databricks / databricks-sdk-py Goto Github PK

View Code? Open in Web Editor NEW

300.0 17.0 96.0 10.13 MB

Databricks SDK for Python (Beta)

Home Page: https://databricks-sdk-py.readthedocs.io/

License: Apache License 2.0

Makefile 0.27% Python 99.73%

databricks python databricks-sdk

databricks-sdk-py's People

Contributors

Stargazers

Watchers

Forkers

octodemo n3n33 erdal-pb johnjdyer dby-tmwctw susodapop reinsight-ai grusin-db rafaelvp-db manish-jain-psl ahcamachod souvik-databricks judahrand arnholdd clementboscoa alexott md8github tanay0nspark pranavchiku lm0007 maiconq michaelspece nodejsmith slakshmangit rakshitpandit scouterdefenseadmin01 alex7c4 jaina15 vadim caffeinelover1012 xicez chaitti smurching budrobundy69 shuhei-fujita jon-proietti-nutrien ahmed-hasib-cityfm amirkdv-stripe williamdphillips s-udhaya shorthillsai vsamoilov future-outlier patrickleahey techinstl dbkegley aashishmalhotra erithtotl mustious mrteale xinjiezhen-db fpgmaas jschuller seshness pawaritl neilbest-db sandormatyas ankitaj281196 weilando ramji1987 cornzyblack sondrebouvet jbrill rportilla-databricks ayush1399 karlinhanikitenko juanlamadrid20 thbeh malvik01 paolorechia kimberlyma alexdy2007 maruschin 0xzra bonnetn mbosmanabn chaos-genius rongyj sdonohoo-db brnassivet carloscaravantsz douglasfraser-ikea falaki anhhchu dakinggg trustaff wsilveira-usp mrjefftang chrisgravel-db nkvuong wanderatdusk6 miles-monoid

databricks-sdk-py's Issues

Notebook export with "direct_download=True" fails

Setting the direct_download flag to True in the API returns an unencoded version of the notebook but throws a non-descript error today.

Can't create Single User Cluster

no way to set data_security_mode via WorkspaceClient.clusters.create

had to fall back to ClusterApi

Use the sdk from within databricks?

Hello,

I am looking for the simplest to automate submitting a large amount of jobs. I see that databricks_cli https://docs.databricks.com/dev-tools/cli/index.html does not seem to be in active development, so I am looking at this project as a solution for automation.

I would like to be able to run this sdk from within my databricks workspace, but when I try the example https://docs.databricks.com/dev-tools/sdk-python.html#get-started-with-the-databricks-sdk-for-python, I get the error message default auth: cannot configure default credentials

I assume that is because databricks-sdk-py is unaware that I am running from within my databricks workspace. My question is, is there a way to make the module recognize that and assume my credentials?

listing networks don't work

here is the code:
account = AccountClient(host="https://accounts.cloud.databricks.com",account_id=db_account_id, username=db_username, password=db_password)

account.networks.list()

I used e2-certification account.

here is the error:

ValueError Traceback (most recent call last)
in <cell line: 8>()
6 verifier = Verifier(db_account_id, db_username, db_password, aws_key, aws_secret)
7
----> 8 verifier.run_private_link_check("3324600082051037")

in run_private_link_check(self, workspace_id)
80
81 def run_private_link_check(self, workspace_id:str):
---> 82 db_network_id = self.get_db_network_id(workspace_id)
83 vpc_id = self.get_vpc_id(workspace_id)
84 vpc_endpoints_ids = self.get_vpc_endpoint_ids(network_id)

in get_db_network_id(self, workspace_id)
24 def get_db_network_id(self, workspace_id:str):
25 #get network id.
---> 26 for network in self.account.networks.list():
27 if network.workspace_id == self.workspace_id:
28 return network.network_id

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/deployment.py in list(self)
1292
1293 json = self._api.do('GET', f'/api/2.0/accounts/{self._api.account_id}/networks')
-> 1294 return [Network.from_dict(v) for v in json]
1295
1296

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/deployment.py in (.0)
1292
1293 json = self._api.do('GET', f'/api/2.0/accounts/{self._api.account_id}/networks')
-> 1294 return [Network.from_dict(v) for v in json]
1295
1296

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/deployment.py in from_dict(cls, d)
653 vpc_id=d.get('vpc_id', None),
654 vpc_status=VpcStatus(d['vpc_status']) if 'vpc_status' in d else None,
--> 655 warning_messages=[NetworkWarning.from_dict(v)
656 for v in d['warning_messages']] if 'warning_messages' in d else None,
657 workspace_id=d.get('workspace_id', None))

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/deployment.py in (.0)
653 vpc_id=d.get('vpc_id', None),
654 vpc_status=VpcStatus(d['vpc_status']) if 'vpc_status' in d else None,
--> 655 warning_messages=[NetworkWarning.from_dict(v)
656 for v in d['warning_messages']] if 'warning_messages' in d else None,
657 workspace_id=d.get('workspace_id', None))

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/deployment.py in from_dict(cls, d)
710 def from_dict(cls, d: Dict[str, any]) -> 'NetworkWarning':
711 return cls(warning_message=d.get('warning_message', None),
--> 712 warning_type=WarningType(d['warning_type']) if 'warning_type' in d else None)
713
714

/usr/lib/python3.9/enum.py in call(cls, value, names, module, qualname, type, start)
358 """
359 if names is None: # simple value lookup
--> 360 return cls.new(cls, value)
361 # otherwise, functional API: we're creating a new Enum type
362 return cls.create(

/usr/lib/python3.9/enum.py in new(cls, value)
676 ve_exc = ValueError("%r is not a valid %s" % (value, cls.qualname))
677 if result is None and exc is None:
--> 678 raise ve_exc
679 elif exc is None:
680 exc = TypeError(

ValueError: 'vpc' is not a valid WarningType

Error using Tables API

Errors received from trying to list tables in the workspace

import os

from databricks.sdk import WorkspaceClient


if __name__ == "__main__":  
    w = WorkspaceClient(
        host="https://2111501043581247.7.gcp.databricks.com/",
        token=os.environ['PAT_JAPAN'])
    w.jobs.delete()

    for t in w.tables.list():
        print(t)

Traceback (most recent call last):
  File "/Users/gant.kuln/test_databricks_py/helloworld.py", line 10, in <module>
    for k in w.tables.list():
  File "/Users/gant.kuln/miniconda3/envs/test_databricks_py/lib/python3.10/site-packages/databricks/sdk/service/unitycatalog.py", line 3081, in list
    json = self._api.do('GET', '/api/2.1/unity-catalog/tables', query=query)
  File "/Users/gant.kuln/miniconda3/envs/test_databricks_py/lib/python3.10/site-packages/databricks/sdk/client.py", line 416, in do
    raise DatabricksError(**response.json())
TypeError: DatabricksError.__init__() got an unexpected keyword argument 'details'
(test_databricks_py) gant.kuln@JDY0V6H3X2 test_databricks_py %

[BUG] in jobs.py: TypeError: Object of type Library is not JSON serializable

databricks-sdk-py/databricks/sdk/service/jobs.py

Line 703 in 1e3612d

if self.libraries: body['libraries'] = [v for v in self.libraries]

the Library objects are put in the dict without modifying. This leads to

TypeError: Object of type Library is not JSON serializable

when calling JobsAPI.create with a Libray object set in a JobTask

instead the line should be

if self.libraries: body['libraries'] = [v.as_dict() for v in self.libraries]

OAuth breaks if a bad .netrc file is present

Description

There is a hidden behaviour in requests where a .netrc will silently override provided authentication headers unless manually overridden.

We first observed this in dbt-databricks because of databricks/dbt-databricks#337. Fixing it however required two changes: one to dbt-databricks and one to databricks-sdk-py. The fix is simple: override the default behaviour of requests by supplying a custom AuthBase class.

The fix to dbt-databricks has to do with Python models which use requests and a REST API (instead of thrift server) to check if the all-purpose cluster is running.
The fix to databricks-sdk-py (this issue) is around OAuth, because the SDK uses requests to perform the OAuth handshake. This step is only needed where OAuth is used to authenticate dbt-databricks

Steps to Reproduce

There is probably an easier way to reproduce this with a short script. But I'm using what I found while developing dbt-databricks.

Checkout this code from dbt-databricks databricks/dbt-databricks#338. This branch incorporates the first fix I described above.
Add an intentionally bad ~/.netrc to your workstation, like this:

machine <my-workspace>.cloud.databricks.com
login token
password <expired_token>

Try to run the Python test_python_uc_sql_endpoint integration test after updating _build_databricks_cluster_target in tests/profiles.py to comment out the "token" key. This forces dbt-databricks to use OAuth instead.
Observe that the test fails with this error:

E           ValueError: b'{"errorCode":"invalid_client","errorSummary":"Invalid value for \'client_id\' parameter.","errorLink":"invalid_client","errorId":"oaeLJQz1r35SrSNVtVcJUig0A","errorCauses":[]}'

This happens because without the override applied, the SDK includes authentication headers for a REST API request that doesn't require authentication and the server kicks back an invalid value for 'client_id' error.

I'm about to open a PR that fixes this.

There is a related issue on databricks-sql-python (which implements its own oauth process). That fix is the same as this one.

Better interface for client credentials

Right now it is quite easy to use OAuthClient. It encapsulates Azure vs AWS differences and it returns a credentials_provider, which is nice. However, ClientCredentials is very crude and does nothing of the above.

One proposal [preferred] would be to encapsulate ClientCredentials within OAuthClient and have a single client to worry about.

An alternative would be to have a ClientCredentialsClient that provides a cloud-agnostic API and returns a credentials_provider.

Make error handling more user friendly for common accidental mistakes by showing resolution steps

This is mostly a nice to have. Example -- running dbutils.help() without setting up cluster throws:
databricks.sdk.core.DatabricksError: Missing required field: cluster_id

We could add something along the lines of how to setup cluster, point to the relevant docs.

Full error stack:

Improve error handling or validation for misconfigured host URL

I mistakenly supplied "https://accounts.gcp.databricks.com/workspaces/2111501043581247" as the host URL for WorkspaceClient. But it wasn't obvious from the error message what was wrong. Would be great if the SDK can detect it and provide actionable error message.

Missing init_scripts in Cluster Creation?

Hello,

I believe the current method to create a cluster in the SDK is missing init_scripts (which is a part of the APIs- Create new cluster). I am currently using a custom class which adds the init_scripts part to the body of the request.

Is adding init_scripts during cluster creation through the SDK planned?

Thank you.

Error while using filter parameter in list operation

Hello!

I'm not sure if I'm doing something wrong, but, while using the groups' list operation from WorkspaceClient:

03/05/2023 04:00:10 PM  :::DEBUG::::   Loaded from environment
03/05/2023 04:00:10 PM  :::DEBUG::::   Attempting to configure auth: pat
03/05/2023 04:00:10 PM  :::DEBUG::::   Starting new HTTPS connection (1): xxxxxxxxxxxx.cloud.databricks.com:443
03/05/2023 04:00:12 PM  :::DEBUG::::   https://xxxxxxxxxxxx.cloud.databricks.com:443 "GET /api/2.0/preview/scim/v2/Groups?filter=displayName%2Beq%2Bmy_group HTTP/1.1" 400 None
03/05/2023 04:00:12 PM  :::DEBUG::::   GET /api/2.0/preview/scim/v2/Groups?filter=displayName+eq+my_group
< 400 Bad Request
< {
<   "detail": "Given filter operator is not supported.",
<   "schemas": [
<     "urn:ietf:params:scim:api:messages:2.0:Error"
<   ],
<   "scimType": "InvalidFilter",
<   "status": "400"
< }
Traceback (most recent call last):
  File "<redacted>/main.py", line 6, in <module>
    databricks_groups = databricks_workspace.groups.list(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/databricks/sdk/service/iam.py", line 1291, in list
    json = self._api.do('GET', '/api/2.0/preview/scim/v2/Groups', query=query)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/databricks/sdk/core.py", line 753, in do
    raise self._make_nicer_error(status_code=response.status_code, **payload) from None
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: ApiClient._make_nicer_error() missing 1 required positional argument: 'message'

If I try the exact same filter using cURL, for example, everything works fine.

This is the main.py:

from databricks.sdk import WorkspaceClient
from modules.config.log import logger

databricks_workspace = WorkspaceClient()

databricks_groups = databricks_workspace.groups.list(
    filter="displayName+eq+my_group"
)

I'm using the SDK version 0.1.2 and Python version 3.11.3.

Log to stdErr from requests.get

When sending the request to get the token, there is a log to stderr that would be awesome to silence as it mixes with clients:

Please loosen the dependency on `requests`

This package currently depends on requests <2.29.0,>=2.28.1, which is a very tight range and unfortunately has published vulnerabilities. Please widen the allowed range (I suggest to requests <3,>=2.28.1 if forwards compatibility is important) so that this package does not force insecure dependent packages on its consumers.

Error "'LEGACY_SINGLE_USER_STANDARD' is not a valid DataSecurityMode" when trying to list available cluster names

Repro steps:

Install version 0.0.1 of the Databricks SDK for Python.
Run the following code:

from databricks.sdk import WorkspaceClient

w = WorkspaceClient(host = "https://db-sme-demo-docs.cloud.databricks.com", token = "REDACTED")

for c in w.clusters.list():
  print(c.cluster_name)

Expected:

A list of available cluster names in my workspace.

Actual:

This error:

Traceback (most recent call last):
  File "/Users/paul.cornell/databricks-python-sdk-demo/main.py", line 5, in <module>
    for c in w.clusters.list():
  File "/Users/paul.cornell/.local/share/virtualenvs/paul.cornell-Otax6dmi/lib/python3.9/site-packages/databricks/sdk/service/clusters.py", line 1908, in list
    return [ClusterInfo.from_dict(v) for v in json['clusters']]
  File "/Users/paul.cornell/.local/share/virtualenvs/paul.cornell-Otax6dmi/lib/python3.9/site-packages/databricks/sdk/service/clusters.py", line 1908, in <listcomp>
    return [ClusterInfo.from_dict(v) for v in json['clusters']]
  File "/Users/paul.cornell/.local/share/virtualenvs/paul.cornell-Otax6dmi/lib/python3.9/site-packages/databricks/sdk/service/clusters.py", line 410, in from_dict
    data_security_mode=DataSecurityMode(d['data_security_mode'])
  File "/usr/local/Cellar/[email protected]/3.9.15/Frameworks/Python.framework/Versions/3.9/lib/python3.9/enum.py", line 384, in __call__
    return cls.__new__(cls, value)
  File "/usr/local/Cellar/[email protected]/3.9.15/Frameworks/Python.framework/Versions/3.9/lib/python3.9/enum.py", line 702, in __new__
    raise ve_exc
ValueError: 'LEGACY_SINGLE_USER_STANDARD' is not a valid DataSecurityMode

Add examples for `grants` and other submodules

I see there are methods in w.grants but they're not documented in the examples path of the repo.

Also would be nice to add a README.md in the examples folder (or add it to the root README) stating that the structure of the SDK is the same as the URLs of the REST API explorer. That would make it easier to know and navigate the complete functionality of the SDK

Query History examples missing

The Examples and the README is not helpful for using the SDK when it comes to Query History APIs.

The documentation and user guide needs to be improved else this is not useful and I have to use the python requests package and do everything by myself from scratch.

[BUG] Storage Credentials create

WorkspaceClient.storage_credentials.create() requires Metastore ID which the API is not expecting. Throws error 'CreateStorageCredential metastore_id can not be provided.' when provided and 'StorageCredentialsAPI. create() missing 1 required positional argument: 'metastore_id' when not provided

Python import error when using `auth_type="oauth-m2m"`

Import error when using auth_type="oauth-m2m"
When running this python statement:

a = AccountClient(auth_type="oauth-m2m", profile="E2CERTACCT")

I found the error below while running in the python debugger.

 File "/Users/xxxxxx.xxxxxxxx/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 23, in <module>
    from .azure import ARM_DATABRICKS_RESOURCE_ID, ENVIRONMENTS, AzureEnvironment
ImportError: attempted relative import with no known parent package

log:

2023-06-09 22:53:01,090 [databricks.sdk][INFO] loading E2CERTACCT profile from ~/.databrickscfg: host, account_id, client_id, client_secret
2023-06-09 22:53:01,090 [databricks.sdk][DEBUG] Ignoring pat auth, because oauth-m2m is preferred
2023-06-09 22:53:01,090 [databricks.sdk][DEBUG] Ignoring basic auth, because oauth-m2m is preferred
2023-06-09 22:53:01,090 [databricks.sdk][DEBUG] Ignoring metadata-service auth, because oauth-m2m is preferred
2023-06-09 22:53:01,090 [databricks.sdk][DEBUG] Attempting to configure auth: oauth-m2m
Traceback (most recent call last):
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/requests/models.py", line 971, in json
    return complexjson.loads(self.text, **kwargs)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/json/__init__.py", line 357, in loads
    return _default_decoder.decode(s)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/json/decoder.py", line 355, in raw_decode
    raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 415, in __call__
    header_factory = provider(cfg)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 60, in wrapper
    return func(cfg)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 114, in oauth_service_principal
    token_url=resp.json()["token_endpoint"],
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/requests/models.py", line 975, in json
    raise RequestsJSONDecodeError(e.msg, e.doc, e.pos)
requests.exceptions.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 775, in _init_auth
    self._header_factory = self._credentials_provider(self)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 421, in __call__
    raise ValueError(f'{auth_type}: {e}') from e
ValueError: oauth-m2m: Expecting value: line 1 column 1 (char 0)

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 496, in __init__
    self._init_auth()
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 780, in _init_auth
    raise ValueError(f'{self._credentials_provider.auth_type()} auth: {e}') from e
ValueError: default auth: oauth-m2m: Expecting value: line 1 column 1 (char 0)

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/douglas.moore/development/dba-helper/permissions-graph/extract_account_principals.py", line 44, in <module>
    a = AccountClient(auth_type="oauth-m2m", profile="E2CERTACCT", debug_headers=True, debug_truncate_bytes=300)
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/__init__.py", line 192, in __init__
    config = client.Config(host=host,
  File "/Users/douglas.moore/.pyenv/versions/3.8.12/lib/python3.8/site-packages/databricks/sdk/core.py", line 501, in __init__
    raise ValueError(message) from e
ValueError: default auth: oauth-m2m: Expecting value: line 1 column 1 (char 0). Config: host=https://accounts.cloud.databricks.com, account_id=deadbeef-deadbeef-deadbeef, client_id=dead999-1234-4321-9999-deadbeef, client_secret=***, profile=E2CERTACCT, auth_type=oauth-m2m, debug_truncate_bytes=300, debug_headers=True

Capture resourceType from GET Groups API

The SCIM Groups API returns a value under meta.resourceType which can have a value of Group (account level group) or WorkspaceGroup (workspace local group).

Currently, the SDK does not capture this piece of information

@dataclass
class Group:
    display_name: str = None
    entitlements: 'List[ComplexValue]' = None
    external_id: str = None
    groups: 'List[ComplexValue]' = None
    id: str = None
    members: 'List[ComplexValue]' = None
    roles: 'List[ComplexValue]' = None

Note that the value is present when accessing the API directly

I propose we add the meta.resourceType field to the Group data class. This functionality is currently leveraged by the UC-Migration project

Note that I'm happy to make an attempt at contributing this functionality. I don't see any contribution guides for this project.

metastore.delete force does not force

I believe the issue is on the API side but
workspace_client.metastores.delete(id=metastoreId',force=True) will not force delete the metastore

I am having to send force as url parameter
workspace_client.metastores.delete(id=metastoreId+'?force=true',force=True) which is adding parameter force=true in url and not message body

Workspace Conf fails to get status

Hello,

When using the workspace_conf.get_status() in the WorkspaceClient, the method does not work and returns an AttributeError: type object 'dict' has no attribute 'from_dict'. I believe WorkspaceConf is just a dictionary and thus, it is raising an Attribute Error. Could you please verify and confirm if this is correct.

e.g. workspace_client.workspace_conf.get_status(keys="enableTokensConfig") does not work

Thank you.

Make databricks-sdk work out of the box in notebooks

Right now two steps are needed to make the sdk work from notebooks:

%pip install
manually passing a token to the sdk by calling internal dbutils APIs to get one

So we don't need to solve #1 right away, but maybe we can just bake the logic from #2 into the SDK itself, just like e.g. mlflow works out of the box without passing a token around manually? Would be nice if you could just type w = WorkspaceClient() in a notebook!

Query History list() does not pass filter_by correctly

Although it's not documented, the query history list endpoint seems to handle filter_by properly only if it is passed in the request body rather than as a query param:

Ran using databricks-sdk v0.1.5, requests v2.28.2:

In [2]: from databricks.sdk import WorkspaceClient
In [3]: client = WorkspaceClient(host=..., token=...)
In [4]: from databricks.sdk.service.sql import QueryFilter
In [6]: filter_by = QueryFilter.from_dict(
   ...:             {
   ...:                 "query_start_time_range": {
   ...:                     "start_time_ms": 0,
   ...:                     "end_time_ms": int(time.time() * 1000),
   ...:                 }
   ...:             }
   ...:         )

In [7]: filter_by
Out[7]: QueryFilter(query_start_time_range=TimeRange(end_time_ms=1683569801679, start_time_ms=0), statuses=None, user_ids=None, warehouse_ids=None)

In [9]: next(client.query_history.list(filter_by=filter_by))
---------------------------------------------------------------------------
DatabricksError                           Traceback (most recent call last)
Cell In[9], line 1
----> 1 next(client.query_history.list(filter_by=filter_by))

File ~/.pyenv/versions/3.10.9/lib/python3.10/site-packages/databricks/sdk/service/sql.py:2824, in QueryHistoryAPI.list(self, filter_by, include_metrics, max_results, page_token, **kwargs)
   2821 if page_token: query['page_token'] = request.page_token
   2823 while True:
-> 2824     json = self._api.do('GET', '/api/2.0/sql/history/queries', query=query)
   2825     if 'res' not in json or not json['res']:
   2826         return

File ~/.pyenv/versions/3.10.9/lib/python3.10/site-packages/databricks/sdk/core.py:753, in ApiClient.do(self, method, path, query, body)
    749 if not response.ok:
    750     # TODO: experiment with traceback pruning for better readability
    751     # See https://stackoverflow.com/a/58821552/277035
    752     payload = response.json()
--> 753     raise self._make_nicer_error(status_code=response.status_code, **payload) from None
    754 if not len(response.content):
    755     return {}

DatabricksError: Could not parse request object: Expected 'START_OBJECT' not 'VALUE_STRING'
 at [Source: (ByteArrayInputStream); line: 1, column: 15]
 at [Source: java.io.ByteArrayInputStream@794544b2; line: 1, column: 15]

but if I call:

In [53]: "res" in client.query_history._api.do('GET', '/api/2.0/sql/history/queries', body={"filter_by": filter_by.as_dict()})
Out[53]: True

the API works as expected.

Second, when using pagination with the query history endpoint, it doesn't seem to allow specifying page_token and filter_by at the same time:

In [55]: client.query_history._api.do('GET', '/api/2.0/sql/history/queries', body={"filter_by": filter_by.as_dict(), "page_token": "abc"})
---------------------------------------------------------------------------
DatabricksError                           Traceback (most recent call last)
Cell In[55], line 1
----> 1 client.query_history._api.do('GET', '/api/2.0/sql/history/queries', body={"filter_by": filter_by.as_dict(), "page_token": "abc"})

File ~/.pyenv/versions/3.10.9/lib/python3.10/site-packages/databricks/sdk/core.py:753, in ApiClient.do(self, method, path, query, body)
    749 if not response.ok:
    750     # TODO: experiment with traceback pruning for better readability
    751     # See https://stackoverflow.com/a/58821552/277035
    752     payload = response.json()
--> 753     raise self._make_nicer_error(status_code=response.status_code, **payload) from None
    754 if not len(response.content):
    755     return {}

DatabricksError: You can provide only one of 'page_token' or 'filter_by'

The current implementation doesn't remove filter_by on subsequent calls and thus doesn't paginate correctly when filter_by is passed in. Here, I patch the do call to pass query params as body, but this time we fail when we attempt to get the second page.

In [56]: original_do = client.api_client.do

In [57]: from unittest.mock import patch

In [62]: def patch_do(method, path, query = None, body = None):
    ...:     print(method,path,query,body)
    ...:     res = original_do(method, path, query = None, body=query)
    ...:     print("RES", res.keys())
    ...:     return res
    ...: 

In [64]: with patch.object(ApiClient, "do") as mock_do:
    ...:     mock_do.side_effect = patch_do
    ...:     it = client.query_history.list(filter_by=filter_by, max_results=1)
    ...:     next(it)
    ...:     next(it)
    ...: 
GET /api/2.0/sql/history/queries {'filter_by': {'query_start_time_range': {'end_time_ms': 1683569801679}}, 'max_results': 1} None
RES dict_keys(['next_page_token', 'has_next_page', 'res'])
GET /api/2.0/sql/history/queries {'filter_by': {'query_start_time_range': {'end_time_ms': 1683569801679}}, 'max_results': 1, 'page_token': 'CkwKJDAxZWRlZGFjLWJjNDktMTgyOS1hM2UwLTYwNDYwZTVkNjU4MBCD2ZHe/zAY09WStKW5tAUiEDcyNTJjYmE1NTlmNDhkZjQo+JgQEgkSBxDP89Hk/zAYAQ=='} None
---------------------------------------------------------------------------
DatabricksError                           Traceback (most recent call last)
Cell In[64], line 5
      3 it = client.query_history.list(filter_by=filter_by, max_results=1)
      4 next(it)
----> 5 next(it)

File ~/.pyenv/versions/3.10.9/lib/python3.10/site-packages/databricks/sdk/service/sql.py:2824, in QueryHistoryAPI.list(self, filter_by, include_metrics, max_results, page_token, **kwargs)
   2821 if page_token: query['page_token'] = request.page_token
   2823 while True:
-> 2824     json = self._api.do('GET', '/api/2.0/sql/history/queries', query=query)
   2825     if 'res' not in json or not json['res']:
   2826         return

