepository, articleRepository: articleRepository, remover: newMediaRemover(imageRepository, storage, variantFiles), recorder: recorder, entries: newEntryMediaScanner(contentTypes, entries), transactor: transactor, logger: logger}
 }
 
 func copyIdsOf(copies []image.Image) []uint32 {
@@ -47,22 +44,8 @@ func copyIdsOf(copies []image.Image) []uint32 {
 	return copyIds
 }
 
-func (u *MergeMediaUseCase) repointSites(copies []image.Image, keptId uint32) (int, error) {
-	return u.sites.RepointImage(copyIdsOf(copies), keptId)
-}
-
-func (u *MergeMediaUseCase) repointAuthors(copies []image.Image, keptId uint32) (int, error) {
-	return u.authors.RepointAvatar(copyIdsOf(copies), keptId)
-}
-
-type bodyRewrite struct {
-	articleId uint32
-	original  string
-	rewritten string
-}
-
 func (u *MergeMediaUseCase) Merge(ctx context.Context, keepId uint32, mergeIds []uint32) (MergeMediaResult, error) {
-	result, err := u.merge(keepId, mergeIds)
+	result, err := u.merge(ctx, keepId, mergeIds, nil)
 	event := audit.NewEvent(audit.ActionMediaMerge, audit.CategoryMedia)
 	if kept, found := u.keptImage(keepId); found {
 		event.Target = imageTarget(kept)
@@ -84,7 +67,7 @@ func (u *MergeMediaUseCase) keptImage(keepId uint32) (image.Image, bool) {
 	return kept, err == nil
 }
 
-func (u *MergeMediaUseCase) merge(keepId uint32, mergeIds []uint32) (MergeMediaResult, error) {
+func (u *MergeMediaUseCase) merge(ctx context.Context, keepId uint32, mergeIds []uint32, index *entryMediaIndex) (MergeMediaResult, error) {
 	if err := validateMergeRequest(keepId, mergeIds); err != nil {
 		return MergeMediaResult{}, err
 	}
@@ -98,51 +81,94 @@ func (u *MergeMediaUseCase) merge(keepId uint32, mergeIds []uint32) (MergeMediaR
 		return MergeMediaResult{}, err
 	}
 
-	rewrites, err := u.planRewrites(kept, copies)
+	planned, err := u.planRewrites(copies)
 	if err != nil {
 		return MergeMediaResult{}, err
 	}
-	if err := u.applyRewrites(rewrites); err != nil {
-		return MergeMediaResult{}, err
-	}
-	repointedArticles, err := u.articleRepository.RepointImage(copyIdsOf(copies), kept.ID)
-	if err != nil {
-		u.restoreBodies(rewrites)
-		return MergeMediaResult{}, err
-	}
-	sitesUpdated, err := u.repointSites(copies, kept.ID)
-	if err != nil {
-		u.restoreBodies(rewrites)
-		return MergeMediaResult{}, err
-	}
-
-	authorsUpdated, err := u.repointAuthors(copies, kept.ID)
-	if err != nil {
-		u.restoreBodies(rewrites)
-		return MergeMediaResult{}, err
-	}
-
-	repointedEntries, err := u.entries.repoint(copyIdsOf(copies), kept.ID)
-	if err != nil {
-		u.restoreBodies(rewrites)
-		return MergeMediaResult{}, err
-	}
-
-	for _, duplicate := range copies {
-		if err := u.remover.remove(duplicate); err != nil {
+	if index == nil {
+		if index, err = u.entries.index(); err != nil {
 			return MergeMediaResult{}, err
 		}
 	}
 
-	updatedIds := updatedArticleIds(rewrites, repointedArticles)
-	return MergeMediaResult{ArticlesUpdated: len(updatedIds), Merged: len(copies), SitesUpdated: sitesUpdated, AuthorsUpdated: authorsUpdated, EntriesUpdated: len(repointedEntries), UpdatedArticleIDs: updatedIds}, nil
+	result, err := u.applyMerge(ctx, kept, copies, planned, index.referencing(copyIdsOf(copies)))
+	if err != nil {
+		return MergeMediaResult{}, err
+	}
+	u.discardCopies(copies)
+	return result, nil
 }
 
