Warp/server/internal/mediaapi/remote.go

package mediaapi

import (
	"context"
	"errors"
	"fmt"
	"mime"
	"net/http"
	"strings"
	"time"

	"github.com/dariuszjaniak/warp/server/internal/federation"
	"github.com/dariuszjaniak/warp/server/internal/httputil"
	"github.com/dariuszjaniak/warp/server/internal/storage"
)

// fetchTimeout bounds one fetch from another server, waiting included.
// It is detached from the request that started it: other requests may be
// waiting on the same fetch, and what it fetches is kept either way.
const fetchTimeout = 10 * time.Minute

// remoteMedia returns another server's media from the cache, fetching it
// first if it is not there.
func (a *API) remoteMedia(ctx context.Context, origin, mediaID string, wait time.Duration) (storage.Media, error) {
	m, err := a.store.Media(ctx, origin, mediaID)
	if err == nil {
		return m, nil
	}
	if !errors.Is(err, storage.ErrNoSuchMedia) {
		return storage.Media{}, err
	}
	remote := a.remoteClient()
	if remote == nil {
		return storage.Media{}, ErrNotFound
	}

	wait = min(max(wait, 0), MaxWait)
	v, err := a.shared(ctx, "download\x00"+mediaName(origin, mediaID), func(ctx context.Context) (any, error) {
		return a.fetchRemote(ctx, remote, origin, mediaID, wait)
	})
	if err != nil {
		return storage.Media{}, err
	}
	return v.(storage.Media), nil
}

// shared runs fn once for every concurrent caller with the same key, on a
// context of its own, and lets each caller stop waiting when its own
// request ends.
func (a *API) shared(ctx context.Context, key string, fn func(context.Context) (any, error)) (any, error) {
	result := a.fetches.DoChan(key, func() (any, error) {
		fctx, cancel := context.WithTimeout(context.WithoutCancel(ctx), fetchTimeout)
		defer cancel()
		return fn(fctx)
	})
	select {
	case r := <-result:
		return r.Val, r.Err
	case <-ctx.Done():
		return nil, ctx.Err()
	}
}

// fetchRemote downloads media from its server and caches it.
func (a *API) fetchRemote(ctx context.Context, remote Remote, origin, mediaID string, wait time.Duration) (storage.Media, error) {
	resp, err := remote.DownloadMedia(ctx, origin, mediaID, wait)
	if err != nil {
		return storage.Media{}, a.remoteError(origin, mediaID, err)
	}
	defer func() { _ = resp.Body.Close() }()

	blob, err := a.blobs.Put(ctx, resp.Body, a.cfg.MaxUploadBytes)
	switch {
	case errors.Is(err, storage.ErrBlobTooLarge):
		return storage.Media{}, ErrTooLarge
	case err != nil:
		a.log.Warn("could not fetch remote media", "media", mediaName(origin, mediaID), "error", err)
		return storage.Media{}, fmt.Errorf("%w: %v", ErrRemote, err)
	}

	m := storage.Media{
		Origin:      origin,
		MediaID:     mediaID,
		ContentType: normaliseContentType(resp.ContentType),
		UploadName:  fileNameFrom(resp.ContentDisposition),
		Size:        blob.Size,
		SHA256:      blob.SHA256,
		CreatedAtMS: a.nowMS(),
	}
	err = a.store.CreateMedia(ctx, m)
	if errors.Is(err, storage.ErrMediaExists) {
		// Fetched twice — by another process, or before a restart. The
		// copy already recorded is the same content.
		return a.store.Media(ctx, origin, mediaID)
	}
	if err != nil {
		return storage.Media{}, err
	}
	a.log.Debug("fetched remote media", "media", mediaName(origin, mediaID),
		"content_type", m.ContentType, "bytes", m.Size)
	return m, nil
}

// fetchRemoteThumbnail asks the media's server for a thumbnail of one of
// the configured sizes, and caches it under that size.
//
// The size asked for is this server's, not the client's: the cache is
// keyed by it, and a client asking for 33 × 33 and another for 34 × 34
// share one thumbnail rather than causing two fetches.
func (a *API) fetchRemoteThumbnail(ctx context.Context, remote Remote, origin, mediaID string, size thumbnailSize, wait time.Duration) (storage.Thumbnail, error) {
	resp, err := remote.MediaThumbnail(ctx, origin, mediaID, federation.ThumbnailRequest{
		Width: size.Width, Height: size.Height, Method: size.Method,
	}, wait)
	if err != nil {
		var httpErr *federation.HTTPError
		if errors.As(err, &httpErr) && httpErr.Status == http.StatusBadRequest {
			return storage.Thumbnail{}, ErrCannotThumbnail
		}
		return storage.Thumbnail{}, a.remoteError(origin, mediaID, err)
	}
	defer func() { _ = resp.Body.Close() }()

	contentType := normaliseContentType(resp.ContentType)
	if !strings.HasPrefix(contentType, "image/") {
		// A thumbnail that is not an image is not one, whatever the
		// other server meant by it.
		return storage.Thumbnail{}, fmt.Errorf("%w: thumbnail of type %q", ErrRemote, contentType)
	}
	blob, err := a.blobs.Put(ctx, resp.Body, a.cfg.MaxUploadBytes)
	switch {
	case errors.Is(err, storage.ErrBlobTooLarge):
		return storage.Thumbnail{}, ErrTooLarge
	case err != nil:
		return storage.Thumbnail{}, fmt.Errorf("%w: %v", ErrRemote, err)
	}
	t := storage.Thumbnail{
		Origin: origin, MediaID: mediaID,
		Width: size.Width, Height: size.Height, Method: size.Method,
		ContentType: contentType, Size: blob.Size, SHA256: blob.SHA256,
		CreatedAtMS: a.nowMS(),
	}
	if err := a.store.StoreThumbnail(ctx, t); err != nil {
		return storage.Thumbnail{}, err
	}
	return t, nil
}

// remoteError maps a failed fetch onto the API's errors.
func (a *API) remoteError(origin, mediaID string, err error) error {
	var httpErr *federation.HTTPError
	switch {
	case errors.Is(err, federation.ErrDisabled), errors.Is(err, federation.ErrNotAllowed):
		return ErrFederationDisabled
	case errors.As(err, &httpErr) && httpErr.Status == http.StatusNotFound:
		return ErrNotFound
	case errors.As(err, &httpErr) && httpErr.Code == httputil.ErrCodeNotYetUploaded:
		return ErrNotYetUploaded
	case errors.As(err, &httpErr) && (httpErr.Status == http.StatusRequestEntityTooLarge || httpErr.Code == httputil.ErrCodeTooLarge):
		return ErrTooLarge
	}
	a.log.Warn("could not fetch remote media", "media", mediaName(origin, mediaID), "error", err)
	return fmt.Errorf("%w: %v", ErrRemote, err)
}

// fileNameFrom reads the file name out of a Content-Disposition header.
func fileNameFrom(disposition string) string {
	if disposition == "" {
		return ""
	}
	_, params, err := mime.ParseMediaType(disposition)
	if err != nil {
		return ""
	}
	name := params["filename"]
	// A file name is a name, not a path: keep only the last element of
	// one a server sent with directories in it.
	if i := strings.LastIndexAny(name, `/\`); i >= 0 {
		name = name[i+1:]
	}
	return name
}