test the media manager a bit, add shutdown logic

internal/media/image.go

@@ -20,16 +20,22 @@ package media
 
 import (
 	"bytes"
+	"context"
 	"errors"
 	"fmt"
 	"image"
 	"image/gif"
 	"image/jpeg"
 	"image/png"
+	"strings"
+	"time"
 
 	"github.com/buckket/go-blurhash"
 	"github.com/nfnt/resize"
 	"github.com/superseriousbusiness/exifremove/pkg/exifremove"
+	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
+	"github.com/superseriousbusiness/gotosocial/internal/id"
+	"github.com/superseriousbusiness/gotosocial/internal/uris"
 )
 
 const (
@@ -38,13 +44,119 @@ const (
 )
 
 type ImageMeta struct {
-	image       []byte
-	contentType string
-	width       int
-	height      int
-	size        int
-	aspect      float64
-	blurhash    string
+	image    []byte
+	width    int
+	height   int
+	size     int
+	aspect   float64
+	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) {
@@ -106,8 +218,12 @@ func decodeImage(b []byte, contentType string) (*ImageMeta, error) {
 // deriveThumbnail returns a byte slice and metadata for a thumbnail
 // of a given jpeg, png, or gif, or an error if something goes wrong.
 //
-// Note that the aspect ratio of the image will be retained,
-// so it will not necessarily be a square, even if x and y are set as the same value.
+// If createBlurhash is true, then a blurhash will also be generated from a tiny
+// version of the image. This costs precious CPU cycles, so only use it if you
+// really need a blurhash and don't have one already.
+//
+// If createBlurhash is false, then the blurhash field on the returned ImageAndMeta
+// will be an empty string.
 func deriveThumbnail(b []byte, contentType string, createBlurhash bool) (*ImageMeta, error) {
 	var i image.Image
 	var err error
@@ -115,21 +231,20 @@ func deriveThumbnail(b []byte, contentType string, createBlurhash bool) (*ImageM
 	switch contentType {
 	case mimeImageJpeg:
 		i, err = jpeg.Decode(bytes.NewReader(b))
-		if err != nil {
-			return nil, err
-		}
 	case mimeImagePng:
 		i, err = png.Decode(bytes.NewReader(b))
-		if err != nil {
-			return nil, err
-		}
 	case mimeImageGif:
 		i, err = gif.Decode(bytes.NewReader(b))
-		if err != nil {
-			return nil, err
-		}
 	default:
-		return nil, fmt.Errorf("content type %s not recognised", contentType)
+		err = fmt.Errorf("content type %s can't be thumbnailed", contentType)
+	}
+
+	if err != nil {
+		return nil, err
+	}
+
+	if i == nil {
+		return nil, errors.New("processed image was nil")
 	}
 
 	thumb := resize.Thumbnail(thumbnailMaxWidth, thumbnailMaxHeight, i, resize.NearestNeighbor)
@@ -146,6 +261,8 @@ func deriveThumbnail(b []byte, contentType string, createBlurhash bool) (*ImageM
 	}
 
 	if createBlurhash {
+		// for generating blurhashes, it's more cost effective to lose detail rather than
+		// pass a big image into the blurhash algorithm, so make a teeny tiny version
 		tiny := resize.Thumbnail(32, 32, thumb, resize.NearestNeighbor)
 		bh, err := blurhash.Encode(4, 3, tiny)
 		if err != nil {
@@ -156,6 +273,7 @@ func deriveThumbnail(b []byte, contentType string, createBlurhash bool) (*ImageM
 
 	out := &bytes.Buffer{}
 	if err := jpeg.Encode(out, thumb, &jpeg.Options{
+		// Quality isn't extremely important for thumbnails, so 75 is "good enough"
 		Quality: 75,
 	}); err != nil {
 		return nil, err

internal/media/manager.go

@@ -24,63 +24,84 @@ import (
 	"fmt"
 	"runtime"
 	"strings"
-	"time"
 
 	"codeberg.org/gruf/go-runners"
 	"codeberg.org/gruf/go-store/kv"
 	"github.com/sirupsen/logrus"
 	"github.com/superseriousbusiness/gotosocial/internal/db"
-	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
-	"github.com/superseriousbusiness/gotosocial/internal/id"
-	"github.com/superseriousbusiness/gotosocial/internal/uris"
 )
 
