bauplan package

Submodules

Module contents

class bauplan.Client(api_key: str | None = None, profile: str | None = None, namespace: str | None = None, **kwargs)

Bases: object

A consistent interface to access Bauplan operations.

Using the client

import bauplan
client = bauplan.Client()

# query the table and return result set as an arrow Table
my_table = client.query('SELECT sum(trips) trips FROM travel_table', branch_name='main')

# efficiently cast the table to a pandas DataFrame
df = my_table.to_pandas()

Notes on authentication

# by default, authenticate from BAUPLAN_API_KEY >> BAUPLAN_PROFILE >> ~/.bauplan/config.yml
client = bauplan.Client()
# client used ~/.bauplan/config.yml profile 'default'

os.environ['BAUPLAN_PROFILE'] = "someprofile"
client = bauplan.Client()
# >> client now uses profile 'someprofile'

os.environ['BAUPLAN_API_KEY'] = "mykey"
client = bauplan.Client()
# >> client now authenticates with api_key value "mykey", because api key > profile

# specify authentication directly - this supercedes BAUPLAN_API_KEY in the environment
client = bauplan.Client(api_key='MY_KEY')

# specify a profile from ~/.bauplan/config.yml - this supercedes BAUPLAN_PROFILE in the environment
client = bauplan.Client(profile='default')

Handling Exceptions

Catalog operations (branch/table methods) raise a subclass of bauplan.exceptions.BauplanError that mirror HTTP status codes.

  • 400: InvalidDataError

  • 401: UnauthorizedError

  • 403: AccessDeniedError

  • 404: ResourceNotFoundError e.g .ID doesn’t match any records

  • 404: ApiRouteError e.g. the given route doesn’t exist

  • 405: ApiMethodError e.g. POST on a route with only GET defined

  • 409: UpdateConflictError e.g. creating a record with a name that already exists

  • 429: TooManyRequestsError

Run/Query/Scan/Import operations raise a subclass of bauplan.exceptions.BauplanError that represents, and also return a RunState object containing details and logs:

  • JobError e.g. something went wrong in a run/query/import/scan; includes error details

Run/import operations also return a state object that includes a job_status and other details. There are two ways to check status for run/import operations:

  1. try/except the JobError exception

  2. check the state.job_status attribute

Examples:

try:
    state = client.run(...)
    state = client.query(...)
    state = client.scan(...)
    state = client.plan_table_creation(...)
except bauplan.exceptions.JobError as e:
    ...

state = client.run(...)
if state.job_status != "success":
    ...
Parameters:
  • api_key – (optional) Your unique Bauplan API key; mutually exclusive with profile. If not provided, fetch precedence is 1) environment BAUPLAN_API_KEY 2) .bauplan/config.yml

  • profile – (optional) The Bauplan config profile name to use to determine api_key; mutually exclusive with api_key.

  • namespace – (optional) The default namespace to use for queries and runs.

apply_table_creation_plan(plan: Dict | TableCreatePlanState, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) TableCreatePlanApplyState

Apply a plan for creating a table. It is done automaticaly during th table plan creation if no schema conflicts exist. Otherwise, if schema conflicts exist, then this function is used to apply them after the schema conflicts are resolved. Most common schema conflict is a two parquet files with the same column name but different datatype

Parameters:
  • plan – The plan to apply.

  • debug – Whether to enable or disable debug mode for the query.

  • args – dict of arbitrary args to pass to the backend.

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Raises:

TableCreatePlanApplyStatusError – if the table creation plan apply fails.

:return The plan state.

branch_exists(branch: str | Branch) bool

Check if a branch exists.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.branch_exists('main')
Parameters:

branch – The name of the branch to check.

Returns:

A boolean for if the branch exists.

create_branch(branch: str | Branch, from_ref: str | Branch | Ref) Branch

Create a new branch at a given ref.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.create_branch(
    branch='myzone.newbranch',
    from_ref='main'
)
Parameters:
  • branch – The name of the new branch.

  • from_ref – The name of the base branch; either a branch like “main” or ref like “main@[sha]”.

Returns:

The created branch object.

create_namespace(namespace: str | Namespace, into_branch: str | Branch) Namespace

Create a new namespace at a given branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.create_namespace(
    branch='myzone.newbranch',
    name='main'
)
Parameters:
  • namespace – The name of the namespace.

  • into_branch – The name of the branch to create the namespace on.

