pass a function into the manager, start work on emoji

6 files changed, 632 insertions, 188 deletions

diff --git a/internal/media/image.go b/internal/media/image.go
index 074dd3839..a5a818206 100644
--- a/internal/media/image.go
+++ b/internal/media/image.go
@@ -27,7 +27,6 @@ import (
 	"image/gif"
 	"image/jpeg"
 	"image/png"
-	"strings"
 	"time"
 
 	"github.com/buckket/go-blurhash"
@@ -52,113 +51,6 @@ type ImageMeta struct {
 	blurhash string
 }
 
-func (m *manager) preProcessImage(ctx context.Context, data []byte, contentType string, accountID string, ai *AdditionalInfo) (*Processing, error) {
-	if !supportedImage(contentType) {
-		return nil, fmt.Errorf("image type %s not supported", contentType)
-	}
-
-	if len(data) == 0 {
-		return nil, errors.New("image was of size 0")
-	}
-
-	id, err := id.NewRandomULID()
-	if err != nil {
-		return nil, err
-	}
-
-	extension := strings.Split(contentType, "/")[1]
-
-	// populate initial fields on the media attachment -- some of these will be overwritten as we proceed
-	attachment := &gtsmodel.MediaAttachment{
-		ID:        id,
-		CreatedAt: time.Now(),
-		UpdatedAt: time.Now(),
-		StatusID:  "",
-		URL:       uris.GenerateURIForAttachment(accountID, string(TypeAttachment), string(SizeOriginal), id, extension),
-		RemoteURL: "",
-		Type:      gtsmodel.FileTypeImage,
-		FileMeta: gtsmodel.FileMeta{
-			Focus: gtsmodel.Focus{
-				X: 0,
-				Y: 0,
-			},
-		},
-		AccountID:         accountID,
-		Description:       "",
-		ScheduledStatusID: "",
-		Blurhash:          "",
-		Processing:        gtsmodel.ProcessingStatusReceived,
-		File: gtsmodel.File{
-			Path:        fmt.Sprintf("%s/%s/%s/%s.%s", accountID, TypeAttachment, SizeOriginal, id, extension),
-			ContentType: contentType,
-			UpdatedAt:   time.Now(),
-		},
-		Thumbnail: gtsmodel.Thumbnail{
-			URL:         uris.GenerateURIForAttachment(accountID, string(TypeAttachment), string(SizeSmall), id, mimeJpeg), // all thumbnails are encoded as jpeg,
-			Path:        fmt.Sprintf("%s/%s/%s/%s.%s", accountID, TypeAttachment, SizeSmall, id, mimeJpeg),                 // all thumbnails are encoded as jpeg,
-			ContentType: mimeJpeg,
-			UpdatedAt:   time.Now(),
-		},
-		Avatar: false,
-		Header: false,
-	}
-
-	// check if we have additional info to add to the attachment,
-	// and overwrite some of the attachment fields if so
-	if ai != nil {
-		if ai.CreatedAt != nil {
-			attachment.CreatedAt = *ai.CreatedAt
-		}
-
-		if ai.StatusID != nil {
-			attachment.StatusID = *ai.StatusID
-		}
-
-		if ai.RemoteURL != nil {
-			attachment.RemoteURL = *ai.RemoteURL
-		}
-
-		if ai.Description != nil {
-			attachment.Description = *ai.Description
-		}
-
-		if ai.ScheduledStatusID != nil {
-			attachment.ScheduledStatusID = *ai.ScheduledStatusID
-		}
-
-		if ai.Blurhash != nil {
-			attachment.Blurhash = *ai.Blurhash
-		}
-
-		if ai.Avatar != nil {
-			attachment.Avatar = *ai.Avatar
-		}
-
-		if ai.Header != nil {
-			attachment.Header = *ai.Header
-		}
-
-		if ai.FocusX != nil {
-			attachment.FileMeta.Focus.X = *ai.FocusX
-		}
-
-		if ai.FocusY != nil {
-			attachment.FileMeta.Focus.Y = *ai.FocusY
-		}
-	}
-
-	media := &Processing{
-		attachment:    attachment,
-		rawData:       data,
-		thumbstate:    received,
-		fullSizeState: received,
-		database:      m.db,
-		storage:       m.storage,
-	}
-
-	return media, nil
-}
-
 func decodeGif(b []byte) (*ImageMeta, error) {
 	gif, err := gif.DecodeAll(bytes.NewReader(b))
 	if err != nil {
diff --git a/internal/media/manager.go b/internal/media/manager.go
index c8642fcb4..e34471591 100644
--- a/internal/media/manager.go
+++ b/internal/media/manager.go
@@ -21,9 +21,7 @@ package media
 import (
 	"context"
 	"errors"
-	"fmt"
 	"runtime"
-	"strings"
 