 // 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).
 	// It will return a pointer to a Media struct upon which further actions can be performed, such as getting
-	// the finished media, thumbnail, decoded bytes, attachment, and setting additional fields.
+	// the finished media, thumbnail, attachment, etc.
 	//
 	// accountID should be the account that the media belongs to.
 	//
-	// RemoteURL is optional, and can be an empty string. Setting this to a non-empty string indicates that
-	// the piece of media originated on a remote instance and has been dereferenced to be cached locally.
-	ProcessMedia(ctx context.Context, data []byte, accountID string, ai *AdditionalInfo) (*Media, error)
-
-	ProcessEmoji(ctx context.Context, data []byte, accountID string) (*Media, error)
+	// 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)
+	// NumWorkers returns the total number of workers available to this manager.
+	NumWorkers() int
+	// QueueSize returns the total capacity of the queue.
+	QueueSize() int
+	// JobsQueued returns the number of jobs currently in the task queue.
+	JobsQueued() int
+	// ActiveWorkers returns the number of workers currently performing jobs.
+	ActiveWorkers() int
+	// Stop stops the underlying worker pool of the manager. It should be called
+	// when closing GoToSocial in order to cleanly finish any in-progress jobs.
+	// It will block until workers are finished processing.
+	Stop() error
 }
 
 type manager struct {
-	db      db.DB
-	storage *kv.KVStore
-	pool    runners.WorkerPool
+	db         db.DB
+	storage    *kv.KVStore
+	pool       runners.WorkerPool
+	numWorkers int
+	queueSize  int
 }
 
-// New returns a media manager with the given db and underlying storage.
-func New(database db.DB, storage *kv.KVStore) (Manager, error) {
-	workers := runtime.NumCPU() / 2
-	queue := workers * 10
-	pool := runners.NewWorkerPool(workers, queue)
-
-	if start := pool.Start(); !start {
-		return nil, errors.New("could not start worker pool")
+// NewManager returns a media manager with the given db and underlying storage.
+//
+// A worker pool will also be initialized for the manager, to ensure that only
+// a limited number of media will be processed in parallel.
+//
+// The number of workers will be the number of CPUs available to the Go runtime,
+// divided by 2 (rounding down, but always at least 1).
+//
+// The length of the queue will be the number of workers multiplied by 10.
+//
+// So for an 8 core machine, the media manager will get 4 workers, and a queue of length 40.
+// For a 4 core machine, this will be 2 workers, and a queue length of 20.
+// For a single or 2-core machine, the media manager will get 1 worker, and a queue of length 10.
+func NewManager(database db.DB, storage *kv.KVStore) (Manager, error) {
+	numWorkers := runtime.NumCPU() / 2
+	// make sure we always have at least 1 worker even on single-core machines
+	if numWorkers == 0 {
+		numWorkers = 1
 	}
-	logrus.Debugf("started media manager worker pool with %d workers and queue capacity of %d", workers, queue)
+	queueSize := numWorkers * 10
 
 	m := &manager{
-		db:      database,
-		storage: storage,
-		pool:    pool,
+		db:         database,
+		storage:    storage,
+		pool:       runners.NewWorkerPool(numWorkers, queueSize),
+		numWorkers: numWorkers,
+		queueSize:  queueSize,
 	}
 
+	if start := m.pool.Start(); !start {
+		return nil, errors.New("could not start worker pool")
+	}
+	logrus.Debugf("started media manager worker pool with %d workers and queue capacity of %d", numWorkers, queueSize)
+
 	return m, nil
 }
 
-/*
-	INTERFACE FUNCTIONS
-*/
-
-func (m *manager) ProcessMedia(ctx context.Context, data []byte, accountID string, ai *AdditionalInfo) (*Media, error) {
+func (m *manager) ProcessMedia(ctx context.Context, data []byte, accountID string, ai *AdditionalInfo) (*Processing, error) {
 	contentType, err := parseContentType(data)
 	if err != nil {
 		return nil, err
@@ -100,16 +121,20 @@ func (m *manager) ProcessMedia(ctx context.Context, data []byte, accountID strin
 			return nil, err
 		}
 
