Ejemplo n.º 1
0
def get_data_details(project_id: str):
    """
    Route to get the details for all data for a given project matching the project_id.
    Raises an HTTPNotFoundError if the project is not found in the database.

    :param project_id: the project_id to get the data details for
    :return: a tuple containing (json response, http status code)
    """
    args = {key: val for key, val in request.args.items()}
    _LOGGER.info(
        "getting all the data for project_id {} and request args {}".format(
            project_id, args))

    args = SearchProjectDataSchema().load(args)

    # Validate project and model
    get_project_by_id(project_id)
    get_project_model_by_project_id(project_id)

    project_data = (ProjectData.select().where(
        ProjectData.project_id == project_id).group_by(ProjectData).order_by(
            ProjectData.created).paginate(args["page"], args["page_length"]))

    resp_data = data_dump_and_validation(ResponseProjectDataSchema(),
                                         {"data": project_data})
    _LOGGER.info("sending project data {}".format(resp_data))

    return jsonify(resp_data), HTTPStatus.OK.value
Ejemplo n.º 2
0
    def _save_project_data(data: ProjectData, path: str, model_path: str):
        validate_model_data(path, model_path)

        try:
            data.setup_filesystem()
            data.file = "{}.npz".format(data.data_id)
            shutil.copy(path, data.file_path)
            # revalidate to make sure the copy worked
            validate_model_data(data.file_path, model_path)
            data.save()
        except Exception as err:
            if data:
                try:
                    os.remove(data.file_path)
                except OSError:
                    pass

                try:
                    data.delete_instance()
                except Exception as rollback_err:
                    _LOGGER.error(
                        "error while rolling back new data: {}".format(rollback_err)
                    )

            _LOGGER.error(
                "error while creating new project data, rolling back: {}".format(err)
            )
            raise err
Ejemplo n.º 3
0
    def _get_project_data(self) -> ProjectData:
        data = ProjectData.get_or_none(ProjectData.data_id == self._data_id)

        if data is None:
            raise ValueError(
                "ProjectData with data_id {} was not found".format(self._data_id)
            )

        return data
Ejemplo n.º 4
0
def get_project_data_by_ids(project_id: str, data_id: str) -> ProjectData:
    query = ProjectData.get_or_none(ProjectData.project_id == project_id,
                                    ProjectData.data_id == data_id)

    if query is None:
        _LOGGER.error(
            "could not find data with data_id {} for project_id {}".format(
                data_id, project_id))
        raise HTTPNotFoundError(
            "could not find data with data_id {} for project_id {}".format(
                data_id, project_id))

    return query
Ejemplo n.º 5
0
 def _create_project_data(self) -> ProjectData:
     original = self._get_project_data()
     data = ProjectData.create(
         project=original.project, source=original.source, job=original.job
     )
     return data
Ejemplo n.º 6
0
def load_data_from_repo(project_id: str):
    """
    Route for loading data file(s) for a project from the Neural Magic model repo.
    Starts a background job in the JobWorker setup to run.
    The state of the job can be checked after.
    Raises an HTTPNotFoundError if the project is not found in the database.

    :param project_id: the id of the project to load the data for
    :return: a tuple containing (json response, http status code)
    """
    _LOGGER.info(
        "loading data from repo for project {} for request json {}".format(
            project_id, request.json))
    project = get_project_by_id(project_id)
    get_project_model_by_project_id(project_id)
    data = SetProjectDataFromSchema().load(request.get_json(force=True))
    project_data = None
    job = None

    try:
        project_data = ProjectData.create(project=project,
                                          source="downloaded_path",
                                          job=None)
        job = Job.create(
            project_id=project.project_id,
            type_=DataFromRepoJobWorker.get_type(),
            worker_args=DataFromRepoJobWorker.format_args(
                data_id=project_data.data_id, uri=data["uri"]),
        )
        project_data.job = job
        project_data.save()
        project_data.setup_filesystem()
        project_data.validate_filesystem()
    except Exception as err:
        if project_data:
            try:
                os.remove(project_data.file_path)
            except OSError:
                pass

            try:
                project_data.delete_instance()
            except Exception as rollback_err:
                _LOGGER.error("error while rolling back new data: {}".format(
                    rollback_err))

        if job:
            try:
                job.delete_instance()
            except Exception as rollback_err:
                _LOGGER.error("error while rolling back new data: {}".format(
                    rollback_err))

        _LOGGER.error(
            "error while creating new project data, rolling back: {}".format(
                err))
        raise err

    # call into JobWorkerManager to kick off job if it's not already running
    JobWorkerManager().refresh()

    resp_data = data_dump_and_validation(ResponseProjectDataSingleSchema(),
                                         {"data": project_data})
    _LOGGER.info("created project data from path {}".format(resp_data))

    return jsonify(resp_data), HTTPStatus.OK.value
Ejemplo n.º 7
0
def upload_data(project_id: str):
    """
    Route for uploading a data file to a project.
    Raises an HTTPNotFoundError if the project is not found in the database.

    :param project_id: the id of the project to upload the data for
    :return: a tuple containing (json response, http status code)
    """
    _LOGGER.info("uploading data file for project {}".format(project_id))
    project = get_project_by_id(project_id)
    project_model = get_project_model_by_project_id(project_id)

    if "data_file" not in request.files:
        _LOGGER.error("missing uploaded file 'data_file'")
        raise ValidationError("missing uploaded file 'data_file'")

    data_file = request.files["data_file"]

    with NamedTemporaryFile() as temp:
        data_path = gettempdir()
        tempname = os.path.join(data_path, temp.name)
        data_file.save(tempname)

        try:
            _LOGGER.info(project_model.file_path)

            validate_model_data(os.path.join(data_path, "*"),
                                project_model.file_path)

            data = CreateUpdateProjectDataSchema().dump({
                "source": "uploaded",
                "job": None
            })
            project_data = ProjectData.create(project=project, **data)
            project_data.file = "{}.npz".format(project_data)
            project_data.setup_filesystem()
            shutil.copy(tempname, project_data.file_path)

            project_data.validate_filesystem()
            validate_model_data(project_data.file_path,
                                project_model.file_path)
            project_data.save()
        except Exception as err:
            if project_data:
                try:
                    os.remove(project_data.file_path)
                except OSError:
                    pass

                try:
                    project_data.delete_instance()
                except Exception as rollback_err:
                    _LOGGER.error(
                        "error while rolling back new data: {}".format(
                            rollback_err))

            _LOGGER.error(
                "error while creating new project data, rolling back: {}".
                format(err))
            raise err

    resp_data = data_dump_and_validation(ResponseProjectDataSingleSchema(),
                                         {"data": project_data})
    _LOGGER.info("created project data {}".format(resp_data))

    return jsonify(resp_data), HTTPStatus.OK.value