 	"codeberg.org/gruf/go-runners"
 	"codeberg.org/gruf/go-store/kv"
@@ -33,15 +31,17 @@ import (
 
 // Manager provides an interface for managing media: parsing, storing, and retrieving media objects like photos, videos, and gifs.
 type Manager interface {
-	// ProcessMedia begins the process of decoding and storing the given data as a piece of media (aka an attachment).
+	// ProcessMedia begins the process of decoding and storing the given data as an attachment.
 	// It will return a pointer to a Media struct upon which further actions can be performed, such as getting
 	// the finished media, thumbnail, attachment, etc.
 	//
+	// data should be a function that the media manager can call to return raw bytes of a piece of media.
+	//
 	// accountID should be the account that the media belongs to.
 	//
 	// ai is optional and can be nil. Any additional information about the attachment provided will be put in the database.
-	ProcessMedia(ctx context.Context, data []byte, accountID string, ai *AdditionalInfo) (*Processing, error)
-	ProcessEmoji(ctx context.Context, data []byte, accountID string) (*Processing, error)
+	ProcessMedia(ctx context.Context, data DataFunc, accountID string, ai *AdditionalMediaInfo) (*ProcessingMedia, error)
+	ProcessEmoji(ctx context.Context, data DataFunc, shortcode string, ai *AdditionalEmojiInfo) (*ProcessingEmoji, error)
 	// NumWorkers returns the total number of workers available to this manager.
 	NumWorkers() int
 	// QueueSize returns the total capacity of the queue.
@@ -101,49 +101,52 @@ func NewManager(database db.DB, storage *kv.KVStore) (Manager, error) {
 	return m, nil
 }
 
-func (m *manager) ProcessMedia(ctx context.Context, data []byte, accountID string, ai *AdditionalInfo) (*Processing, error) {
-	contentType, err := parseContentType(data)
+func (m *manager) ProcessMedia(ctx context.Context, data DataFunc, accountID string, ai *AdditionalMediaInfo) (*ProcessingMedia, error) {
+	processingMedia, err := m.preProcessMedia(ctx, data, accountID, ai)
 	if err != nil {
 		return nil, err
 	}
 
-	split := strings.Split(contentType, "/")
-	if len(split) != 2 {
-		return nil, fmt.Errorf("content type %s malformed", contentType)
-	}
-
-	mainType := split[0]
-
-	switch mainType {
-	case mimeImage:
-		media, err := m.preProcessImage(ctx, data, contentType, accountID, ai)
-		if err != nil {
-			return nil, err
+	logrus.Tracef("ProcessMedia: about to enqueue media with attachmentID %s, queue length is %d", processingMedia.AttachmentID(), m.pool.Queue())
+	m.pool.Enqueue(func(innerCtx context.Context) {
+		select {
+		case <-innerCtx.Done():
+			// if the inner context is done that means the worker pool is closing, so we should just return
+			return
+		default:
+			// start loading the media already for the caller's convenience
+			if _, err := processingMedia.LoadAttachment(innerCtx); err != nil {
+				logrus.Errorf("ProcessMedia: error processing media with attachmentID %s: %s", processingMedia.AttachmentID(), err)
+			}
 		}
+	})
+	logrus.Tracef("ProcessMedia: succesfully queued media with attachmentID %s, queue length is %d", processingMedia.AttachmentID(), m.pool.Queue())
 
-		logrus.Tracef("ProcessMedia: about to enqueue media with attachmentID %s, queue length is %d", media.AttachmentID(), m.pool.Queue())
-		m.pool.Enqueue(func(innerCtx context.Context) {
-			select {
-			case <-innerCtx.Done():
-				// if the inner context is done that means the worker pool is closing, so we should just return
-				return
-			default:
-				// start loading the media already for the caller's convenience
-				if _, err := media.Load(innerCtx); err != nil {
-					logrus.Errorf("ProcessMedia: error processing media with attachmentID %s: %s", media.AttachmentID(), err)
-				}
-			}
-		})
-		logrus.Tracef("ProcessMedia: succesfully queued media with attachmentID %s, queue length is %d", media.AttachmentID(), m.pool.Queue())
+	return processingMedia, nil
+}
 
-		return media, nil
-	default:
-		return nil, fmt.Errorf("content type %s not (yet) supported", contentType)
+func (m *manager) ProcessEmoji(ctx context.Context, data DataFunc, shortcode string, ai *AdditionalEmojiInfo) (*ProcessingEmoji, error) {
+	processingEmoji, err := m.preProcessEmoji(ctx, data, shortcode, ai)
+	if err != nil {
+		return nil, err
 	}
-}
 