+		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 preloading the media for the caller's convenience
-				media.preLoad(innerCtx)
+				// 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 media, nil
 	default:
@@ -117,112 +142,32 @@ func (m *manager) ProcessMedia(ctx context.Context, data []byte, accountID strin
 	}
 }
 
-func (m *manager) ProcessEmoji(ctx context.Context, data []byte, accountID string) (*Media, error) {
+func (m *manager) ProcessEmoji(ctx context.Context, data []byte, accountID string) (*Processing, error) {
 	return nil, nil
 }
 
-// preProcessImage initializes processing
-func (m *manager) preProcessImage(ctx context.Context, data []byte, contentType string, accountID string, ai *AdditionalInfo) (*Media, 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]
-
-	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:        0,
-		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
-	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 := &Media{
-		attachment:    attachment,
-		rawData:       data,
-		thumbstate:    received,
-		fullSizeState: received,
-		database:      m.db,
-		storage:       m.storage,
-	}
-
-	return media, nil
+func (m *manager) NumWorkers() int {
+	return m.numWorkers
+}
+
+func (m *manager) QueueSize() int {
+	return m.queueSize
+}
+
+func (m *manager) JobsQueued() int {
+	return m.pool.Queue()
+}
+
+func (m *manager) ActiveWorkers() int {
+	return m.pool.Workers()
+}
+
+func (m *manager) Stop() error {
+	logrus.Info("stopping media manager worker pool")
+
+	stopped := m.pool.Stop()
+	if !stopped {
+		return errors.New("could not stop media manager worker pool")
+	}
+	return nil
 }

internal/media/manager_test.go

@@ -19,36 +19,238 @@
 package media_test
 
 import (
-	"codeberg.org/gruf/go-store/kv"
+	"context"
+	"fmt"
+	"os"
+	"testing"
+	"time"
+
 	"github.com/stretchr/testify/suite"
-	"github.com/superseriousbusiness/gotosocial/internal/db"
+	gtsmodel "github.com/superseriousbusiness/gotosocial/internal/db/bundb/migrations/20211113114307_init"
 	"github.com/superseriousbusiness/gotosocial/internal/media"
-	"github.com/superseriousbusiness/gotosocial/testrig"
 )
 
-type MediaManagerStandardTestSuite struct {
-	suite.Suite
-
-	db      db.DB
-	storage *kv.KVStore
-	manager media.Manager
+type ManagerTestSuite struct {
+	MediaStandardTestSuite
 }
 
-func (suite *MediaManagerStandardTestSuite) SetupSuite() {
-	testrig.InitTestLog()
-	testrig.InitTestConfig()
+func (suite *ManagerTestSuite) TestSimpleJpegProcessBlocking() {
+	ctx := context.Background()
 
-	suite.db = testrig.NewTestDB()
-	suite.storage = testrig.NewTestStorage()
+	// load bytes from a test image
+	testBytes, err := os.ReadFile("./test/test-jpeg.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(testBytes)
+
+	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
+
+	// process the media with no additional info provided
+	processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, 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)
+	suite.NoError(err)
+	suite.NotNil(attachment)
+
+	// make sure it's got the stuff set on it that we expect
+	// the attachment ID and accountID we expect
+	suite.Equal(attachmentID, attachment.ID)
+	suite.Equal(accountID, attachment.AccountID)
+
+	// file meta should be correctly derived from the image
+	suite.EqualValues(gtsmodel.Original{
+		Width: 1920, Height: 1080, Size: 2073600, Aspect: 1.7777777777777777,
+	}, attachment.FileMeta.Original)
+	suite.EqualValues(gtsmodel.Small{
+		Width: 512, Height: 288, Size: 147456, Aspect: 1.7777777777777777,
+	}, attachment.FileMeta.Small)
+	suite.Equal("image/jpeg", attachment.File.ContentType)
+	suite.Equal("LjBzUo#6RQR._NvzRjWF?urqV@a$", attachment.Blurhash)
+
+	// now make sure the attachment is in the database
+	dbAttachment, err := suite.db.GetAttachmentByID(ctx, attachmentID)
+	suite.NoError(err)
+	suite.NotNil(dbAttachment)
+
+	// make sure the processed file is in storage
+	processedFullBytes, err := suite.storage.Get(attachment.File.Path)
+	suite.NoError(err)
+	suite.NotEmpty(processedFullBytes)
+
+	// load the processed bytes from our test folder, to compare
+	processedFullBytesExpected, err := os.ReadFile("./test/test-jpeg-processed.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(processedFullBytesExpected)
+
+	// the bytes in storage should be what we expected
+	suite.Equal(processedFullBytesExpected, processedFullBytes)
+
+	// now do the same for the thumbnail and make sure it's what we expected
+	processedThumbnailBytes, err := suite.storage.Get(attachment.Thumbnail.Path)
+	suite.NoError(err)
+	suite.NotEmpty(processedThumbnailBytes)
+
+	processedThumbnailBytesExpected, err := os.ReadFile("./test/test-jpeg-thumbnail.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(processedThumbnailBytesExpected)
+
+	suite.Equal(processedThumbnailBytesExpected, processedThumbnailBytes)
 }
 
