-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
RFC: dagstermill io use io manager for output notebook (#4490)
depends on #4500 prior context: https://dagster.phacility.com/D6145 **User-facing changes:** (1) Deprecating `output_notebook` in favor of `output_notebook_name`. - Using the old property `output_notebook` would require "file_manager" and result in a FileHandle output - no breaking change to existing users. - Using the new property `output_notebook_name` would result in a bytes output and require "output_notebook_io_manager", see details in (2) (2) With the new param, it requires a dedicated IO manager for output notebook. When `output_notebook` or `output_notebook_name` is specified, "output_notebook_io_manager" is required as a resource. - we provide a built-in `local_output_notebook_io_manager` for handling local output notebook materialization. - **New capabilities:** Users now can customize their own "output_notebook_io_manager" by extending `OutputNotebookIOManager`. This enables use cases: - users who want better asset key control can override `get_output_asset_key`. - users who want precise IO control over the output notebook can customize their own `handle_output` and `load_input`, e.g. they can control the file name of the output notebook. - users who want to attach more meaningful metadata can yield EventMetadataEntry in their own `handle_output` method this is also fixing #3380
- Loading branch information
Showing
6 changed files
with
293 additions
and
140 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
74 changes: 74 additions & 0 deletions
74
python_modules/libraries/dagstermill/dagstermill/io_managers.py
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,74 @@ | ||
| import os | ||
| from pathlib import Path | ||
| from typing import Any, List, Optional | ||
|
|
||
| from dagster import check | ||
| from dagster.config.field import Field | ||
| from dagster.core.definitions.event_metadata import EventMetadataEntry | ||
| from dagster.core.definitions.events import AssetKey | ||
| from dagster.core.execution.context.input import InputContext | ||
| from dagster.core.execution.context.output import OutputContext | ||
| from dagster.core.storage.io_manager import IOManager, io_manager | ||
| from dagster.utils import mkdir_p | ||
|
|
||
|
|
||
| class OutputNotebookIOManager(IOManager): | ||
| def __init__(self, asset_key_prefix: Optional[List[str]] = None): | ||
| self.asset_key_prefix = asset_key_prefix if asset_key_prefix else [] | ||
|
|
||
| def get_output_asset_key(self, context: OutputContext): | ||
| return AssetKey([*self.asset_key_prefix, f"{context.step_key}_output_notebook"]) | ||
|
|
||
| def handle_output(self, context: OutputContext, obj: bytes): | ||
| raise NotImplementedError | ||
|
|
||
| def load_input(self, context: InputContext) -> Any: | ||
| raise NotImplementedError | ||
|
|
||
|
|
||
| class LocalOutputNotebookIOManager(OutputNotebookIOManager): | ||
| """Built-in IO Manager for handling output notebook.""" | ||
|
|
||
| def __init__(self, base_dir: str, asset_key_prefix: Optional[List[str]] = None): | ||
| super(LocalOutputNotebookIOManager, self).__init__(asset_key_prefix=asset_key_prefix) | ||
| self.base_dir = base_dir | ||
| self.write_mode = "wb" | ||
| self.read_mode = "rb" | ||
|
|
||
| def _get_path(self, context: OutputContext) -> str: | ||
| """Automatically construct filepath.""" | ||
| keys = context.get_run_scoped_output_identifier() | ||
| return str(Path(self.base_dir, *keys).with_suffix(".ipynb")) | ||
|
|
||
| def handle_output(self, context: OutputContext, obj: bytes): | ||
| """obj: bytes""" | ||
| check.inst_param(context, "context", OutputContext) | ||
|
|
||
| # the output notebook itself is stored at output_file_path | ||
| output_notebook_path = self._get_path(context) | ||
| mkdir_p(os.path.dirname(output_notebook_path)) | ||
| with open(output_notebook_path, self.write_mode) as dest_file_obj: | ||
| dest_file_obj.write(obj) | ||
| yield EventMetadataEntry.fspath(path=output_notebook_path, label="path") | ||
|
|
||
| def load_input(self, context) -> bytes: | ||
| check.inst_param(context, "context", InputContext) | ||
| # pass output notebook to downstream solids as File Object | ||
| with open(self._get_path(context.upstream_output), self.read_mode) as file_obj: | ||
| return file_obj.read() | ||
|
|
||
|
|
||
| @io_manager( | ||
| config_schema={ | ||
| "asset_key_prefix": Field(str, is_required=False), | ||
| "base_dir": Field(str, is_required=False), | ||
| }, | ||
| ) | ||
| def local_output_notebook_io_manager(init_context): | ||
| """Built-in IO Manager that handles output notebooks.""" | ||
| return LocalOutputNotebookIOManager( | ||
| base_dir=init_context.resource_config.get( | ||
| "base_dir", init_context.instance.storage_directory() | ||
| ), | ||
| asset_key_prefix=init_context.resource_config.get("asset_key_prefix", []), | ||
| ) |
Oops, something went wrong.
b678a82There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Successfully deployed to the following URLs: