Config
zenml.config
special
The config module contains classes and functions that manage user-specific configuration.
ZenML's configuration is stored in a file called
config.yaml, located on the user's directory for configuration files.
(The exact location differs from operating system to operating system.)
The GlobalConfiguration class is the main class in this module. It provides
a Pydantic configuration object that is used to store and retrieve
configuration. This GlobalConfiguration object handles the serialization and
deserialization of the configuration options that are stored in the file in
order to persist the configuration across sessions.
base_settings
Base class for all ZenML settings.
BaseSettings (SecretReferenceMixin)
pydantic-model
Base class for settings.
The LEVEL class variable defines on which level the settings can be
specified. By default, subclasses can be defined on both pipelines and
steps.
Source code in zenml/config/base_settings.py
class BaseSettings(SecretReferenceMixin):
"""Base class for settings.
The `LEVEL` class variable defines on which level the settings can be
specified. By default, subclasses can be defined on both pipelines and
steps.
"""
LEVEL: ClassVar[ConfigurationLevel] = (
ConfigurationLevel.PIPELINE | ConfigurationLevel.STEP
)
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# allow extra attributes so this class can be used to parse dicts
# of arbitrary subclasses
extra = Extra.allow
Config
Pydantic configuration class.
Source code in zenml/config/base_settings.py
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# allow extra attributes so this class can be used to parse dicts
# of arbitrary subclasses
extra = Extra.allow
ConfigurationLevel (IntFlag)
Settings configuration level.
Bit flag that can be used to specify where a BaseSettings subclass
can be specified.
Source code in zenml/config/base_settings.py
class ConfigurationLevel(IntFlag):
"""Settings configuration level.
Bit flag that can be used to specify where a `BaseSettings` subclass
can be specified.
"""
STEP = auto()
PIPELINE = auto()
build_configuration
Build configuration class.
BuildConfiguration (BaseModel)
pydantic-model
Configuration of Docker builds.
Attributes:
| Name | Type | Description |
|---|---|---|
key |
str |
The key to store the build. |
settings |
DockerSettings |
Settings for the build. |
step_name |
Optional[str] |
Name of the step for which this image will be built. |
entrypoint |
Optional[str] |
Optional entrypoint for the image. |
extra_files |
Dict[str, str] |
Extra files to include in the Docker image. |
Source code in zenml/config/build_configuration.py
class BuildConfiguration(BaseModel):
"""Configuration of Docker builds.
Attributes:
key: The key to store the build.
settings: Settings for the build.
step_name: Name of the step for which this image will be built.
entrypoint: Optional entrypoint for the image.
extra_files: Extra files to include in the Docker image.
"""
key: str
settings: DockerSettings
step_name: Optional[str] = None
entrypoint: Optional[str] = None
extra_files: Dict[str, str] = {}
def compute_settings_checksum(
self,
stack: "Stack",
code_repository: Optional["BaseCodeRepository"] = None,
) -> str:
"""Checksum for all build settings.
Args:
stack: The stack for which to compute the checksum. This is needed
to gather the stack integration requirements in case the
Docker settings specify to install them.
code_repository: Optional code repository that will be used to
download files inside the image.
Returns:
The checksum.
"""
hash_ = hashlib.md5()
hash_.update(self.settings.json().encode())
if self.entrypoint:
hash_.update(self.entrypoint.encode())
for destination, source in self.extra_files.items():
hash_.update(destination.encode())
hash_.update(source.encode())
from zenml.utils.pipeline_docker_image_builder import (
PipelineDockerImageBuilder,
)
pass_code_repo = self.should_download_files(
code_repository=code_repository
)
requirements_files = (
PipelineDockerImageBuilder.gather_requirements_files(
docker_settings=self.settings,
stack=stack,
code_repository=code_repository if pass_code_repo else None,
log=False,
)
)
for _, requirements, _ in requirements_files:
hash_.update(requirements.encode())
return hash_.hexdigest()
def should_include_files(
self,
code_repository: Optional["BaseCodeRepository"],
) -> bool:
"""Whether files should be included in the image.
Args:
code_repository: Code repository that can be used to download files
inside the image.
Returns:
Whether files should be included in the image.
"""
if self.settings.source_files == SourceFileMode.INCLUDE:
return True
if (
self.settings.source_files == SourceFileMode.DOWNLOAD_OR_INCLUDE
and not code_repository
):
return True
return False
def should_download_files(
self,
code_repository: Optional["BaseCodeRepository"],
) -> bool:
"""Whether files should be downloaded in the image.
Args:
code_repository: Code repository that can be used to download files
inside the image.
Returns:
Whether files should be downloaded in the image.
"""
if not code_repository:
return False
return self.settings.source_files in {
SourceFileMode.DOWNLOAD,
SourceFileMode.DOWNLOAD_OR_INCLUDE,
}
compute_settings_checksum(self, stack, code_repository=None)
Checksum for all build settings.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
stack |
Stack |
The stack for which to compute the checksum. This is needed to gather the stack integration requirements in case the Docker settings specify to install them. |
required |
code_repository |
Optional[BaseCodeRepository] |
Optional code repository that will be used to download files inside the image. |
None |
Returns:
| Type | Description |
|---|---|
str |
The checksum. |
Source code in zenml/config/build_configuration.py
def compute_settings_checksum(
self,
stack: "Stack",
code_repository: Optional["BaseCodeRepository"] = None,
) -> str:
"""Checksum for all build settings.
Args:
stack: The stack for which to compute the checksum. This is needed
to gather the stack integration requirements in case the
Docker settings specify to install them.
code_repository: Optional code repository that will be used to
download files inside the image.
Returns:
The checksum.
"""
hash_ = hashlib.md5()
hash_.update(self.settings.json().encode())
if self.entrypoint:
hash_.update(self.entrypoint.encode())
for destination, source in self.extra_files.items():
hash_.update(destination.encode())
hash_.update(source.encode())
from zenml.utils.pipeline_docker_image_builder import (
PipelineDockerImageBuilder,
)
pass_code_repo = self.should_download_files(
code_repository=code_repository
)
requirements_files = (
PipelineDockerImageBuilder.gather_requirements_files(
docker_settings=self.settings,
stack=stack,
code_repository=code_repository if pass_code_repo else None,
log=False,
)
)
for _, requirements, _ in requirements_files:
hash_.update(requirements.encode())
return hash_.hexdigest()
should_download_files(self, code_repository)
Whether files should be downloaded in the image.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
code_repository |
Optional[BaseCodeRepository] |
Code repository that can be used to download files inside the image. |
required |
Returns:
| Type | Description |
|---|---|
bool |
Whether files should be downloaded in the image. |
Source code in zenml/config/build_configuration.py
def should_download_files(
self,
code_repository: Optional["BaseCodeRepository"],
) -> bool:
"""Whether files should be downloaded in the image.
Args:
code_repository: Code repository that can be used to download files
inside the image.
Returns:
Whether files should be downloaded in the image.
"""
if not code_repository:
return False
return self.settings.source_files in {
SourceFileMode.DOWNLOAD,
SourceFileMode.DOWNLOAD_OR_INCLUDE,
}
should_include_files(self, code_repository)
Whether files should be included in the image.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
code_repository |
Optional[BaseCodeRepository] |
Code repository that can be used to download files inside the image. |
required |
Returns:
| Type | Description |
|---|---|
bool |
Whether files should be included in the image. |
Source code in zenml/config/build_configuration.py
def should_include_files(
self,
code_repository: Optional["BaseCodeRepository"],
) -> bool:
"""Whether files should be included in the image.
Args:
code_repository: Code repository that can be used to download files
inside the image.
Returns:
Whether files should be included in the image.
"""
if self.settings.source_files == SourceFileMode.INCLUDE:
return True
if (
self.settings.source_files == SourceFileMode.DOWNLOAD_OR_INCLUDE
and not code_repository
):
return True
return False
compiler
Class for compiling ZenML pipelines into a serializable format.
Compiler
Compiles ZenML pipelines to serializable representations.
Source code in zenml/config/compiler.py
class Compiler:
"""Compiles ZenML pipelines to serializable representations."""
def compile(
self,
pipeline: "Pipeline",
stack: "Stack",
run_configuration: PipelineRunConfiguration,
) -> Tuple[PipelineDeploymentBaseModel, PipelineSpec]:
"""Compiles a ZenML pipeline to a serializable representation.
Args:
pipeline: The pipeline to compile.
stack: The stack on which the pipeline will run.
run_configuration: The run configuration for this pipeline.
Returns:
The compiled pipeline deployment and spec
"""
logger.debug("Compiling pipeline `%s`.", pipeline.name)
# Copy the pipeline before we apply any run-level configurations, so
# we don't mess with the pipeline object/step objects in any way
pipeline = copy.deepcopy(pipeline)
self._apply_run_configuration(
pipeline=pipeline, config=run_configuration
)
self._apply_stack_default_settings(pipeline=pipeline, stack=stack)
if run_configuration.run_name:
self._verify_run_name(run_configuration.run_name)
pipeline_settings = self._filter_and_validate_settings(
settings=pipeline.configuration.settings,
configuration_level=ConfigurationLevel.PIPELINE,
stack=stack,
)
pipeline.configure(settings=pipeline_settings, merge=False)
settings_to_passdown = {
key: settings
for key, settings in pipeline_settings.items()
if ConfigurationLevel.STEP in settings.LEVEL
}
steps = {
invocation_id: self._compile_step_invocation(
invocation=invocation,
pipeline_settings=settings_to_passdown,
pipeline_extra=pipeline.configuration.extra,
stack=stack,
step_config=run_configuration.steps.get(invocation_id),
pipeline_failure_hook_source=pipeline.configuration.failure_hook_source,
pipeline_success_hook_source=pipeline.configuration.success_hook_source,
)
for invocation_id, invocation in self._get_sorted_invocations(
pipeline=pipeline
)
}
self._ensure_required_stack_components_exist(stack=stack, steps=steps)
run_name = run_configuration.run_name or self._get_default_run_name(
pipeline_name=pipeline.name
)
deployment = PipelineDeploymentBaseModel(
run_name_template=run_name,
pipeline_configuration=pipeline.configuration,
step_configurations=steps,
client_environment=get_run_environment_dict(),
)
step_specs = [step.spec for step in steps.values()]
pipeline_spec = self._compute_pipeline_spec(
pipeline=pipeline, step_specs=step_specs
)
logger.debug("Compiled pipeline deployment: %s", deployment)
logger.debug("Compiled pipeline spec: %s", pipeline_spec)
return deployment, pipeline_spec
def compile_spec(self, pipeline: "Pipeline") -> PipelineSpec:
"""Compiles a ZenML pipeline to a pipeline spec.
This method can be used when a pipeline spec is needed but the full
deployment including stack information is not required.
Args:
pipeline: The pipeline to compile.
Returns:
The compiled pipeline spec.
"""
logger.debug(
"Compiling pipeline spec for pipeline `%s`.", pipeline.name
)
# Copy the pipeline before we connect the steps, so we don't mess with
# the pipeline object/step objects in any way
pipeline = copy.deepcopy(pipeline)
invocations = [
self._get_step_spec(
invocation=invocation,
)
for _, invocation in self._get_sorted_invocations(
pipeline=pipeline
)
]
pipeline_spec = self._compute_pipeline_spec(
pipeline=pipeline, step_specs=invocations
)
logger.debug("Compiled pipeline spec: %s", pipeline_spec)
return pipeline_spec
def _apply_run_configuration(
self, pipeline: "Pipeline", config: PipelineRunConfiguration
) -> None:
"""Applies run configurations to the pipeline and its steps.
Args:
pipeline: The pipeline to configure.
config: The run configurations.
Raises:
KeyError: If the run configuration contains options for a
non-existent step.
"""
pipeline.configure(
enable_cache=config.enable_cache,
enable_artifact_metadata=config.enable_artifact_metadata,
enable_artifact_visualization=config.enable_artifact_visualization,
enable_step_logs=config.enable_step_logs,
settings=config.settings,
extra=config.extra,
)
for invocation_id in config.steps:
if invocation_id not in pipeline.invocations:
raise KeyError(f"No step invocation with id {invocation_id}.")
# Override `enable_cache` of all steps if set at run level
if config.enable_cache is not None:
for invocation in pipeline.invocations.values():
invocation.step.configure(enable_cache=config.enable_cache)
# Override `enable_artifact_metadata` of all steps if set at run level
if config.enable_artifact_metadata is not None:
for invocation in pipeline.invocations.values():
invocation.step.configure(
enable_artifact_metadata=config.enable_artifact_metadata
)
# Override `enable_artifact_visualization` if set at run level
if config.enable_artifact_visualization is not None:
for invocation in pipeline.invocations.values():
invocation.step.configure(
enable_artifact_visualization=config.enable_artifact_visualization
)
# Override `enable_step_logs` if set at run level
if config.enable_step_logs is not None:
for invocation in pipeline.invocations.values():
invocation.step.configure(
enable_step_logs=config.enable_step_logs
)
def _apply_stack_default_settings(
self, pipeline: "Pipeline", stack: "Stack"
) -> None:
"""Applies stack default settings to a pipeline.
Args:
pipeline: The pipeline to which to apply the default settings.
stack: The stack containing potential default settings.
"""
pipeline_settings = pipeline.configuration.settings
for component in stack.components.values():
if not component.settings_class:
continue
settings_key = settings_utils.get_stack_component_setting_key(
component
)
default_settings = self._get_default_settings(component)
if settings_key in pipeline_settings:
combined_settings = pydantic_utils.update_model(
default_settings, update=pipeline_settings[settings_key]
)
pipeline_settings[settings_key] = combined_settings
else:
pipeline_settings[settings_key] = default_settings
pipeline.configure(settings=pipeline_settings, merge=False)
def _get_default_settings(
self,
stack_component: "StackComponent",
) -> "BaseSettings":
"""Gets default settings configured on a stack component.
Args:
stack_component: The stack component for which to get the settings.
Returns:
The settings configured on the stack component.
"""
assert stack_component.settings_class
# Exclude additional config attributes that aren't part of the settings
field_names = set(stack_component.settings_class.__fields__)
default_settings = stack_component.settings_class.parse_obj(
stack_component.config.dict(
include=field_names, exclude_unset=True, exclude_defaults=True
)
)
return default_settings
@staticmethod
def _verify_run_name(run_name: str) -> None:
"""Verifies that the run name contains only valid placeholders.
Args:
run_name: The run name to verify.
Raises:
ValueError: If the run name contains invalid placeholders.
"""
valid_placeholder_names = {"date", "time"}
placeholders = {
v[1] for v in string.Formatter().parse(run_name) if v[1]
}
if not placeholders.issubset(valid_placeholder_names):
raise ValueError(
f"Invalid run name {run_name}. Only the placeholders "
f"{valid_placeholder_names} are allowed in run names."
)
def _verify_upstream_steps(
self, invocation: "StepInvocation", pipeline: "Pipeline"
) -> None:
"""Verifies the upstream steps for a step invocation.
Args:
invocation: The step invocation for which to verify the upstream
steps.
pipeline: The parent pipeline of the invocation.
Raises:
RuntimeError: If an upstream step is missing.
"""
available_steps = set(pipeline.invocations)
invalid_upstream_steps = invocation.upstream_steps - available_steps
if invalid_upstream_steps:
raise RuntimeError(
f"Invalid upstream steps: {invalid_upstream_steps}. Available "
f"steps in this pipeline: {available_steps}."
)
def _filter_and_validate_settings(
self,
settings: Dict[str, "BaseSettings"],
configuration_level: ConfigurationLevel,
stack: "Stack",
) -> Dict[str, "BaseSettings"]:
"""Filters and validates settings.
Args:
settings: The settings to check.
configuration_level: The level on which these settings
were configured.
stack: The stack on which the pipeline will run.
Raises:
TypeError: If settings with an unsupported configuration
level were specified.
Returns:
The filtered settings.
"""
validated_settings = {}
for key, settings_instance in settings.items():
resolver = SettingsResolver(key=key, settings=settings_instance)
try:
settings_instance = resolver.resolve(stack=stack)
except KeyError:
logger.info(
"Not including stack component settings with key `%s`.",
key,
)
continue
if configuration_level not in settings_instance.LEVEL:
raise TypeError(
f"The settings class {settings_instance.__class__} can not "
f"be specified on a {configuration_level.name} level."
)
validated_settings[key] = settings_instance
return validated_settings
def _get_step_spec(
self,
invocation: "StepInvocation",
) -> StepSpec:
"""Gets the spec for a step invocation.
Args:
invocation: The invocation for which to get the spec.
Returns:
The step spec.
"""
inputs = {
key: InputSpec(
step_name=artifact.invocation_id,
output_name=artifact.output_name,
)
for key, artifact in invocation.input_artifacts.items()
}
return StepSpec(
source=invocation.step.resolve(),
upstream_steps=sorted(invocation.upstream_steps),
inputs=inputs,
pipeline_parameter_name=invocation.id,
)
def _compile_step_invocation(
self,
invocation: "StepInvocation",
pipeline_settings: Dict[str, "BaseSettings"],
pipeline_extra: Dict[str, Any],
stack: "Stack",
step_config: Optional["StepConfigurationUpdate"],
pipeline_failure_hook_source: Optional["Source"] = None,
pipeline_success_hook_source: Optional["Source"] = None,
) -> Step:
"""Compiles a ZenML step.
Args:
invocation: The step invocation to compile.
pipeline_settings: settings configured on the
pipeline of the step.
pipeline_extra: Extra values configured on the pipeline of the step.
stack: The stack on which the pipeline will be run.
step_config: Run configuration for the step.
pipeline_failure_hook_source: Source for the failure hook.
pipeline_success_hook_source: Source for the success hook.
Returns:
The compiled step.
"""
# Copy the invocation (including its referenced step) before we apply
# the step configuration which is exclusive to this invocation.
invocation = copy.deepcopy(invocation)
step = invocation.step
if step_config:
step._apply_configuration(step_config)
step_spec = self._get_step_spec(invocation=invocation)
step_settings = self._filter_and_validate_settings(
settings=step.configuration.settings,
configuration_level=ConfigurationLevel.STEP,
stack=stack,
)
step_extra = step.configuration.extra
step_on_failure_hook_source = step.configuration.failure_hook_source
step_on_success_hook_source = step.configuration.success_hook_source
step.configure(
settings=pipeline_settings,
extra=pipeline_extra,
on_failure=pipeline_failure_hook_source,
on_success=pipeline_success_hook_source,
merge=False,
)
step.configure(
settings=step_settings,
extra=step_extra,
on_failure=step_on_failure_hook_source,
on_success=step_on_success_hook_source,
merge=True,
)
parameters_to_ignore = (
set(step_config.parameters) if step_config else set()
)
complete_step_configuration = invocation.finalize(
parameters_to_ignore=parameters_to_ignore
)
return Step(spec=step_spec, config=complete_step_configuration)
@staticmethod
def _get_default_run_name(pipeline_name: str) -> str:
"""Gets the default name for a pipeline run.
Args:
pipeline_name: Name of the pipeline which will be run.
Returns:
Run name.
"""
return f"{pipeline_name}-{{date}}-{{time}}"
def _get_sorted_invocations(
self,
pipeline: "Pipeline",
) -> List[Tuple[str, "StepInvocation"]]:
"""Sorts the step invocations of a pipeline using topological sort.
The resulting list of invocations will be in an order that can be
executed sequentially without any conflicts.
Args:
pipeline: The pipeline of which to sort the invocations
Returns:
The sorted steps.
"""
from zenml.orchestrators.dag_runner import reverse_dag
from zenml.orchestrators.topsort import topsorted_layers
# Sort step names using topological sort
dag: Dict[str, List[str]] = {}
for name, step in pipeline.invocations.items():
self._verify_upstream_steps(invocation=step, pipeline=pipeline)
dag[name] = list(step.upstream_steps)
reversed_dag: Dict[str, List[str]] = reverse_dag(dag)
layers = topsorted_layers(
nodes=list(dag),
get_node_id_fn=lambda node: node,
get_parent_nodes=lambda node: dag[node],
get_child_nodes=lambda node: reversed_dag[node],
)
sorted_step_names = [step for layer in layers for step in layer]
sorted_invocations: List[Tuple[str, "StepInvocation"]] = [
(name_in_pipeline, pipeline.invocations[name_in_pipeline])
for name_in_pipeline in sorted_step_names
]
return sorted_invocations
@staticmethod
def _ensure_required_stack_components_exist(
stack: "Stack", steps: Mapping[str, "Step"]
) -> None:
"""Ensures that the stack components required for each step exist.
Args:
stack: The stack on which the pipeline should be deployed.
steps: The steps of the pipeline.
Raises:
StackValidationError: If a required stack component is missing.
"""
available_step_operators = (
{stack.step_operator.name} if stack.step_operator else set()
)
available_experiment_trackers = (
{stack.experiment_tracker.name}
if stack.experiment_tracker
else set()
)
for name, step in steps.items():
step_operator = step.config.step_operator
if step_operator and step_operator not in available_step_operators:
raise StackValidationError(
f"Step '{name}' requires step operator "
f"'{step_operator}' which is not configured in "
f"the stack '{stack.name}'. Available step operators: "
f"{available_step_operators}."
)
experiment_tracker = step.config.experiment_tracker
if (
experiment_tracker
and experiment_tracker not in available_experiment_trackers
):
raise StackValidationError(
f"Step '{name}' requires experiment tracker "
f"'{experiment_tracker}' which is not "
f"configured in the stack '{stack.name}'. Available "
f"experiment trackers: {available_experiment_trackers}."
)
@staticmethod
def _compute_pipeline_spec(
pipeline: "Pipeline", step_specs: List["StepSpec"]
) -> "PipelineSpec":
"""Computes the pipeline spec.
Args:
pipeline: The pipeline for which to compute the spec.
step_specs: The step specs for the pipeline.
Returns:
The pipeline spec.
"""
from zenml.pipelines import BasePipeline
additional_spec_args: Dict[str, Any] = {}
if isinstance(pipeline, BasePipeline):
# use older spec version for legacy pipelines
additional_spec_args["version"] = "0.3"
else:
additional_spec_args["source"] = pipeline.resolve()
additional_spec_args["parameters"] = pipeline._parameters
return PipelineSpec(steps=step_specs, **additional_spec_args)
compile(self, pipeline, stack, run_configuration)
Compiles a ZenML pipeline to a serializable representation.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
pipeline |
Pipeline |
The pipeline to compile. |
required |
stack |
Stack |
The stack on which the pipeline will run. |
required |
run_configuration |
PipelineRunConfiguration |
The run configuration for this pipeline. |
required |
Returns:
| Type | Description |
|---|---|
Tuple[zenml.models.pipeline_deployment_models.PipelineDeploymentBaseModel, zenml.config.pipeline_spec.PipelineSpec] |
The compiled pipeline deployment and spec |
Source code in zenml/config/compiler.py
def compile(
self,
pipeline: "Pipeline",
stack: "Stack",
run_configuration: PipelineRunConfiguration,
) -> Tuple[PipelineDeploymentBaseModel, PipelineSpec]:
"""Compiles a ZenML pipeline to a serializable representation.
Args:
pipeline: The pipeline to compile.
stack: The stack on which the pipeline will run.
run_configuration: The run configuration for this pipeline.
Returns:
The compiled pipeline deployment and spec
"""
logger.debug("Compiling pipeline `%s`.", pipeline.name)
# Copy the pipeline before we apply any run-level configurations, so
# we don't mess with the pipeline object/step objects in any way
pipeline = copy.deepcopy(pipeline)
self._apply_run_configuration(
pipeline=pipeline, config=run_configuration
)
self._apply_stack_default_settings(pipeline=pipeline, stack=stack)
if run_configuration.run_name:
self._verify_run_name(run_configuration.run_name)
pipeline_settings = self._filter_and_validate_settings(
settings=pipeline.configuration.settings,
configuration_level=ConfigurationLevel.PIPELINE,
stack=stack,
)
pipeline.configure(settings=pipeline_settings, merge=False)
settings_to_passdown = {
key: settings
for key, settings in pipeline_settings.items()
if ConfigurationLevel.STEP in settings.LEVEL
}
steps = {
invocation_id: self._compile_step_invocation(
invocation=invocation,
pipeline_settings=settings_to_passdown,
pipeline_extra=pipeline.configuration.extra,
stack=stack,
step_config=run_configuration.steps.get(invocation_id),
pipeline_failure_hook_source=pipeline.configuration.failure_hook_source,
pipeline_success_hook_source=pipeline.configuration.success_hook_source,
)
for invocation_id, invocation in self._get_sorted_invocations(
pipeline=pipeline
)
}
self._ensure_required_stack_components_exist(stack=stack, steps=steps)
run_name = run_configuration.run_name or self._get_default_run_name(
pipeline_name=pipeline.name
)
deployment = PipelineDeploymentBaseModel(
run_name_template=run_name,
pipeline_configuration=pipeline.configuration,
step_configurations=steps,
client_environment=get_run_environment_dict(),
)
step_specs = [step.spec for step in steps.values()]
pipeline_spec = self._compute_pipeline_spec(
pipeline=pipeline, step_specs=step_specs
)
logger.debug("Compiled pipeline deployment: %s", deployment)
logger.debug("Compiled pipeline spec: %s", pipeline_spec)
return deployment, pipeline_spec
compile_spec(self, pipeline)
Compiles a ZenML pipeline to a pipeline spec.
This method can be used when a pipeline spec is needed but the full deployment including stack information is not required.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
pipeline |
Pipeline |
The pipeline to compile. |
required |
Returns:
| Type | Description |
|---|---|
PipelineSpec |
The compiled pipeline spec. |
Source code in zenml/config/compiler.py
def compile_spec(self, pipeline: "Pipeline") -> PipelineSpec:
"""Compiles a ZenML pipeline to a pipeline spec.
This method can be used when a pipeline spec is needed but the full
deployment including stack information is not required.
Args:
pipeline: The pipeline to compile.
Returns:
The compiled pipeline spec.
"""
logger.debug(
"Compiling pipeline spec for pipeline `%s`.", pipeline.name
)
# Copy the pipeline before we connect the steps, so we don't mess with
# the pipeline object/step objects in any way
pipeline = copy.deepcopy(pipeline)
invocations = [
self._get_step_spec(
invocation=invocation,
)
for _, invocation in self._get_sorted_invocations(
pipeline=pipeline
)
]
pipeline_spec = self._compute_pipeline_spec(
pipeline=pipeline, step_specs=invocations
)
logger.debug("Compiled pipeline spec: %s", pipeline_spec)
return pipeline_spec
constants
ZenML settings constants.
docker_settings
Docker settings.
DockerSettings (BaseSettings)
pydantic-model
Settings for building Docker images to run ZenML pipelines.
Build process:
- No
dockerfilespecified: If any of the options regarding requirements, environment variables or copying files require us to build an image, ZenML will build this image. Otherwise theparent_imagewill be used to run the pipeline. dockerfilespecified: ZenML will first build an image based on the specified Dockerfile. If any of the options regarding requirements, environment variables or copying files require an additional image built on top of that, ZenML will build a second image. If not, the image build from the specified Dockerfile will be used to run the pipeline.
Requirements installation order:
Depending on the configuration of this object, requirements will be
installed in the following order (each step optional):
- The packages installed in your local python environment
- The packages specified via the requirements attribute
- The packages specified via the required_integrations and potentially
stack requirements
- The packages specified via the required_hub_plugins attribute
Attributes:
| Name | Type | Description |
|---|---|---|
parent_image |
Optional[str] |
Full name of the Docker image that should be used as the parent for the image that will be built. Defaults to a ZenML image built for the active Python and ZenML version. Additional notes:
* If you specify a custom image here, you need to make sure it has
ZenML installed.
* If this is a non-local image, the environment which is running
the pipeline and building the Docker image needs to be able to pull
this image.
* If a custom |
dockerfile |
Optional[str] |
Path to a custom Dockerfile that should be built. Depending on the other values you specify in this object, the resulting image will be used directly to run your pipeline or ZenML will use it as a parent image to build on top of. See the general docstring of this class for more information. Additional notes:
* If you specify this, the |
build_context_root |
Optional[str] |
Build context root for the Docker build, only used
when the |
build_options |
Dict[str, Any] |
Additional options that will be passed unmodified to the
Docker build call when building an image using the specified
|
skip_build |
bool |
If set to |
target_repository |
str |
Name of the Docker repository to which the image should be pushed. This repository will be appended to the registry URI of the container registry of your stack and should therefore not include any registry. |
replicate_local_python_environment |
Union[List[str], zenml.config.docker_settings.PythonEnvironmentExportMethod] |
If not |
requirements |
Union[NoneType, str, List[str]] |
Path to a requirements file or a list of required pip
packages. During the image build, these requirements will be
installed using pip. If you need to use a different tool to
resolve and/or install your packages, please use a custom parent
image or specify a custom |
required_integrations |
List[str] |
List of ZenML integrations that should be installed. All requirements for the specified integrations will be installed inside the Docker image. |
required_hub_plugins |
List[str] |
List of ZenML Hub plugins to install.
Expected format: '( |
install_stack_requirements |
bool |
If |
apt_packages |
List[str] |
APT packages to install inside the Docker image. |
environment |
Dict[str, Any] |
Dictionary of environment variables to set inside the Docker image. |
dockerignore |
Optional[str] |
Path to a dockerignore file to use when building the Docker image. |
copy_files |
bool |
DEPRECATED, use the |
copy_global_config |
bool |
DEPRECATED/UNUSED. |
user |
Optional[str] |
If not |
source_files |
SourceFileMode |
Defines how the user source files will be handled when building the Docker image. * INCLUDE: The files will be included in the Docker image. * DOWNLOAD: The files will be downloaded when running the image. If this is specified, the files must be inside a registered code repository and the repository must have no local changes, otherwise the build will fail. * DOWNLOAD_OR_INCLUDE: The files will be downloaded if they're inside a registered code repository and the repository has no local changes, otherwise they will be included in the image. * IGNORE: The files will not be included or downloaded in the image. If you use this option, you're responsible that all the files to run your steps exist in the right place. |
Source code in zenml/config/docker_settings.py
class DockerSettings(BaseSettings):
"""Settings for building Docker images to run ZenML pipelines.
Build process:
--------------
* No `dockerfile` specified: If any of the options regarding
requirements, environment variables or copying files require us to build an
image, ZenML will build this image. Otherwise the `parent_image` will be
used to run the pipeline.
* `dockerfile` specified: ZenML will first build an image based on the
specified Dockerfile. If any of the options regarding
requirements, environment variables or copying files require an additional
image built on top of that, ZenML will build a second image. If not, the
image build from the specified Dockerfile will be used to run the pipeline.
Requirements installation order:
--------------------------------
Depending on the configuration of this object, requirements will be
installed in the following order (each step optional):
- The packages installed in your local python environment
- The packages specified via the `requirements` attribute
- The packages specified via the `required_integrations` and potentially
stack requirements
- The packages specified via the `required_hub_plugins` attribute
Attributes:
parent_image: Full name of the Docker image that should be
used as the parent for the image that will be built. Defaults to
a ZenML image built for the active Python and ZenML version.