File ~/.pyenv/versions/3.10.9/lib/python3.10/unittest/mock.py:1114, in CallableMixin.__call__(self, *args, **kwargs)
   1112 self._mock_check_sig(*args, **kwargs)
   1113 self._increment_mock_call(*args, **kwargs)
-> 1114 return self._mock_call(*args, **kwargs)

File ~/.pyenv/versions/3.10.9/lib/python3.10/unittest/mock.py:1118, in CallableMixin._mock_call(self, *args, **kwargs)
   1117 def _mock_call(self, /, *args, **kwargs):
-> 1118     return self._execute_mock_call(*args, **kwargs)

File ~/.pyenv/versions/3.10.9/lib/python3.10/unittest/mock.py:1179, in CallableMixin._execute_mock_call(self, *args, **kwargs)
   1177         raise result
   1178 else:
-> 1179     result = effect(*args, **kwargs)
   1181 if result is not DEFAULT:
   1182     return result

Cell In[62], line 3, in patch_do(method, path, query, body)
      1 def patch_do(method, path, query = None, body = None):
      2     print(method,path,query,body)
----> 3     res = original_do(method, path, query = None, body=query)
      4     print("RES", res.keys())
      5     return res

File ~/.pyenv/versions/3.10.9/lib/python3.10/site-packages/databricks/sdk/core.py:753, in ApiClient.do(self, method, path, query, body)
    749 if not response.ok:
    750     # TODO: experiment with traceback pruning for better readability
    751     # See https://stackoverflow.com/a/58821552/277035
    752     payload = response.json()
--> 753     raise self._make_nicer_error(status_code=response.status_code, **payload) from None
    754 if not len(response.content):
    755     return {}

DatabricksError: You can provide only one of 'page_token' or 'filter_by'

To summarize (sorry for the long post!), if my assumptions about the query history API are correct, then:

All params should be passed in the body, even though it's a GET request
The filter_by param should be removed when querying with page_token

Error when creating single node job clusters

When creating a single node job cluster, an error is thrown if the num_workers is set to 0. However, this is the same setting used when inspecting the JSON structure of a single node job cluster.

Error message:
databricks.sdk.core.DatabricksError: Cluster validation error: Missing required field: settings.cluster_spec.new_cluster.size

Cluster configuraiton:

{
"job_cluster_key": "ascend_ingest",
"new_cluster": {
"spark_version": "12.2.x-scala2.12",
"spark_conf": {
"spark.databricks.delta.preview.enabled": "true",
"spark.master": "local[*, 4]",
"spark.databricks.cluster.profile": "singleNode",
},
"azure_attributes": {
"first_on_demand": 1,
"availability": "ON_DEMAND_AZURE",
"spot_bid_max_price": -1,
},
"node_type_id": "Standard_DS3_v2",
"custom_tags": {"ResourceClass": "SingleNode"},
"spark_env_vars": {"PYSPARK_PYTHON": "/databricks/python3/bin/python3"},
"enable_elastic_disk": True,
"data_security_mode": "SINGLE_USER",
"runtime_engine": "STANDARD",

            ## Seems theres a bug that won't allow for single node clusters
            "num_workers": 0,
        },
    }

Note that changing num_workers to 1 resolves the issue.

w.workspace.list('/'): 'MLFLOW_EXPERIMENT' is not a valid ObjectType

When listing files in a workspace folder that contains MFflow experiments (files with a glass bottle icon), exception is thrown.

how to reproduce:

put mlflow experiment type of file in a workspace root
w.workspace.list('/')

Exception thrown:

ValueError                                Traceback (most recent call last)
<command-3538949529974781> in <cell line: 1>()
----> 1 w.workspace.list('/')

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/workspace.py in list(self, path, notebooks_modified_after, **kwargs)
    283 
    284         json = self._api.do('GET', '/api/2.0/workspace/list', query=query)
--> 285         return [ObjectInfo.from_dict(v) for v in json['objects']]
    286 
    287     def mkdirs(self, path: str, **kwargs):

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/workspace.py in <listcomp>(.0)
    283 
    284         json = self._api.do('GET', '/api/2.0/workspace/list', query=query)
--> 285         return [ObjectInfo.from_dict(v) for v in json['objects']]
    286 
    287     def mkdirs(self, path: str, **kwargs):

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/workspace.py in from_dict(cls, d)
    166                    modified_at=d.get('modified_at', None),
    167                    object_id=d.get('object_id', None),
--> 168                    object_type=ObjectType(d['object_type']) if 'object_type' in d else None,
    169                    path=d.get('path', None),
    170                    size=d.get('size', None))

/usr/lib/python3.9/enum.py in __call__(cls, value, names, module, qualname, type, start)
    358         """
    359         if names is None:  # simple value lookup
--> 360             return cls.__new__(cls, value)
    361         # otherwise, functional API: we're creating a new Enum type
    362         return cls._create_(

/usr/lib/python3.9/enum.py in __new__(cls, value)
    676                 ve_exc = ValueError("%r is not a valid %s" % (value, cls.__qualname__))
    677                 if result is None and exc is None:
--> 678                     raise ve_exc
    679                 elif exc is None:
    680                     exc = TypeError(

ValueError: 'MLFLOW_EXPERIMENT' is not a valid ObjectType

w.dbfs.copy("file:/...","dbfs:/...") is much slower than using REST api with requests library

While this library is easy to use,
I found copying files ranging from 200 kb to 5MB files from local laptop storage to dbfs to be much slower with this library when compared to the requests library or curl.
Curl was about 15.3 seconds for a 5MB file.

This library didn't finish the 5MB file after 20 minutes.

[FEATURE] Add support for Azure Managed Identity authentication

the azure_use_msi is documented in README, but marked as implemented. Adding support for managed identity could simplify deployment of Python-based applications

groups parameter for `service_principals.create(...)` is ignored?

Please see below process where the service_principals.create function fails to add the specified group. Am I using the wrong datatype?

dbw=WorkspaceClient(...)

groups = {group.display_name:group for group in dbw.groups.list()}
print(groups)

{
 'users': Group(id='...', display_name='users', ...),
 'service_principals': Group(id='...', display_name='service_principals',...),
 'admins': Group(id='...', display_name='admins', ...)
}

sp = dbw.service_principals.create(
    id=secret_client.get_secret("...").value,
    application_id=secret_client.get_secret("...").value,
    display_name="...",
    groups=[
        groups["service_principals"] # this seems to be ignored and it gets added to users instead.
    ]
)
print(sp)

ServicePrincipal(
id='...',
active=True,
application_id='...',
display_name='...',
entitlements=None,
external_id=None,
groups=None, # <-------------------- ???
roles=None
)

the expected value for groups is

groups=[ComplexValue(display='service_principals', primary=None, type='direct', value='...')],

many thanks

'/api/2.0/secrets/get' is not in the official docs

Looking through the code, I noticed you're using a call to /api/2.0/secrets/get to retrieve secrets remotely. That API endpoint doesn't seem to be documented in Secrets API 2.0

Could it please be added?