-func updatedArticleIds(rewrites []bodyRewrite, repointed []uint32) []uint32 {
-	ids := slices.Clone(repointed)
-	for _, plan := range rewrites {
-		ids = append(ids, plan.articleId)
+func (u *MergeMediaUseCase) applyMerge(ctx context.Context, kept image.Image, copies []image.Image, planned []uint32, candidates []mediaEntry) (MergeMediaResult, error) {
+	var result MergeMediaResult
+	copyIds := copyIdsOf(copies)
+	err := u.transactor.Run(ctx, func(tx transaction.Repositories) error {
+		rewritten, err := rewriteBodies(tx.Articles, planned, copies, kept)
+		if err != nil {
+			return err
+		}
+		repointedArticles, err := tx.Articles.RepointImage(copyIds, kept.ID)
+		if err != nil {
+			return err
+		}
+		if result.SitesUpdated, err = tx.Sites.RepointImage(copyIds, kept.ID); err != nil {
+			return err
+		}
+		if result.AuthorsUpdated, err = tx.Authors.RepointAvatar(copyIds, kept.ID); err != nil {
+			return err
+		}
+		if result.EntriesUpdated, err = repointEntries(tx.Entries, candidates, copyIds, kept.ID); err != nil {
+			return err
+		}
+		for _, duplicate := range copies {
+			if err := tx.Images.Delete(duplicate.ID); err != nil {
+				return err
+			}
+		}
+		result.UpdatedArticleIDs = updatedArticleIds(rewritten, repointedArticles)
+		result.ArticlesUpdated = len(result.UpdatedArticleIDs)
+		result.Merged = len(copies)
+		return nil
+	})
+	return result, err
+}
+
+func (u *MergeMediaUseCase) discardCopies(copies []image.Image) {
+	for _, duplicate := range copies {
+		if err := u.remover.discardFiles(duplicate); err != nil {
+			u.logger.Printf("media merge: could not remove the files of %s: %v", image.StoredName(duplicate.Path), err)
+		}
 	}
+}
+
+func rewriteBodies(articles article.Repository, planned []uint32, copies []image.Image, kept image.Image) ([]uint32, error) {
+	keptName := image.StoredName(kept.Path)
+	var rewritten []uint32
+	for _, id := range planned {
+		current, err := articles.Get(id)
+		if errors.Is(err, article.ErrArticleNotFound) {
+			continue
+		}
+		if err != nil {
+			return nil, err
+		}
+		body := current.Body
+		for _, duplicate := range copies {
+			body = image.ReplaceStoredName(body, image.StoredName(duplicate.Path), keptName)
+		}
+		if body == current.Body {
+			continue
+		}
+		if _, err := articles.UpdateBody(id, body); err != nil {
+			return nil, err
+		}
+		rewritten = append(rewritten, id)
+	}
+	return rewritten, nil
+}
+
+func updatedArticleIds(rewritten []uint32, repointed []uint32) []uint32 {
+	ids := append(slices.Clone(repointed), rewritten...)
 	slices.Sort(ids)
 	return slices.Compact(ids)
 }
@@ -179,10 +205,8 @@ func (u *MergeMediaUseCase) loadCopies(kept image.Image, mergeIds []uint32) ([]i
 	return copies, nil
 }
 
-func (u *MergeMediaUseCase) planRewrites(kept image.Image, copies []image.Image) ([]bodyRewrite, error) {
-	keptName := image.StoredName(kept.Path)
-	rewritten := map[uint32]*bodyRewrite{}
-
+func (u *MergeMediaUseCase) planRewrites(copies []image.Image) ([]uint32, error) {
+	seen := map[uint32]bool{}
 	for _, duplicate := range copies {
 		copyName := image.StoredName(duplicate.Path)
 		articles, err := u.articleRepository.FindByBodyContaining(copyName)
@@ -190,39 +214,15 @@ func (u *MergeMediaUseCase) planRewrites(kept image.Image, copies []image.Image)
 			return nil, err
 		}
 		for _, a := range articles {
-			if !strings.Contains(a.Body, copyName) {
-				continue
+			if image.ContainsStoredName(a.Body, copyName) {
+				seen[a.ID] = true
 			}
-			plan, ok := rewritten[a.ID]
-			if !ok {
-				plan = &bodyRewrite{articleId: a.ID, original: a.Body, rewritten: a.Body}
-				rewritten[a.ID] = plan
-			}
-			plan.rewritten = strings.ReplaceAll(plan.rewritten, copyName, keptName)
 		}
 	}
-
-	plans := make([]bodyRewrite, 0, len(rewritten))
-	for _, plan := range rewritten {
-		plans = append(plans, *plan)
-	}
-	slices.SortFunc(plans, func(a, b bodyRewrite) int { return cmp.Compare(a.articleId, b.articleId) })
-	return plans, nil
-}
-
-func (u *MergeMediaUseCase) applyRewrites(rewrites []bodyRewrite) error {
-	for applied, plan := range rewrites {
-		if _, err := u.articleRepository.UpdateBody(plan.articleId, plan.rewritten); err != nil {
-			u.restoreBodies(rewrites[:applied])
-			return err
-		}
-	}
-	return nil
-}
-
-func (u *MergeMediaUseCase) restoreBodies(applied []bodyRewrite) {
-	for _, plan := range applied {
-		_, err := u.articleRepository.UpdateBody(plan.articleId, plan.original)
-		logCompensation("restore article body "+strconv.FormatUint(uint64(plan.articleId), 10), err)
+	planned := make([]uint32, 0, len(seen))
+	for id := range seen {
+		planned = append(planned, id)
 	}
+	slices.Sort(planned)
+	return planned, nil
 }