Additional notes:
* If you specify a custom image here, you need to make sure it has
ZenML installed.
* If this is a non-local image, the environment which is running
the pipeline and building the Docker image needs to be able to pull
this image.
* If a custom `dockerfile` is specified for this settings
object, this parent image will be ignored.
dockerfile: Path to a custom Dockerfile that should be built. Depending
on the other values you specify in this object, the resulting
image will be used directly to run your pipeline or ZenML will use
it as a parent image to build on top of. See the general docstring
of this class for more information.
Additional notes:
* If you specify this, the `parent_image` attribute will be ignored.
* If you specify this, the image built from this Dockerfile needs
to have ZenML installed.
build_context_root: Build context root for the Docker build, only used
when the `dockerfile` attribute is set. If this is left empty, the
build context will only contain the Dockerfile.
build_options: Additional options that will be passed unmodified to the
Docker build call when building an image using the specified
`dockerfile`. You can use this to for example specify build
args or a target stage. See
https://docker-py.readthedocs.io/en/stable/images.html#docker.models.images.ImageCollection.build
for a full list of available options.
skip_build: If set to `True`, the parent image will be used directly to
run the steps of your pipeline.
target_repository: Name of the Docker repository to which the
image should be pushed. This repository will be appended to the
registry URI of the container registry of your stack and should
therefore **not** include any registry.
replicate_local_python_environment: If not `None`, ZenML will use the
specified method to generate a requirements file that replicates
the packages installed in the currently running python environment.
This requirements file will then be installed in the Docker image.
requirements: Path to a requirements file or a list of required pip
packages. During the image build, these requirements will be
installed using pip. If you need to use a different tool to
resolve and/or install your packages, please use a custom parent
image or specify a custom `dockerfile`.
required_integrations: List of ZenML integrations that should be
installed. All requirements for the specified integrations will
be installed inside the Docker image.
required_hub_plugins: List of ZenML Hub plugins to install.
Expected format: '(<author_username>/)<plugin_name>==<version>'.
If no version is specified, the latest version is taken. The
packages of required plugins and all their dependencies will be
installed inside the Docker image.
install_stack_requirements: If `True`, ZenML will automatically detect
if components of your active stack are part of a ZenML integration
and install the corresponding requirements and apt packages.
If you set this to `False` or use custom components in your stack,
you need to make sure these get installed by specifying them in
the `requirements` and `apt_packages` attributes.
apt_packages: APT packages to install inside the Docker image.
environment: Dictionary of environment variables to set inside the
Docker image.
dockerignore: Path to a dockerignore file to use when building the
Docker image.
copy_files: DEPRECATED, use the `source_files` attribute instead.
copy_global_config: DEPRECATED/UNUSED.
user: If not `None`, will set the user, make it owner of the `/app`
directory which contains all the user code and run the container
entrypoint as this user.
source_files: Defines how the user source files will be handled when
building the Docker image.
* INCLUDE: The files will be included in the Docker image.
* DOWNLOAD: The files will be downloaded when running the image. If
this is specified, the files must be inside a registered code
repository and the repository must have no local changes,
otherwise the build will fail.
* DOWNLOAD_OR_INCLUDE: The files will be downloaded if they're
inside a registered code repository and the repository has no
local changes, otherwise they will be included in the image.
* IGNORE: The files will not be included or downloaded in the image.
If you use this option, you're responsible that all the files
to run your steps exist in the right place.
"""
parent_image: Optional[str] = None
dockerfile: Optional[str] = None
build_context_root: Optional[str] = None
build_options: Dict[str, Any] = {}
skip_build: bool = False
target_repository: str = "zenml"
replicate_local_python_environment: Optional[
Union[List[str], PythonEnvironmentExportMethod]
] = None
requirements: Union[None, str, List[str]] = None
required_integrations: List[str] = []
required_hub_plugins: List[str] = []
install_stack_requirements: bool = True
apt_packages: List[str] = []
environment: Dict[str, Any] = {}
dockerignore: Optional[str] = None
copy_files: bool = True
copy_global_config: bool = True
user: Optional[str] = None
source_files: SourceFileMode = SourceFileMode.DOWNLOAD_OR_INCLUDE
_deprecation_validator = deprecation_utils.deprecate_pydantic_attributes(
"copy_files", "copy_global_config"
)
@root_validator(pre=True)
def _migrate_copy_files(cls, values: Dict[str, Any]) -> Dict[str, Any]:
"""Migrates the value from the old copy_files attribute.
Args:
values: The settings values.
Returns:
The migrated settings values.
"""
copy_files = values.get("copy_files", None)
if copy_files is None:
return values
if values.get("source_files", None):
# Ignore the copy files value in favor of the new source files
logger.warning(
"Both `copy_files` and `source_files` specified for the "
"DockerSettings, ignoring the `copy_files` value."
)
elif copy_files is True:
values["source_files"] = SourceFileMode.INCLUDE
elif copy_files is False:
values["source_files"] = SourceFileMode.IGNORE
return values
@root_validator
def _validate_skip_build(cls, values: Dict[str, Any]) -> Dict[str, Any]:
"""Ensures that a parent image is passed when trying to skip the build.
Args:
values: The settings values.
Returns:
The validated settings values.
Raises:
ValueError: If the build should be skipped but no parent image
was specified.
"""
skip_build = values.get("skip_build", False)
parent_image = values.get("parent_image")
if skip_build and not parent_image:
raise ValueError(
"Docker settings that specify `skip_build=True` must always "
"contain a `parent_image`. This parent image will be used "
"to run the steps of your pipeline directly without additional "
"Docker builds on top of it."
)
return values
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# prevent extra attributes during model initialization
extra = Extra.forbid
Config
Pydantic configuration class.
Source code in zenml/config/docker_settings.py
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# prevent extra attributes during model initialization
extra = Extra.forbid
PythonEnvironmentExportMethod (Enum)
Different methods to export the local Python environment.
Source code in zenml/config/docker_settings.py
class PythonEnvironmentExportMethod(Enum):
"""Different methods to export the local Python environment."""
PIP_FREEZE = "pip_freeze"
POETRY_EXPORT = "poetry_export"
@property
def command(self) -> str:
"""Shell command that outputs local python packages.
The output string must be something that can be interpreted as a
requirements file for pip once it's written to a file.
Returns:
Shell command.
"""
return {
PythonEnvironmentExportMethod.PIP_FREEZE: "pip freeze",
PythonEnvironmentExportMethod.POETRY_EXPORT: "poetry export --format=requirements.txt",
}[self]
SourceFileMode (Enum)
Different methods to handle source files in Docker images.
Source code in zenml/config/docker_settings.py
class SourceFileMode(Enum):
"""Different methods to handle source files in Docker images."""
INCLUDE = "include"
DOWNLOAD_OR_INCLUDE = "download_or_include"
DOWNLOAD = "download"
IGNORE = "ignore"
global_config
Functionality to support ZenML GlobalConfiguration.
GlobalConfigMetaClass (ModelMetaclass)
Global configuration metaclass.
This metaclass is used to enforce a singleton instance of the GlobalConfiguration class with the following additional properties:
- the GlobalConfiguration is initialized automatically on import with the default configuration, if no config file exists yet.
- the GlobalConfiguration undergoes a schema migration if the version of the config file is older than the current version of the ZenML package.
- a default store is set if no store is configured yet.
Source code in zenml/config/global_config.py
class GlobalConfigMetaClass(ModelMetaclass):
"""Global configuration metaclass.
This metaclass is used to enforce a singleton instance of the
GlobalConfiguration class with the following additional properties:
* the GlobalConfiguration is initialized automatically on import with the
default configuration, if no config file exists yet.
* the GlobalConfiguration undergoes a schema migration if the version of the
config file is older than the current version of the ZenML package.
* a default store is set if no store is configured yet.
"""
def __init__(cls, *args: Any, **kwargs: Any) -> None:
"""Initialize a singleton class.
Args:
*args: positional arguments
**kwargs: keyword arguments
"""
super().__init__(*args, **kwargs)
cls._global_config: Optional["GlobalConfiguration"] = None
def __call__(cls, *args: Any, **kwargs: Any) -> "GlobalConfiguration":
"""Create or return the default global config instance.
If the GlobalConfiguration constructor is called with custom arguments,
the singleton functionality of the metaclass is bypassed: a new
GlobalConfiguration instance is created and returned immediately and
without saving it as the global GlobalConfiguration singleton.
Args:
*args: positional arguments
**kwargs: keyword arguments
Returns:
The global GlobalConfiguration instance.
"""
if args or kwargs:
return cast(
"GlobalConfiguration", super().__call__(*args, **kwargs)
)
if not cls._global_config:
cls._global_config = cast(
"GlobalConfiguration", super().__call__(*args, **kwargs)
)
cls._global_config._migrate_config()
return cls._global_config
__call__(cls, *args, **kwargs)
special
Create or return the default global config instance.
If the GlobalConfiguration constructor is called with custom arguments, the singleton functionality of the metaclass is bypassed: a new GlobalConfiguration instance is created and returned immediately and without saving it as the global GlobalConfiguration singleton.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
*args |
Any |
positional arguments |
() |
**kwargs |
Any |
keyword arguments |
{} |
Returns:
| Type | Description |
|---|---|
GlobalConfiguration |
The global GlobalConfiguration instance. |
Source code in zenml/config/global_config.py
def __call__(cls, *args: Any, **kwargs: Any) -> "GlobalConfiguration":
"""Create or return the default global config instance.
If the GlobalConfiguration constructor is called with custom arguments,
the singleton functionality of the metaclass is bypassed: a new
GlobalConfiguration instance is created and returned immediately and
without saving it as the global GlobalConfiguration singleton.
Args:
*args: positional arguments
**kwargs: keyword arguments
Returns:
The global GlobalConfiguration instance.
"""
if args or kwargs:
return cast(
"GlobalConfiguration", super().__call__(*args, **kwargs)
)
if not cls._global_config:
cls._global_config = cast(
"GlobalConfiguration", super().__call__(*args, **kwargs)
)
cls._global_config._migrate_config()
return cls._global_config
__init__(cls, *args, **kwargs)
special
Initialize a singleton class.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
*args |
Any |
positional arguments |
() |
**kwargs |
Any |
keyword arguments |
{} |
Source code in zenml/config/global_config.py
def __init__(cls, *args: Any, **kwargs: Any) -> None:
"""Initialize a singleton class.
Args:
*args: positional arguments
**kwargs: keyword arguments
"""
super().__init__(*args, **kwargs)
cls._global_config: Optional["GlobalConfiguration"] = None
GlobalConfiguration (BaseModel)
pydantic-model
Stores global configuration options.