Returns:

The created namespace.

create_table(table: str | Table, search_uri: str, branch: str | Branch | None = None, namespace: str | Namespace | None = None, replace: bool | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) Table

Create a table from an S3 location.

This operation will attempt to create a table based of schemas of N parquet files found by a given search uri.

import bauplan
client = bauplan.Client()

plan_state = client.create_table(
    table='newtablename',
    search_uri='s3://path/to/my/files/*.parquet',
    branch='main',
)
if plan_state.error:
    plan_error_action(...)
success_action(plan_state.plan)
Parameters:
  • table – The table which will be created.

  • search_uri – The location of the files to scan for schema.

  • branch – The branch name in which to create the table in.

  • namespace – Optional argument specifying the namespace. If not. specified, it will be inferred based on table location or the default. namespace

  • replace – Replace the table if it already exists.

  • debug – Whether to enable or disable debug mode for the query.

  • args – dict of arbitrary args to pass to the backend.

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Raises:
Returns:

The plan state.

delete_branch(branch: str | Branch) bool

Delete a branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.delete_branch(branch='mybranch')
Parameters:

branch – The name of the branch to delete.

Returns:

A boolean for if the branch was deleted.

delete_namespace(namespace: str | Namespace, form_branch: str | Branch) bool

Delete a namespace.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.delete_namespace(
    branch='mybranch',
    name='mynamespace',
)
Parameters:
  • namespace – The name of the namespace to delete.

  • form_branch – The name of the branch to delete the namespace from.

Returns:

A boolean for if the namespace was deleted.

delete_table(table: str | Table, branch: str | Branch) bool

Drop a table.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.delete_table(table='mytable', branch='mybranch')
Parameters:
  • table – The table to delete.

  • branch – The branch on which the table is stored.

Returns:

A boolean for if the table was deleted.

get_branch(branch: str | Branch) Branch

Get the branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

# retrieve only the tables as tuples of (name, kind)
branch = client.get_branch('main')
print(branch.hash)
Parameters:

branch – The name of the branch to retrieve.

Returns:

A Branch object.

get_branches(name: str | None = None, user: str | None = None, limit: int | None = None, itersize: int | None = None) Generator[Branch, None, None]

Get the available data branches in the Bauplan catalog.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

for branch in client.get_branches():
    print(branch.name, branch.hash)
Parameters:
  • name – Filter the branches by name.

  • user – Filter the branches by user.

  • itersize – int 1-500.

  • limit – int > 0.

Yield:

A Branch object.

get_namespace(namespace: str | Namespace, ref: str | Branch | Ref) Namespace

Get a namespace.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

namespace =  client.get_namespace('my_namespace', 'main')
Parameters:
  • namespace – The name of the namespace to get.

  • ref – The ref or branch to check the namespace on.

Returns:

A Namespace object.

get_namespaces(ref: str | Branch | Ref, filter_by_name: str | None = None, itersize: int | None = None, limit: int | None = None) Generator[Namespace, None, None]

Get the available data namespaces in the Bauplan catalog branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

for namespace in client.get_namespaces():
    print(namespace.name)
Parameters:
  • ref – The ref or branch to retrieve the namespaces from.

  • filter_by_name – Filter the namespaces by name.

  • itersize – int 1-500.

  • limit – int > 0.

Yield:

A Namespace object.

get_table(table: str | Table, branch: str | Branch, include_raw: bool = False) Table

Get the table data and metadata for a table in the target branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

# get the fields and metadata for the taxi_zones table in the main branch
table = client.get_table_with_metadata(branch='main', table='taxi_zones')

# loop through the fields and print their name, required, and type
for c in table.fields:
    print(c.name, c.required, c.type)

# show the number of records in the table
print(table.records)
Parameters:
  • branch – The branch to get the table from.

  • table – The table to retrieve.

  • include_raw – Whether or not to include the raw metadata.json object as a nested dict.

Returns:

a TableWithMetadata object, optionally including the raw metadata.json object.

get_tables(ref: str | Branch | Ref, namespace: str | Namespace | None = None, limit: int | None = None, itersize: int | None = None) Generator[Table, None, None]

Get the tables and views in the target branch.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

# retrieve only the tables as tuples of (name, kind)
tables = client.get_tables('main')
for table in tables:
    print(table.name, table.kind)
Parameters:
  • ref – The ref or branch to get the tables from.

  • namespace – The namespace to get the tables from.

  • limit – int > 0.

