[chore] media and emoji refactoring (#3000)

* start updating media manager interface ready for storing attachments / emoji right away

* store emoji and media as uncached immediately, then (re-)cache on Processing{}.Load()

* remove now unused media workers

* fix tests and issues

* fix another test!

* fix emoji activitypub uri setting behaviour, fix remainder of test compilation issues

* fix more tests

* fix (most of) remaining tests, add debouncing to repeatedly failing media / emojis

* whoops, rebase issue

* remove kim's whacky experiments

* do some reshuffling, ensure emoji uri gets set

* ensure marked as not cached on cleanup

* tweaks to media / emoji processing to handle context canceled better

* ensure newly fetched emojis actually get set in returned slice

* use different varnames to be a bit more obvious

* move emoji refresh rate limiting to dereferencer

* add exported dereferencer functions for remote media, use these for recaching in processor

* add check for nil attachment in updateAttachment()

* remove unused emoji and media fields + columns

* see previous commit

* fix old migrations expecting image_updated_at to exists (from copies of old models)

* remove freshness checking code (seems to be broken...)

* fix error arg causing nil ptr exception

* finish documentating functions with comments, slight tweaks to media / emoji deref error logic

* remove some extra unneeded boolean checking

* finish writing documentation (code comments) for exported media manager methods

* undo changes to migration snapshot gtsmodels, updated failing migration to have its own snapshot

* move doesColumnExist() to util.go in migrations package

internal/processing/account/account_test.go

@@ -111,7 +111,7 @@ func (suite *AccountStandardTestSuite) SetupTest() {
 	suite.emailSender = testrig.NewEmailSender("../../../web/template/", suite.sentEmails)
 
 	filter := visibility.NewFilter(&suite.state)
-	common := common.New(&suite.state, suite.tc, suite.federator, filter)
+	common := common.New(&suite.state, suite.mediaManager, suite.tc, suite.federator, filter)
 	suite.accountProcessor = account.New(&common, &suite.state, suite.tc, suite.mediaManager, suite.federator, filter, processing.GetParseMentionFunc(&suite.state, suite.federator))
 	testrig.StandardDBSetup(suite.db, nil)
 	testrig.StandardStorageSetup(suite.storage, "../../../testrig/media")

internal/processing/account/update.go

@@ -19,10 +19,12 @@ package account
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"io"
 	"mime/multipart"
 
+	"codeberg.org/gruf/go-bytesize"
 	"github.com/superseriousbusiness/gotosocial/internal/ap"
 	apimodel "github.com/superseriousbusiness/gotosocial/internal/api/model"
 	"github.com/superseriousbusiness/gotosocial/internal/config"
@@ -203,9 +205,13 @@ func (p *Processor) Update(ctx context.Context, account *gtsmodel.Account, form
 	}
 
 	if form.Avatar != nil && form.Avatar.Size != 0 {
-		avatarInfo, err := p.UpdateAvatar(ctx, form.Avatar, nil, account.ID)
-		if err != nil {
-			return nil, gtserror.NewErrorBadRequest(err)
+		avatarInfo, errWithCode := p.UpdateAvatar(ctx,
+			account,
+			form.Avatar,
+			nil,
+		)
+		if errWithCode != nil {
+			return nil, errWithCode
 		}
 		account.AvatarMediaAttachmentID = avatarInfo.ID
 		account.AvatarMediaAttachment = avatarInfo
@@ -213,9 +219,13 @@ func (p *Processor) Update(ctx context.Context, account *gtsmodel.Account, form
 	}
 
 	if form.Header != nil && form.Header.Size != 0 {
-		headerInfo, err := p.UpdateHeader(ctx, form.Header, nil, account.ID)
-		if err != nil {
-			return nil, gtserror.NewErrorBadRequest(err)
+		headerInfo, errWithCode := p.UpdateHeader(ctx,
+			account,
+			form.Header,
+			nil,
+		)
+		if errWithCode != nil {
+			return nil, errWithCode
 		}
 		account.HeaderMediaAttachmentID = headerInfo.ID
 		account.HeaderMediaAttachment = headerInfo
@@ -316,35 +326,33 @@ func (p *Processor) Update(ctx context.Context, account *gtsmodel.Account, form
 // for this to become the account's new avatar.
 func (p *Processor) UpdateAvatar(
 	ctx context.Context,
+	account *gtsmodel.Account,
 	avatar *multipart.FileHeader,
 	description *string,
-	accountID string,
-) (*gtsmodel.MediaAttachment, error) {
-	maxImageSize := config.GetMediaImageMaxSize()
-	if avatar.Size > int64(maxImageSize) {
-		return nil, gtserror.Newf("size %d exceeded max media size of %d bytes", avatar.Size, maxImageSize)
+) (
+	*gtsmodel.MediaAttachment,
+	gtserror.WithCode,
+) {
+	max := config.GetMediaImageMaxSize()
+	if sz := bytesize.Size(avatar.Size); sz > max {
+		text := fmt.Sprintf("size %s exceeds max media size %s", sz, max)
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 
-	data := func(innerCtx context.Context) (io.ReadCloser, int64, error) {
+	data := func(_ context.Context) (io.ReadCloser, int64, error) {
 		f, err := avatar.Open()
 		return f, avatar.Size, err
 	}
 
-	// Process the media attachment and load it immediately.
-	media := p.mediaManager.PreProcessMedia(data, accountID, &media.AdditionalMediaInfo{
-		Avatar:      util.Ptr(true),
-		Description: description,
-	})
-
-	attachment, err := media.LoadAttachment(ctx)
-	if err != nil {
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
-	} else if attachment.Type == gtsmodel.FileTypeUnknown {
-		err = gtserror.Newf("could not process uploaded file with extension %s", attachment.File.ContentType)
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
-	}
-
-	return attachment, nil
+	// Write to instance storage.
+	return p.c.StoreLocalMedia(ctx,
+		account.ID,
+		data,
+		media.AdditionalMediaInfo{
+			Avatar:      util.Ptr(true),
+			Description: description,
+		},
+	)
 }
 
 // UpdateHeader does the dirty work of checking the header
@@ -353,33 +361,31 @@ func (p *Processor) UpdateAvatar(
 // for this to become the account's new header.
 func (p *Processor) UpdateHeader(
 	ctx context.Context,
+	account *gtsmodel.Account,
 	header *multipart.FileHeader,
 	description *string,
-	accountID string,
-) (*gtsmodel.MediaAttachment, error) {
-	maxImageSize := config.GetMediaImageMaxSize()
-	if header.Size > int64(maxImageSize) {
-		return nil, gtserror.Newf("size %d exceeded max media size of %d bytes", header.Size, maxImageSize)
+) (
+	*gtsmodel.MediaAttachment,
+	gtserror.WithCode,
+) {
+	max := config.GetMediaImageMaxSize()
+	if sz := bytesize.Size(header.Size); sz > max {
+		text := fmt.Sprintf("size %s exceeds max media size %s", sz, max)
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 
-	data := func(innerCtx context.Context) (io.ReadCloser, int64, error) {
+	data := func(_ context.Context) (io.ReadCloser, int64, error) {
 		f, err := header.Open()
 		return f, header.Size, err
 	}
 
-	// Process the media attachment and load it immediately.
-	media := p.mediaManager.PreProcessMedia(data, accountID, &media.AdditionalMediaInfo{
-		Header:      util.Ptr(true),
-		Description: description,
-	})
-
-	attachment, err := media.LoadAttachment(ctx)
-	if err != nil {
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
-	} else if attachment.Type == gtsmodel.FileTypeUnknown {
-		err = gtserror.Newf("could not process uploaded file with extension %s", attachment.File.ContentType)
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
-	}
-
-	return attachment, nil
+	// Write to instance storage.
+	return p.c.StoreLocalMedia(ctx,
+		account.ID,
+		data,
+		media.AdditionalMediaInfo{
+			Header:      util.Ptr(true),
+			Description: description,
+		},
+	)
 }

internal/processing/admin/admin.go

@@ -20,20 +20,26 @@ package admin
 import (
 	"github.com/superseriousbusiness/gotosocial/internal/cleaner"
 	"github.com/superseriousbusiness/gotosocial/internal/email"
+	"github.com/superseriousbusiness/gotosocial/internal/federation"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
+	"github.com/superseriousbusiness/gotosocial/internal/processing/common"
 	"github.com/superseriousbusiness/gotosocial/internal/state"
 	"github.com/superseriousbusiness/gotosocial/internal/transport"
 	"github.com/superseriousbusiness/gotosocial/internal/typeutils"
 )
 
 type Processor struct {
-	state               *state.State
-	cleaner             *cleaner.Cleaner
-	converter           *typeutils.Converter
-	mediaManager        *media.Manager
-	transportController transport.Controller
-	emailSender         email.Sender
+	// common processor logic
+	c *common.Processor
+
+	state     *state.State
+	cleaner   *cleaner.Cleaner
+	converter *typeutils.Converter
+	federator *federation.Federator
+	media     *media.Manager
+	transport transport.Controller
+	email     email.Sender
 
 	// admin Actions currently
 	// undergoing processing
@@ -46,21 +52,24 @@ func (p *Processor) Actions() *Actions {
 
 // New returns a new admin processor.
 func New(
+	common *common.Processor,
 	state *state.State,
 	cleaner *cleaner.Cleaner,
+	federator *federation.Federator,
 	converter *typeutils.Converter,
 	mediaManager *media.Manager,
 	transportController transport.Controller,
 	emailSender email.Sender,
 ) Processor {
 	return Processor{
-		state:               state,
-		cleaner:             cleaner,
-		converter:           converter,
-		mediaManager:        mediaManager,
-		transportController: transportController,
-		emailSender:         emailSender,
-
+		c:         common,
+		state:     state,
+		cleaner:   cleaner,
+		converter: converter,
+		federator: federator,
+		media:     mediaManager,
+		transport: transportController,
+		email:     emailSender,
 		actions: &Actions{
 			r:     make(map[string]*gtsmodel.AdminAction),
 			state: state,

internal/processing/admin/debug_apurl.go

@@ -78,7 +78,7 @@ func (p *Processor) DebugAPUrl(
 	}
 
 	// All looks fine. Prepare the transport and (signed) GET request.
-	tsport, err := p.transportController.NewTransportForUsername(ctx, adminAcct.Username)
+	tsport, err := p.transport.NewTransportForUsername(ctx, adminAcct.Username)
 	if err != nil {
 		err = gtserror.Newf("error creating transport: %w", err)
 		return nil, gtserror.NewErrorInternalError(err, err.Error())

internal/processing/admin/email.go

@@ -55,7 +55,7 @@ func (p *Processor) EmailTest(
 		InstanceName:    instance.Title,
 	}
 
-	if err := p.emailSender.SendTestEmail(toAddress, testData); err != nil {
+	if err := p.email.SendTestEmail(toAddress, testData); err != nil {
 		if gtserror.IsSMTP(err) {
 			// An error occurred during the SMTP part.
 			// We should indicate this to the caller, as

internal/processing/admin/emoji.go

@@ -31,7 +31,6 @@ import (
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
 	"github.com/superseriousbusiness/gotosocial/internal/id"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
-	"github.com/superseriousbusiness/gotosocial/internal/uris"
 	"github.com/superseriousbusiness/gotosocial/internal/util"
 )
 
@@ -41,64 +40,21 @@ func (p *Processor) EmojiCreate(
 	account *gtsmodel.Account,
 	form *apimodel.EmojiCreateRequest,
 ) (*apimodel.Emoji, gtserror.WithCode) {
-	// Ensure emoji with this shortcode
-	// doesn't already exist on the instance.
-	maybeExisting, err := p.state.DB.GetEmojiByShortcodeDomain(ctx, form.Shortcode, "")
-	if err != nil && !errors.Is(err, db.ErrNoEntries) {
-		err := gtserror.Newf("error checking existence of emoji with shortcode %s: %w", form.Shortcode, err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
 
-	if maybeExisting != nil {
-		err := fmt.Errorf("emoji with shortcode %s already exists", form.Shortcode)
-		return nil, gtserror.NewErrorConflict(err, err.Error())
-	}
-
-	// Prepare data function for emoji processing
-	// (just read data from the submitted form).
-	data := func(innerCtx context.Context) (io.ReadCloser, int64, error) {
+	// Simply read provided form data for emoji data source.
+	data := func(_ context.Context) (io.ReadCloser, int64, error) {
 		f, err := form.Image.Open()
 		return f, form.Image.Size, err
 	}
 
-	// If category was supplied on the form,
-	// ensure the category exists and provide
-	// it as additional info to emoji processing.
-	var ai *media.AdditionalEmojiInfo
-	if form.CategoryName != "" {
-		category, err := p.getOrCreateEmojiCategory(ctx, form.CategoryName)
-		if err != nil {
-			return nil, gtserror.NewErrorInternalError(err)
-		}
-
-		ai = &media.AdditionalEmojiInfo{
-			CategoryID: &category.ID,
-		}
-	}
-
-	// Generate new emoji ID and URI.
-	emojiID, err := id.NewRandomULID()
-	if err != nil {
-		err := gtserror.Newf("error creating id for new emoji: %w", err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	emojiURI := uris.URIForEmoji(emojiID)
-
-	// Begin media processing.
-	processingEmoji, err := p.mediaManager.PreProcessEmoji(ctx,
-		data, form.Shortcode, emojiID, emojiURI, ai, false,
+	// Attempt to create the new local emoji.
+	emoji, errWithCode := p.createEmoji(ctx,
+		form.Shortcode,
+		form.CategoryName,
+		data,
 	)
-	if err != nil {
-		err := gtserror.Newf("error processing emoji: %w", err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	// Complete processing immediately.
-	emoji, err := processingEmoji.LoadEmoji(ctx)
-	if err != nil {
-		err := gtserror.Newf("error loading emoji: %w", err)
-		return nil, gtserror.NewErrorInternalError(err)
+	if errWithCode != nil {
+		return nil, errWithCode
 	}
 
 	apiEmoji, err := p.converter.EmojiToAPIEmoji(ctx, emoji)
@@ -110,53 +66,6 @@ func (p *Processor) EmojiCreate(
 	return &apiEmoji, nil
 }
 
-// emojisGetFilterParams builds extra
-// query parameters to return as part
-// of an Emojis pageable response.
-//
-// The returned string will look like:
-//
-// "filter=domain:all,enabled,shortcode:example"
-func emojisGetFilterParams(
-	shortcode string,
-	domain string,
-	includeDisabled bool,
-	includeEnabled bool,
-) string {
-	var filterBuilder strings.Builder
-	filterBuilder.WriteString("filter=")
-
-	switch domain {
-	case "", "local":
-		// Local emojis only.
-		filterBuilder.WriteString("domain:local")
-
-	case db.EmojiAllDomains:
-		// Local or remote.
-		filterBuilder.WriteString("domain:all")
-
-	default:
-		// Specific domain only.
-		filterBuilder.WriteString("domain:" + domain)
-	}
-
-	if includeDisabled != includeEnabled {
-		if includeDisabled {
-			filterBuilder.WriteString(",disabled")
-		}
-		if includeEnabled {
-			filterBuilder.WriteString(",enabled")
-		}
-	}
-
-	if shortcode != "" {
-		// Specific shortcode only.
-		filterBuilder.WriteString(",shortcode:" + shortcode)
-	}
-
-	return filterBuilder.String()
-}
-
 // EmojisGet returns an admin view of custom
 // emojis, filtered with the given parameters.
 func (p *Processor) EmojisGet(
@@ -287,21 +196,24 @@ func (p *Processor) EmojiDelete(
 // given id, using the provided form parameters.
 func (p *Processor) EmojiUpdate(
 	ctx context.Context,
-	id string,
+	emojiID string,
 	form *apimodel.EmojiUpdateRequest,
 ) (*apimodel.AdminEmoji, gtserror.WithCode) {
-	emoji, err := p.state.DB.GetEmojiByID(ctx, id)
+
+	// Get the emoji with given ID from the database.
+	emoji, err := p.state.DB.GetEmojiByID(ctx, emojiID)
 	if err != nil && !errors.Is(err, db.ErrNoEntries) {
-		err := gtserror.Newf("db error: %w", err)
+		err := gtserror.Newf("error fetching emoji from db: %w", err)
 		return nil, gtserror.NewErrorInternalError(err)
 	}
 