Configuration options are read from a config file, but can be overwritten
by environment variables. See GlobalConfiguration.__getattribute__ for
more details.
Attributes:
| Name | Type | Description |
|---|---|---|
user_id |
Unique user id. |
|
user_email |
Email address associated with this client. |
|
user_email_opt_in |
Whether the user has opted in to email communication. |
|
analytics_opt_in |
If a user agreed to sending analytics or not. |
|
version |
Version of ZenML that was last used to create or update the global config. |
|
store |
Store configuration. |
|
active_stack_id |
The ID of the active stack. |
|
active_workspace_name |
The name of the active workspace. |
|
jwt_secret_key |
The secret key used to sign and verify JWT tokens. |
|
_config_path |
Directory where the global config file is stored. |
Source code in zenml/config/global_config.py
class GlobalConfiguration(BaseModel, metaclass=GlobalConfigMetaClass):
"""Stores global configuration options.
Configuration options are read from a config file, but can be overwritten
by environment variables. See `GlobalConfiguration.__getattribute__` for
more details.
Attributes:
user_id: Unique user id.
user_email: Email address associated with this client.
user_email_opt_in: Whether the user has opted in to email communication.
analytics_opt_in: If a user agreed to sending analytics or not.
version: Version of ZenML that was last used to create or update the
global config.
store: Store configuration.
active_stack_id: The ID of the active stack.
active_workspace_name: The name of the active workspace.
jwt_secret_key: The secret key used to sign and verify JWT tokens.
_config_path: Directory where the global config file is stored.
"""
user_id: uuid.UUID = Field(default_factory=uuid.uuid4)
user_email: Optional[str] = None
user_email_opt_in: Optional[bool] = None
analytics_opt_in: bool = True
version: Optional[str]
store: Optional[StoreConfiguration]
active_stack_id: Optional[uuid.UUID]
active_workspace_name: Optional[str]
jwt_secret_key: str = Field(default_factory=generate_jwt_secret_key)
_config_path: str
_zen_store: Optional["BaseZenStore"] = None
_active_workspace: Optional["WorkspaceResponseModel"] = None
def __init__(
self, config_path: Optional[str] = None, **kwargs: Any
) -> None:
"""Initializes a GlobalConfiguration using values from the config file.
GlobalConfiguration is a singleton class: only one instance can exist.
Calling this constructor multiple times will always yield the same
instance (see the exception below).
The `config_path` argument is only meant for internal use and testing
purposes. User code must never pass it to the constructor. When a custom
`config_path` value is passed, an anonymous GlobalConfiguration instance
is created and returned independently of the GlobalConfiguration
singleton and that will have no effect as far as the rest of the ZenML
core code is concerned.
If the config file doesn't exist yet, we try to read values from the
legacy (ZenML version < 0.6) config file.
Args:
config_path: (internal use) custom config file path. When not
specified, the default global configuration path is used and the
global configuration singleton instance is returned. Only used
to create configuration copies for transfer to different
runtime environments.
**kwargs: keyword arguments
"""
self._config_path = config_path or self.default_config_directory()
config_values = self._read_config()
config_values.update(**kwargs)
super().__init__(**config_values)
if not fileio.exists(self._config_file(config_path)):
self._write_config()
@classmethod
def get_instance(cls) -> Optional["GlobalConfiguration"]:
"""Return the GlobalConfiguration singleton instance.
Returns:
The GlobalConfiguration singleton instance or None, if the
GlobalConfiguration hasn't been initialized yet.
"""
return cls._global_config
@classmethod
def _reset_instance(
cls, config: Optional["GlobalConfiguration"] = None
) -> None:
"""Reset the GlobalConfiguration singleton instance.
This method is only meant for internal use and testing purposes.
Args:
config: The GlobalConfiguration instance to set as the global
singleton. If None, the global GlobalConfiguration singleton is
reset to an empty value.
"""
cls._global_config = config
if config:
config._write_config()
@validator("version")
def _validate_version(cls, v: Optional[str]) -> Optional[str]:
"""Validate the version attribute.
Args:
v: The version attribute value.
Returns:
The version attribute value.
Raises:
RuntimeError: If the version parsing fails.
"""
if v is None:
return v
if not isinstance(version.parse(v), version.Version):
# If the version parsing fails, it returns a `LegacyVersion`
# instead. Check to make sure it's an actual `Version` object
# which represents a valid version.
raise RuntimeError(
f"Invalid version in global configuration: {v}."
)
return v
def __setattr__(self, key: str, value: Any) -> None:
"""Sets an attribute and persists it in the global configuration.
Args:
key: The attribute name.
value: The attribute value.
"""
super().__setattr__(key, value)
if key.startswith("_"):
return
self._write_config()
def __custom_getattribute__(self, key: str) -> Any:
"""Gets an attribute value for a specific key.
If a value for this attribute was specified using an environment
variable called `$(CONFIG_ENV_VAR_PREFIX)$(ATTRIBUTE_NAME)` and its
value can be parsed to the attribute type, the value from this
environment variable is returned instead.
Args:
key: The attribute name.
Returns:
The attribute value.
"""
value = super().__getattribute__(key)
if key.startswith("_") or key not in type(self).__fields__:
return value
environment_variable_name = f"{CONFIG_ENV_VAR_PREFIX}{key.upper()}"
try:
environment_variable_value = os.environ[environment_variable_name]
# set the environment variable value to leverage Pydantic's type
# conversion and validation
super().__setattr__(key, environment_variable_value)
return_value = super().__getattribute__(key)
# set back the old value as we don't want to permanently store
# the environment variable value here
super().__setattr__(key, value)
return return_value
except (ValidationError, KeyError, TypeError):
return value
if not TYPE_CHECKING:
# When defining __getattribute__, mypy allows accessing non-existent
# attributes without failing
# (see https://github.com/python/mypy/issues/13319).
__getattribute__ = __custom_getattribute__
def _migrate_config(self) -> None:
"""Migrates the global config to the latest version."""
curr_version = version.parse(__version__)
if self.version is None:
logger.info(
"Initializing the ZenML global configuration version to %s",
curr_version,
)
else:
config_version = version.parse(self.version)
if config_version > curr_version:
logger.error(
"The ZenML global configuration version (%s) is higher "
"than the version of ZenML currently being used (%s). "
"Read more about this issue and how to solve it here: "
"`https://docs.zenml.io/user-guide/advanced-guide/environment-management/global-settings-of-zenml#version-mismatch-downgrading`",
config_version,
curr_version,
)
# TODO [ENG-899]: Give more detailed instruction on how to
# resolve version mismatch.
return
if config_version == curr_version:
return
logger.info(
"Migrating the ZenML global configuration from version %s "
"to version %s...",
config_version,
curr_version,
)
# this will also trigger rewriting the config file to disk
# to ensure the schema migration results are persisted
self.version = __version__
def _read_config(self) -> Dict[str, Any]:
"""Reads configuration options from disk.
If the config file doesn't exist yet, this method returns an empty
dictionary.
Returns:
A dictionary containing the configuration options.
"""
config_values = {}
if fileio.exists(self._config_file()):
config_values = cast(
Dict[str, Any],
yaml_utils.read_yaml(self._config_file()),
)
return config_values
def _write_config(self, config_path: Optional[str] = None) -> None:
"""Writes the global configuration options to disk.
Args:
config_path: custom config file path. When not specified, the
default global configuration path is used.
"""
config_file = self._config_file(config_path)
yaml_dict = json.loads(self.json(exclude_none=True))
logger.debug(f"Writing config to {config_file}")
if not fileio.exists(config_file):
io_utils.create_dir_recursive_if_not_exists(
config_path or self.config_directory
)
yaml_utils.write_yaml(config_file, yaml_dict)
def _configure_store(
self,
config: StoreConfiguration,
skip_default_registrations: bool = False,
**kwargs: Any,
) -> None:
"""Configure the global zen store.
This method creates and initializes the global store according to the
supplied configuration.
Args:
config: The new store configuration to use.
skip_default_registrations: If `True`, the creation of the default
stack and user in the store will be skipped.
**kwargs: Additional keyword arguments to pass to the store
constructor.
"""
from zenml.zen_stores.base_zen_store import BaseZenStore
if self.store == config and self._zen_store:
# TODO: Do we actually need to create/initialize the store here
# or can we just return instead? We think this is just getting
# called for default registrations.
BaseZenStore.create_store(
config, skip_default_registrations, **kwargs
)
return
# TODO: Revisit the flow regarding the registration of the default
# entities once the analytics v1 is removed.
store = BaseZenStore.create_store(config, True, **kwargs)
logger.debug(f"Configuring the global store to {store.config}")
self.store = store.config
self._zen_store = store
if not skip_default_registrations:
store._initialize_database()
# Sanitize the global configuration to reflect the new store
self._sanitize_config()
self._write_config()
local_stores_path = Path(self.local_stores_path)
local_stores_path.mkdir(parents=True, exist_ok=True)
def _sanitize_config(self) -> None:
"""Sanitize and save the global configuration.
This method is called to ensure that the active stack and workspace
are set to their default values, if possible.
"""
active_workspace, active_stack = self.zen_store.validate_active_config(
self.active_workspace_name,
self.active_stack_id,
config_name="global",
)
self.active_workspace_name = active_workspace.name
self._active_workspace = active_workspace
self.set_active_stack(active_stack)
@staticmethod
def default_config_directory() -> str:
"""Path to the default global configuration directory.
Returns:
The default global configuration directory.
"""
return io_utils.get_global_config_directory()
def _config_file(self, config_path: Optional[str] = None) -> str:
"""Path to the file where global configuration options are stored.
Args:
config_path: custom config file path. When not specified, the
default global configuration path is used.
Returns:
The path to the global configuration file.
"""
return os.path.join(config_path or self._config_path, "config.yaml")
def copy_configuration(
self,
config_path: str,
load_config_path: Optional[PurePath] = None,
store_config: Optional[StoreConfiguration] = None,
empty_store: bool = False,
) -> "GlobalConfiguration":
"""Create a copy of the global config using a different config path.
This method is used to copy the global configuration and store it in a
different configuration path, where it can be loaded in the context of a
new environment, such as a container image.
The configuration files accompanying the store configuration are also
copied to the new configuration path (e.g. certificates etc.)
unless a custom store configuration is provided or the `empty_store`
flag is set to `True`.
If the default local store is currently in use, it will not be included
in the configuration copy. This is the same as explicitly setting the
`empty_store` flag to `True`.
