julwin opened a new issue, #28304:
URL: https://github.com/apache/airflow/issues/28304
### Apache Airflow version
2.5.0
### What happened
When using dynamic task mapping, I tried to update an airflow dataset for
every file in a S3 location.
The URI for the dataset should be generated inside of a Taskflow task,
containing the basename of the processed file.
The
```python
@task()
def extract_s3_destination():
dataset = Dataset("s3://dataset-bucket/example.csv")
return {"filename": path, "dest_key": dest_key, "outlets": [dataset]}
```
When handing this over to `expand_kwargs` (even with strict=False), this
error occurs:
```python
filenames = extract_s3_destination.expand(path=XComArg(list_modified_files))
copy_to_s3 = LocalFilesystemToS3Operator.partial(
task_id="copy_file_to_s3",
aws_conn_id="s3",
gzip=True,
replace=True,
dest_bucket="{{ conn.s3.login }}",
).expand_kwargs(filenames, strict=False)
```
```
{manager.py:56} WARNING - DatasetModel
Dataset(uri='unprocessed/customfilename20221212.zip', extra=None) not found
```
The same happens when
### What you think should happen instead
The Dataset should be updated if not exists or created, the way you would
expect it after reading these examples from the Documentation:
```
from airflow import Dataset
with DAG(...):
MyOperator(
# this task updates example.csv
outlets=[Dataset("s3://dataset-bucket/example.csv")],
...,
)
with DAG(
# this DAG should be run when example.csv is updated (by dag1)
schedule=[Dataset("s3://dataset-bucket/example.csv")],
...,
):
...
```
### How to reproduce
_No response_
### Operating System
kubernetes / airflow executor / default image
### Versions of Apache Airflow Providers
_No response_
### Deployment
Official Apache Airflow Helm Chart
### Deployment details
_No response_
### Anything else
_No response_
### Are you willing to submit PR?
- [X] Yes I am willing to submit a PR!
### Code of Conduct
- [X] I agree to follow this project's [Code of
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