-func (m *manager) ProcessEmoji(ctx context.Context, data []byte, accountID string) (*Processing, error) {
-	return nil, nil
+	logrus.Tracef("ProcessEmoji: about to enqueue emoji with id %s, queue length is %d", processingEmoji.EmojiID(), m.pool.Queue())
+	m.pool.Enqueue(func(innerCtx context.Context) {
+		select {
+		case <-innerCtx.Done():
+			// if the inner context is done that means the worker pool is closing, so we should just return
+			return
+		default:
+			// start loading the emoji already for the caller's convenience
+			if _, err := processingEmoji.LoadEmoji(innerCtx); err != nil {
+				logrus.Errorf("ProcessEmoji: error processing emoji with id %s: %s", processingEmoji.EmojiID(), err)
+			}
+		}
+	})
+	logrus.Tracef("ProcessEmoji: succesfully queued emoji with id %s, queue length is %d", processingEmoji.EmojiID(), m.pool.Queue())
+
+	return processingEmoji, nil
 }
 
 func (m *manager) NumWorkers() int {
diff --git a/internal/media/manager_test.go b/internal/media/manager_test.go
index 74d0c3008..0fadceb37 100644
--- a/internal/media/manager_test.go
+++ b/internal/media/manager_test.go
@@ -37,21 +37,21 @@ type ManagerTestSuite struct {
 func (suite *ManagerTestSuite) TestSimpleJpegProcessBlocking() {
 	ctx := context.Background()
 
-	// load bytes from a test image
-	testBytes, err := os.ReadFile("./test/test-jpeg.jpg")
-	suite.NoError(err)
-	suite.NotEmpty(testBytes)
+	data := func(_ context.Context) ([]byte, error) {
+		// load bytes from a test image
+		return os.ReadFile("./test/test-jpeg.jpg")
+	}
 
 	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
 
 	// process the media with no additional info provided
-	processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, accountID, nil)
+	processingMedia, err := suite.manager.ProcessMedia(ctx, data, accountID, nil)
 	suite.NoError(err)
 	// fetch the attachment id from the processing media
 	attachmentID := processingMedia.AttachmentID()
 
 	// do a blocking call to fetch the attachment
-	attachment, err := processingMedia.Load(ctx)
+	attachment, err := processingMedia.LoadAttachment(ctx)
 	suite.NoError(err)
 	suite.NotNil(attachment)
 
@@ -103,15 +103,15 @@ func (suite *ManagerTestSuite) TestSimpleJpegProcessBlocking() {
 func (suite *ManagerTestSuite) TestSimpleJpegProcessAsync() {
 	ctx := context.Background()
 
-	// load bytes from a test image
-	testBytes, err := os.ReadFile("./test/test-jpeg.jpg")
-	suite.NoError(err)
-	suite.NotEmpty(testBytes)
+	data := func(_ context.Context) ([]byte, error) {
+		// load bytes from a test image
+		return os.ReadFile("./test/test-jpeg.jpg")
+	}
 
 	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
 
 	// process the media with no additional info provided
-	processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, accountID, nil)
+	processingMedia, err := suite.manager.ProcessMedia(ctx, data, accountID, nil)
 	suite.NoError(err)
 	// fetch the attachment id from the processing media
 	attachmentID := processingMedia.AttachmentID()
@@ -183,13 +183,17 @@ func (suite *ManagerTestSuite) TestSimpleJpegQueueSpamming() {
 	suite.NoError(err)
 	suite.NotEmpty(testBytes)
 
+	data := func(_ context.Context) ([]byte, error) {
+		return testBytes, nil
+	}
+
 	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
 
 	spam := 50
-	inProcess := []*media.Processing{}
+	inProcess := []*media.ProcessingMedia{}
 	for i := 0; i < spam; i++ {
 		// process the media with no additional info provided
-		processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, accountID, nil)
+		processingMedia, err := suite.manager.ProcessMedia(ctx, data, accountID, nil)
 		suite.NoError(err)
 		inProcess = append(inProcess, processingMedia)
 	}
@@ -201,7 +205,7 @@ func (suite *ManagerTestSuite) TestSimpleJpegQueueSpamming() {
 		attachmentID := processingMedia.AttachmentID()
 