Args:
config_path: path where the configuration copy should be saved
load_config_path: absolute path that will be used to load the copied
configuration. This can be set to a value different from
`config_path` if the configuration copy will be loaded from
a different environment, e.g. when the configuration is copied
to a container image and loaded using a different absolute path.
This will be reflected in the paths and URLs encoded in the
copied configuration.
store_config: custom store configuration to use for the copied
global configuration. If not specified, the current global store
configuration is used.
empty_store: if `True`, an empty store configuration is used for the
copied global configuration. This means that the copied global
configuration will be initialized to the default local store in
the new environment.
Returns:
A new global configuration object copied to the specified path.
"""
from zenml.zen_stores.base_zen_store import BaseZenStore
self._write_config(config_path)
config_copy = GlobalConfiguration(config_path=config_path)
store: Optional[StoreConfiguration] = None
if store_config is not None:
store = store_config
elif empty_store or self.uses_default_store():
store = None
elif self.store:
store_config_class = BaseZenStore.get_store_config_class(
self.store.type
)
store_config_copy = store_config_class.copy_configuration(
self.store, config_path, load_config_path
)
store = store_config_copy
config_copy.store = store
return config_copy
@property
def config_directory(self) -> str:
"""Directory where the global configuration file is located.
Returns:
The directory where the global configuration file is located.
"""
return self._config_path
@property
def local_stores_path(self) -> str:
"""Path where local stores information is stored.
Returns:
The path where local stores information is stored.
"""
if ENV_ZENML_LOCAL_STORES_PATH in os.environ:
return os.environ[ENV_ZENML_LOCAL_STORES_PATH]
return os.path.join(
self.config_directory,
LOCAL_STORES_DIRECTORY_NAME,
)
def get_default_store(self) -> StoreConfiguration:
"""Get the default store configuration.
Returns:
The default store configuration.
"""
from zenml.zen_stores.base_zen_store import BaseZenStore
env_store_config: Dict[str, str] = {}
env_secrets_store_config: Dict[str, str] = {}
for k, v in os.environ.items():
if v == "":
continue
if k.startswith(ENV_ZENML_STORE_PREFIX):
env_store_config[k[len(ENV_ZENML_STORE_PREFIX) :].lower()] = v
elif k.startswith(ENV_ZENML_SECRETS_STORE_PREFIX):
env_secrets_store_config[
k[len(ENV_ZENML_SECRETS_STORE_PREFIX) :].lower()
] = v
if len(env_store_config):
if "type" not in env_store_config and "url" in env_store_config:
env_store_config["type"] = BaseZenStore.get_store_type(
env_store_config["url"]
)
logger.debug(
"Using environment variables to configure the default store"
)
config = StoreConfiguration(
**env_store_config,
)
else:
config = BaseZenStore.get_default_store_config(
path=os.path.join(
self.local_stores_path,
DEFAULT_STORE_DIRECTORY_NAME,
)
)
if len(env_secrets_store_config):
if "type" not in env_secrets_store_config:
env_secrets_store_config["type"] = config.type.value
logger.debug(
"Using environment variables to configure the secrets store"
)
config.secrets_store = SecretsStoreConfiguration(
**env_secrets_store_config
)
return config
def set_default_store(self) -> None:
"""Creates and sets the default store configuration.
Call this method to initialize or revert the store configuration to the
default store.
"""
with event_handler(
AnalyticsEvent.INITIALIZED_STORE
) as analytics_handler:
default_store_cfg = self.get_default_store()
self._configure_store(default_store_cfg)
logger.debug("Using the default store for the global config.")
analytics_handler.metadata = {
"store_type": default_store_cfg.type.value
}
def uses_default_store(self) -> bool:
"""Check if the global configuration uses the default store.
Returns:
`True` if the global configuration uses the default store.
"""
return (
self.store is not None
and self.store.url == self.get_default_store().url
)
def set_store(
self,
config: StoreConfiguration,
skip_default_registrations: bool = False,
**kwargs: Any,
) -> None:
"""Update the active store configuration.
Call this method to validate and update the active store configuration.
Args:
config: The new store configuration to use.
skip_default_registrations: If `True`, the creation of the default
stack and user in the store will be skipped.
**kwargs: Additional keyword arguments to pass to the store
constructor.
"""
with event_handler(
event=AnalyticsEvent.INITIALIZED_STORE,
metadata={"store_type": config.type.value},
):
self._configure_store(config, skip_default_registrations, **kwargs)
logger.info("Updated the global store configuration.")
if self.zen_store.type == StoreType.REST:
# Every time a client connects to a ZenML server, we want to
# group the client ID and the server ID together. This records
# only that a particular client has successfully connected to a
# particular server at least once, but no information about the
# user account is recorded here.
with event_handler(
event=AnalyticsEvent.ZENML_SERVER_CONNECTED
):
server_info = self.zen_store.get_store_info()
identify_group(
AnalyticsGroup.ZENML_SERVER_GROUP,
group_id=server_info.id,
group_metadata={
"version": server_info.version,
"deployment_type": str(
server_info.deployment_type
),
"database_type": str(server_info.database_type),
},
v2=True,
)
@property
def zen_store(self) -> "BaseZenStore":
"""Initialize and/or return the global zen store.
If the store hasn't been initialized yet, it is initialized when this
property is first accessed according to the global store configuration.
Returns:
The current zen store.
"""
if not self.store:
self.set_default_store()
elif self._zen_store is None:
self._configure_store(self.store)
assert self._zen_store is not None
return self._zen_store
def set_active_workspace(
self, workspace: "WorkspaceResponseModel"
) -> "WorkspaceResponseModel":
"""Set the workspace for the local client.
Args:
workspace: The workspace to set active.
Returns:
The workspace that was set active.
"""
self.active_workspace_name = workspace.name
self._active_workspace = workspace
# Sanitize the global configuration to reflect the new workspace
self._sanitize_config()
return workspace
def set_active_stack(self, stack: "StackResponseModel") -> None:
"""Set the active stack for the local client.
Args:
stack: The model of the stack to set active.
"""
self.active_stack_id = stack.id
def get_active_workspace(self) -> "WorkspaceResponseModel":
"""Get a model of the active workspace for the local client.
Returns:
The model of the active workspace.
"""
workspace_name = self.get_active_workspace_name()
if self._active_workspace is not None:
return self._active_workspace
workspace = self.zen_store.get_workspace(
workspace_name_or_id=workspace_name,
)
return self.set_active_workspace(workspace)
def get_active_workspace_name(self) -> str:
"""Get the name of the active workspace.
If the active workspace doesn't exist yet, the ZenStore is reinitialized.
Returns:
The name of the active workspace.
"""
if self.active_workspace_name is None:
_ = self.zen_store
assert self.active_workspace_name is not None
return self.active_workspace_name
def get_active_stack_id(self) -> UUID:
"""Get the ID of the active stack.
If the active stack doesn't exist yet, the ZenStore is reinitialized.
Returns:
The active stack ID.
"""
if self.active_stack_id is None:
_ = self.zen_store
assert self.active_stack_id is not None
return self.active_stack_id
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes from configs of previous ZenML versions to
# permit downgrading
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
# This is needed to allow correct handling of SecretStr values during
# serialization.
json_encoders = {
SecretStr: lambda v: v.get_secret_value() if v else None
}
config_directory: str
property
readonly
Directory where the global configuration file is located.
Returns:
| Type | Description |
|---|---|
str |
The directory where the global configuration file is located. |
local_stores_path: str
property
readonly
Path where local stores information is stored.
Returns:
| Type | Description |
|---|---|
str |
The path where local stores information is stored. |
zen_store: BaseZenStore
property
readonly
Initialize and/or return the global zen store.
If the store hasn't been initialized yet, it is initialized when this property is first accessed according to the global store configuration.
Returns:
| Type | Description |
|---|---|
BaseZenStore |
The current zen store. |
Config
Pydantic configuration class.
Source code in zenml/config/global_config.py
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes from configs of previous ZenML versions to
# permit downgrading
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
# This is needed to allow correct handling of SecretStr values during
# serialization.
json_encoders = {
SecretStr: lambda v: v.get_secret_value() if v else None
}
__custom_getattribute__(self, key)
special
Gets an attribute value for a specific key.
If a value for this attribute was specified using an environment
variable called $(CONFIG_ENV_VAR_PREFIX)$(ATTRIBUTE_NAME) and its
value can be parsed to the attribute type, the value from this
environment variable is returned instead.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The attribute name. |
required |
Returns:
| Type | Description |
|---|---|
Any |
The attribute value. |
Source code in zenml/config/global_config.py
def __custom_getattribute__(self, key: str) -> Any:
"""Gets an attribute value for a specific key.
If a value for this attribute was specified using an environment
variable called `$(CONFIG_ENV_VAR_PREFIX)$(ATTRIBUTE_NAME)` and its
value can be parsed to the attribute type, the value from this
environment variable is returned instead.
Args:
key: The attribute name.
Returns:
The attribute value.
"""
value = super().__getattribute__(key)
if key.startswith("_") or key not in type(self).__fields__:
return value
environment_variable_name = f"{CONFIG_ENV_VAR_PREFIX}{key.upper()}"
try:
environment_variable_value = os.environ[environment_variable_name]
# set the environment variable value to leverage Pydantic's type
# conversion and validation
super().__setattr__(key, environment_variable_value)
return_value = super().__getattribute__(key)
# set back the old value as we don't want to permanently store
# the environment variable value here
super().__setattr__(key, value)
return return_value
except (ValidationError, KeyError, TypeError):
return value
__getattribute__(self, key)
special
Gets an attribute value for a specific key.
If a value for this attribute was specified using an environment
variable called $(CONFIG_ENV_VAR_PREFIX)$(ATTRIBUTE_NAME) and its
value can be parsed to the attribute type, the value from this
environment variable is returned instead.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The attribute name. |
required |
Returns:
| Type | Description |
|---|---|
Any |
The attribute value. |
Source code in zenml/config/global_config.py
def __custom_getattribute__(self, key: str) -> Any:
"""Gets an attribute value for a specific key.
If a value for this attribute was specified using an environment
variable called `$(CONFIG_ENV_VAR_PREFIX)$(ATTRIBUTE_NAME)` and its
value can be parsed to the attribute type, the value from this
environment variable is returned instead.
Args:
key: The attribute name.
Returns:
The attribute value.