-func (suite *MediaManagerStandardTestSuite) SetupTest() {
-	testrig.StandardStorageSetup(suite.storage, "../../testrig/media")
-	testrig.StandardDBSetup(suite.db, nil)
-	suite.manager = testrig.NewTestMediaManager(suite.db, suite.storage)
+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)
+
+	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
+
+	// process the media with no additional info provided
+	processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, accountID, nil)
+	suite.NoError(err)
+	// fetch the attachment id from the processing media
+	attachmentID := processingMedia.AttachmentID()
+
+	// wait for the media to finish processing
+	for finished := processingMedia.Finished(); !finished; finished = processingMedia.Finished() {
+		time.Sleep(10 * time.Millisecond)
+		fmt.Printf("\n\nnot finished yet...\n\n")
+	}
+	fmt.Printf("\n\nfinished!\n\n")
+
+	// fetch the attachment from the database
+	attachment, err := suite.db.GetAttachmentByID(ctx, attachmentID)
+	suite.NoError(err)
+	suite.NotNil(attachment)
+
+	// make sure it's got the stuff set on it that we expect
+	// the attachment ID and accountID we expect
+	suite.Equal(attachmentID, attachment.ID)
+	suite.Equal(accountID, attachment.AccountID)
+
+	// file meta should be correctly derived from the image
+	suite.EqualValues(gtsmodel.Original{
+		Width: 1920, Height: 1080, Size: 2073600, Aspect: 1.7777777777777777,
+	}, attachment.FileMeta.Original)
+	suite.EqualValues(gtsmodel.Small{
+		Width: 512, Height: 288, Size: 147456, Aspect: 1.7777777777777777,
+	}, attachment.FileMeta.Small)
+	suite.Equal("image/jpeg", attachment.File.ContentType)
+	suite.Equal("LjBzUo#6RQR._NvzRjWF?urqV@a$", attachment.Blurhash)
+
+	// now make sure the attachment is in the database
+	dbAttachment, err := suite.db.GetAttachmentByID(ctx, attachmentID)
+	suite.NoError(err)
+	suite.NotNil(dbAttachment)
+
+	// make sure the processed file is in storage
+	processedFullBytes, err := suite.storage.Get(attachment.File.Path)
+	suite.NoError(err)
+	suite.NotEmpty(processedFullBytes)
+
+	// load the processed bytes from our test folder, to compare
+	processedFullBytesExpected, err := os.ReadFile("./test/test-jpeg-processed.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(processedFullBytesExpected)
+
+	// the bytes in storage should be what we expected
+	suite.Equal(processedFullBytesExpected, processedFullBytes)
+
+	// now do the same for the thumbnail and make sure it's what we expected
+	processedThumbnailBytes, err := suite.storage.Get(attachment.Thumbnail.Path)
+	suite.NoError(err)
+	suite.NotEmpty(processedThumbnailBytes)
+
+	processedThumbnailBytesExpected, err := os.ReadFile("./test/test-jpeg-thumbnail.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(processedThumbnailBytesExpected)
+
+	suite.Equal(processedThumbnailBytesExpected, processedThumbnailBytes)
 }
 
-func (suite *MediaManagerStandardTestSuite) TearDownTest() {
-	testrig.StandardDBTeardown(suite.db)
-	testrig.StandardStorageTeardown(suite.storage)
+func (suite *ManagerTestSuite) TestSimpleJpegQueueSpamming() {
+	// in this test, we spam the manager queue with 50 new media requests, just to see how it holds up
+
+	ctx := context.Background()
+
+	// load bytes from a test image
+	testBytes, err := os.ReadFile("./test/test-jpeg.jpg")
+	suite.NoError(err)
+	suite.NotEmpty(testBytes)
+
+	accountID := "01FS1X72SK9ZPW0J1QQ68BD264"
+
+	spam := 50
+	inProcess := []*media.Processing{}
+	for i := 0; i < spam; i++ {
+		// process the media with no additional info provided
+		processingMedia, err := suite.manager.ProcessMedia(ctx, testBytes, accountID, nil)
+		suite.NoError(err)
+		inProcess = append(inProcess, processingMedia)
+	}
+
+	for _, processingMedia := range inProcess {
+		fmt.Printf("\n\n\nactive workers: %d, queue length: %d\n\n\n", suite.manager.ActiveWorkers(), suite.manager.JobsQueued())
+
+		// fetch the attachment id from the processing media
+		attachmentID := processingMedia.AttachmentID()
+
+		// do a blocking call to fetch the attachment
+		attachment, err := processingMedia.Load(ctx)
+		suite.NoError(err)
+		suite.NotNil(attachment)
+
+		// make sure it's got the stuff set on it that we expect
+		// the attachment ID and accountID we expect
+		suite.Equal(attachmentID, attachment.ID)
+		suite.Equal(accountID, attachment.AccountID)
+
+		// file meta should be correctly derived from the image
+		suite.EqualValues(gtsmodel.Original{
+			Width: 1920, Height: 1080, Size: 2073600, Aspect: 1.7777777777777777,
+		}, attachment.FileMeta.Original)
+		suite.EqualValues(gtsmodel.Small{
+			Width: 512, Height: 288, Size: 147456, Aspect: 1.7777777777777777,
+		}, attachment.FileMeta.Small)
+		suite.Equal("image/jpeg", attachment.File.ContentType)
+		suite.Equal("LjBzUo#6RQR._NvzRjWF?urqV@a$", attachment.Blurhash)
+
+		// now make sure the attachment is in the database
+		dbAttachment, err := suite.db.GetAttachmentByID(ctx, attachmentID)
+		suite.NoError(err)
+		suite.NotNil(dbAttachment)
+
+		// make sure the processed file is in storage
+		processedFullBytes, err := suite.storage.Get(attachment.File.Path)
+		suite.NoError(err)
+		suite.NotEmpty(processedFullBytes)
+
+		// load the processed bytes from our test folder, to compare
+		processedFullBytesExpected, err := os.ReadFile("./test/test-jpeg-processed.jpg")
+		suite.NoError(err)
+		suite.NotEmpty(processedFullBytesExpected)
+
+		// the bytes in storage should be what we expected
+		suite.Equal(processedFullBytesExpected, processedFullBytes)
+
+		// now do the same for the thumbnail and make sure it's what we expected
+		processedThumbnailBytes, err := suite.storage.Get(attachment.Thumbnail.Path)
+		suite.NoError(err)
+		suite.NotEmpty(processedThumbnailBytes)
+
+		processedThumbnailBytesExpected, err := os.ReadFile("./test/test-jpeg-thumbnail.jpg")
+		suite.NoError(err)
+		suite.NotEmpty(processedThumbnailBytesExpected)
+
+		suite.Equal(processedThumbnailBytesExpected, processedThumbnailBytes)
+	}
+}
+
+func TestManagerTestSuite(t *testing.T) {
+	suite.Run(t, &ManagerTestSuite{})
 }