+	// Check found.
 	if emoji == nil {
-		err := gtserror.Newf("no emoji with id %s found in the db", id)
-		return nil, gtserror.NewErrorNotFound(err)
+		const text = "emoji not found"
+		return nil, gtserror.NewErrorNotFound(errors.New(text), text)
 	}
 
-	switch t := form.Type; t {
+	switch form.Type {
 
 	case apimodel.EmojiUpdateCopy:
 		return p.emojiUpdateCopy(ctx, emoji, form.Shortcode, form.CategoryName)
@@ -313,8 +225,8 @@ func (p *Processor) EmojiUpdate(
 		return p.emojiUpdateModify(ctx, emoji, form.Image, form.CategoryName)
 
 	default:
-		err := fmt.Errorf("unrecognized emoji action type %s", t)
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
+		const text = "unrecognized emoji update action type"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 }
 
@@ -342,56 +254,6 @@ func (p *Processor) EmojiCategoriesGet(
 	return apiCategories, nil
 }
 
-/*
-	UTIL FUNCTIONS
-*/
-
-// getOrCreateEmojiCategory either gets an existing
-// category with the given name from the database,
-// or, if the category doesn't yet exist, it creates
-// the category and then returns it.
-func (p *Processor) getOrCreateEmojiCategory(
-	ctx context.Context,
-	name string,
-) (*gtsmodel.EmojiCategory, error) {
-	category, err := p.state.DB.GetEmojiCategoryByName(ctx, name)
-	if err != nil && !errors.Is(err, db.ErrNoEntries) {
-		return nil, gtserror.Newf(
-			"database error trying get emoji category %s: %w",
-			name, err,
-		)
-	}
-
-	if category != nil {
-		// We had it already.
-		return category, nil
-	}
-
-	// We don't have the category yet,
-	// create it with the given name.
-	categoryID, err := id.NewRandomULID()
-	if err != nil {
-		return nil, gtserror.Newf(
-			"error generating id for new emoji category %s: %w",
-			name, err,
-		)
-	}
-
-	category = &gtsmodel.EmojiCategory{
-		ID:   categoryID,
-		Name: name,
-	}
-
-	if err := p.state.DB.PutEmojiCategory(ctx, category); err != nil {
-		return nil, gtserror.Newf(
-			"db error putting new emoji category %s: %w",
-			name, err,
-		)
-	}
-
-	return category, nil
-}
-
 // emojiUpdateCopy copies and stores the given
 // *remote* emoji as a *local* emoji, preserving
 // the same image, and using the provided shortcode.
@@ -400,99 +262,56 @@ func (p *Processor) getOrCreateEmojiCategory(
 // emoji already stored in the database + storage.
 func (p *Processor) emojiUpdateCopy(
 	ctx context.Context,
-	targetEmoji *gtsmodel.Emoji,
+	target *gtsmodel.Emoji,
 	shortcode *string,
-	category *string,
+	categoryName *string,
 ) (*apimodel.AdminEmoji, gtserror.WithCode) {
-	if targetEmoji.IsLocal() {
-		err := fmt.Errorf("emoji %s is not a remote emoji, cannot copy it to local", targetEmoji.ID)
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
+	if target.IsLocal() {
+		const text = "target emoji is not remote; cannot copy to local"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 
-	if shortcode == nil {
-		err := errors.New("no shortcode provided")
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
-	}
+	// Ensure target emoji is locally cached.
+	target, err := p.federator.RefreshEmoji(
+		ctx,
+		target,
 
-	sc := *shortcode
-	if sc == "" {
-		err := errors.New("empty shortcode provided")
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
+		// no changes we want to make.
+		media.AdditionalEmojiInfo{},
+		false,
+	)
+	if err != nil {
+		err := gtserror.Newf("error recaching emoji %s: %w", target.ImageRemoteURL, err)
+		return nil, gtserror.NewErrorNotFound(err)
 	}
 
-	// Ensure we don't already have an emoji
-	// stored locally with this shortcode.
-	maybeExisting, err := p.state.DB.GetEmojiByShortcodeDomain(ctx, sc, "")
-	if err != nil && !errors.Is(err, db.ErrNoEntries) {
-		err := gtserror.Newf("db error checking for emoji with shortcode %s: %w", sc, err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	if maybeExisting != nil {
-		err := fmt.Errorf("emoji with shortcode %s already exists on this instance", sc)
-		return nil, gtserror.NewErrorConflict(err, err.Error())
-	}
-
-	// We don't have an emoji with this
-	// shortcode yet! Prepare to create it.
-
 	// Data function for copying just streams media
 	// out of storage into an additional location.
 	//
 	// This means that data for the copy persists even
 	// if the remote copied emoji gets deleted at some point.
 	data := func(ctx context.Context) (io.ReadCloser, int64, error) {
-		rc, err := p.state.Storage.GetStream(ctx, targetEmoji.ImagePath)
-		return rc, int64(targetEmoji.ImageFileSize), err
+		rc, err := p.state.Storage.GetStream(ctx, target.ImagePath)
+		return rc, int64(target.ImageFileSize), err
 	}
 
-	// Generate new emoji ID and URI.
-	emojiID, err := id.NewRandomULID()
-	if err != nil {
-		err := gtserror.Newf("error creating id for new emoji: %w", err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	emojiURI := uris.URIForEmoji(emojiID)
-
-	// If category was supplied, ensure the
-	// category exists and provide it as
-	// additional info to emoji processing.
-	var ai *media.AdditionalEmojiInfo
-	if category != nil && *category != "" {
-		category, err := p.getOrCreateEmojiCategory(ctx, *category)
-		if err != nil {
-			return nil, gtserror.NewErrorInternalError(err)
-		}
-
-		ai = &media.AdditionalEmojiInfo{
-			CategoryID: &category.ID,
-		}
-	}
-
-	// Begin media processing.
-	processingEmoji, err := p.mediaManager.PreProcessEmoji(ctx,
-		data, sc, emojiID, emojiURI, ai, false,
+	// Attempt to create the new local emoji.
+	emoji, errWithCode := p.createEmoji(ctx,
+		util.PtrValueOr(shortcode, ""),
+		util.PtrValueOr(categoryName, ""),
+		data,
 	)
+	if errWithCode != nil {
+		return nil, errWithCode
+	}
+
+	apiEmoji, err := p.converter.EmojiToAdminAPIEmoji(ctx, emoji)
 	if err != nil {
-		err := gtserror.Newf("error processing emoji: %w", err)
+		err := gtserror.Newf("error converting emoji: %w", err)
 		return nil, gtserror.NewErrorInternalError(err)
 	}
 
-	// Complete processing immediately.
-	newEmoji, err := processingEmoji.LoadEmoji(ctx)
-	if err != nil {
-		err := gtserror.Newf("error loading emoji: %w", err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	adminEmoji, err := p.converter.EmojiToAdminAPIEmoji(ctx, newEmoji)
-	if err != nil {
-		err := gtserror.Newf("error converting emoji %s to admin emoji: %w", newEmoji.ID, err)
-		return nil, gtserror.NewErrorInternalError(err)
-	}
-
-	return adminEmoji, nil
+	return apiEmoji, nil
 }
 
 // emojiUpdateDisable marks the given *remote*
@@ -521,7 +340,7 @@ func (p *Processor) emojiUpdateDisable(
 
 	adminEmoji, err := p.converter.EmojiToAdminAPIEmoji(ctx, emoji)
 	if err != nil {
-		err := gtserror.Newf("error converting emoji %s to admin emoji: %w", emoji.ID, err)
+		err := gtserror.Newf("error converting emoji: %w", err)
 		return nil, gtserror.NewErrorInternalError(err)
 	}
 
