cvat: Downloaded xml's have unexpected tokens at top.

Hello Katarina, I have found out that downloaded files are zip files. You can save it then unzip the file and save again. It is not the best solution though but you can use the following code if you like to directly dump to xml. You need the zipfile package. You should change the same named function of CLI class.

    def tasks_dump(self, task_id, fileformat, filename, **kwargs):
        """ Download annotations for a task in the specified format
        (e.g. 'YOLO ZIP 1.0')."""
        url = self.api.tasks_id(task_id)
        response = self.session.get(url)
        response.raise_for_status()
        response_json = response.json()

        url = self.api.tasks_id_annotations_filename(task_id,
                                                     response_json['name'],
                                                     fileformat)
        while True:
            response = self.session.get(url)
            response.raise_for_status()
            if response.status_code == 201:
                break

        response = self.session.get(url + '&action=download')
        response.raise_for_status()
        fbytes = BytesIO(response.content)
        zipf = ZipFile(fbytes)
        with codecs.open(filename, 'wb') as fp:
            fp.write(zipf.read('annotations.xml'))

Best Barış

response header content type appears as ‘application/zip’ , shouldn’t it be something different? I cannot encode the byte array to utf-8

UPDATE: Apparently this is because response is a zip file, everything is fine when I contents as a zip file. But in CLI page the example is with an xml file which was the case in previous ones. Check, https://github.com/opencv/cvat/blob/develop/utils/cli/README.md

From CLI, I would prefer unzipped versions or an option to choose if possible.