internal/infrastructure/useCases/PurgeScheduler.go

@@ -9,7 +9,7 @@ import (
 )
 
 type PurgeRunner interface {
-	Execute(fingerprint *string) (purge.RunResult, error)
+	Execute(ctx context.Context, fingerprint *string) (purge.RunResult, error)
 }
 
 type PurgeSchedulerUseCase struct {
@@ -22,7 +22,7 @@ func NewPurgeSchedulerUseCase(repository purge.Repository, runner PurgeRunner, r
 	return &PurgeSchedulerUseCase{repository, runner, recorder}
 }
 
-func (u *PurgeSchedulerUseCase) Tick(now time.Time) (purge.RunResult, bool, error) {
+func (u *PurgeSchedulerUseCase) Tick(ctx context.Context, now time.Time) (purge.RunResult, bool, error) {
 	settings, err := u.repository.Get()
 	if err != nil {
 		return purge.RunResult{}, false, err
@@ -31,7 +31,7 @@ func (u *PurgeSchedulerUseCase) Tick(now time.Time) (purge.RunResult, bool, erro
 	if next == nil || now.Before(*next) {
 		return purge.RunResult{}, false, nil
 	}
-	result, err := u.runner.Execute(nil)
+	result, err := u.runner.Execute(ctx, nil)
 	if err != nil {
 		u.recordFailedRun(err)
 		return purge.RunResult{}, false, err

internal/infrastructure/useCases/RunPurge.go

@@ -36,7 +36,7 @@ func NewRunPurgeUseCase(imageRepository image.Repository, articleRepository arti
 }
 
 func (u *RunPurgeUseCase) Run(ctx context.Context, fingerprint *string) (purge.RunResult, error) {
-	result, err := u.Execute(fingerprint)
+	result, err := u.Execute(ctx, fingerprint)
 	event := audit.NewEvent(audit.ActionMediaPurge, audit.CategoryMedia)
 	if err != nil {
 		recordFailure(ctx, u.recorder, event, err)
@@ -55,7 +55,7 @@ func purgeParams(result purge.RunResult) map[string]any {
 	return params
 }
 
-func (u *RunPurgeUseCase) Execute(fingerprint *string) (purge.RunResult, error) {
+func (u *RunPurgeUseCase) Execute(ctx context.Context, fingerprint *string) (purge.RunResult, error) {
 	if !u.guard.TryLock() {
 		return purge.RunResult{}, purge.ErrPurgeBusy
 	}
@@ -69,7 +69,10 @@ func (u *RunPurgeUseCase) Execute(fingerprint *string) (purge.RunResult, error)
 		return purge.RunResult{}, purge.ErrPurgeConflict
 	}
 
-	result := u.mergeGroups(plan)
+	result, err := u.mergeGroups(ctx, plan)
+	if err != nil {
+		return purge.RunResult{}, err
+	}
 	if err := u.recordRun(result); err != nil {
 		u.logger.Printf("purge: could not record the run result: %v", err)
 		return purge.RunResult{}, err
@@ -77,11 +80,14 @@ func (u *RunPurgeUseCase) Execute(fingerprint *string) (purge.RunResult, error)
 	return result, nil
 }
 