@@ -541,104 +360,222 @@ func (p *Processor) emojiUpdateModify(
 	ctx context.Context,
 	emoji *gtsmodel.Emoji,
 	image *multipart.FileHeader,
-	category *string,
+	categoryName *string,
 ) (*apimodel.AdminEmoji, gtserror.WithCode) {
 	if !emoji.IsLocal() {
-		err := fmt.Errorf("emoji %s is not a local emoji, cannot update it via this endpoint", emoji.ID)
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
+		const text = "cannot modify remote emoji"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 
 	// Ensure there's actually something to update.
-	if image == nil && category == nil {
-		err := errors.New("neither new category nor new image set, cannot update")
-		return nil, gtserror.NewErrorBadRequest(err, err.Error())
+	if image == nil && categoryName == nil {
+		const text = "no changes were provided"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
 	}
 
-	// Only update category
-	// if it's changed.
-	var (
-		newCategory      *gtsmodel.EmojiCategory
-		newCategoryID    string
-		updateCategoryID bool
-	)
-
-	if category != nil {
-		catName := *category
-		if catName != "" {
-			// Set new category.
-			var err error
-			newCategory, err = p.getOrCreateEmojiCategory(ctx, catName)
-			if err != nil {
-				err := gtserror.Newf("error getting or creating category: %w", err)
-				return nil, gtserror.NewErrorInternalError(err)
+	if categoryName != nil {
+		if *categoryName != "" {
+			// A category was provided, get / create relevant emoji category.
+			category, errWithCode := p.mustGetEmojiCategory(ctx, *categoryName)
+			if errWithCode != nil {
+				return nil, errWithCode
 			}
 
-			newCategoryID = newCategory.ID
+			if category.ID == emoji.CategoryID {
+				// There was no change,
+				// indicate this by unsetting
+				// the category name pointer.
+				categoryName = nil
+			} else {
+				// Update emoji category.
+				emoji.CategoryID = category.ID
+				emoji.Category = category
+			}
 		} else {
-			// Clear existing category.
-			newCategoryID = ""
+			// Emoji category was unset.
+			emoji.CategoryID = ""
+			emoji.Category = nil
 		}
-
-		updateCategoryID = emoji.CategoryID != newCategoryID
 	}
 
-	// Only update image
-	// if one is provided.
-	var updateImage bool
-	if image != nil && image.Size != 0 {
-		updateImage = true
-	}
+	// Check whether any image changes were requested.
+	imageUpdated := (image != nil && image.Size > 0)
 
-	if updateCategoryID && !updateImage {
-		// Only updating category; we only
-		// need to do a db update for this.
-		emoji.CategoryID = newCategoryID
-		emoji.Category = newCategory
+	if !imageUpdated && categoryName != nil {
+		// Only updating category; only a single database update required.
 		if err := p.state.DB.UpdateEmoji(ctx, emoji, "category_id"); err != nil {
-			err := gtserror.Newf("db error updating emoji %s: %w", emoji.ID, err)
+			err := gtserror.Newf("error updating emoji in db: %w", err)
 			return nil, gtserror.NewErrorInternalError(err)
 		}
-	} else if updateImage {
+	} else if imageUpdated {
+		var err error
+
 		// Updating image and maybe categoryID.
 		// We can do both at the same time :)
 
-		// Set data function to provided image.
-		data := func(ctx context.Context) (io.ReadCloser, int64, error) {
-			i, err := image.Open()
-			return i, image.Size, err
+		// Simply read provided form data for emoji data source.
+		data := func(_ context.Context) (io.ReadCloser, int64, error) {
+			f, err := image.Open()
+			return f, image.Size, err
 		}
 
-		// If necessary, include
-		// update to categoryID too.
-		var ai *media.AdditionalEmojiInfo
-		if updateCategoryID {
-			ai = &media.AdditionalEmojiInfo{
-				CategoryID: &newCategoryID,
-			}
-		}
+		// Prepare emoji model for recache from new data.
+		processing := p.media.RecacheEmoji(emoji, data)
 
-		// Begin media processing.
-		processingEmoji, err := p.mediaManager.PreProcessEmoji(ctx,
-			data, emoji.Shortcode, emoji.ID, emoji.URI, ai, false,
-		)
+		// Load to trigger update + write.
+		emoji, err = processing.Load(ctx)
 		if err != nil {
-			err := gtserror.Newf("error processing emoji: %w", err)
-			return nil, gtserror.NewErrorInternalError(err)
-		}
-
-		// Replace emoji ptr with newly-processed version.
-		emoji, err = processingEmoji.LoadEmoji(ctx)
-		if err != nil {
-			err := gtserror.Newf("error loading emoji: %w", err)
+			err := gtserror.Newf("error processing emoji %s: %w", emoji.Shortcode, err)
 			return nil, gtserror.NewErrorInternalError(err)
 		}
 	}
 
 	adminEmoji, err := p.converter.EmojiToAdminAPIEmoji(ctx, emoji)
 	if err != nil {
-		err := gtserror.Newf("error converting emoji %s to admin emoji: %w", emoji.ID, err)
+		err := gtserror.Newf("error converting emoji: %w", err)
 		return nil, gtserror.NewErrorInternalError(err)
 	}
 
 	return adminEmoji, nil
 }
+
+// createEmoji will create a new local emoji
+// with the given shortcode, attached category
+// name (if any) and data source function.
+func (p *Processor) createEmoji(
+	ctx context.Context,
+	shortcode string,
+	categoryName string,
+	data media.DataFunc,
+) (
+	*gtsmodel.Emoji,
+	gtserror.WithCode,
+) {
+	// Validate shortcode.
+	if shortcode == "" {
+		const text = "empty shortcode name"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
+	}
+
+	// Look for an existing local emoji with shortcode to ensure this is new.
+	existing, err := p.state.DB.GetEmojiByShortcodeDomain(ctx, shortcode, "")
+	if err != nil && !errors.Is(err, db.ErrNoEntries) {
+		err := gtserror.Newf("error fetching emoji from db: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	} else if existing != nil {
+		const text = "emoji with shortcode already exists"
+		return nil, gtserror.NewErrorConflict(errors.New(text), text)
+	}
+
+	var categoryID *string
+
+	if categoryName != "" {
+		// A category was provided, get / create relevant emoji category.
+		category, errWithCode := p.mustGetEmojiCategory(ctx, categoryName)
+		if errWithCode != nil {
+			return nil, errWithCode
+		}
+
+		// Set category ID for emoji.
+		categoryID = &category.ID
+	}
+
+	// Store to instance storage.
+	return p.c.StoreLocalEmoji(
+		ctx,
+		shortcode,
+		data,
+		media.AdditionalEmojiInfo{
+			CategoryID: categoryID,
+		},
+	)
+}
+
+// mustGetEmojiCategory either gets an existing
+// category with the given name from the database,
+// or, if the category doesn't yet exist, it creates
+// the category and then returns it.
+func (p *Processor) mustGetEmojiCategory(
+	ctx context.Context,
+	name string,
+) (
+	*gtsmodel.EmojiCategory,
+	gtserror.WithCode,
+) {
+	// Look for an existing emoji category with name.
+	category, err := p.state.DB.GetEmojiCategoryByName(ctx, name)
+	if err != nil && !errors.Is(err, db.ErrNoEntries) {
+		err := gtserror.Newf("error fetching emoji category from db: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	if category != nil {
+		// We had it already.
+		return category, nil
+	}
+
+	// Create new ID.
+	id := id.NewULID()
+
+	// Prepare new category for insertion.
+	category = &gtsmodel.EmojiCategory{
+		ID:   id,
+		Name: name,
+	}
+
+	// Insert new category into the database.
+	err = p.state.DB.PutEmojiCategory(ctx, category)
+	if err != nil {
+		err := gtserror.Newf("error inserting emoji category into db: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	return category, nil
+}
+
+// emojisGetFilterParams builds extra
+// query parameters to return as part
+// of an Emojis pageable response.
+//
+// The returned string will look like:
+//
+// "filter=domain:all,enabled,shortcode:example"
+func emojisGetFilterParams(
+	shortcode string,
+	domain string,
+	includeDisabled bool,
+	includeEnabled bool,
+) string {
+	var filterBuilder strings.Builder
+	filterBuilder.WriteString("filter=")
+
+	switch domain {
+	case "", "local":
+		// Local emojis only.
+		filterBuilder.WriteString("domain:local")
+
+	case db.EmojiAllDomains:
+		// Local or remote.
+		filterBuilder.WriteString("domain:all")
+
+	default:
+		// Specific domain only.
+		filterBuilder.WriteString("domain:" + domain)
+	}
+
+	if includeDisabled != includeEnabled {
+		if includeDisabled {
+			filterBuilder.WriteString(",disabled")
+		}
+		if includeEnabled {
+			filterBuilder.WriteString(",enabled")
+		}
+	}
+
+	if shortcode != "" {
+		// Specific shortcode only.
+		filterBuilder.WriteString(",shortcode:" + shortcode)
+	}
+
+	return filterBuilder.String()
+}

internal/processing/admin/media.go

@@ -28,7 +28,7 @@ import (
 
 // MediaRefetch forces a refetch of remote emojis.
 func (p *Processor) MediaRefetch(ctx context.Context, requestingAccount *gtsmodel.Account, domain string) gtserror.WithCode {
-	transport, err := p.transportController.NewTransportForUsername(ctx, requestingAccount.Username)
+	transport, err := p.transport.NewTransportForUsername(ctx, requestingAccount.Username)
 	if err != nil {
 		err = fmt.Errorf("error getting transport for user %s during media refetch request: %w", requestingAccount.Username, err)
 		return gtserror.NewErrorInternalError(err)
@@ -36,7 +36,7 @@ func (p *Processor) MediaRefetch(ctx context.Context, requestingAccount *gtsmode
 
 	go func() {
 		log.Info(ctx, "starting emoji refetch")
-		refetched, err := p.mediaManager.RefetchEmojis(context.Background(), domain, transport.DereferenceMedia)
+		refetched, err := p.media.RefetchEmojis(context.Background(), domain, transport.DereferenceMedia)
 		if err != nil {
 			log.Errorf(ctx, "error refetching emojis: %s", err)
 		} else {

internal/processing/common/common.go

@@ -20,6 +20,7 @@ package common
 import (
 	"github.com/superseriousbusiness/gotosocial/internal/federation"
 	"github.com/superseriousbusiness/gotosocial/internal/filter/visibility"
+	"github.com/superseriousbusiness/gotosocial/internal/media"
 	"github.com/superseriousbusiness/gotosocial/internal/state"
 	"github.com/superseriousbusiness/gotosocial/internal/typeutils"
 )
@@ -29,6 +30,7 @@ import (
 // processing subsection of the codebase.
 type Processor struct {
 	state     *state.State
+	media     *media.Manager
 	converter *typeutils.Converter
 	federator *federation.Federator
 	filter    *visibility.Filter
@@ -37,12 +39,14 @@ type Processor struct {
 // New returns a new Processor instance.
 func New(
 	state *state.State,
+	media *media.Manager,
 	converter *typeutils.Converter,
 	federator *federation.Federator,
 	filter *visibility.Filter,
 ) Processor {
 	return Processor{
 		state:     state,
+		media:     media,
 		converter: converter,
 		federator: federator,
 		filter:    filter,

internal/processing/common/media.go

@@ -0,0 +1,98 @@
+// GoToSocial
+// Copyright (C) GoToSocial Authors admin@gotosocial.org
+// SPDX-License-Identifier: AGPL-3.0-or-later
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package common
+
+import (
+	"context"
+	"errors"
+	"fmt"
+
+	"github.com/superseriousbusiness/gotosocial/internal/gtserror"
+	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
+	"github.com/superseriousbusiness/gotosocial/internal/media"
+)
+
+// StoreLocalMedia is a wrapper around CreateMedia() and
+// ProcessingMedia{}.Load() with appropriate error responses.
+func (p *Processor) StoreLocalMedia(
+	ctx context.Context,
+	accountID string,
+	data media.DataFunc,
+	info media.AdditionalMediaInfo,
+) (
+	*gtsmodel.MediaAttachment,
+	gtserror.WithCode,
+) {
+	// Create a new processing media attachment.
+	processing, err := p.media.CreateMedia(ctx,
+		accountID,
+		data,
+		info,
+	)
+	if err != nil {
+		err := gtserror.Newf("error creating media: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	// Immediately trigger write to storage.
+	attachment, err := processing.Load(ctx)
+	if err != nil {
+		const text = "error processing emoji"
+		err := gtserror.Newf("error processing media: %w", err)
+		return nil, gtserror.NewErrorUnprocessableEntity(err, text)
+	} else if attachment.Type == gtsmodel.FileTypeUnknown {
+		text := fmt.Sprintf("could not process %s type media", attachment.File.ContentType)
+		return nil, gtserror.NewErrorUnprocessableEntity(errors.New(text), text)
+	}
+
+	return attachment, nil
+}
+
+// StoreLocalMedia is a wrapper around CreateMedia() and
+// ProcessingMedia{}.Load() with appropriate error responses.
+func (p *Processor) StoreLocalEmoji(
+	ctx context.Context,
+	shortcode string,
+	data media.DataFunc,
+	info media.AdditionalEmojiInfo,
+) (
+	*gtsmodel.Emoji,
+	gtserror.WithCode,
+) {
+	// Create a new processing emoji media.
+	processing, err := p.media.CreateEmoji(ctx,
+		shortcode,
+		"", // domain = "" -> local
+		data,
+		info,
+	)
+	if err != nil {
+		err := gtserror.Newf("error creating emoji: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	// Immediately write to storage.
+	emoji, err := processing.Load(ctx)
+	if err != nil {
+		const text = "error processing emoji"
+		err := gtserror.Newf("error processing emoji %s: %w", shortcode, err)
+		return nil, gtserror.NewErrorUnprocessableEntity(err, text)
+	}
+
+	return emoji, nil
+}

internal/processing/instance.go

@@ -246,9 +246,13 @@ func (p *Processor) InstancePatch(ctx context.Context, form *apimodel.InstanceSe
 
 	if form.Avatar != nil && form.Avatar.Size != 0 {
 		// Process instance avatar image + description.
-		avatarInfo, err := p.account.UpdateAvatar(ctx, form.Avatar, form.AvatarDescription, instanceAcc.ID)
-		if err != nil {
-			return nil, gtserror.NewErrorBadRequest(err, "error processing avatar")
+		avatarInfo, errWithCode := p.account.UpdateAvatar(ctx,
+			instanceAcc,
+			form.Avatar,
+			form.AvatarDescription,
+		)
+		if errWithCode != nil {
+			return nil, errWithCode
 		}
 		instanceAcc.AvatarMediaAttachmentID = avatarInfo.ID
 		instanceAcc.AvatarMediaAttachment = avatarInfo
@@ -264,9 +268,13 @@ func (p *Processor) InstancePatch(ctx context.Context, form *apimodel.InstanceSe
 
 	if form.Header != nil && form.Header.Size != 0 {
 		// process instance header image
-		headerInfo, err := p.account.UpdateHeader(ctx, form.Header, nil, instanceAcc.ID)
-		if err != nil {
-			return nil, gtserror.NewErrorBadRequest(err, "error processing header")
+		headerInfo, errWithCode := p.account.UpdateHeader(ctx,
+			instanceAcc,
+			form.Header,
+			nil,
+		)
+		if errWithCode != nil {
+			return nil, errWithCode
 		}
 		instanceAcc.HeaderMediaAttachmentID = headerInfo.ID
 		instanceAcc.HeaderMediaAttachment = headerInfo

internal/processing/media/create.go

@@ -30,7 +30,7 @@ import (
 
 // Create creates a new media attachment belonging to the given account, using the request form.
 func (p *Processor) Create(ctx context.Context, account *gtsmodel.Account, form *apimodel.AttachmentRequest) (*apimodel.Attachment, gtserror.WithCode) {
-	data := func(innerCtx context.Context) (io.ReadCloser, int64, error) {
+	data := func(_ context.Context) (io.ReadCloser, int64, error) {
 		f, err := form.File.Open()
 		return f, form.File.Size, err
 	}
@@ -41,19 +41,18 @@ func (p *Processor) Create(ctx context.Context, account *gtsmodel.Account, form
 		return nil, gtserror.NewErrorBadRequest(err, err.Error())
 	}
 