internal/media/media_test.go

@@ -0,0 +1,54 @@
+/*
+   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_test
+
+import (
+	"codeberg.org/gruf/go-store/kv"
+	"github.com/stretchr/testify/suite"
+	"github.com/superseriousbusiness/gotosocial/internal/db"
+	"github.com/superseriousbusiness/gotosocial/internal/media"
+	"github.com/superseriousbusiness/gotosocial/testrig"
+)
+
+type MediaStandardTestSuite struct {
+	suite.Suite
+
+	db      db.DB
+	storage *kv.KVStore
+	manager media.Manager
+}
+
+func (suite *MediaStandardTestSuite) SetupSuite() {
+	testrig.InitTestConfig()
+	testrig.InitTestLog()
+
+	suite.db = testrig.NewTestDB()
+	suite.storage = testrig.NewTestStorage()
+}
+
+func (suite *MediaStandardTestSuite) SetupTest() {
+	testrig.StandardStorageSetup(suite.storage, "../../testrig/media")
+	testrig.StandardDBSetup(suite.db, nil)
+	suite.manager = testrig.NewTestMediaManager(suite.db, suite.storage)
+}
+
+func (suite *MediaStandardTestSuite) TearDownTest() {
+	testrig.StandardDBTeardown(suite.db)
+	testrig.StandardStorageTeardown(suite.storage)
+}

internal/media/processing.go

@@ -37,7 +37,9 @@ const (
 	errored                      // processing order has been completed with an error
 )
 
-type Media struct {
+// Processing represents a piece of media that is currently being processed. It exposes
+// various functions for retrieving data from the process.
+type Processing struct {
 	mu sync.Mutex
 
 	/*
@@ -75,173 +77,169 @@ type Media struct {
 	err error // error created during processing, if any
 }
 
-func (m *Media) Thumb(ctx context.Context) (*ImageMeta, error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
+func (p *Processing) Thumb(ctx context.Context) (*ImageMeta, error) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
 
-	switch m.thumbstate {
+	switch p.thumbstate {
 	case received:
 		// we haven't processed a thumbnail for this media yet so do it now
 
 		// check if we need to create a blurhash or if there's already one set
 		var createBlurhash bool
-		if m.attachment.Blurhash == "" {
+		if p.attachment.Blurhash == "" {
 			// no blurhash created yet
 			createBlurhash = true
 		}
 
-		thumb, err := deriveThumbnail(m.rawData, m.attachment.File.ContentType, createBlurhash)
+		thumb, err := deriveThumbnail(p.rawData, p.attachment.File.ContentType, createBlurhash)
 		if err != nil {
-			m.err = fmt.Errorf("error deriving thumbnail: %s", err)
-			m.thumbstate = errored
-			return nil, m.err
+			p.err = fmt.Errorf("error deriving thumbnail: %s", err)
+			p.thumbstate = errored
+			return nil, p.err
 		}
 
 		// put the thumbnail in storage
-		if err := m.storage.Put(m.attachment.Thumbnail.Path, thumb.image); err != nil {
-			m.err = fmt.Errorf("error storing thumbnail: %s", err)
-			m.thumbstate = errored
-			return nil, m.err
+		if err := p.storage.Put(p.attachment.Thumbnail.Path, thumb.image); err != nil {
+			p.err = fmt.Errorf("error storing thumbnail: %s", err)
+			p.thumbstate = errored
+			return nil, p.err
 		}
 
 		// set appropriate fields on the attachment based on the thumbnail we derived
 		if createBlurhash {
-			m.attachment.Blurhash = thumb.blurhash
+			p.attachment.Blurhash = thumb.blurhash
 		}
 
-		m.attachment.FileMeta.Small = gtsmodel.Small{
+		p.attachment.FileMeta.Small = gtsmodel.Small{
 			Width:  thumb.width,
 			Height: thumb.height,
 			Size:   thumb.size,
 			Aspect: thumb.aspect,
 		}
-		m.attachment.Thumbnail.FileSize = thumb.size
+		p.attachment.Thumbnail.FileSize = thumb.size
 
-		if err := putOrUpdateAttachment(ctx, m.database, m.attachment); err != nil {
-			m.err = err
-			m.thumbstate = errored
+		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
-		m.thumb = thumb
+		p.thumb = thumb
 
 		// we're done processing the thumbnail!
-		m.thumbstate = complete
+		p.thumbstate = complete
 		fallthrough
 	case complete:
-		return m.thumb, nil
+		return p.thumb, nil
 	case errored:
-		return nil, m.err
+		return nil, p.err
 	}
 