  • itersize – int 1-500.

Yield:

A Table object.

import_data(table: str | Table, search_uri: str, branch: str | Branch | None = None, namespace: str | Namespace | None = None, continue_on_error: bool = False, import_duplicate_files: bool = False, best_effort: bool = False, preview: bool | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) TableDataImportState

Imports data into an already existing table.

import bauplan
client = bauplan.Client()

s3_path = 's3://path/to/my/files/*.parquet'
plan_state = client.data_import(
    table='newtablename',
    search_uri=s3_path,
    branch='main'
)
if plan_state.error:
    plan_error_action(...)
success_action(plan_state.plan)
Parameters:
  • table – Previously created table in into which data will be imported.

  • search_uri – Uri which to scan for files to import.

  • branch – Branch in which to import the table.

  • namespace – Namespace of the table. If not specified, namespace will be infered from table name or default settings.

  • continue_on_error – Do not fail the import even if 1 data import fails.

  • import_duplicate_files – Ignore prevention of importing s3 files that were already imported.

  • best_effort – Don’t fail if schema of table does not match.

  • transformation_query – Optional duckdb compliant query applied on each parquet file. Use original_table as the table in the query.

  • preview – Whether to enable or disable preview mode for the query.

  • debug – Whether to enable or disable debug mode for the query.

  • args – dict of arbitrary args to pass to the backend.

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The plan state.

merge_branch(source_ref: str | Branch | Ref, into_branch: str | Branch) bool

Merge one branch into another.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert merge_branch(
    into_branch='myzone.somebranch',
    source_ref='myzone.oldbranch'
)
Parameters:
  • source_ref – The name of the merge source; either a branch like “main” or ref like “main@[sha]”.

  • into_branch – The name of the merge target.

Returns:

a boolean for whether the merge worked.

namespace_exists(namespace: str | Namespace, ref: str | Branch | Ref) bool

Check if a namespace exists.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.namespace_exists('my_namespace', 'main')
Parameters:
  • namespace – The name of the namespace to check.

  • ref – The ref or branch to check the namespace on.

Returns:

A boolean for if the namespace exists.

plan_table_creation(table: str | Table, search_uri: str, branch: str | Branch | None = None, namespace: str | Namespace | None = None, replace: bool | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) TableCreatePlanState

Create a table import plan from an S3 location.

This operation will attempt to create a table based of schemas of N parquet files found by a given search uri. A YAML file containing the schema and plan is returns and if there are no conflicts, it is automatically applied.

import bauplan
client = bauplan.Client()

plan_state = client.plan_table_creation(
    search_uri='s3://path/to/my/files/*.parquet',
    table='newtablename',
    branch='main',
)
if plan_state.error:
    plan_error_action(...)
success_action(plan_state.plan)
Parameters:
  • table – The table which will be created.

  • search_uri – The location of the files to scan for schema.

  • branch – The branch name in which to create the table in.

  • namespace – Optional argument specifying the namespace. If not. specified, it will be inferred based on table location or the default. namespace

  • replace – Replace the table if it already exists.

  • debug – Whether to enable or disable debug mode for the query.

  • args – dict of arbitrary args to pass to the backend.

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Raises:

TableCreatePlanStatusError – if the table creation plan fails.

Returns:

The plan state.

query(query: str, ref: str | Branch | Ref | None = None, max_rows: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) Table

Execute a SQL query and return the results as a pyarrow.Table. Note that this function uses Arrow also internally, resulting in a fast data transfer.

If you prefer to return the results as a pandas DataFrame, use the to_pandas function of pyarrow.Table.

import bauplan

client = bauplan.Client()

# query the table and return result set as an arrow Table
my_table = client.query('SELECT c1 FROM my_table', ref='main')

# efficiently cast the table to a pandas DataFrame
df = mytable.to_pandas()
Parameters:
  • query – The Bauplan query to execute.

  • ref – The ref or branch name to read data from.

  • max_rows – The maximum number of rows to return; default: None (no limit).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the query in. If not set, the query will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • args – Additional arguments to pass to the query (default: None).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The query results as a pyarrow.Table.

query_to_csv_file(filename: str, query: str, ref: str | Branch | Ref | None = None, max_rows: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None, **kwargs: Any) str

Export the results of a SQL query to a file in CSV format.

import bauplan
client = bauplan.Client()

# query the table and iterate through the results one row at a time
client.query_to_csv_file('./my.csv', 'SELECT c1 FROM my_table'):
Parameters:
  • filename – The name or path of the file csv to write the results to.