-func (u *RunPurgeUseCase) mergeGroups(plan purgePlan) purge.RunResult {
+func (u *RunPurgeUseCase) mergeGroups(ctx context.Context, plan purgePlan) (purge.RunResult, error) {
 	result := purge.RunResult{GroupsTotal: len(plan.groups)}
 	updated := map[uint32]bool{}
 	for _, group := range plan.groups {
-		merged, err := u.merge.merge(group.kept.ID, slices.Clone(group.mergedIds))
+		if err := ctx.Err(); err != nil {
+			return purge.RunResult{}, err
+		}
+		merged, err := u.merge.merge(ctx, group.kept.ID, slices.Clone(group.mergedIds), plan.entries)
 		if err != nil {
 			result.FailedGroups++
 			continue
@@ -95,7 +101,7 @@ func (u *RunPurgeUseCase) mergeGroups(plan purgePlan) purge.RunResult {
 		}
 	}
 	result.ArticlesUpdated = len(updated)
-	return result
+	return result, nil
 }
 
 func (u *RunPurgeUseCase) recordRun(result purge.RunResult) error {

internal/infrastructure/useCases/entryMedia.go

@@ -6,6 +6,7 @@ import (
 	"RenewCMS/internal/domain/image"
 	"cmp"
 	"errors"
+	"maps"
 	"slices"
 )
 
@@ -30,16 +31,12 @@ func newMediaResolver(images image.Repository) *mediaResolver {
 
 func (r *mediaResolver) resolve(owner contenttype.ContentType, record entry.Entry) (map[string]EntryMedia, error) {
 	resolved := map[string]EntryMedia{}
-	for _, field := range contenttype.MediaFields(owner.Fields) {
-		id, set := contenttype.MediaID(record.Data[field.Key])
-		if !set {
-			continue
-		}
-		media, err := r.lookup(id)
+	for _, ref := range contenttype.MediaRefs(owner.Fields, record.Data) {
+		media, err := r.lookup(ref.ID)
 		if err != nil {
 			return nil, err
 		}
-		resolved[field.Key] = media
+		resolved[ref.Field.Key] = media
 	}
 	return resolved, nil
 }
@@ -73,6 +70,63 @@ type mediaEntry struct {
 	record entry.Entry
 }
 
+func (m mediaEntry) keysPointingAt(ids map[uint32]bool) []string {
+	keys := []string{}
+	for _, ref := range contenttype.MediaRefs(m.owner.Fields, m.record.Data) {
+		if ids[ref.ID] {
+			keys = append(keys, ref.Field.Key)
+		}
+	}
+	return keys
+}
+
+func compareMediaEntries(a, b mediaEntry) int {
+	return cmp.Or(cmp.Compare(a.owner.Name, b.owner.Name), cmp.Compare(a.owner.ID, b.owner.ID), cmp.Compare(a.record.ID, b.record.ID))
+}
+
+type entryMediaIndex struct {
+	byImage map[uint32][]mediaEntry
+}
+
+func newEntryMediaIndex(found []mediaEntry) *entryMediaIndex {
+	index := &entryMediaIndex{byImage: map[uint32][]mediaEntry{}}
+	for _, candidate := range found {
+		distinct := map[uint32]bool{}
+		for _, ref := range contenttype.MediaRefs(candidate.owner.Fields, candidate.record.Data) {
+			distinct[ref.ID] = true
+		}
+		for id := range distinct {
+			index.byImage[id] = append(index.byImage[id], candidate)
+		}
+	}
+	return index
+}
+
+func (i *entryMediaIndex) usageCounts() map[uint32]int {
+	counts := make(map[uint32]int, len(i.byImage))
+	for id, found := range i.byImage {
+		counts[id] = len(found)
+	}
+	return counts
+}
+
+func (i *entryMediaIndex) referencing(imageIDs []uint32) []mediaEntry {
+	type entryKey struct{ typeID, entryID uint32 }
+	seen := map[entryKey]bool{}
+	var found []mediaEntry
+	for _, id := range imageIDs {
+		for _, candidate := range i.byImage[id] {
+			key := entryKey{candidate.owner.ID, candidate.record.ID}
+			if !seen[key] {
+				seen[key] = true
+				found = append(found, candidate)
+			}
+		}
+	}
+	slices.SortFunc(found, compareMediaEntries)
+	return found
+}
+
 type entryMediaScanner struct {
 	contentTypes contenttype.Repository
 	entries      entry.Repository
@@ -82,7 +136,7 @@ func newEntryMediaScanner(contentTypes contenttype.Repository, entries entry.Rep
 	return &entryMediaScanner{contentTypes: contentTypes, entries: entries}
 }
 