Workspace import method name

Hello! I'm wondering why the import_ method has a trailing underscore, but export doesn't - was that intentional?

databricks-sdk-py/databricks/sdk/service/workspace.py

Line 1145 in c578001

def import_(self,

`AttributeError: 'NoneType' object has no attribute 'debug_truncate_bytes'` when instantiating an `ApiClient` with `cfg = None`

Tried to run the following test:

from databricks.sdk import JobsAPI

    client = ApiClient()
    api = JobsAPI(client)
    cluster = JobCluster(
        job_cluster_key = "cluster1"
    )

    task1 = JobTaskSettings(
        task_key = "task1",
    )

    api.create(
        job_clusters = [cluster],
        tasks = [task1]
    )

This causes the following error:

self = <databricks.sdk.core.ApiClient object at 0x1265bcbb0>, cfg = None

    def __init__(self, cfg: Config = None):
        self._cfg = Config() if not cfg else cfg
>       self._debug_truncate_bytes = cfg.debug_truncate_bytes if cfg.debug_truncate_bytes else 96
E       AttributeError: 'NoneType' object has no attribute 'debug_truncate_bytes'

I've fixed the error in a separate branch and will submit a PR. Changes done:

Replaced:

def __init__(self, cfg: Config = None):
    self._cfg = Config() if not cfg else cfg
    self._debug_truncate_bytes = cfg.debug_truncate_bytes if cfg.debug_truncate_bytes else 96

With:

if cfg:
      self._cfg = cfg
      self._debug_truncate_bytes = cfg.debug_truncate_bytes if cfg.debug_truncate_bytes else 96
      self._user_agent_base = cfg.user_agent
else:
      self._cfg = Config()
      self._debug_truncate_bytes = 96
      self._user_agent_base = None

Relax requests dependency constraint once psf/requests issue 6432 is resolved

Currently we add a strict upper bound of <2.29 to the requests library. Requests 2.30+ is incompatible with urllib <2, but users of databricks-sdk may still depend on older versions of urllib. Once psf/requests#6432 is resolved, we should relax the upper bound to allow more recent versions of the requests library which incorporate the most recent release of urllib. This should improve the security posture of the SDK.

[BUG] Always return `None` from `model_registry.get_model()`

When calling API WorkspaceClient.model_registry.get_model(), it always returns GetModelResponse(registered_model=None).

However, the returned value is correct when I call the Web API directly with requests by passing the param.

url = f"{host}/api/2.0/mlflow/databricks/registered-models/get"
resp = requests.get(url, headers=headers, params={"name": "model_name"})

Could you please take a look?

OAuth Cross-origin error

Getting Cross-origin token redemption is permitted only for the 'Single-Page Application' client-type. Request origin: 'http://localhost:8020/'. when using OAuth client with Azure.

It works if I remove the code for both u2m and m2m:

        if 'microsoft' in self._client.token_url:
            # Tokens issued for the 'Single-Page Application' client-type may
            # only be redeemed via cross-origin requests
            headers = {'Origin': self._client.redirect_url}

Enum Generator: allow passing strings for parameters which are enums to make calling more pythonic

Problem

Generated Enum types does not have support of being instantiated from string, hence it forces end users to not use non pythonic syntax of:

from databricks.sdk.service.workspace import ExportFormat

w.workspace.export('/my_notebook', format=ExportFormat.SOURCE)

instead of neat, short, and pythonic:

w.workspace.export('/my_notebook', format='SOURCE')

throws an exception:

AttributeError                            Traceback (most recent call last)
<command-3538949529976490> in <cell line: 1>()
----> 1 w.workspace.export('/Users/[email protected]/auto-dlt/generated/gr_products/gr_bronze-autoloader', format='SOURCE')

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/workspace.py in export(self, path, direct_download, format, **kwargs)
    225         query = {}
    226         if direct_download: query['direct_download'] = request.direct_download
--> 227         if format: query['format'] = request.format.value
    228         if path: query['path'] = request.path
    229 

AttributeError: 'str' object has no attribute 'value'

Example

For example:

help(w.workspace.export)

has signature of:

export(path: str, *, direct_download: bool = None, format: databricks.sdk.service.workspace.ExportFormat = None, **kwargs) -> databricks.sdk.service.workspace.ExportResponse method of databricks.sdk.service.workspace.WorkspaceAPI instance

in turn parameter format is of Enum type defined in databricks.sdk.service.workspace.ExportFormat as

class ExportFormat(Enum):
    """This specifies the format of the file to be imported. By default, this is `SOURCE`. However it
    may be one of: `SOURCE`, `HTML`, `JUPYTER`, `DBC`. The value is case sensitive."""

    DBC = 'DBC'
    HTML = 'HTML'
    JUPYTER = 'JUPYTER'
    R_MARKDOWN = 'R_MARKDOWN'
    SOURCE = 'SOURCE'

[BUG] in jobs.py: TypeError: Object of type AccessControlRequest is not JSON serializable

databricks-sdk-py/databricks/sdk/service/jobs.py

Line 236 in 1e3612d

 if self.access_control_list: body['access_control_list'] = [v for v in self.access_control_list] 

the AccessControlRequest objects are put in the dict without modifying. This leads to

TypeError: Object of type AccessControlRequest is not JSON serializable

when calling JobsAPI.create

instead the line should be

if self.access_control_list: body['access_control_list'] = [v.as_dict() for v in self.access_control_list]

`databricks.sdk.service.compute.BaseClusterInfo` is missing `docker_image` field

This issue makes it impossible to use a custom container for a Cluster as documented here: https://docs.databricks.com/clusters/custom-containers.html

[Question] Are we going to support asynchronous API calls with this new client

This is an ask from one of my customers if we are planning to support a asynchronous client for our SDK?

As the current client

from databricks_cli.sdk.api_client import ApiClient

doesn't supports asynchronous calls

Error when getting Git credentials

Hi,

I'm trying to use the Python SDK to programatically create (if not already existing) a set of Git credentials on my Databricks environment. Getting the credentials gives the traceback below. I'm calling the SDK as such:

from databricks.sdk import GitCredentialsAPI
from databricks_cli.sdk.api_client import ApiClient

self.git_creds = GitCredentialsAPI(self.api_client)
self.git_creds.list()

Is the ApiClient you are using by any chance not the same class as the one I'm importing?