  • query – The Bauplan query to execute.

  • ref – The ref or branch name to read data from.

  • max_rows – The maximum number of rows to return; default: None (no limit).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the query in. If not set, the query will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • args – Additional arguments to pass to the query (default: None).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The path of the file written.

query_to_generator(query: str, ref: str | Branch | Ref | None = None, max_rows: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, as_json: bool | None = False, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) Generator[Dict[str, Any], None, None]

Execute a SQL query and return the results as a generator, where each row is a Python dictionary.

import bauplan
client = bauplan.Client()

# query the table and iterate through the results one row at a time
for row in client.query_to_generator('SELECT c1 FROM my_table', ref='main'):
    # do logic
Parameters:
  • query – The Bauplan query to execute.

  • ref – The ref or branch name to read data from.

  • max_rows – The maximum number of rows to return; default: None (no limit).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the query in. If not set, the query will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • as_json – Whether to return the results as a JSON-compatible string (default: False).

  • args – Additional arguments to pass to the query (default: None).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Yield:

A dictionary representing a row of query results.

query_to_json_file(filename: str, query: str, file_format: Literal['json', 'jsonl'] | None = 'json', ref: str | Branch | Ref | None = None, max_rows: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) str

Export the results of a SQL query to a file in JSON format.

import bauplan
client = bauplan.Client()

# query the table and iterate through the results one row at a time
client.query_to_file('./my.json', 'SELECT c1 FROM my_table'):
Parameters:
  • filename – The name or path of the file json to write the results to.

  • query – The Bauplan query to execute.

  • file_format – The format to write the results in; default: json. Allowed values are ‘json’ and ‘jsonl’.

  • ref – The ref or branch name to read data from.

  • max_rows – The maximum number of rows to return; default: None (no limit).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the query in. If not set, the query will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • args – Additional arguments to pass to the query (default: None).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The path of the file written.

query_to_parquet_file(filename: str, query: str, ref: str | Branch | Ref | None = None, max_rows: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None, **kwargs: Any) str

Export the results of a SQL query to a file in Parquet format.

import bauplan
client = bauplan.Client()

# query the table and iterate through the results one row at a time
client.query_to_file('./my.json', 'SELECT c1 FROM my_table'):
Parameters:
  • filename – The name or path of the file parquet to write the results to.

  • query – The Bauplan query to execute.

  • ref – The ref or branch name to read data from.

  • max_rows – The maximum number of rows to return; default: None (no limit).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the query in. If not set, the query will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • args – Additional arguments to pass to the query (default: None).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The path of the file written.

rerun(job_id: str, ref: str | Branch | Ref | None = None, namespace: str | Namespace | None = None, cache: Literal['on', 'off'] | None = None, transaction: Literal['on', 'off'] | None = None, dry_run: bool | None = None, strict: Literal['on', 'off'] | None = None, preview: bool | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) RunState

Re run a Bauplan project by its ID and return the state of the run. This is the equivalent of running through the CLI the bauplan rerun command.

Parameters:
  • job_id – The Job ID of the previous run. This can be used to re-run a previous run, e.g., on a different branch.

  • ref – The ref or branch name to read.

  • namespace – The Namespace to run the job in. If not set, the job will be run in the default namespace.

  • cache – Whether to enable or disable caching for the run.

  • transaction – Whether to enable or disable transaction mode for the run.

  • dry_run – Whether to enable or disable dry-run mode for the run; models are not materialized.

  • strict – Whether to enable or disable strict schema validation.

  • preview – Whether to enable or disable preview mode for the run.

  • debug – Whether to enable or disable debug mode for the run.

  • args – Additional arguments (optional).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The state of the run.

run(project_dir: str | None = None, ref: str | Branch | Ref | None = None, namespace: str | Namespace | None = None, parameters: Dict[str, str | int | float | bool] | None = None, cache: Literal['on', 'off'] | None = None, transaction: Literal['on', 'off'] | None = None, dry_run: bool | None = None, strict: Literal['on', 'off'] | None = None, preview: bool | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None) RunState

Run a Bauplan project and return the state of the run. This is the equivalent of running through the CLI the bauplan run command.

Parameters:
  • project_dir – The directory of the project (where the bauplan_project.yml file is located).

  • ref – The ref or branch name to read.

  • namespace – The Namespace to run the job in. If not set, the job will be run in the default namespace.