 		// do a blocking call to fetch the attachment
-		attachment, err := processingMedia.Load(ctx)
+		attachment, err := processingMedia.LoadAttachment(ctx)
 		suite.NoError(err)
 		suite.NotNil(attachment)
 
diff --git a/internal/media/processingemoji.go b/internal/media/processingemoji.go
new file mode 100644
index 000000000..7e2d4f31f
--- /dev/null
+++ b/internal/media/processingemoji.go
@@ -0,0 +1,382 @@
+/*
+   GoToSocial
+   Copyright (C) 2021-2022 GoToSocial Authors admin@gotosocial.org
+
+   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 media
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"sync"
+	"time"
+
+	"codeberg.org/gruf/go-store/kv"
+	"github.com/superseriousbusiness/gotosocial/internal/db"
+	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
+	"github.com/superseriousbusiness/gotosocial/internal/id"
+	"github.com/superseriousbusiness/gotosocial/internal/uris"
+)
+
+// ProcessingEmoji represents an emoji currently processing. It exposes
+// various functions for retrieving data from the process.
+type ProcessingEmoji struct {
+	mu sync.Mutex
+
+	/*
+		below fields should be set on newly created media;
+		emoji will be updated incrementally as media goes through processing
+	*/
+
+	emoji *gtsmodel.Emoji
+	data  DataFunc
+
+	rawData []byte // will be set once the fetchRawData function has been called
+
+	/*
+		below fields represent the processing state of the static version of the emoji
+	*/
+
+	staticState processState
+	static      *ImageMeta
+
+	/*
+		below fields represent the processing state of the emoji image
+	*/
+
+	fullSizeState processState
+	fullSize      *ImageMeta
+
+	/*
+		below pointers to database and storage are maintained so that
+		the media can store and update itself during processing steps
+	*/
+
+	database db.DB
+	storage  *kv.KVStore
+
+	err error // error created during processing, if any
+}
+
+// EmojiID returns the ID of the underlying emoji without blocking processing.
+func (p *ProcessingEmoji) EmojiID() string {
+	return p.emoji.ID
+}
+
+// LoadEmoji blocks until the static and fullsize image
+// has been processed, and then returns the completed emoji.
+func (p *ProcessingEmoji) LoadEmoji(ctx context.Context) (*gtsmodel.Emoji, error) {
+	if err := p.fetchRawData(ctx); err != nil {
+		return nil, err
+	}
+
+	if _, err := p.loadStatic(ctx); err != nil {
+		return nil, err
+	}
+
+	if _, err := p.loadFullSize(ctx); err != nil {
+		return nil, err
+	}
+
+	return p.emoji, nil
+}
+
+// Finished returns true if processing has finished for both the thumbnail
+// and full fized version of this piece of media.
+func (p *ProcessingEmoji) Finished() bool {
+	return p.staticState == complete && p.fullSizeState == complete
+}
+
+func (p *ProcessingEmoji) loadStatic(ctx context.Context) (*ImageMeta, error) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	switch p.staticState {
+	case received:
+		// we haven't processed a static version of this emoji yet so do it now
+		static, err := deriveStaticEmoji(p.rawData, p.emoji.ImageContentType)
+		if err != nil {
+			p.err = fmt.Errorf("error deriving static: %s", err)
+			p.staticState = errored
+			return nil, p.err
+		}
+
+		// put the static in storage
+		if err := p.storage.Put(p.emoji.ImageStaticPath, static.image); err != nil {
+			p.err = fmt.Errorf("error storing static: %s", err)
+			p.staticState = errored
+			return nil, p.err
+		}
+
+		// set appropriate fields on the emoji based on the static version we derived
+		p.attachment.FileMeta.Small = gtsmodel.Small{
+			Width:  static.width,
+			Height: static.height,
+			Size:   static.size,
+			Aspect: static.aspect,
+		}
+		p.attachment.Thumbnail.FileSize = static.size
+
+		if err := putOrUpdateAttachment(ctx, p.database, p.attachment); err != nil {
+			p.err = err
+			p.thumbstate = errored
+			return nil, err
+		}
+
+		// set the thumbnail of this media
+		p.thumb = static
+
+		// we're done processing the thumbnail!
+		p.thumbstate = complete
+		fallthrough
+	case complete:
+		return p.thumb, nil
+	case errored:
+		return nil, p.err
+	}
+
+	return nil, fmt.Errorf("thumbnail processing status %d unknown", p.thumbstate)
+}
+
+func (p *ProcessingEmoji) loadFullSize(ctx context.Context) (*ImageMeta, error) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	switch p.fullSizeState {
+	case received:
+		var clean []byte
+		var err error
+		var decoded *ImageMeta
+
+		ct := p.attachment.File.ContentType
+		switch ct {
+		case mimeImageJpeg, mimeImagePng:
+			// first 'clean' image by purging exif data from it
+			var exifErr error
+			if clean, exifErr = purgeExif(p.rawData); exifErr != nil {
+				err = exifErr
+				break
+			}
+			decoded, err = decodeImage(clean, ct)
+		case mimeImageGif:
+			// gifs are already clean - no exif data to remove
+			clean = p.rawData
+			decoded, err = decodeGif(clean)
+		default:
+			err = fmt.Errorf("content type %s not a processible image type", ct)
+		}
+
+		if err != nil {
+			p.err = err
+			p.fullSizeState = errored
+			return nil, err
+		}
+
+		// put the full size in storage
+		if err := p.storage.Put(p.attachment.File.Path, decoded.image); err != nil {
+			p.err = fmt.Errorf("error storing full size image: %s", err)
+			p.fullSizeState = errored
+			return nil, p.err
+		}
+
+		// set appropriate fields on the attachment based on the image we derived
+		p.attachment.FileMeta.Original = gtsmodel.Original{
+			Width:  decoded.width,
+			Height: decoded.height,
+			Size:   decoded.size,
+			Aspect: decoded.aspect,
+		}
+		p.attachment.File.FileSize = decoded.size
+		p.attachment.File.UpdatedAt = time.Now()
+		p.attachment.Processing = gtsmodel.ProcessingStatusProcessed
+
+		if err := putOrUpdateAttachment(ctx, p.database, p.attachment); err != nil {
+			p.err = err
+			p.fullSizeState = errored
+			return nil, err
+		}
+
+		// set the fullsize of this media
+		p.fullSize = decoded
+
+		// we're done processing the full-size image
+		p.fullSizeState = complete
+		fallthrough
+	case complete:
+		return p.fullSize, nil
+	case errored:
+		return nil, p.err
+	}
+
+	return nil, fmt.Errorf("full size processing status %d unknown", p.fullSizeState)
+}
+
+// fetchRawData calls the data function attached to p if it hasn't been called yet,
+// and updates the underlying attachment fields as necessary.
+// It should only be called from within a function that already has a lock on p!
+func (p *ProcessingEmoji) fetchRawData(ctx context.Context) error {
+	// check if we've already done this and bail early if we have
+	if p.rawData != nil {
+		return nil
+	}
+
+	// execute the data function and pin the raw bytes for further processing
+	b, err := p.data(ctx)
+	if err != nil {
+		return fmt.Errorf("fetchRawData: error executing data function: %s", err)
+	}
+	p.rawData = b
+
+	// now we have the data we can work out the content type
+	contentType, err := parseContentType(p.rawData)
+	if err != nil {
+		return fmt.Errorf("fetchRawData: error parsing content type: %s", err)
+	}
+
+	if !supportedEmoji(contentType) {
+		return fmt.Errorf("fetchRawData: content type %s was not valid for an emoji", contentType)
+	}
+
+	split := strings.Split(contentType, "/")
+	mainType := split[0]  // something like 'image'
+	extension := split[1] // something like 'gif'
+
+	// set some additional fields on the emoji now that
+	// we know more about what the underlying image actually is
+	p.emoji.ImageURL = uris.GenerateURIForAttachment(p.attachment.AccountID, string(TypeAttachment), string(SizeOriginal), p.attachment.ID, extension)
+	p.attachment.File.Path = fmt.Sprintf("%s/%s/%s/%s.%s", p.attachment.AccountID, TypeAttachment, SizeOriginal, p.attachment.ID, extension)
+	p.attachment.File.ContentType = contentType
+
+	switch mainType {
+	case mimeImage:
+		if extension == mimeGif {
+			p.attachment.Type = gtsmodel.FileTypeGif
+		} else {
+			p.attachment.Type = gtsmodel.FileTypeImage
+		}
+	default:
+		return fmt.Errorf("fetchRawData: cannot process mime type %s (yet)", mainType)
+	}
+
+	return nil
+}
+
+// putOrUpdateEmoji is just a convenience function for first trying to PUT the emoji in the database,
+// and then if that doesn't work because the emoji already exists, updating it instead.
+func putOrUpdateEmoji(ctx context.Context, database db.DB, emoji *gtsmodel.Emoji) error {
+	if err := database.Put(ctx, emoji); err != nil {
+		if err != db.ErrAlreadyExists {
+			return fmt.Errorf("putOrUpdateEmoji: proper error while putting emoji: %s", err)
+		}
+		if err := database.UpdateByPrimaryKey(ctx, emoji); err != nil {
+			return fmt.Errorf("putOrUpdateEmoji: error while updating emoji: %s", err)
+		}
+	}
+
+	return nil
+}
+
+func (m *manager) preProcessEmoji(ctx context.Context, data DataFunc, shortcode string, ai *AdditionalEmojiInfo) (*ProcessingEmoji, error) {
+	instanceAccount, err := m.db.GetInstanceAccount(ctx, "")
+	if err != nil {
+		return nil, fmt.Errorf("preProcessEmoji: error fetching this instance account from the db: %s", err)
+	}
+
+	id, err := id.NewRandomULID()
+	if err != nil {
+		return nil, err
+	}
+
+	// populate initial fields on the emoji -- some of these will be overwritten as we proceed
+	emoji := &gtsmodel.Emoji{
+		ID:                     id,
+		CreatedAt:              time.Now(),
+		UpdatedAt:              time.Now(),
+		Shortcode:              shortcode,
+		Domain:                 "", // assume our own domain unless told otherwise
+		ImageRemoteURL:         "",
+		ImageStaticRemoteURL:   "",
+		ImageURL:               "",                                                                                                    // we don't know yet
+		ImageStaticURL:         uris.GenerateURIForAttachment(instanceAccount.ID, string(TypeEmoji), string(SizeStatic), id, mimePng), // all static emojis are encoded as png
+		ImagePath:              "",                                                                                                    // we don't know yet
+		ImageStaticPath:        fmt.Sprintf("%s/%s/%s/%s.%s", instanceAccount.ID, TypeEmoji, SizeStatic, id, mimePng),                 // all static emojis are encoded as png
+		ImageContentType:       "",                                                                                                    // we don't know yet
+		ImageStaticContentType: mimeImagePng,                                                                                          // all static emojis are encoded as png
+		ImageFileSize:          0,
+		ImageStaticFileSize:    0,
+		ImageUpdatedAt:         time.Now(),
+		Disabled:               false,
+		URI:                    "", // we don't know yet
+		VisibleInPicker:        true,
+		CategoryID:             "",
+	}
+
+	// check if we have additional info to add to the emoji,
+	// and overwrite some of the emoji fields if so
+	if ai != nil {
+		if ai.CreatedAt != nil {
+			attachment.CreatedAt = *ai.CreatedAt
+		}
+
+		if ai.StatusID != nil {
+			attachment.StatusID = *ai.StatusID
+		}
+
+		if ai.RemoteURL != nil {
+			attachment.RemoteURL = *ai.RemoteURL
+		}
+
+		if ai.Description != nil {
+			attachment.Description = *ai.Description
+		}
+
+		if ai.ScheduledStatusID != nil {
+			attachment.ScheduledStatusID = *ai.ScheduledStatusID
+		}
+
+		if ai.Blurhash != nil {
+			attachment.Blurhash = *ai.Blurhash
+		}
+
+		if ai.Avatar != nil {
+			attachment.Avatar = *ai.Avatar
+		}
+
+		if ai.Header != nil {
+			attachment.Header = *ai.Header
+		}
+
+		if ai.FocusX != nil {
+			attachment.FileMeta.Focus.X = *ai.FocusX
+		}
+
+		if ai.FocusY != nil {
+			attachment.FileMeta.Focus.Y = *ai.FocusY
+		}
+	}
+
+	processingEmoji := &ProcessingEmoji{
+		emoji:         emoji,
+		data:          data,
+		staticState:   received,
+		fullSizeState: received,
+		database:      m.db,
+		storage:       m.storage,
+	}
+
+	return processingEmoji, nil
+}
diff --git a/internal/media/processing.go b/internal/media/processingmedia.go
index 3f9fc2bfc..a6e45034f 100644
--- a/internal/media/processing.go
+++ b/internal/media/processingmedia.go
@@ -21,12 +21,15 @@ package media
 import (
 	"context"
 	"fmt"
+	"strings"
 	"sync"
 	"time"
 