"""
value = super().__getattribute__(key)
if key.startswith("_") or key not in type(self).__fields__:
return value
environment_variable_name = f"{CONFIG_ENV_VAR_PREFIX}{key.upper()}"
try:
environment_variable_value = os.environ[environment_variable_name]
# set the environment variable value to leverage Pydantic's type
# conversion and validation
super().__setattr__(key, environment_variable_value)
return_value = super().__getattribute__(key)
# set back the old value as we don't want to permanently store
# the environment variable value here
super().__setattr__(key, value)
return return_value
except (ValidationError, KeyError, TypeError):
return value
__init__(self, config_path=None, **kwargs)
special
Initializes a GlobalConfiguration using values from the config file.
GlobalConfiguration is a singleton class: only one instance can exist. Calling this constructor multiple times will always yield the same instance (see the exception below).
The config_path argument is only meant for internal use and testing
purposes. User code must never pass it to the constructor. When a custom
config_path value is passed, an anonymous GlobalConfiguration instance
is created and returned independently of the GlobalConfiguration
singleton and that will have no effect as far as the rest of the ZenML
core code is concerned.
If the config file doesn't exist yet, we try to read values from the legacy (ZenML version < 0.6) config file.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
config_path |
Optional[str] |
(internal use) custom config file path. When not specified, the default global configuration path is used and the global configuration singleton instance is returned. Only used to create configuration copies for transfer to different runtime environments. |
None |
**kwargs |
Any |
keyword arguments |
{} |
Source code in zenml/config/global_config.py
def __init__(
self, config_path: Optional[str] = None, **kwargs: Any
) -> None:
"""Initializes a GlobalConfiguration using values from the config file.
GlobalConfiguration is a singleton class: only one instance can exist.
Calling this constructor multiple times will always yield the same
instance (see the exception below).
The `config_path` argument is only meant for internal use and testing
purposes. User code must never pass it to the constructor. When a custom
`config_path` value is passed, an anonymous GlobalConfiguration instance
is created and returned independently of the GlobalConfiguration
singleton and that will have no effect as far as the rest of the ZenML
core code is concerned.
If the config file doesn't exist yet, we try to read values from the
legacy (ZenML version < 0.6) config file.
Args:
config_path: (internal use) custom config file path. When not
specified, the default global configuration path is used and the
global configuration singleton instance is returned. Only used
to create configuration copies for transfer to different
runtime environments.
**kwargs: keyword arguments
"""
self._config_path = config_path or self.default_config_directory()
config_values = self._read_config()
config_values.update(**kwargs)
super().__init__(**config_values)
if not fileio.exists(self._config_file(config_path)):
self._write_config()
__json_encoder__(obj)
special
staticmethod
partial(func, args, *keywords) - new function with partial application of the given arguments and keywords.
__setattr__(self, key, value)
special
Sets an attribute and persists it in the global configuration.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The attribute name. |
required |
value |
Any |
The attribute value. |
required |
Source code in zenml/config/global_config.py
def __setattr__(self, key: str, value: Any) -> None:
"""Sets an attribute and persists it in the global configuration.
Args:
key: The attribute name.
value: The attribute value.
"""
super().__setattr__(key, value)
if key.startswith("_"):
return
self._write_config()
copy_configuration(self, config_path, load_config_path=None, store_config=None, empty_store=False)
Create a copy of the global config using a different config path.
This method is used to copy the global configuration and store it in a different configuration path, where it can be loaded in the context of a new environment, such as a container image.
The configuration files accompanying the store configuration are also
copied to the new configuration path (e.g. certificates etc.)
unless a custom store configuration is provided or the empty_store
flag is set to True.
If the default local store is currently in use, it will not be included
in the configuration copy. This is the same as explicitly setting the
empty_store flag to True.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
config_path |
str |
path where the configuration copy should be saved |
required |
load_config_path |
Optional[pathlib.PurePath] |
absolute path that will be used to load the copied
configuration. This can be set to a value different from
|
None |
store_config |
Optional[zenml.config.store_config.StoreConfiguration] |
custom store configuration to use for the copied global configuration. If not specified, the current global store configuration is used. |
None |
empty_store |
bool |
if |
False |
Returns:
| Type | Description |
|---|---|
GlobalConfiguration |
A new global configuration object copied to the specified path. |
Source code in zenml/config/global_config.py
def copy_configuration(
self,
config_path: str,
load_config_path: Optional[PurePath] = None,
store_config: Optional[StoreConfiguration] = None,
empty_store: bool = False,
) -> "GlobalConfiguration":
"""Create a copy of the global config using a different config path.
This method is used to copy the global configuration and store it in a
different configuration path, where it can be loaded in the context of a
new environment, such as a container image.
The configuration files accompanying the store configuration are also
copied to the new configuration path (e.g. certificates etc.)
unless a custom store configuration is provided or the `empty_store`
flag is set to `True`.
If the default local store is currently in use, it will not be included
in the configuration copy. This is the same as explicitly setting the
`empty_store` flag to `True`.
Args:
config_path: path where the configuration copy should be saved
load_config_path: absolute path that will be used to load the copied
configuration. This can be set to a value different from
`config_path` if the configuration copy will be loaded from
a different environment, e.g. when the configuration is copied
to a container image and loaded using a different absolute path.
This will be reflected in the paths and URLs encoded in the
copied configuration.
store_config: custom store configuration to use for the copied
global configuration. If not specified, the current global store
configuration is used.
empty_store: if `True`, an empty store configuration is used for the
copied global configuration. This means that the copied global
configuration will be initialized to the default local store in
the new environment.
Returns:
A new global configuration object copied to the specified path.
"""
from zenml.zen_stores.base_zen_store import BaseZenStore
self._write_config(config_path)
config_copy = GlobalConfiguration(config_path=config_path)
store: Optional[StoreConfiguration] = None
if store_config is not None:
store = store_config
elif empty_store or self.uses_default_store():
store = None
elif self.store:
store_config_class = BaseZenStore.get_store_config_class(
self.store.type
)
store_config_copy = store_config_class.copy_configuration(
self.store, config_path, load_config_path
)
store = store_config_copy
config_copy.store = store
return config_copy
default_config_directory()
staticmethod
Path to the default global configuration directory.
Returns:
| Type | Description |
|---|---|
str |
The default global configuration directory. |
Source code in zenml/config/global_config.py
@staticmethod
def default_config_directory() -> str:
"""Path to the default global configuration directory.
Returns:
The default global configuration directory.
"""
return io_utils.get_global_config_directory()
get_active_stack_id(self)
Get the ID of the active stack.
If the active stack doesn't exist yet, the ZenStore is reinitialized.
Returns:
| Type | Description |
|---|---|
UUID |
The active stack ID. |
Source code in zenml/config/global_config.py
def get_active_stack_id(self) -> UUID:
"""Get the ID of the active stack.
If the active stack doesn't exist yet, the ZenStore is reinitialized.
Returns:
The active stack ID.
"""
if self.active_stack_id is None:
_ = self.zen_store
assert self.active_stack_id is not None
return self.active_stack_id
get_active_workspace(self)
Get a model of the active workspace for the local client.
Returns:
| Type | Description |
|---|---|
WorkspaceResponseModel |
The model of the active workspace. |
Source code in zenml/config/global_config.py
def get_active_workspace(self) -> "WorkspaceResponseModel":
"""Get a model of the active workspace for the local client.
Returns:
The model of the active workspace.
"""
workspace_name = self.get_active_workspace_name()
if self._active_workspace is not None:
return self._active_workspace
workspace = self.zen_store.get_workspace(
workspace_name_or_id=workspace_name,
)
return self.set_active_workspace(workspace)
get_active_workspace_name(self)
Get the name of the active workspace.
If the active workspace doesn't exist yet, the ZenStore is reinitialized.
Returns:
| Type | Description |
|---|---|
str |
The name of the active workspace. |
Source code in zenml/config/global_config.py
def get_active_workspace_name(self) -> str:
"""Get the name of the active workspace.
If the active workspace doesn't exist yet, the ZenStore is reinitialized.
Returns:
The name of the active workspace.
"""
if self.active_workspace_name is None:
_ = self.zen_store
assert self.active_workspace_name is not None
return self.active_workspace_name
get_default_store(self)
Get the default store configuration.
Returns:
| Type | Description |
|---|---|
StoreConfiguration |
The default store configuration. |
Source code in zenml/config/global_config.py
def get_default_store(self) -> StoreConfiguration:
"""Get the default store configuration.
Returns:
The default store configuration.
"""
from zenml.zen_stores.base_zen_store import BaseZenStore
env_store_config: Dict[str, str] = {}
env_secrets_store_config: Dict[str, str] = {}
for k, v in os.environ.items():
if v == "":
continue
if k.startswith(ENV_ZENML_STORE_PREFIX):
env_store_config[k[len(ENV_ZENML_STORE_PREFIX) :].lower()] = v
elif k.startswith(ENV_ZENML_SECRETS_STORE_PREFIX):
env_secrets_store_config[
k[len(ENV_ZENML_SECRETS_STORE_PREFIX) :].lower()
] = v
if len(env_store_config):
if "type" not in env_store_config and "url" in env_store_config:
env_store_config["type"] = BaseZenStore.get_store_type(
env_store_config["url"]
)
logger.debug(
"Using environment variables to configure the default store"
)
config = StoreConfiguration(
**env_store_config,
)
else:
config = BaseZenStore.get_default_store_config(
path=os.path.join(
self.local_stores_path,
DEFAULT_STORE_DIRECTORY_NAME,
)
)
if len(env_secrets_store_config):
if "type" not in env_secrets_store_config:
env_secrets_store_config["type"] = config.type.value
logger.debug(
"Using environment variables to configure the secrets store"
)
config.secrets_store = SecretsStoreConfiguration(
**env_secrets_store_config
)
return config
get_instance()
classmethod
Return the GlobalConfiguration singleton instance.
Returns:
| Type | Description |
|---|---|
Optional[GlobalConfiguration] |
The GlobalConfiguration singleton instance or None, if the GlobalConfiguration hasn't been initialized yet. |
Source code in zenml/config/global_config.py
@classmethod
def get_instance(cls) -> Optional["GlobalConfiguration"]:
"""Return the GlobalConfiguration singleton instance.
Returns:
The GlobalConfiguration singleton instance or None, if the
GlobalConfiguration hasn't been initialized yet.
"""
return cls._global_config
set_active_stack(self, stack)
Set the active stack for the local client.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
stack |
StackResponseModel |
The model of the stack to set active. |
required |
Source code in zenml/config/global_config.py
def set_active_stack(self, stack: "StackResponseModel") -> None:
"""Set the active stack for the local client.