-	// process the media attachment and load it immediately
-	media := p.mediaManager.PreProcessMedia(data, account.ID, &media.AdditionalMediaInfo{
-		Description: &form.Description,
-		FocusX:      &focusX,
-		FocusY:      &focusY,
-	})
-
-	attachment, err := media.LoadAttachment(ctx)
-	if err != nil {
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
-	} else if attachment.Type == gtsmodel.FileTypeUnknown {
-		err = gtserror.Newf("could not process uploaded file with extension %s", attachment.File.ContentType)
-		return nil, gtserror.NewErrorUnprocessableEntity(err, err.Error())
+	// Create local media and write to instance storage.
+	attachment, errWithCode := p.c.StoreLocalMedia(ctx,
+		account.ID,
+		data,
+		media.AdditionalMediaInfo{
+			Description: &form.Description,
+			FocusX:      &focusX,
+			FocusY:      &focusY,
+		},
+	)
+	if errWithCode != nil {
+		return nil, errWithCode
 	}
 
 	apiAttachment, err := p.converter.AttachmentToAPIAttachment(ctx, attachment)

internal/processing/media/getfile.go

@@ -19,14 +19,14 @@ package media
 
 import (
 	"context"
+	"errors"
 	"fmt"
-	"io"
 	"net/url"
 	"strings"
 	"time"
 
 	apimodel "github.com/superseriousbusiness/gotosocial/internal/api/model"
-	"github.com/superseriousbusiness/gotosocial/internal/gtscontext"
+	"github.com/superseriousbusiness/gotosocial/internal/db"
 	"github.com/superseriousbusiness/gotosocial/internal/gtserror"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
@@ -38,7 +38,7 @@ import (
 // to the caller via an io.reader embedded in *apimodel.Content.
 func (p *Processor) GetFile(
 	ctx context.Context,
-	requestingAccount *gtsmodel.Account,
+	requester *gtsmodel.Account,
 	form *apimodel.GetContentRequestForm,
 ) (*apimodel.Content, gtserror.WithCode) {
 	// parse the form fields
@@ -69,13 +69,13 @@ func (p *Processor) GetFile(
 	}
 
 	// make sure the requesting account and the media account don't block each other
-	if requestingAccount != nil {
-		blocked, err := p.state.DB.IsEitherBlocked(ctx, requestingAccount.ID, owningAccountID)
+	if requester != nil {
+		blocked, err := p.state.DB.IsEitherBlocked(ctx, requester.ID, owningAccountID)
 		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("block status could not be established between accounts %s and %s: %s", owningAccountID, requestingAccount.ID, err))
+			return nil, gtserror.NewErrorNotFound(fmt.Errorf("block status could not be established between accounts %s and %s: %s", owningAccountID, requester.ID, err))
 		}
 		if blocked {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("block exists between accounts %s and %s", owningAccountID, requestingAccount.ID))
+			return nil, gtserror.NewErrorNotFound(fmt.Errorf("block exists between accounts %s and %s", owningAccountID, requester.ID))
 		}
 	}
 
@@ -83,17 +83,254 @@ func (p *Processor) GetFile(
 	// so we need to take different steps depending on the media type being requested
 	switch mediaType {
 	case media.TypeEmoji:
-		return p.getEmojiContent(ctx, wantedMediaID, owningAccountID, mediaSize)
+		return p.getEmojiContent(ctx,
+			owningAccountID,
+			wantedMediaID,
+			mediaSize,
+		)
 	case media.TypeAttachment, media.TypeHeader, media.TypeAvatar:
-		return p.getAttachmentContent(ctx, requestingAccount, wantedMediaID, owningAccountID, mediaSize)
+		return p.getAttachmentContent(ctx,
+			requester,
+			owningAccountID,
+			wantedMediaID,
+			mediaSize,
+		)
 	default:
 		return nil, gtserror.NewErrorNotFound(fmt.Errorf("media type %s not recognized", mediaType))
 	}
 }
 
-/*
-	UTIL FUNCTIONS
-*/
+func (p *Processor) getAttachmentContent(
+	ctx context.Context,
+	requester *gtsmodel.Account,
+	ownerID string,
+	mediaID string,
+	sizeStr media.Size,
+) (
+	*apimodel.Content,
+	gtserror.WithCode,
+) {
+	// Search for media with given ID in the database.
+	attach, err := p.state.DB.GetAttachmentByID(ctx, mediaID)
+	if err != nil && !errors.Is(err, db.ErrNoEntries) {
+		err := gtserror.Newf("error fetching media from database: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	if attach == nil {
+		const text = "media not found"
+		return nil, gtserror.NewErrorNotFound(errors.New(text), text)
+	}
+
+	// Ensure the 'owner' owns media.
+	if attach.AccountID != ownerID {
+		const text = "media was not owned by passed account id"
+		return nil, gtserror.NewErrorNotFound(errors.New(text) /* no help text! */)
+	}
+
+	var remoteURL *url.URL
+	if attach.RemoteURL != "" {
+
+		// Parse media remote URL to valid URL object.
+		remoteURL, err = url.Parse(attach.RemoteURL)
+		if err != nil {
+			err := gtserror.Newf("invalid media remote url %s: %w", attach.RemoteURL, err)
+			return nil, gtserror.NewErrorInternalError(err)
+		}
+	}
+
+	// Uknown file types indicate no *locally*
+	// stored data we can serve. Handle separately.
+	if attach.Type == gtsmodel.FileTypeUnknown {
+		if remoteURL == nil {
+			err := gtserror.Newf("missing remote url for unknown type media %s: %w", attach.ID, err)
+			return nil, gtserror.NewErrorInternalError(err)
+		}
+
+		// If this is an "Unknown" file type, ie., one we
+		// tried to process and couldn't, or one we refused
+		// to process because it wasn't supported, then we
+		// can skip a lot of steps here by simply forwarding
+		// the request to the remote URL.
+		url := &storage.PresignedURL{
+			URL: remoteURL,
+
+			// We might manage to cache the media
+			// at some point, so set a low-ish expiry.
+			Expiry: time.Now().Add(2 * time.Hour),
+		}
+
+		return &apimodel.Content{URL: url}, nil
+	}
+
+	var requestUser string
+
+	if requester != nil {
+		// Set requesting acc username.
+		requestUser = requester.Username
+	}
+
+	// Ensure that stored media is cached.
+	// (this handles local media / recaches).
+	attach, err = p.federator.RefreshMedia(
+		ctx,
+		requestUser,
+		attach,
+		media.AdditionalMediaInfo{},
+		false,
+	)
+	if err != nil {
+		err := gtserror.Newf("error recaching media: %w", err)
+		return nil, gtserror.NewErrorNotFound(err)
+	}
+
+	// Start preparing API content model.
+	apiContent := &apimodel.Content{
+		ContentUpdated: attach.UpdatedAt,
+	}
+
+	// Retrieve appropriate
+	// size file from storage.
+	switch sizeStr {
+
+	case media.SizeOriginal:
+		apiContent.ContentType = attach.File.ContentType
+		apiContent.ContentLength = int64(attach.File.FileSize)
+		return p.getContent(ctx,
+			attach.File.Path,
+			apiContent,
+		)
+
+	case media.SizeSmall:
+		apiContent.ContentType = attach.Thumbnail.ContentType
+		apiContent.ContentLength = int64(attach.Thumbnail.FileSize)
+		return p.getContent(ctx,
+			attach.Thumbnail.Path,
+			apiContent,
+		)
+
+	default:
+		const text = "invalid media attachment size"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
+	}
+}
+
+func (p *Processor) getEmojiContent(
+	ctx context.Context,
+
+	ownerID string,
+	emojiID string,
+	sizeStr media.Size,
+) (
+	*apimodel.Content,
+	gtserror.WithCode,
+) {
+	// Reconstruct static emoji image URL to search for it.
+	// As refreshed emojis use a newly generated path ID to
+	// differentiate them (cache-wise) from the original.
+	staticURL := uris.URIForAttachment(
+		ownerID,
+		string(media.TypeEmoji),
+		string(media.SizeStatic),
+		emojiID,
+		"png",
+	)
+
+	// Search for emoji with given static URL in the database.
+	emoji, err := p.state.DB.GetEmojiByStaticURL(ctx, staticURL)
+	if err != nil && !errors.Is(err, db.ErrNoEntries) {
+		err := gtserror.Newf("error fetching emoji from database: %w", err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	if emoji == nil {
+		const text = "emoji not found"
+		return nil, gtserror.NewErrorNotFound(errors.New(text), text)
+	}
+
+	if *emoji.Disabled {
+		const text = "emoji has been disabled"
+		return nil, gtserror.NewErrorNotFound(errors.New(text), text)
+	}
+
+	// Ensure that stored emoji is cached.
+	// (this handles local emoji / recaches).
+	emoji, err = p.federator.RefreshEmoji(
+		ctx,
+		emoji,
+		media.AdditionalEmojiInfo{},
+		false,
+	)
+	if err != nil {
+		err := gtserror.Newf("error recaching emoji: %w", err)
+		return nil, gtserror.NewErrorNotFound(err)
+	}
+
+	// Start preparing API content model.
+	apiContent := &apimodel.Content{}
+
+	// Retrieve appropriate
+	// size file from storage.
+	switch sizeStr {
+
+	case media.SizeOriginal:
+		apiContent.ContentType = emoji.ImageContentType
+		apiContent.ContentLength = int64(emoji.ImageFileSize)
+		return p.getContent(ctx,
+			emoji.ImagePath,
+			apiContent,
+		)
+
+	case media.SizeStatic:
+		apiContent.ContentType = emoji.ImageStaticContentType
+		apiContent.ContentLength = int64(emoji.ImageStaticFileSize)
+		return p.getContent(ctx,
+			emoji.ImageStaticPath,
+			apiContent,
+		)
+
+	default:
+		const text = "invalid media attachment size"
+		return nil, gtserror.NewErrorBadRequest(errors.New(text), text)
+	}
+}
+
+// getContent performs the final file fetching of
+// stored content at path in storage. This is
+// populated in the apimodel.Content{} and returned.
+// (note: this also handles un-proxied S3 storage).
+func (p *Processor) getContent(
+	ctx context.Context,
+	path string,
+	content *apimodel.Content,
+) (
+	*apimodel.Content,
+	gtserror.WithCode,
+) {
+	// If running on S3 storage with proxying disabled then
+	// just fetch pre-signed URL instead of the content.
+	if url := p.state.Storage.URL(ctx, path); url != nil {
+		content.URL = url
+		return content, nil
+	}
+
+	// Fetch file stream for the stored media at path.
+	rc, err := p.state.Storage.GetStream(ctx, path)
+	if err != nil && !storage.IsNotFound(err) {
+		err := gtserror.Newf("error getting file %s from storage: %w", path, err)
+		return nil, gtserror.NewErrorInternalError(err)
+	}
+
+	// Ensure found.
+	if rc == nil {
+		const text = "file not found"
+		return nil, gtserror.NewErrorNotFound(errors.New(text), text)
+	}
+
+	// Return with stream.
+	content.Content = rc
+	return content, nil
+}
 
 func parseType(s string) (media.Type, error) {
 	switch s {
@@ -120,198 +357,3 @@ func parseSize(s string) (media.Size, error) {
 	}
 	return "", fmt.Errorf("%s not a recognized media.Size", s)
 }
-
-func (p *Processor) getAttachmentContent(ctx context.Context, requestingAccount *gtsmodel.Account, wantedMediaID string, owningAccountID string, mediaSize media.Size) (*apimodel.Content, gtserror.WithCode) {
-	// retrieve attachment from the database and do basic checks on it
-	a, err := p.state.DB.GetAttachmentByID(ctx, wantedMediaID)
-	if err != nil {
-		err = gtserror.Newf("attachment %s could not be taken from the db: %w", wantedMediaID, err)
-		return nil, gtserror.NewErrorNotFound(err)
-	}
-
-	if a.AccountID != owningAccountID {
-		err = gtserror.Newf("attachment %s is not owned by %s", wantedMediaID, owningAccountID)
-		return nil, gtserror.NewErrorNotFound(err)
-	}
-
-	// If this is an "Unknown" file type, ie., one we
-	// tried to process and couldn't, or one we refused
-	// to process because it wasn't supported, then we
-	// can skip a lot of steps here by simply forwarding
-	// the request to the remote URL.
-	if a.Type == gtsmodel.FileTypeUnknown {
-		remoteURL, err := url.Parse(a.RemoteURL)
-		if err != nil {
-			err = gtserror.Newf("error parsing remote URL of 'Unknown'-type attachment for redirection: %w", err)
-			return nil, gtserror.NewErrorInternalError(err)
-		}
-
-		url := &storage.PresignedURL{
-			URL: remoteURL,
-			// We might manage to cache the media
-			// at some point, so set a low-ish expiry.
-			Expiry: time.Now().Add(2 * time.Hour),
-		}
-
-		return &apimodel.Content{URL: url}, nil
-	}
-
-	if !*a.Cached {
-		// if we don't have it cached, then we can assume two things:
-		// 1. this is remote media, since local media should never be uncached
-		// 2. we need to fetch it again using a transport and the media manager
-		remoteMediaIRI, err := url.Parse(a.RemoteURL)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error parsing remote media iri %s: %w", a.RemoteURL, err))
-		}
-
-		// use an empty string as requestingUsername to use the instance account, unless the request for this
-		// media has been http signed, then use the requesting account to make the request to remote server
-		var requestingUsername string
-		if requestingAccount != nil {
-			requestingUsername = requestingAccount.Username
-		}
-
-		// Pour one out for tobi's original streamed recache
-		// (streaming data both to the client and storage).
-		// Gone and forever missed <3
-		//
-		// [
-		//   the reason it was removed was because a slow
-		//   client connection could hold open a storage
-		//   recache operation -> holding open a media worker.
-		// ]
-
-		dataFn := func(ctx context.Context) (io.ReadCloser, int64, error) {
-			t, err := p.transportController.NewTransportForUsername(ctx, requestingUsername)
-			if err != nil {
-				return nil, 0, err
-			}
-			return t.DereferenceMedia(gtscontext.SetFastFail(ctx), remoteMediaIRI)
-		}
-
-		// Start recaching this media with the prepared data function.
-		processingMedia, err := p.mediaManager.PreProcessMediaRecache(ctx, dataFn, wantedMediaID)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error recaching media: %w", err))
-		}
-
-		// Load attachment and block until complete
-		a, err = processingMedia.LoadAttachment(ctx)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error loading recached attachment: %w", err))
-		}
-	}
-
-	var (
-		storagePath       string
-		attachmentContent = &apimodel.Content{
-			ContentUpdated: a.UpdatedAt,
-		}
-	)
-
-	// get file information from the attachment depending on the requested media size
-	switch mediaSize {
-	case media.SizeOriginal:
-		attachmentContent.ContentType = a.File.ContentType
-		attachmentContent.ContentLength = int64(a.File.FileSize)
-		storagePath = a.File.Path
-	case media.SizeSmall:
-		attachmentContent.ContentType = a.Thumbnail.ContentType
-		attachmentContent.ContentLength = int64(a.Thumbnail.FileSize)
-		storagePath = a.Thumbnail.Path
-	default:
-		return nil, gtserror.NewErrorNotFound(fmt.Errorf("media size %s not recognized for attachment", mediaSize))
-	}
-
-	// ... so now we can safely return it
-	return p.retrieveFromStorage(ctx, storagePath, attachmentContent)
-}
-
-func (p *Processor) getEmojiContent(ctx context.Context, fileName string, owningAccountID string, emojiSize media.Size) (*apimodel.Content, gtserror.WithCode) {
-	emojiContent := &apimodel.Content{}
-	var storagePath string
-
-	// reconstruct the static emoji image url -- reason
-	// for using the static URL rather than full size url
-	// is that static emojis are always encoded as png,
-	// so this is more reliable than using full size url
-	imageStaticURL := uris.URIForAttachment(
-		owningAccountID,
-		string(media.TypeEmoji),
-		string(media.SizeStatic),
-		fileName,
-		"png",
-	)
-
-	e, err := p.state.DB.GetEmojiByStaticURL(ctx, imageStaticURL)
-	if err != nil {
-		return nil, gtserror.NewErrorNotFound(fmt.Errorf("emoji %s could not be taken from the db: %w", fileName, err))
-	}
-
-	if *e.Disabled {
-		return nil, gtserror.NewErrorNotFound(fmt.Errorf("emoji %s has been disabled", fileName))
-	}
-
-	if !*e.Cached {
-		// if we don't have it cached, then we can assume two things:
-		// 1. this is remote emoji, since local emoji should never be uncached
-		// 2. we need to fetch it again using a transport and the media manager
-		remoteURL, err := url.Parse(e.ImageRemoteURL)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error parsing remote emoji iri %s: %w", e.ImageRemoteURL, err))
-		}
-
-		dataFn := func(ctx context.Context) (io.ReadCloser, int64, error) {
-			t, err := p.transportController.NewTransportForUsername(ctx, "")
-			if err != nil {
-				return nil, 0, err
-			}
-			return t.DereferenceMedia(gtscontext.SetFastFail(ctx), remoteURL)
-		}
-
-		// Start recaching this emoji with the prepared data function.
-		processingEmoji, err := p.mediaManager.PreProcessEmojiRecache(ctx, dataFn, e.ID)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error recaching emoji: %w", err))
-		}
-
-		// Load attachment and block until complete
-		e, err = processingEmoji.LoadEmoji(ctx)
-		if err != nil {
-			return nil, gtserror.NewErrorNotFound(fmt.Errorf("error loading recached emoji: %w", err))
-		}
-	}
-
-	switch emojiSize {
-	case media.SizeOriginal:
-		emojiContent.ContentType = e.ImageContentType
-		emojiContent.ContentLength = int64(e.ImageFileSize)
-		storagePath = e.ImagePath
-	case media.SizeStatic:
-		emojiContent.ContentType = e.ImageStaticContentType
-		emojiContent.ContentLength = int64(e.ImageStaticFileSize)
-		storagePath = e.ImageStaticPath
-	default:
-		return nil, gtserror.NewErrorNotFound(fmt.Errorf("media size %s not recognized for emoji", emojiSize))
-	}
-
-	return p.retrieveFromStorage(ctx, storagePath, emojiContent)
-}
-
-func (p *Processor) retrieveFromStorage(ctx context.Context, storagePath string, content *apimodel.Content) (*apimodel.Content, gtserror.WithCode) {
-	// If running on S3 storage with proxying disabled then
-	// just fetch a pre-signed URL instead of serving the content.
-	if url := p.state.Storage.URL(ctx, storagePath); url != nil {
-		content.URL = url
-		return content, nil
-	}
-
-	reader, err := p.state.Storage.GetStream(ctx, storagePath)
-	if err != nil {
-		return nil, gtserror.NewErrorNotFound(fmt.Errorf("error retrieving from storage: %s", err))
-	}
-
-	content.Content = reader
-	return content, nil
-}

