This package offers both synchronous and asynchronous implementations of a standardized Python API to communicate with the dtool lookup server.
Project description
Python API for interacting with dtool lookup server.
This package offers a class-based asynchronous lookup API within dtool_lookup_api.core.LookupClient, a simple class-less wrapper around it at dtool_lookup_api.asynchronous, and a synchronous interface on top at dtool_lookup_api.synchronous.
Direct imports of utility functions from dtool_lookup_api in the examples below forward to the synchronous API variant.
Installation
To install the dtool_lookup_api package.
pip install dtool_lookup_api
This package depends on a dtool-lookup-server instance to talk to.
Configuration
The API needs to know the URL of the lookup server
export DTOOL_LOOKUP_SERVER_URL=https://localhost:5000
You may also need specify an access token generated on the server
export DTOOL_LOOKUP_SERVER_TOKEN=$(flask user token testuser)
Instead of specifying the access token directly, it is also possible to provide
export DTOOL_LOOKUP_SERVER_TOKEN_GENERATOR_URL=https://localhost:5001
export DTOOL_LOOKUP_SERVER_USERNAME=my-username
export DTOOL_LOOKUP_SERVER_PASSWORD=my-password
for the API to request a token. This, however, is intended only for testing purposes and strongly discouraged in a production environment, as your password would reside within environment variables or the dtool config file as clear text.
Our recommended setup is a combination of
export DTOOL_LOOKUP_SERVER_URL=https://localhost:5000
export DTOOL_LOOKUP_SERVER_TOKEN_GENERATOR_URL=https://localhost:5001
in the config. If used interactively, the API will then ask for your credentials at the first interaction and cache the provided values for this session, i.e.
In [1]: from dtool_lookup_api import query
...: res = query(
...: {
...: 'readme.owners.name': {'$regex': '^Testing User$'},
...: }
...: )
Authentication URL https://localhost:5001/token username:my-username
Authentication URL https://localhost:5001/token password:
In [2]: res
Out[2]:
[{'base_uri': 'smb://test-share',
'created_at': 'Sun, 08 Nov 2020 18:38:40 GMT',
'creator_username': 'jotelha',
'dtoolcore_version': '3.17.0',
'frozen_at': 'Wed, 11 Nov 2020 17:20:30 GMT',
'name': 'simple_test_dataset',
'tags': [],
'type': 'dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
In [3]: from dtool_lookup_api import all
...: all()
Out[4]:
[{'base_uri': 'smb://test-share',
'created_at': 1604860720.736269,
'creator_username': 'jotelha',
'frozen_at': 1604921621.719575,
'name': 'simple_test_dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
Credentials caching and interactive prompting are turned off with
In [1]: import dtool_lookup_api.core.config
...: dtool_lookup_api.core.config.Config.interactive = False
...: dtool_lookup_api.core.config.Config.cache = False
In [2]: from dtool_lookup_api import all
...: all()
...
RuntimeError: Authentication failed
For testing purposes, it is possible to disable SSL certificates validation with
export DTOOL_LOOKUP_SERVER_VERIFY_SSL=false
As usual, these settings may be specified within the default dtool configuration file as well, i.e. at ~/.config/dtool/dtool.json
{
"DTOOL_LOOKUP_SERVER_TOKEN_GENERATOR_URL": "https://localhost:5001/token",
"DTOOL_LOOKUP_SERVER_URL": "https://localhost:5000"
}
List all datasets
To list all registered datasets
In [1]: from dtool_lookup_api import all
...: res = all()
In [2]: res
Out[2]:
[{'base_uri': 'smb://test-share',
'created_at': 1604860720.736269,
'creator_username': 'jotelha',
'frozen_at': 1604921621.719575,
'name': 'simple_test_dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
Looking up datasets by UUID
To lookup URIs from a dataset UUID within Python
In [1]: from dtool_lookup_api import lookup
...: uuid = "1a1f9fad-8589-413e-9602-5bbd66bfe675"
...: res = lookup(uuid)
In [2]: res
Out[2]:
[{'base_uri': 'smb://test-share',
'created_at': 1604860720.736269,
'creator_username': 'jotelha',
'frozen_at': 1604921621.719575,
'name': 'simple_test_dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
Full text searching
Full text search for the word “test”
In [1]: from dtool_lookup_api import search
...: res = search("test")
In [2]: res
Out[2]:
[{'base_uri': 'smb://test-share',
'created_at': 1604860720.736,
'creator_username': 'jotelha',
'dtoolcore_version': '3.17.0',
'frozen_at': 1605027357.308,
'name': 'simple_test_dataset',
'tags': [],
'type': 'dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
Manifest
Request the manifest of a particular dataset by URI
In [1]: from dtool_lookup_api import manifest
...: uri = 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675'
...: res = manifest(uri)
In [2]: res
Out[2]:
{'dtoolcore_version': '3.17.0',
'hash_function': 'md5sum_hexdigest',