Thanks!

Traceback (most recent call last):
File "/home/georgelpreput/Source/pushcart-deploy/.venv/bin/pushcart-deploy", line 6, in
sys.exit(deploy())
^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/.venv/lib/python3.11/site-packages/click/core.py", line 1130, in call
return self.main(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/.venv/lib/python3.11/site-packages/click/core.py", line 1055, in main
rv = self.invoke(ctx)
^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/.venv/lib/python3.11/site-packages/click/core.py", line 1404, in invoke
return ctx.invoke(self.callback, **ctx.params)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/.venv/lib/python3.11/site-packages/click/core.py", line 760, in invoke
return __callback(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/src/pushcart_deploy/setup.py", line 118, in deploy
d.deploy()
File "/home/georgelpreput/Source/pushcart-deploy/src/pushcart_deploy/setup.py", line 69, in deploy
_ = self.repos.get_or_create_git_credentials()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/src/pushcart_deploy/databricks_api/repos_wrapper.py", line 81, in get_or_create_git_credentials
c for c in self.git_creds.list() if c["git_username"] == git_username
^^^^^^^^^^^^^^^^^^^^^
File "/home/georgelpreput/Source/pushcart-deploy/.venv/lib/python3.11/site-packages/databricks/sdk/service/workspace.py", line 759, in list
json = self._api.do('GET', '/api/2.0/git-credentials')
^^^^^^^^^^^^
AttributeError: 'ApiClient' object has no attribute 'do'

PAT auth and config profile auth not recognized

With version 0.0.2 of the SDK installed locally, I have a local config profile named DEFAULT defined in my ~/.databrickscfg file. I have no local DATABRICKS_* environment variables defined.

The following code returns the error databricks.sdk.client.DatabricksError: No auth configured:

from databricks.sdk import WorkspaceClient

w = WorkspaceClient()

for c in w.clusters.list():
  print(c.cluster_name)

So does the following code:

from databricks.sdk import WorkspaceClient

w = WorkspaceClient(profile = 'DEFAULT')

for c in w.clusters.list():
  print(c.cluster_name)

If I set my local DATABRICKS_HOST and DATABRICKS_TOKEN environment variables however, the following code runs as expected with no errors:

from databricks.sdk import WorkspaceClient

w = WorkspaceClient()

for c in w.clusters.list():
  print(c.cluster_name)

I would expect the SDK to recognize my local DEFAULT profile to be recognized without having to specify profile = 'DEFAULT'. Also, I should be able to set my local DEFAULT profile and not be forced to set local DATABRICKS_* environment variables if I don't want to.

Code completion works in unexpected ways when the Databricks extension for Visual Studio Code is also installed

Repro steps:

Install the Databricks extension for Visual Studio Code and enable globals, which imports a databricks.sdk namespace so that you I access the dbutils global from my Python code.
Install version 0.0.1 of the Python wheel for the Databricks SDK for Python.

Expect:

Get IntelliSense (code completion) for both dbutils and the Databricks SDK for Python classes.

Actual:

I get IntelliSense as I type for this: from databricks.sdk.runtime import dbutils. But then databricks.sdk.runtime squiggles with Import "databricks.sdk.runtime" could not be resolved from sourcePylancereportMissingModuleSource.
This doesn't work at all for IntelliSense as I type: from databricks.sdk import WorkspaceClient. However, any code that relies on this import runs as expected.
When I type from databricks.sdk import and then press Ctrl + Enter, I only get a drop-down with runtime. I was expecting a longer list with Databricks SDK for Python classes, for example AccountClient and WorkspaceClient.

Token Cache Support

Because of the many tools we own, it would be greatly beneficial to have a single way to store tokens locally. keyring or encrypted file.

Create job API call returns TypeError of not JSON serializable

Code:

from databricks.sdk import WorkspaceClient
from databricks.sdk.service.jobs import JobTaskSettings, NotebookTask, NotebookTaskSource

w = WorkspaceClient()

job_name            = input("Some short name for the job (for example, my-job): ")
description         = input("Some short description for the job (for example, My job): ")
existing_cluster_id = input("ID of the existing cluster in the workspace to run the job on (for example, 1234-567890-ab123cd4): ")
notebook_path       = input("Workspace path of the notebook to run (for example, /Users/[email protected]/my-notebook): ")
task_key            = input("Some key to apply to the job's tasks (for example, my-key): ")

print("Attempting to run the job. Please wait...\n")

j = w.jobs.create(
  name = job_name,
  tasks = [
    JobTaskSettings(
      description = description,
      existing_cluster_id = existing_cluster_id,
      notebook_task = NotebookTask(
        base_parameters = {""},
        notebook_path = notebook_path,
        source = NotebookTaskSource("WORKSPACE")
      ),
      task_key = task_key
    )
  ]
)

print(f"View the job at {w.config.host}/#job/{j.job_id}\n")

Input:

Some short name for the job (for example, my-job): my-job
Some short description for the job (for example, My job): My job
ID of the existing cluster in the workspace to run the job on (for example, 1234-567890-ab123cd4): <CLUSTER-ID-REDACTED>
Workspace path of the notebook to run (for example, /Users/[email protected]/my-notebook): /Users/<FULL-USERNAME-REDACTED>/hello
Some key to apply to the job's tasks (for example, my-key): my-key
Attempting to run the job. Please wait...

Error:

Traceback (most recent call last):
  File "/Users/paul.cornell/databricks-sdk-py-demo/run-job.py", line 14, in <module>
    j = w.jobs.create(
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/databricks/sdk/service/jobs.py", line 2074, in create
    json = self._api.do('POST', '/api/2.1/jobs/create', body=body)
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/databricks/sdk/core.py", line 686, in do
    response = self.request(method, f"{self._cfg.host}{path}", params=query, json=body, headers=headers)
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/requests/sessions.py", line 573, in request
    prep = self.prepare_request(req)
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/requests/sessions.py", line 484, in prepare_request
    p.prepare(
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/requests/models.py", line 371, in prepare
    self.prepare_body(data, files, json)
  File "/Users/paul.cornell/databricks-sdk-py-demo/.venv/lib/python3.10/site-packages/requests/models.py", line 511, in prepare_body
    body = complexjson.dumps(json, allow_nan=False)
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/json/__init__.py", line 238, in dumps
    **kw).encode(obj)
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/json/encoder.py", line 199, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/json/encoder.py", line 257, in iterencode
    return _iterencode(o, 0)
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/json/encoder.py", line 179, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '
TypeError: Object of type set is not JSON serializable

jobs.list_runs does not return 'latest_repair_id'

Hello,

I wrote a script that can repair all the runs for a given job_id. However, when I have to repair for a second time, I come across this error:

w.jobs.repair_run(j.run_id, rerun_all_failed_tasks=True)
databricks.sdk.core.DatabricksError: The latest repair ID needs to be provided in order to create a new repair

I cannot find any output in jobs object that has this information. Can you provide some insight here? Thanks!