internal/processing/media/media.go

@@ -18,24 +18,39 @@
 package media
 
 import (
+	"github.com/superseriousbusiness/gotosocial/internal/federation"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
+	"github.com/superseriousbusiness/gotosocial/internal/processing/common"
 	"github.com/superseriousbusiness/gotosocial/internal/state"
 	"github.com/superseriousbusiness/gotosocial/internal/transport"
 	"github.com/superseriousbusiness/gotosocial/internal/typeutils"
 )
 
 type Processor struct {
+	// common processor logic
+	c *common.Processor
+
 	state               *state.State
 	converter           *typeutils.Converter
+	federator           *federation.Federator
 	mediaManager        *media.Manager
 	transportController transport.Controller
 }
 
 // New returns a new media processor.
-func New(state *state.State, converter *typeutils.Converter, mediaManager *media.Manager, transportController transport.Controller) Processor {
+func New(
+	common *common.Processor,
+	state *state.State,
+	converter *typeutils.Converter,
+	federator *federation.Federator,
+	mediaManager *media.Manager,
+	transportController transport.Controller,
+) Processor {
 	return Processor{
+		c:                   common,
 		state:               state,
 		converter:           converter,
+		federator:           federator,
 		mediaManager:        mediaManager,
 		transportController: transportController,
 	}

internal/processing/media/media_test.go

@@ -20,8 +20,10 @@ package media_test
 import (
 	"github.com/stretchr/testify/suite"
 	"github.com/superseriousbusiness/gotosocial/internal/db"
+	"github.com/superseriousbusiness/gotosocial/internal/filter/visibility"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
+	"github.com/superseriousbusiness/gotosocial/internal/processing/common"
 	mediaprocessing "github.com/superseriousbusiness/gotosocial/internal/processing/media"
 	"github.com/superseriousbusiness/gotosocial/internal/state"
 	"github.com/superseriousbusiness/gotosocial/internal/storage"
@@ -78,7 +80,12 @@ func (suite *MediaStandardTestSuite) SetupTest() {
 	suite.state.Storage = suite.storage
 	suite.mediaManager = testrig.NewTestMediaManager(&suite.state)
 	suite.transportController = testrig.NewTestTransportController(&suite.state, testrig.NewMockHTTPClient(nil, "../../../testrig/media"))
-	suite.mediaProcessor = mediaprocessing.New(&suite.state, suite.tc, suite.mediaManager, suite.transportController)
+
+	federator := testrig.NewTestFederator(&suite.state, suite.transportController, suite.mediaManager)
+	filter := visibility.NewFilter(&suite.state)
+	common := common.New(&suite.state, suite.mediaManager, suite.tc, federator, filter)
+
+	suite.mediaProcessor = mediaprocessing.New(&common, &suite.state, suite.tc, federator, suite.mediaManager, suite.transportController)
 	testrig.StandardDBSetup(suite.db, nil)
 	testrig.StandardStorageSetup(suite.storage, "../../../testrig/media")
 }

internal/processing/polls/poll_test.go

@@ -57,7 +57,7 @@ func (suite *PollTestSuite) SetupTest() {
 	mediaMgr := media.NewManager(&suite.state)
 	federator := testrig.NewTestFederator(&suite.state, controller, mediaMgr)
 	suite.filter = visibility.NewFilter(&suite.state)
-	common := common.New(&suite.state, converter, federator, suite.filter)
+	common := common.New(&suite.state, mediaMgr, converter, federator, suite.filter)
 	suite.polls = polls.New(&common, &suite.state, converter)
 }
 

internal/processing/processor.go

@@ -179,15 +179,15 @@ func NewProcessor(
 	//
 	// Start with sub processors that will
 	// be required by the workers processor.
-	common := common.New(state, converter, federator, filter)
+	common := common.New(state, mediaManager, converter, federator, filter)
 	processor.account = account.New(&common, state, converter, mediaManager, federator, filter, parseMentionFunc)
-	processor.media = media.New(state, converter, mediaManager, federator.TransportController())
+	processor.media = media.New(&common, state, converter, federator, mediaManager, federator.TransportController())
 	processor.stream = stream.New(state, oauthServer)
 
 	// Instantiate the rest of the sub
 	// processors + pin them to this struct.
 	processor.account = account.New(&common, state, converter, mediaManager, federator, filter, parseMentionFunc)
-	processor.admin = admin.New(state, cleaner, converter, mediaManager, federator.TransportController(), emailSender)
+	processor.admin = admin.New(&common, state, cleaner, federator, converter, mediaManager, federator.TransportController(), emailSender)
 	processor.fedi = fedi.New(state, &common, converter, federator, filter)
 	processor.filtersv1 = filtersv1.New(state, converter, &processor.stream)
 	processor.filtersv2 = filtersv2.New(state, converter, &processor.stream)

internal/processing/status/status_test.go

@@ -96,7 +96,7 @@ func (suite *StatusStandardTestSuite) SetupTest() {
 		suite.typeConverter,
 	)
 
-	common := common.New(&suite.state, suite.typeConverter, suite.federator, filter)
+	common := common.New(&suite.state, suite.mediaManager, suite.typeConverter, suite.federator, filter)
 	polls := polls.New(&common, &suite.state, suite.typeConverter)
 	suite.status = status.New(&suite.state, &common, &polls, suite.federator, suite.typeConverter, filter, processing.GetParseMentionFunc(&suite.state, suite.federator))