 	"codeberg.org/gruf/go-store/kv"
 	"github.com/superseriousbusiness/gotosocial/internal/db"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
+	"github.com/superseriousbusiness/gotosocial/internal/id"
+	"github.com/superseriousbusiness/gotosocial/internal/uris"
 )
 
 type processState int
@@ -37,9 +40,9 @@ const (
 	errored                      // processing order has been completed with an error
 )
 
-// Processing represents a piece of media that is currently being processed. It exposes
+// ProcessingMedia represents a piece of media that is currently being processed. It exposes
 // various functions for retrieving data from the process.
-type Processing struct {
+type ProcessingMedia struct {
 	mu sync.Mutex
 
 	/*
@@ -47,10 +50,10 @@ type Processing struct {
 		attachment will be updated incrementally as media goes through processing
 	*/
 
-	attachment *gtsmodel.MediaAttachment // will only be set if the media is an attachment
-	emoji      *gtsmodel.Emoji           // will only be set if the media is an emoji
+	attachment *gtsmodel.MediaAttachment
+	data       DataFunc
 
-	rawData []byte
+	rawData []byte // will be set once the fetchRawData function has been called
 
 	/*
 		below fields represent the processing state of the media thumbnail
@@ -77,7 +80,40 @@ type Processing struct {
 	err error // error created during processing, if any
 }
 