Likely incomplete isAzure function

The code checks for only .azuredatabricks.net, but I am pretty sure we support other host patterns. Some of which are:

".azuredatabricks.net",
".databricks.azure.cn",
".databricks.azure.us",

dbutils.fs.mount generates wrong proxy call

When calling "dbutils.fs.mount" the proxy call towards the cluster is using the wrong dictionary names for the parameters.
The generated code looks like this:

        import json
        (args, kwargs) = json.loads('[[], {"source": "<someUrl>", "mountPoint": "/mnt/<myMountPoint", "encryptionType": "", "owner": "", "extraConfigs": null}]')
        result = dbutils.fs.mount(*args, **kwargs)
        dbutils.notebook.exit(json.dumps(result))

When this is executed in the cluster, the following error is thrown:
TypeError: DBUtils.FSHandler.mount() got an unexpected keyword argument 'mountPoint'
Further more this error is not proxied back - instead you get TypeError: the JSON object must be str, bytes or bytearray, not NoneType (dbutils.py:245)

The actual problem are the wrong parameter names in the Method-Signatur of "mount".
The original DataBricks-API expects the parameters with underscores instead of camelCase (see https://docs.databricks.com/dbfs/mounts.html):

dbutils.fs.mount(
  source: str,
  mount_point: str,
  encryption_type: Optional[str] = "",
  extra_configs: Optional[dict[str:str]] = None
)

Same probably applies to the other Mount-Methods.

w.alerts.list(): 'datetime-range' is not a valid ParameterType

log into https://e2-demo-field-eng.cloud.databricks.com/?o=1444828305810485
run: w.alerts.list()

I am not sure which alert causes this issue, it would be great if debug=True (or something of this sort) was passable as kwargs to enable debugging of element that caused exception (or is there better way to do it from notebooks?)

Exception:

ValueError                                Traceback (most recent call last)
<command-3538949529976490> in <cell line: 1>()
----> 1 w.alerts.list()

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in list(self)
   2198 
   2199         json = self._api.do('GET', '/api/2.0/preview/sql/alerts')
-> 2200         return [Alert.from_dict(v) for v in json]
   2201 
   2202     def list_schedules(self, alert_id: str, **kwargs) -> Iterator[RefreshSchedule]:

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in <listcomp>(.0)
   2198 
   2199         json = self._api.do('GET', '/api/2.0/preview/sql/alerts')
-> 2200         return [Alert.from_dict(v) for v in json]
   2201 
   2202     def list_schedules(self, alert_id: str, **kwargs) -> Iterator[RefreshSchedule]:

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in from_dict(cls, d)
     70                    name=d.get('name', None),
     71                    options=AlertOptions.from_dict(d['options']) if 'options' in d else None,
---> 72                    query=Query.from_dict(d['query']) if 'query' in d else None,
     73                    rearm=d.get('rearm', None),
     74                    state=AlertState(d['state']) if 'state' in d else None,

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in from_dict(cls, d)
   1201             latest_query_data_id=d.get('latest_query_data_id', None),
   1202             name=d.get('name', None),
-> 1203             options=QueryOptions.from_dict(d['options']) if 'options' in d else None,
   1204             permission_tier=PermissionLevel(d['permission_tier']) if 'permission_tier' in d else None,
   1205             query=d.get('query', None),

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in from_dict(cls, d)
   1454     def from_dict(cls, d: Dict[str, any]) -> 'QueryOptions':
   1455         return cls(moved_to_trash_at=d.get('moved_to_trash_at', None),
-> 1456                    parameters=[Parameter.from_dict(v)
   1457                                for v in d['parameters']] if 'parameters' in d else None)
   1458 

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in <listcomp>(.0)
   1454     def from_dict(cls, d: Dict[str, any]) -> 'QueryOptions':
   1455         return cls(moved_to_trash_at=d.get('moved_to_trash_at', None),
-> 1456                    parameters=[Parameter.from_dict(v)
   1457                                for v in d['parameters']] if 'parameters' in d else None)
   1458 

/local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/databricks/sdk/service/sql.py in from_dict(cls, d)
   1101         return cls(name=d.get('name', None),
   1102                    title=d.get('title', None),
-> 1103                    type=ParameterType(d['type']) if 'type' in d else None,
   1104                    value=d.get('value', None))
   1105 

/usr/lib/python3.9/enum.py in __call__(cls, value, names, module, qualname, type, start)
    358         """
    359         if names is None:  # simple value lookup
--> 360             return cls.__new__(cls, value)
    361         # otherwise, functional API: we're creating a new Enum type
    362         return cls._create_(

/usr/lib/python3.9/enum.py in __new__(cls, value)
    676                 ve_exc = ValueError("%r is not a valid %s" % (value, cls.__qualname__))
    677                 if result is None and exc is None:
--> 678                     raise ve_exc
    679                 elif exc is None:
    680                     exc = TypeError(

ValueError: 'datetime-range' is not a valid ParameterType

Make Conda package

Missing "not applicable" positional arguments for job task settings

With the following code that uses 0.0.1 of the SDK:

import os
from databricks.sdk import WorkspaceClient

host  = os.getenv('DATABRICKS_HOST')
token = os.getenv('DATABRICKS_TOKEN')

w = WorkspaceClient(host  = host, token = token, auth_type = "pat")

w.jobs.create(
  job_name = 'my-job',
  tasks = my_tasks
)

The my_tasks declaration doesn't seem to work no matter what syntax I use. For example:

Creating a list returns AttributeError: 'dict' object has no attribute 'as_dict':

my_tasks = [
  {
    "description": "My job.",
    "existing_cluster_id": "1128-232547-p64vrmx2",
    "notebook_task": {
      "notebook_path": "/Users/[email protected]/go-fakedata"
    },
    "task_key": "my-key"
  }
]

Creating a dictionary returns AttributeError: 'str' object has no attribute 'as_dict':

my_tasks = {
  "description": "My job.",
  "existing_cluster_id": "1128-232547-p64vrmx2",
  "notebook_task": {
    "notebook_path": "/Users/[email protected]/go-fakedata"
  },
  "task_key": "my-key"
}

databricks / databricks-sdk-py Goto Github PK

databricks-sdk-py's People

Contributors

Stargazers

Watchers

Forkers

databricks-sdk-py's Issues

here is the error:

Description

Steps to Reproduce

Problem

Example

Recommend Projects

Recommend Topics

Recommend Org