onedrive: fix occasional 416 errors on multipart uploads

Before this change, when uploading multipart files, onedrive would
sometimes return an unexpected 416 error and rclone would abort the
transfer.

This is usually after a 500 error which caused rclone to do a retry.

This change checks the upload position on a 416 error and works how
much of the current chunk to skip, then retries (or skips) the current
chunk as appropriate.

If the position is before the current chunk or after the current chunk
then rclone will abort the transfer.

See: https://forum.rclone.org/t/fragment-overlap-error-with-encrypted-onedrive/14001

Fixes #3131

backend/onedrive/onedrive.go

@@ -1540,21 +1540,74 @@ func (o *Object) createUploadSession(ctx context.Context, modTime time.Time) (re
 	return response, err
 }
 
+// getPosition gets the current position in a multipart upload
+func (o *Object) getPosition(ctx context.Context, url string) (pos int64, err error) {
+	opts := rest.Opts{
+		Method:  "GET",
+		RootURL: url,
+	}
+	var info api.UploadFragmentResponse
+	var resp *http.Response
+	err = o.fs.pacer.Call(func() (bool, error) {
+		resp, err = o.fs.srv.CallJSON(ctx, &opts, nil, &info)
+		return shouldRetry(resp, err)
+	})
+	if err != nil {
+		return 0, err
+	}
+	if len(info.NextExpectedRanges) != 1 {
+		return 0, errors.Errorf("bad number of ranges in upload position: %v", info.NextExpectedRanges)
+	}
+	position := info.NextExpectedRanges[0]
+	i := strings.IndexByte(position, '-')
+	if i < 0 {
+		return 0, errors.Errorf("no '-' in next expected range: %q", position)
+	}
+	position = position[:i]
+	pos, err = strconv.ParseInt(position, 10, 64)
+	if err != nil {
+		return 0, errors.Wrapf(err, "bad expected range: %q", position)
+	}
+	return pos, nil
+}
+
 // uploadFragment uploads a part
 func (o *Object) uploadFragment(ctx context.Context, url string, start int64, totalSize int64, chunk io.ReadSeeker, chunkSize int64) (info *api.Item, err error) {
-	opts := rest.Opts{
-		Method:        "PUT",
-		RootURL:       url,
-		ContentLength: &chunkSize,
-		ContentRange:  fmt.Sprintf("bytes %d-%d/%d", start, start+chunkSize-1, totalSize),
-		Body:          chunk,
-	}
 	//	var response api.UploadFragmentResponse
 	var resp *http.Response
 	var body []byte
+	var skip = int64(0)
 	err = o.fs.pacer.Call(func() (bool, error) {
-		_, _ = chunk.Seek(0, io.SeekStart)
+		toSend := chunkSize - skip
+		opts := rest.Opts{
+			Method:        "PUT",
+			RootURL:       url,
+			ContentLength: &toSend,
+			ContentRange:  fmt.Sprintf("bytes %d-%d/%d", start+skip, start+chunkSize-1, totalSize),
+			Body:          chunk,
+		}
+		_, _ = chunk.Seek(skip, io.SeekStart)
 		resp, err = o.fs.srv.Call(ctx, &opts)
+		if err != nil && resp != nil && resp.StatusCode == http.StatusRequestedRangeNotSatisfiable {
+			fs.Debugf(o, "Received 416 error - reading current position from server: %v", err)
+			pos, posErr := o.getPosition(ctx, url)
+			if posErr != nil {
+				fs.Debugf(o, "Failed to read position: %v", posErr)
+				return false, posErr
+			}
+			skip = pos - start
+			fs.Debugf(o, "Read position %d, chunk is %d..%d, bytes to skip = %d", pos, start, start+chunkSize, skip)
+			switch {
+			case skip < 0:
+				return false, errors.Wrapf(err, "sent block already (skip %d < 0), can't rewind", skip)
+			case skip > chunkSize:
+				return false, errors.Wrapf(err, "position is in the future (skip %d > chunkSize %d), can't skip forward", skip, chunkSize)
+			case skip == chunkSize:
+				fs.Debugf(o, "Skipping chunk as already sent (skip %d == chunkSize %d)", skip, chunkSize)
+				return false, nil
+			}
+			return true, errors.Wrapf(err, "retry this chunk skipping %d bytes", skip)
+		}
 		if err != nil {
 			return shouldRetry(resp, err)
 		}