  • parameters – Parameters for templating into SQL or Python models.

  • cache – Whether to enable or disable caching for the run.

  • transaction – Whether to enable or disable transaction mode for the run.

  • dry_run – Whether to enable or disable dry-run mode for the run; models are not materialized.

  • strict – Whether to enable or disable strict schema validation.

  • preview – Whether to enable or disable preview mode for the run.

  • debug – Whether to enable or disable debug mode for the run.

  • args – Additional arguments (optional).

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The state of the run.

scan(table: str | Table, ref: str | Branch | Ref | None = None, columns: List[str] | None = None, filters: str | None = None, limit: int | None = None, cache: Literal['on', 'off'] | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, namespace: str | Namespace | None = None, debug: bool | None = None, args: Dict[str, str] | None = None, client_timeout: int | float | None = None, **kwargs: Any) Table

Execute a table scan (with optional filters) and return the results as an arrow Table.

Note that this function uses SQLGlot to compose a safe SQL query, and then internally defer to the query_to_arrow function for the actual scan.

import bauplan
client = bauplan.Client()

# run a table scan over the data lake
# filters are passed as a string
my_table = client.scan(
    table='my_table',
    ref='main',
    columns=['c1'],
    filter='c2 > 10',
)
Parameters:
  • table – The table to scan.

  • ref – The ref or branch name to read data from.

  • columns – The columns to return (default: None).

  • filters – The filters to apply (default: None).

  • limit – The maximum number of rows to return (default: None).

  • cache – Whether to enable or disable caching for the query.

  • connector – The connector type for the model (defaults to Bauplan). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

  • namespace – The Namespace to run the scan in. If not set, the scan will be run in the default namespace for your account.

  • debug – Whether to enable or disable debug mode for the query.

  • args – dict of arbitrary args to pass to the backend.

  • client_timeout – seconds to timeout; this also cancels the remote job execution.

Returns:

The scan results as a pyarrow.Table.

table_exists(table: str | Table, branch: str | Branch) bool

Check if a table exists.

Upon failure, raises bauplan.exceptions.BauplanError

import bauplan
client = bauplan.Client()

assert client.table_exists('table_foo', 'main')
Parameters:
  • branch – The branch to get the table from.

  • table – The table to retrieve.

Returns:

A boolean for if the table exists.

class bauplan.JobStatus(canceled: 'str' = 'CANCELLED', cancelled: 'str' = 'CANCELLED', failed: 'str' = 'FAILED', rejected: 'str' = 'REJECTED', success: 'str' = 'SUCCESS', timeout: 'str' = 'TIMEOUT', unknown: 'str' = 'UNKNOWN')

Bases: object

canceled: str = 'CANCELLED'
cancelled: str = 'CANCELLED'
failed: str = 'FAILED'
rejected: str = 'REJECTED'
success: str = 'SUCCESS'
timeout: str = 'TIMEOUT'
unknown: str = 'UNKNOWN'
class bauplan.Model(name: str, columns: List[str] | None = None, filter: str | None = None, ref: str | None = None, connector: str | None = None, connector_config_key: str | None = None, connector_config_uri: str | None = None, **kwargs: Any)

Bases: object

Represents a model (dataframe/table representing a DAG step) as an input to a function.

e.g.

@bauplan.model()
def some_parent_model():
    return pyarrow.Table.from_pydict({'bar': [1, 2, 3]})

@bauplan.model()
def your_cool_model(
    # parent models are passed as inputs, using bauplan.Model
    # class
    parent_0=bauplan.Model(
        'some_parent_model',
        columns=['bar'],
        filter='bar > 1',
    )
):
    # Can return a pandas dataframe or a pyarrow table
    return pyarrow.Table.from_pandas(
        pd.DataFrame({
            'foo': parent_0['bar'] * 2,
        })
    )

Bauplan can wrap other engines for the processing of some models, exposing a common interface and unified API for the user while dispatching the relevant operations to the underlying engine.

The authentication and authorization happens securely and transparently through ssm; the user is asked to specify a connector type and the credentials through the relevant keywords:

@bauplan.model()
def your_cool_model(
    parent_0=bauplan.Model(
        'some_parent_model',
        columns=['bar'],
        filter='bar > 1',
        connector='dremio',
        connector_config_key='bauplan',
    )
):
    # parent_0 inside the function body
    # will still be an Arrow table: the user code
    # should still be the same as the data is moved
    # transparently by Bauplan from an engine to the function.
    return pyarrow.Table.from_pandas(
        pd.DataFrame({
            'foo': parent_0['bar'] * 2,
        })
    )
Parameters:
  • name – The name of the model.