-	return nil, fmt.Errorf("thumbnail processing status %d unknown", m.thumbstate)
+	return nil, fmt.Errorf("thumbnail processing status %d unknown", p.thumbstate)
 }
 
-func (m *Media) FullSize(ctx context.Context) (*ImageMeta, error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
+func (p *Processing) FullSize(ctx context.Context) (*ImageMeta, error) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
 
-	switch m.fullSizeState {
+	switch p.fullSizeState {
 	case received:
 		var clean []byte
 		var err error
 		var decoded *ImageMeta
 
-		ct := m.attachment.File.ContentType
+		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(m.rawData); exifErr != nil {
+			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 = m.rawData
+			clean = p.rawData
 			decoded, err = decodeGif(clean)
 		default:
 			err = fmt.Errorf("content type %s not a processible image type", ct)
 		}
 
 		if err != nil {
-			m.err = err
-			m.fullSizeState = errored
+			p.err = err
+			p.fullSizeState = errored
 			return nil, err
 		}
 
 		// put the full size in storage
-		if err := m.storage.Put(m.attachment.File.Path, decoded.image); err != nil {
-			m.err = fmt.Errorf("error storing full size image: %s", err)
-			m.fullSizeState = errored
-			return nil, m.err
+		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
-		m.attachment.FileMeta.Original = gtsmodel.Original{
+		p.attachment.FileMeta.Original = gtsmodel.Original{
 			Width:  decoded.width,
 			Height: decoded.height,
 			Size:   decoded.size,
 			Aspect: decoded.aspect,
 		}
-		m.attachment.File.FileSize = decoded.size
-		m.attachment.File.UpdatedAt = time.Now()
-		m.attachment.Processing = gtsmodel.ProcessingStatusProcessed
+		p.attachment.File.FileSize = decoded.size
+		p.attachment.File.UpdatedAt = time.Now()
+		p.attachment.Processing = gtsmodel.ProcessingStatusProcessed
 
-		if err := putOrUpdateAttachment(ctx, m.database, m.attachment); err != nil {
-			m.err = err
-			m.fullSizeState = errored
+		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
-		m.fullSize = decoded
+		p.fullSize = decoded
 
 		// we're done processing the full-size image
-		m.fullSizeState = complete
+		p.fullSizeState = complete
 		fallthrough
 	case complete:
-		return m.fullSize, nil
+		return p.fullSize, nil
 	case errored:
-		return nil, m.err
+		return nil, p.err
 	}
 
-	return nil, fmt.Errorf("full size processing status %d unknown", m.fullSizeState)
+	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 (m *Media) AttachmentID() string {
-	return m.attachment.ID
+func (p *Processing) AttachmentID() string {
+	return p.attachment.ID
 }
 
-// preLoad begins the process of deriving the thumbnail and encoding the full-size image.
-// It does this in a non-blocking way, so you can call it and then come back later and check
-// if it's finished.
-func (m *Media) preLoad(ctx context.Context) {
-	go m.Thumb(ctx)
-	go m.FullSize(ctx)
-}
-
-// Load is the blocking equivalent of pre-load. It makes sure the thumbnail and full-size
-// image have been processed, then it returns the completed attachment.
-func (m *Media) LoadAttachment(ctx context.Context) (*gtsmodel.MediaAttachment, error) {
-	if _, err := m.Thumb(ctx); err != 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
 	}
 
-	if _, err := m.FullSize(ctx); err != nil {
+	if _, err := p.FullSize(ctx); err != nil {
 		return nil, err
 	}
 
-	return m.attachment, nil
+	return p.attachment, nil
 }
 
-func (m *Media) LoadEmoji(ctx context.Context) (*gtsmodel.Emoji, error) {
+func (p *Processing) LoadEmoji(ctx context.Context) (*gtsmodel.Emoji, error) {
 	return nil, nil
 }
 
+func (p *Processing) Finished() bool {
+	return p.thumbstate == complete && p.fullSizeState == complete
+}
+
 // putOrUpdateAttachment is just a convenience function for first trying to PUT the attachment in the database,
 // and then if that doesn't work because the attachment already exists, updating it instead.
 func putOrUpdateAttachment(ctx context.Context, database db.DB, attachment *gtsmodel.MediaAttachment) error {