Class interface for running custom algorithms and workflows on STAC Items
Project description
STAC Task (stac-task)
- Quickstart for Creating New Tasks
- Task Input
- Full Process Definition Example
- Migration
- Development
- Contributing
This Python library consists of the Task class, which is used to create custom tasks based on a "STAC In, STAC Out" approach. The Task class acts as wrapper around custom code and provides several convenience methods for modifying STAC Items, creating derived Items, and providing a CLI.
This library is based on a branch of cirrus-lib except aims to be more generic.
Quickstart for Creating New Tasks
from typing import Any
from stactask import Task, DownloadConfig
class MyTask(Task):
name = "my-task"
description = "this task does it all"
def validate(self, payload: dict[str, Any]) -> bool:
return len(self.items) == 1
def process(self, **kwargs: Any) -> list[dict[str, Any]]:
item = self.items[0]
# download a datafile
item = self.download_item_assets(
item,
config=DownloadConfig(include=['data'])
)
# operate on the local file to create a new asset
item = self.upload_item_assets_to_s3(item)
# this task returns a single item
return [item.to_dict(include_self_link=True, transform_hrefs=False)]
Task Input
Field Name | Type | Description |
---|---|---|
type | string | Must be FeatureCollection |
features | [Item] | A list of STAC Item |
process | ProcessDefinition | A Process Definition |
ProcessDefinition Object
A STAC task can be provided additional configuration via the 'process' field in the input ItemCollection.
Field Name | Type | Description |
---|---|---|
description | string | Optional description of the process configuration |
upload_options | UploadOptions | Options used when uploading assets to a remote server |
tasks | Map<str, Map> | Dictionary of task configurations. A list of task configurations is supported for backwards compatibility reasons, but a dictionary should be preferred. |
UploadOptions Object
Field Name | Type | Description |
---|---|---|
path_template | string | REQUIRED A string template for specifying the location of uploaded assets |
public_assets | [str] | A list of asset keys that should be marked as public when uploaded |
headers | Map<str, str> | A set of key, value headers to send when uploading data to s3 |
collections | Map<str, str> | A mapping of output collection name to a JSONPath pattern (for matching Items) |
s3_urls | bool | Controls if the final published URLs should be an s3 (s3://bucket/key) or https URL |
path_template
The path_template string is a way to control the output location of uploaded assets from a STAC Item using metadata from the Item itself.
The template can contain fixed strings along with variables used for substitution.
See the PySTAC documentation for LayoutTemplate
for a list of supported template variables and their meaning.
collections
The collections dictionary provides a collection ID and JSONPath pattern for matching against STAC Items. At the end of processing, before the final STAC Items are returned, the Task class can be used to assign all of the Items to specific collection IDs. For each Item the JSONPath pattern for all collections will be compared. The first match will cause the Item's Collection ID to be set to the provided value.
For example:
"collections": {
"landsat-c2l2": "$[?(@.id =~ 'LC08.*')]"
}
In this example, the task will set any STAC Items that have an ID beginning with "LC08" to the landsat-c2l2
collection.
See JSONPath Online Evaluator to experiment with JSONPath and regex101 to experiment with regex.
tasks
The tasks field is a dictionary with an optional key for each task. If present, it contains
a dictionary that is converted to a set of keywords and passed to the Task's process
function.
The documentation for each task will provide the list of available parameters.
{
"tasks": {
"task-a": {
"param1": "value1"
},
"task-c": {
"param2": "value2"
}
}
}
In the example above a task named task-a
would have the param1=value1
passed as a keyword, while task-c
would have param2=value2
passed. If there were a task-b
to be run it would not be passed any keywords.
TaskConfig Object
DEPRECATED: tasks
should be a dictionary of parameters, with task names as keys. See tasks for more information.
A Task Configuration contains information for running a specific task.
Field Name | Type | Description |
---|---|---|
name | str | REQUIRED Name of the task |
parameters | Map<str, str> | Dictionary of keyword parameters that will be passed to the Tasks process function |
Full Process Definition Example
Process definitions are sometimes called "Payloads":
{
"description": "My process configuration",
"collections": {
"landsat-c2l2": "$[?(@.id =~ 'LC08.*')]"
},
"upload_options": {
"path_template": "s3://my-bucket/${collection}/${year}/${month}/${day}/${id}"
},
"tasks": {
"task-name": {
"param": "value"
}
}
}
Migration
0.4.x -> 0.5.x
In 0.5.0, the previous use of fsspec to download Item Assets has been replaced with the stac-asset library. This has necessitated a change in the parameters that the download methods accept.
The primary change is that the Task methods download_item_assets
and
download_items_assets
(items plural) now accept fewer explicit and implicit
(kwargs) parameters.
Previously, the methods looked like:
def download_item_assets(
self,
item: Item,
path_template: str = "${collection}/${id}",
keep_original_filenames: bool = False,
**kwargs: Any,
) -> Item:
but now look like:
def download_item_assets(
self,
item: Item,
path_template: str = "${collection}/${id}",
config: Optional[DownloadConfig] = None,
) -> Item:
Similarly, the asset_io
package methods were previously:
async def download_item_assets(
item: Item,
assets: Optional[list[str]] = None,
save_item: bool = True,
overwrite: bool = False,
path_template: str = "${collection}/${id}",
absolute_path: bool = False,
keep_original_filenames: bool = False,
**kwargs: Any,
) -> Item:
and are now:
async def download_item_assets(
item: Item,
path_template: str = "${collection}/${id}",
config: Optional[DownloadConfig] = None,
) -> Item:
Additionally, kwargs
keys were set to pass configuration through to fsspec. The most common
parameter was requester_pays
, to set the Requester Pays flag in AWS S3 requests.
Many of these parameters can be directly translated into configuration passed in a
DownloadConfig
object, which is just a wrapper over the stac_asset.Config
object.
Migration of these various parameters to DownloadConfig
are as follows:
assets
: setinclude
requester_pays
: sets3_requester_pays
= Truekeep_original_filenames
: setfile_name_strategy
toFileNameStrategy.FILE_NAME
if True orFileNameStrategy.KEY
if Falseoverwrite
: setoverwrite
save_item
: none, Item is always savedabsolute_path
: none. To create or retrieve the Asset hrefs as absolute paths, use eitherItem#make_all_asset_hrefs_absolute()
orAsset#get_absolute_href()
Development
Clone, install in editable mode with development requirements, and install the pre-commit hooks:
git clone https://github.com/stac-utils/stac-task
cd stac-task
pip install -e '.[dev]'
pre-commit install
To run the tests:
pytest
To lint all the files:
pre-commit run --all-files
Contributing
Use Github issues and pull requests.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.