-func (p *Processing) Thumb(ctx context.Context) (*ImageMeta, error) {
+// AttachmentID returns the ID of the underlying media attachment without blocking processing.
+func (p *ProcessingMedia) AttachmentID() string {
+	return p.attachment.ID
+}
+
+// LoadAttachment blocks until the thumbnail and fullsize content
+// has been processed, and then returns the completed attachment.
+func (p *ProcessingMedia) LoadAttachment(ctx context.Context) (*gtsmodel.MediaAttachment, error) {
+	if err := p.fetchRawData(ctx); err != nil {
+		return nil, err
+	}
+
+	if _, err := p.loadThumb(ctx); err != nil {
+		return nil, err
+	}
+
+	if _, err := p.loadFullSize(ctx); err != nil {
+		return nil, err
+	}
+
+	return p.attachment, nil
+}
+
+func (p *ProcessingMedia) LoadEmoji(ctx context.Context) (*gtsmodel.Emoji, error) {
+	return nil, nil
+}
+
+// Finished returns true if processing has finished for both the thumbnail
+// and full fized version of this piece of media.
+func (p *ProcessingMedia) Finished() bool {
+	return p.thumbstate == complete && p.fullSizeState == complete
+}
+
+func (p *ProcessingMedia) loadThumb(ctx context.Context) (*ImageMeta, error) {
 	p.mu.Lock()
 	defer p.mu.Unlock()
 
@@ -140,7 +176,7 @@ func (p *Processing) Thumb(ctx context.Context) (*ImageMeta, error) {
 	return nil, fmt.Errorf("thumbnail processing status %d unknown", p.thumbstate)
 }
 
-func (p *Processing) FullSize(ctx context.Context) (*ImageMeta, error) {
+func (p *ProcessingMedia) loadFullSize(ctx context.Context) (*ImageMeta, error) {
 	p.mu.Lock()
 	defer p.mu.Unlock()
 
@@ -213,31 +249,54 @@ func (p *Processing) FullSize(ctx context.Context) (*ImageMeta, error) {
 	return nil, fmt.Errorf("full size processing status %d unknown", p.fullSizeState)
 }
 
-// AttachmentID returns the ID of the underlying media attachment without blocking processing.
-func (p *Processing) AttachmentID() string {
-	return p.attachment.ID
-}
+// fetchRawData calls the data function attached to p if it hasn't been called yet,
+// and updates the underlying attachment fields as necessary.
+// It should only be called from within a function that already has a lock on p!
+func (p *ProcessingMedia) fetchRawData(ctx context.Context) error {
+	// check if we've already done this and bail early if we have
+	if p.rawData != nil {
+		return nil
+	}
 
-// Load blocks until the thumbnail and fullsize content has been processed, and then
-// returns the completed attachment.
-func (p *Processing) Load(ctx context.Context) (*gtsmodel.MediaAttachment, error) {
-	if _, err := p.Thumb(ctx); err != nil {
-		return nil, err
+	// execute the data function and pin the raw bytes for further processing
+	b, err := p.data(ctx)
+	if err != nil {
+		return fmt.Errorf("fetchRawData: error executing data function: %s", err)
 	}
+	p.rawData = b
 
-	if _, err := p.FullSize(ctx); err != nil {
-		return nil, err
+	// now we have the data we can work out the content type
+	contentType, err := parseContentType(p.rawData)
+	if err != nil {
+		return fmt.Errorf("fetchRawData: error parsing content type: %s", err)
 	}
 
-	return p.attachment, nil
-}
+	split := strings.Split(contentType, "/")
+	if len(split) != 2 {
+		return fmt.Errorf("fetchRawData: content type %s was not valid", contentType)
+	}
 
-func (p *Processing) LoadEmoji(ctx context.Context) (*gtsmodel.Emoji, error) {
-	return nil, nil
-}
+	mainType := split[0]  // something like 'image'
+	extension := split[1] // something like 'jpeg'
+
+	// set some additional fields on the attachment now that
+	// we know more about what the underlying media actually is
+	p.attachment.URL = uris.GenerateURIForAttachment(p.attachment.AccountID, string(TypeAttachment), string(SizeOriginal), p.attachment.ID, extension)
+	p.attachment.File.Path = fmt.Sprintf("%s/%s/%s/%s.%s", p.attachment.AccountID, TypeAttachment, SizeOriginal, p.attachment.ID, extension)
+	p.attachment.File.ContentType = contentType
+
+	switch mainType {
+	case mimeImage:
+		if extension == mimeGif {
+			p.attachment.Type = gtsmodel.FileTypeGif
+		} else {
+			p.attachment.Type = gtsmodel.FileTypeImage
+		}
+	default:
+		return fmt.Errorf("fetchRawData: cannot process mime type %s (yet)", mainType)
+	}
 
-func (p *Processing) Finished() bool {
-	return p.thumbstate == complete && p.fullSizeState == complete
+	return nil
 }
 
 // putOrUpdateAttachment is just a convenience function for first trying to PUT the attachment in the database,
@@ -254,3 +313,99 @@ func putOrUpdateAttachment(ctx context.Context, database db.DB, attachment *gtsm
 
 	return nil
 }
+
+func (m *manager) preProcessMedia(ctx context.Context, data DataFunc, accountID string, ai *AdditionalMediaInfo) (*ProcessingMedia, error) {
+	id, err := id.NewRandomULID()
+	if err != nil {
+		return nil, err
+	}
+
+	file := gtsmodel.File{
+		Path:        "", // we don't know yet because it depends on the uncalled DataFunc
+		ContentType: "", // we don't know yet because it depends on the uncalled DataFunc
+		UpdatedAt:   time.Now(),
+	}
+
+	thumbnail := gtsmodel.Thumbnail{
+		URL:         uris.GenerateURIForAttachment(accountID, string(TypeAttachment), string(SizeSmall), id, mimeJpeg), // all thumbnails are encoded as jpeg,
+		Path:        fmt.Sprintf("%s/%s/%s/%s.%s", accountID, TypeAttachment, SizeSmall, id, mimeJpeg),                 // all thumbnails are encoded as jpeg,
+		ContentType: mimeJpeg,
+		UpdatedAt:   time.Now(),
+	}
+
+	// populate initial fields on the media attachment -- some of these will be overwritten as we proceed
+	attachment := &gtsmodel.MediaAttachment{
+		ID:                id,
+		CreatedAt:         time.Now(),
+		UpdatedAt:         time.Now(),
+		StatusID:          "",
+		URL:               "", // we don't know yet because it depends on the uncalled DataFunc
+		RemoteURL:         "",
+		Type:              gtsmodel.FileTypeUnknown, // we don't know yet because it depends on the uncalled DataFunc
+		FileMeta:          gtsmodel.FileMeta{},
+		AccountID:         accountID,
+		Description:       "",
+		ScheduledStatusID: "",
+		Blurhash:          "",
+		Processing:        gtsmodel.ProcessingStatusReceived,
+		File:              file,
+		Thumbnail:         thumbnail,
+		Avatar:            false,
+		Header:            false,
+	}
+
+	// check if we have additional info to add to the attachment,
+	// and overwrite some of the attachment fields if so
+	if ai != nil {
+		if ai.CreatedAt != nil {
+			attachment.CreatedAt = *ai.CreatedAt
+		}
+
+		if ai.StatusID != nil {
+			attachment.StatusID = *ai.StatusID
+		}
+
+		if ai.RemoteURL != nil {
+			attachment.RemoteURL = *ai.RemoteURL
+		}
+
+		if ai.Description != nil {
+			attachment.Description = *ai.Description
+		}
+
+		if ai.ScheduledStatusID != nil {
+			attachment.ScheduledStatusID = *ai.ScheduledStatusID
+		}
+
+		if ai.Blurhash != nil {
+			attachment.Blurhash = *ai.Blurhash
+		}
+
+		if ai.Avatar != nil {
+			attachment.Avatar = *ai.Avatar
+		}
+
+		if ai.Header != nil {
+			attachment.Header = *ai.Header
+		}
+
+		if ai.FocusX != nil {
+			attachment.FileMeta.Focus.X = *ai.FocusX
+		}
+
+		if ai.FocusY != nil {
+			attachment.FileMeta.Focus.Y = *ai.FocusY
+		}
+	}
+
+	processingMedia := &ProcessingMedia{
+		attachment:    attachment,
+		data:          data,
+		thumbstate:    received,
+		fullSizeState: received,
+		database:      m.db,
+		storage:       m.storage,
+	}
+
+	return processingMedia, nil
+}
diff --git a/internal/media/types.go b/internal/media/types.go
index aaf423682..6426223d1 100644
--- a/internal/media/types.go
+++ b/internal/media/types.go
@@ -20,6 +20,7 @@ package media
 
 import (
 	"bytes"
+	"context"
 	"errors"
 	"fmt"
 	"time"
@@ -68,9 +69,9 @@ const (
 	TypeEmoji      Type = "emoji"      // TypeEmoji is the key for emoji type requests
 )
 
-// AdditionalInfo represents additional information that should be added to an attachment
+// AdditionalMediaInfo represents additional information that should be added to an attachment
 // when processing a piece of media.
-type AdditionalInfo struct {
+type AdditionalMediaInfo struct {
 	// Time that this media was created; defaults to time.Now().
 	CreatedAt *time.Time
 	// ID of the status to which this media is attached; defaults to "".
@@ -93,6 +94,13 @@ type AdditionalInfo struct {
 	FocusY *float32
 }
 
+type AdditionalEmojiInfo struct {
+	
+}
+
+// DataFunc represents a function used to retrieve the raw bytes of a piece of media.
+type DataFunc func(ctx context.Context) ([]byte, error)
+
 // parseContentType parses the MIME content type from a file, returning it as a string in the form (eg., "image/jpeg").
 // Returns an error if the content type is not something we can process.
 func parseContentType(content []byte) (string, error) {