  • columns – The list of columns in the model. If the arg is not provided, the model will load all columns.

  • filter – The optional filter for the model. Defaults to None.

  • ref – The optional reference to the model. Defaults to None.

  • connector – The connector type for the model (defaults to Bauplan SQL). Allowed values are ‘snowflake’ and ‘dremio’.

  • connector_config_key – The key name if the SSM key is custom with the pattern bauplan/connectors/<connector_type>/<key>.

  • connector_config_uri – Full SSM uri if completely custom path, e.g. ssm://us-west-2/123456789012/baubau/dremio.

bauplan.expectation(**kwargs: Any) Callable

Decorator that defines a Bauplan expectation.

An expectation is a function from one (or more) dataframe-like object(s) to a boolean: it is commonly used to perform data validation and data quality checks when running a pipeline. Expectations takes as input the table(s) they are validating and return a boolean indicating whether the expectation is met or not. A Python expectation needs a Python environment to run, which is defined using the python decorator, e.g.:

@bauplan.expectation()
@bauplan.python('3.10')
def test_joined_dataset(
    data=bauplan.Model(
        'join_dataset',
        columns=['anomaly']
    )
):
    # your data validation code here
    return expect_column_no_nulls(data, 'anomaly')
Parameters:

f – The function to decorate.

bauplan.model(name: str | None = None, columns: List[str] | None = None, materialize: bool | None = None, internet_access: bool | None = None, partitioned_by: str | List[str] | Tuple[str, ...] | None = None, materialization_strategy: Literal['NONE', 'REPLACE', 'APPEND'] | None = None, cache_strategy: Literal['NONE', 'DEFAULT'] | None = None, **kwargs: Any) Callable

Decorator that specifies a Bauplan model.

A model is a function from one (or more) dataframe-like object(s) to another dataframe-like object: it is used to define a transformation in a pipeline. Models are chained together implicitly by using them as inputs to their children. A Python model needs a Python environment to run, which is defined using the python decorator, e.g.:

@bauplan.model(
    columns=['*'],
    materialize=False
)
@bauplan.python('3.11')
def source_scan(
    data=bauplan.Model(
        'iot_kaggle',
        columns=['*'],
        filter="motion='false'"
    )
):
    # your code here
    return data
Parameters:
  • name – the name of the model (e.g. ‘users’); if missing the function name is used.

  • columns – the columns of the output dataframe after the model runs (e.g. [‘id’, ‘name’, ‘email’]). Use [‘*’] as a wildcard.

  • materialize – whether the model should be materialized.

  • internet_access – whether the model requires internet access.

  • partitioned_by – the columns to partition the data by.

  • materialization_strategy – the materialization strategy to use.

  • cache_strategy – the cache strategy to use.

bauplan.pyspark(version: str | None = None, conf: Dict[str, str] | None = None, **kwargs: Any) Callable

Decorator that makes a pyspark session available to a Bauplan function (a model or an expectation). Add a spark=None parameter to the function model args

Parameters:
  • version – the version string of pyspark

  • conf – A dict containing the pyspark config

bauplan.python(version: str | None = None, pip: Dict[str, str] | None = None, **kwargs: Any) Callable

Decorator that defines a Python environment for a Bauplan function (e.g. a model or expectation). It is used to specify directly in code the configuration of the Python environment required to run the function, i.e. the Python version and the Python packages required.

Parameters:
  • version – The python version for the interpreter (e.g. '3.11').

  • pip – A dictionary of dependencies (and versions) required by the function (e.g. {'requests': '2.26.0'}).

bauplan.resources(cpus: int | float | None = None, memory: int | str | None = None, memory_swap: int | str | None = None, timeout: int | None = None, **kwargs: Any) Callable

Decorator that defines the resources required by a Bauplan function (e.g. a model or expectation). It is used to specify directly in code the configuration of the resources required to run the function.

Parameters:
  • cpus – The number of CPUs required by the function (e.g: `0.5`)

  • memory – The amount of memory required by the function (e.g: `1G`, `1000`)

  • memory_swap – The amount of swap memory required by the function (e.g: `1G`, `1000`)

  • timeout – The maximum time the function is allowed to run (e.g: `60`)