malevich.interpreter.core

class malevich.interpreter.core.CoreInterpreter(core_auth: tuple[str, str], core_host: str = 'https://core.malevich.ai/')

Interpret flows to be run on Malevich Core

Malevich Core is a computational cloud of Malevich. It provides low-level access to the computational resources of Malevich. This interpreter utilizes its API to run your flows.

Note

The interpreter can operate with dependencies installed with both image and space installer.

Prepare

The prepare hook is a simple proxy to the task_prepare function of Malevich Core API. All the parameters are optional.

Options:

• with_logs (bool):

  If set, return prepare logs if True after end

• debug_mode (bool):

  If set, displays additional information about errors

• info_url (str):

  URL to which the request is sent. If not specified, the default url is used. Rewrite msg_url from configuration if exist

• core_manage (bool):

  If set, the requests will be managed by Core, otherwise by the DAG Manager.

• with_show (bool):

  Show results (like for each operation, default equals with_logs arg)

• profile_mode (str):

  If set, provides more information in logs. Possible modes: no, all, time, df_info, df_show

Run

The run hook is a simple proxy to the task_run function of Malevich Core API. All the parameters are optional.

Options:

Interpreter-specific parameters

• run_id (str):

  A custom identifier for the run. If not specified, a random identifier is generated.

• overrides (dict[str, str]):

  A dictionary of overrides for the collections. The keys are collection names, the values are collection IDs. The option is managed by Runners, but can be provided manually.

Core-API parameters

• with_logs (bool):

  If set return prepare logs if True after end

• debug_mode (bool):

  If set, displays additional information about errors

• with_show (bool):

  Show results in logs

• profile_mode (str):

  If set, provides more information in logs. Possible modes: no, all, time, df_info, df_show

Stop

The stop hook is a simple proxy to the task_stop function. It stops the task and does not have any options.

Results

Results is represented as a list of malevich.results.core.CoreResult objects. Each object contains a payload which can be an asset, a dataframe or a list of such. Use .get_df(), .get_binary(), .get_dfs() and .get_binary_dir() to extract the results.

Configure

You can mark an operation to be run on a specific platform. To do so, set platform and platform_settings parameters in the .configure() method. The platform parameter can be base or vast. The platform_settings parameter is a dictionary of settings for the platform. The settings are platform-specific.