'items': {'eb58eb70ebcddf630feeea28834f5256c207edfd': {'hash': '2f7d9c3e0cfd47e8fcab0c12447b2bf0',
'relpath': 'simple_text_file.txt',
'size_in_bytes': 17,
'utc_timestamp': 1605027357.284966}}}
Readme
Request the readme cotent of a particular dataset by URI
In [1]: from dtool_lookup_api import readme
..: res = readme('smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675')
In [2]: res
Out[2]:
{'creation_date': '2020-11-08',
'description': 'testing description',
'expiration_date': '2022-11-08',
'funders': [{'code': 'testing_code',
'organization': 'testing_organization',
'program': 'testing_program'}],
'owners': [{'email': 'testing@test.edu',
'name': 'Testing User',
'orcid': 'testing_orcid',
'username': 'testing_user'}],
'project': 'testing project'}
Direct mongo language queries
To list all datasets at a certain base URI with their name matching some regular expression pattern, send a direct mongo language query to the server with
In [15]: from dtool_lookup_api import query
...: res = query(
...: {
...: 'base_uri': 'smb://test-share',
...: 'name': {'$regex': 'test'},
...: }
...: )
In [16]: res
Out[16]:
[{'base_uri': 'smb://test-share',
'created_at': 'Sun, 08 Nov 2020 18:38:40 GMT',
'creator_username': 'jotelha',
'dtoolcore_version': '3.17.0',
'frozen_at': 'Tue, 10 Nov 2020 16:55:57 GMT',
'name': 'simple_test_dataset',
'tags': [],
'type': 'dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
It is possible to search readme content via
In [21]: from dtool_lookup_api import query
...: res = query(
...: {
...: 'readme.owners.name': {'$regex': '^Testing User$'},
...: }
...: )
In [22]: res
Out[22]:
[{'base_uri': 'smb://test-share',
'created_at': 'Sun, 08 Nov 2020 18:38:40 GMT',
'creator_username': 'jotelha',
'dtoolcore_version': '3.17.0',
'frozen_at': 'Tue, 10 Nov 2020 16:55:57 GMT',
'name': 'simple_test_dataset',
'tags': [],
'type': 'dataset',
'uri': 'smb://test-share/1a1f9fad-8589-413e-9602-5bbd66bfe675',
'uuid': '1a1f9fad-8589-413e-9602-5bbd66bfe675'}]
This requires the server-side dtool-lookup-server-direct-mongo-plugin.
TODO: Response from server-side direct mongo plugin still yields dates as strings. Fix within https://github.com/IMTEK-Simulation/dtool-lookup-server-direct-mongo-plugin.
Usage on Jupyter notebook
The current implementation via asgiref.async_to_sync (https://github.com/django/asgiref) hinders the use of the synchronous interface within Jupyter notebooks. Directly use the asynchronous api instead
import dtool_lookup_api.asynchronous as dl
res = await dl.query({
'base_uri': 'smb://test-share',
'name': {'$regex': 'test'},
})
The drawback of the above approach is that the same code doesn’t work in python and in jupyter (await outsite of a function is a syntax error in non-interactive python context). The code below can be executed in both contexts:
import dtool_lookup_api.asynchronous as dl
if asyncio.get_event_loop().is_running():
# then we are in jupyter notebook
# this allows nested event loops, i.e. calls to asyncio.run inside the notebook as well
# This way, the same code works in notebook and python
import nest_asyncio
nest_asyncio.apply()
def query(query_dict):
return asyncio.run(dl.query(query_dict))
query({
'base_uri': 'smb://test-share',
'name': {'$regex': 'test'},
})
See https://github.com/jupyter/notebook/issues/3397#issuecomment-419386811, https://ipython.readthedocs.io/en/stable/interactive/autoawait.html
Testing
Tests require the presence of a working dtool lookup server ecosystem.
The testing workflow within .github/workflows/test.yml
uses the
dtool-lookup-server-container-composition
to provide a mock ecosystem. It is possible to run the workflow locally
with the help of docker and
act.
After installing and configuring act, run
act -P ubuntu-latest=catthehacker/ubuntu:full-latest -s GITHUB_TOKEN=$GITHUB_TOKEN -W .github/workflows/test.yml --bind
from within this repository. $GITHUB_TOKEN
must hold a valid
access token.
The user must be member of the docker
group.
The --bind
option avoids quirky permission errors by running
the test in the current directory. This will however result in the
local creation of two subdirectories dtool-lookup-server-container-composition
and workflow
during testing, which may be removed with
rm -rf dtool-lookup-server-container-composition
sudo rm -rf workflow
eventually. All tests have been confirmed to work with the
catthehacker/ubuntu:full-20.04
runner.
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.
Source Distribution
Built Distribution
Hashes for dtool_lookup_api-0.7.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | ef9949dfeb85141c5a444ec6b21719835cf6988773ed343a600733716df91029 |
|
MD5 | 15c4e3678f998fd07777bf24088a77d9 |
|
BLAKE2b-256 | a56e8c3edee873648c9108bbd9b0abde5fb726f78a95c156bbca18d46135f3e8 |