-func (s *entryMediaScanner) scan() ([]mediaEntry, error) {
+func (s *entryMediaScanner) index() (*entryMediaIndex, error) {
 	types, err := s.contentTypes.List()
 	if err != nil {
 		return nil, err
@@ -100,91 +154,51 @@ func (s *entryMediaScanner) scan() ([]mediaEntry, error) {
 			found = append(found, mediaEntry{owner: owner, record: record})
 		}
 	}
-	slices.SortFunc(found, func(a, b mediaEntry) int {
-		return cmp.Or(cmp.Compare(a.owner.Name, b.owner.Name), cmp.Compare(a.owner.ID, b.owner.ID), cmp.Compare(a.record.ID, b.record.ID))
-	})
-	return found, nil
-}
-
-func (m mediaEntry) keysPointingAt(ids map[uint32]bool) []string {
-	keys := []string{}
-	for _, field := range contenttype.MediaFields(m.owner.Fields) {
-		if id, set := contenttype.MediaID(m.record.Data[field.Key]); set && ids[id] {
-			keys = append(keys, field.Key)
-		}
-	}
-	return keys
+	return newEntryMediaIndex(found), nil
 }
 
 func (s *entryMediaScanner) usageCounts() (map[uint32]int, error) {
-	found, err := s.scan()
+	index, err := s.index()
 	if err != nil {
 		return nil, err
 	}
-	counts := map[uint32]int{}
-	for _, candidate := range found {
-		distinct := map[uint32]bool{}
-		for _, id := range entry.MediaIDs(candidate.owner.Fields, candidate.record.Data) {
-			distinct[id] = true
-		}
-		for id := range distinct {
-			counts[id]++
-		}
-	}
-	return counts, nil
+	return index.usageCounts(), nil
 }
 
 func (s *entryMediaScanner) referencing(imageID uint32) ([]mediaEntry, error) {
-	found, err := s.scan()
+	index, err := s.index()
 	if err != nil {
 		return nil, err
 	}
-	targets := map[uint32]bool{imageID: true}
-	return slices.DeleteFunc(found, func(candidate mediaEntry) bool {
-		return len(candidate.keysPointingAt(targets)) == 0
-	}), nil
+	return index.referencing([]uint32{imageID}), nil
 }
 
-type entryRepoint struct {
-	original entry.Entry
-	updated  entry.Entry
-}
-
-func (s *entryMediaScanner) repoint(copyIDs []uint32, keptID uint32) ([]entryRepoint, error) {
-	found, err := s.scan()
-	if err != nil {
-		return nil, err
-	}
+func repointEntries(entries entry.Repository, candidates []mediaEntry, copyIDs []uint32, keptID uint32) (int, error) {
 	copies := map[uint32]bool{}
 	for _, id := range copyIDs {
 		copies[id] = true
 	}
-	applied := []entryRepoint{}
-	for _, candidate := range found {
-		keys := candidate.keysPointingAt(copies)
+	updated := 0
+	for _, candidate := range candidates {
+		current, err := entries.Get(candidate.owner.ID, candidate.record.ID)
+		if errors.Is(err, entry.ErrEntryNotFound) {
+			continue
+		}
+		if err != nil {
+			return updated, err
+		}
+		keys := mediaEntry{owner: candidate.owner, record: current}.keysPointingAt(copies)
 		if len(keys) == 0 {
 			continue
 		}
-		change := entryRepoint{original: candidate.record, updated: candidate.record}
-		change.updated.Data = make(map[string]any, len(candidate.record.Data))
-		for key, value := range candidate.record.Data {
-			change.updated.Data[key] = value
-		}
+		current.Data = maps.Clone(current.Data)
 		for _, key := range keys {
-			change.updated.Data[key] = float64(keptID)
+			current.Data[key] = float64(keptID)
 		}
-		if _, err := s.entries.Update(change.updated); err != nil {
-			s.restore(applied)
-			return nil, err
+		if _, err := entries.Update(current); err != nil {
+			return updated, err
 		}
-		applied = append(applied, change)
-	}
-	return applied, nil
-}
-
-func (s *entryMediaScanner) restore(applied []entryRepoint) {
-	for _, change := range applied {
-		_, err := s.entries.Update(change.original)
-		logCompensation("restore entry media", err)
+		updated++
 	}
+	return updated, nil
 }

internal/infrastructure/useCases/entrySupport.go

@@ -24,26 +24,18 @@ func typedEntry(owner contenttype.ContentType, record entry.Entry, resolver *med
 
 func ensureMediaExists(images image.Repository, owner contenttype.ContentType, data map[string]any, stored map[string]any) error {
 	storedIDs := entry.MediaIDs(owner.Fields, stored)
-	for _, field := range contenttype.MediaFields(owner.Fields) {
-		id, set := contenttype.MediaID(data[field.Key])
-		if !set {
+	for _, ref := range contenttype.MediaRefs(owner.Fields, data) {
+		unchanged := stored != nil && storedIDs[ref.Field.Key] == ref.ID
+		if unchanged && !ref.Field.Required {
 			continue
 		}
-		unchanged := stored != nil && storedIDs[field.Key] == id
-		if unchanged && !field.Required {
-			continue
-		}
-		exists, err := imageExists(images, id)
+		exists, err := imageExists(images, ref.ID)
 		if err != nil {
 			return err
 		}
-		if exists {
-			continue
+		if !exists {
+			return entry.MissingMedia(ref.Field, unchanged)
 		}
-		if unchanged {
-			return entry.InvalidField(field.Key, entry.RuleRequired)
-		}
-		return entry.InvalidField(field.Key, contenttype.RuleType)
 	}
 	return nil
 }

internal/infrastructure/useCases/mediaRemover.go

@@ -1,6 +1,9 @@
 package useCases
 
-import "RenewCMS/internal/domain/image"
+import (
+	"RenewCMS/internal/domain/image"
+	"errors"
+)
 
 type mediaRemover struct {
 	imageRepository image.Repository
@@ -22,3 +25,8 @@ func (r mediaRemover) remove(img image.Image) error {
 	}
 	return r.imageRepository.Delete(img.ID)
 }
+
+func (r mediaRemover) discardFiles(img image.Image) error {
+	storedName := image.StoredName(img.Path)
+	return errors.Join(r.generated.purgeSource(storedName), r.storage.Delete(storedName))
+}

internal/infrastructure/useCases/purgePlan.go

@@ -23,6 +23,7 @@ type purgePlan struct {
 	copies      int
 	articles    int
 	fingerprint string
+	entries     *entryMediaIndex
 }
 
 type purgePlanner struct {
@@ -43,7 +44,15 @@ func (p *purgePlanner) plan() (purgePlan, error) {
 		return purgePlan{}, err
 	}
 
-	siteUsage, err := imageUsageCount(p.sites, p.authors, p.entries)
+	var entryIndex *entryMediaIndex
+	siteUsage, err := combinedImageUsage(p.sites, p.authors, func() (map[uint32]int, error) {
+		built, err := p.entries.index()
+		if err != nil {
+			return nil, err
+		}
+		entryIndex = built
+		return built.usageCounts(), nil
+	})
 	if err != nil {
 		return purgePlan{}, err
 	}
@@ -64,7 +73,7 @@ func (p *purgePlanner) plan() (purgePlan, error) {
 	}
 	slices.SortFunc(groups, func(a, b purgeGroup) int { return int(a.kept.ID) - int(b.kept.ID) })
 
-	return purgePlan{groups: groups, copies: copies, articles: len(touched), fingerprint: fingerprintOf(groups)}, nil
+	return purgePlan{groups: groups, copies: copies, articles: len(touched), fingerprint: fingerprintOf(groups), entries: entryIndex}, nil
 }
 
 func groupByHash(images []image.Image) [][]image.Image {

internal/infrastructure/useCases/siteImageUsage.go

@@ -20,6 +20,10 @@ func siteImageUsage(sites site.Repository) (map[uint32]int, error) {
 }
 
 func imageUsageCount(sites site.Repository, authors author.Repository, entries *entryMediaScanner) (map[uint32]int, error) {
+	return combinedImageUsage(sites, authors, entries.usageCounts)
+}
+
+func combinedImageUsage(sites site.Repository, authors author.Repository, entryCounts func() (map[uint32]int, error)) (map[uint32]int, error) {
 	usage, err := siteImageUsage(sites)
 	if err != nil {
 		return nil, err
@@ -31,11 +35,11 @@ func imageUsageCount(sites site.Repository, authors author.Repository, entries *
 	for imageID, total := range avatars {
 		usage[imageID] += total
 	}
-	entryCounts, err := entries.usageCounts()
+	counts, err := entryCounts()
 	if err != nil {
 		return nil, err
 	}
-	for imageID, total := range entryCounts {
+	for imageID, total := range counts {
 		usage[imageID] += total
 	}
 	return usage, nil