Args:
stack: The model of the stack to set active.
"""
self.active_stack_id = stack.id
set_active_workspace(self, workspace)
Set the workspace for the local client.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
workspace |
WorkspaceResponseModel |
The workspace to set active. |
required |
Returns:
| Type | Description |
|---|---|
WorkspaceResponseModel |
The workspace that was set active. |
Source code in zenml/config/global_config.py
def set_active_workspace(
self, workspace: "WorkspaceResponseModel"
) -> "WorkspaceResponseModel":
"""Set the workspace for the local client.
Args:
workspace: The workspace to set active.
Returns:
The workspace that was set active.
"""
self.active_workspace_name = workspace.name
self._active_workspace = workspace
# Sanitize the global configuration to reflect the new workspace
self._sanitize_config()
return workspace
set_default_store(self)
Creates and sets the default store configuration.
Call this method to initialize or revert the store configuration to the default store.
Source code in zenml/config/global_config.py
def set_default_store(self) -> None:
"""Creates and sets the default store configuration.
Call this method to initialize or revert the store configuration to the
default store.
"""
with event_handler(
AnalyticsEvent.INITIALIZED_STORE
) as analytics_handler:
default_store_cfg = self.get_default_store()
self._configure_store(default_store_cfg)
logger.debug("Using the default store for the global config.")
analytics_handler.metadata = {
"store_type": default_store_cfg.type.value
}
set_store(self, config, skip_default_registrations=False, **kwargs)
Update the active store configuration.
Call this method to validate and update the active store configuration.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
config |
StoreConfiguration |
The new store configuration to use. |
required |
skip_default_registrations |
bool |
If |
False |
**kwargs |
Any |
Additional keyword arguments to pass to the store constructor. |
{} |
Source code in zenml/config/global_config.py
def set_store(
self,
config: StoreConfiguration,
skip_default_registrations: bool = False,
**kwargs: Any,
) -> None:
"""Update the active store configuration.
Call this method to validate and update the active store configuration.
Args:
config: The new store configuration to use.
skip_default_registrations: If `True`, the creation of the default
stack and user in the store will be skipped.
**kwargs: Additional keyword arguments to pass to the store
constructor.
"""
with event_handler(
event=AnalyticsEvent.INITIALIZED_STORE,
metadata={"store_type": config.type.value},
):
self._configure_store(config, skip_default_registrations, **kwargs)
logger.info("Updated the global store configuration.")
if self.zen_store.type == StoreType.REST:
# Every time a client connects to a ZenML server, we want to
# group the client ID and the server ID together. This records
# only that a particular client has successfully connected to a
# particular server at least once, but no information about the
# user account is recorded here.
with event_handler(
event=AnalyticsEvent.ZENML_SERVER_CONNECTED
):
server_info = self.zen_store.get_store_info()
identify_group(
AnalyticsGroup.ZENML_SERVER_GROUP,
group_id=server_info.id,
group_metadata={
"version": server_info.version,
"deployment_type": str(
server_info.deployment_type
),
"database_type": str(server_info.database_type),
},
v2=True,
)
uses_default_store(self)
Check if the global configuration uses the default store.
Returns:
| Type | Description |
|---|---|
bool |
|
Source code in zenml/config/global_config.py
def uses_default_store(self) -> bool:
"""Check if the global configuration uses the default store.
Returns:
`True` if the global configuration uses the default store.
"""
return (
self.store is not None
and self.store.url == self.get_default_store().url
)
generate_jwt_secret_key()
Generate a random JWT secret key.
This key is used to sign and verify generated JWT tokens.
Returns:
| Type | Description |
|---|---|
str |
A random JWT secret key. |
Source code in zenml/config/global_config.py
def generate_jwt_secret_key() -> str:
"""Generate a random JWT secret key.
This key is used to sign and verify generated JWT tokens.
Returns:
A random JWT secret key.
"""
return token_hex(32)
pipeline_configurations
Pipeline configuration classes.
PipelineConfiguration (PipelineConfigurationUpdate)
pydantic-model
Pipeline configuration class.
Source code in zenml/config/pipeline_configurations.py
class PipelineConfiguration(PipelineConfigurationUpdate):
"""Pipeline configuration class."""
name: str
@validator("name")
def ensure_pipeline_name_allowed(cls, name: str) -> str:
"""Ensures the pipeline name is allowed.
Args:
name: Name of the pipeline.
Returns:
The validated name of the pipeline.
Raises:
ValueError: If the name is not allowed.
"""
if name in DISALLOWED_PIPELINE_NAMES:
raise ValueError(
f"Pipeline name '{name}' is not allowed since '{name}' is a "
"reserved key word. Please choose another name."
)
return name
@property
def docker_settings(self) -> "DockerSettings":
"""Docker settings of this pipeline.
Returns:
The Docker settings of this pipeline.
"""
from zenml.config import DockerSettings
model_or_dict: SettingsOrDict = self.settings.get(
DOCKER_SETTINGS_KEY, {}
)
return DockerSettings.parse_obj(model_or_dict)
docker_settings: DockerSettings
property
readonly
Docker settings of this pipeline.
Returns:
| Type | Description |
|---|---|
DockerSettings |
The Docker settings of this pipeline. |
ensure_pipeline_name_allowed(name)
classmethod
Ensures the pipeline name is allowed.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
name |
str |
Name of the pipeline. |
required |
Returns:
| Type | Description |
|---|---|
str |
The validated name of the pipeline. |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If the name is not allowed. |
Source code in zenml/config/pipeline_configurations.py
@validator("name")
def ensure_pipeline_name_allowed(cls, name: str) -> str:
"""Ensures the pipeline name is allowed.
Args:
name: Name of the pipeline.
Returns:
The validated name of the pipeline.
Raises:
ValueError: If the name is not allowed.
"""
if name in DISALLOWED_PIPELINE_NAMES:
raise ValueError(
f"Pipeline name '{name}' is not allowed since '{name}' is a "
"reserved key word. Please choose another name."
)
return name
PipelineConfigurationUpdate (StrictBaseModel)
pydantic-model
Class for pipeline configuration updates.
Source code in zenml/config/pipeline_configurations.py
class PipelineConfigurationUpdate(StrictBaseModel):
"""Class for pipeline configuration updates."""
enable_cache: Optional[bool] = None
enable_artifact_metadata: Optional[bool] = None
enable_artifact_visualization: Optional[bool] = None
enable_step_logs: Optional[bool] = None
settings: Dict[str, BaseSettings] = {}
extra: Dict[str, Any] = {}
failure_hook_source: Optional[Source] = None
success_hook_source: Optional[Source] = None
_convert_source = convert_source_validator(
"failure_hook_source", "success_hook_source"
)
pipeline_run_configuration
Pipeline run configuration class.
PipelineRunConfiguration (StrictBaseModel, YAMLSerializationMixin)
pydantic-model
Class for pipeline run configurations.
Source code in zenml/config/pipeline_run_configuration.py
class PipelineRunConfiguration(
StrictBaseModel, pydantic_utils.YAMLSerializationMixin
):
"""Class for pipeline run configurations."""
run_name: Optional[str] = None
enable_cache: Optional[bool] = None
enable_artifact_metadata: Optional[bool] = None
enable_artifact_visualization: Optional[bool] = None
enable_step_logs: Optional[bool] = None
schedule: Optional[Schedule] = None
build: Union[PipelineBuildBaseModel, UUID, None] = None
steps: Dict[str, StepConfigurationUpdate] = {}
settings: Dict[str, BaseSettings] = {}
extra: Dict[str, Any] = {}
pipeline_spec
Pipeline configuration classes.
PipelineSpec (StrictBaseModel)
pydantic-model
Specification of a pipeline.
Source code in zenml/config/pipeline_spec.py
class PipelineSpec(StrictBaseModel):
"""Specification of a pipeline."""
# Versions:
# - 0.2: Legacy BasePipeline in release <=0.39.1, the upstream steps and
# inputs in the step specs refer to the step names, not the pipeline
# parameter names
# - 0.3: Legacy BasePipeline in release >0.39.1, the upstream steps and
# inputs in the step specs refer to the pipeline parameter names
# - 0.4: New Pipeline class, the upstream steps and
# inputs in the step specs refer to the pipeline parameter names
version: str = "0.4"
source: Optional[Source] = None
parameters: Dict[str, Any] = {}
steps: List[StepSpec]
def __eq__(self, other: Any) -> bool:
"""Returns whether the other object is referring to the same pipeline.
Args:
other: The other object to compare to.
Returns:
True if the other object is referring to the same pipeline.
"""
if isinstance(other, PipelineSpec):
return self.steps == other.steps
return NotImplemented
@property
def json_with_string_sources(self) -> str:
"""JSON representation with sources replaced by their import path.
Returns:
The JSON representation.
"""
from packaging import version
dict_ = self.dict()
if self.source:
dict_["source"] = self.source.import_path
for step_dict in dict_["steps"]:
step_dict["source"] = Source.parse_obj(
step_dict["source"]
).import_path
if version.parse(self.version) < version.parse("0.4"):
# Keep backwards compatibility with old pipeline versions
dict_.pop("source")
dict_.pop("parameters")
return json.dumps(dict_, sort_keys=False, default=pydantic_encoder)
json_with_string_sources: str
property
readonly
JSON representation with sources replaced by their import path.
Returns:
| Type | Description |
|---|---|
str |
The JSON representation. |
__eq__(self, other)
special
Returns whether the other object is referring to the same pipeline.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
other |
Any |
The other object to compare to. |
required |
Returns:
| Type | Description |
|---|---|
bool |
True if the other object is referring to the same pipeline. |
Source code in zenml/config/pipeline_spec.py
def __eq__(self, other: Any) -> bool:
"""Returns whether the other object is referring to the same pipeline.
Args:
other: The other object to compare to.
Returns:
True if the other object is referring to the same pipeline.
"""
if isinstance(other, PipelineSpec):
return self.steps == other.steps
return NotImplemented
resource_settings
Resource settings class used to specify resources for a step.
ByteUnit (Enum)
Enum for byte units.
Source code in zenml/config/resource_settings.py
class ByteUnit(Enum):
"""Enum for byte units."""
KB = "KB"
KIB = "KiB"
MB = "MB"
MIB = "MiB"
GB = "GB"
GIB = "GiB"
TB = "TB"
TIB = "TiB"
PB = "PB"
PIB = "PiB"
@property
def byte_value(self) -> int:
"""Returns the amount of bytes that this unit represents.
Returns:
The byte value of this unit.
"""
return {
ByteUnit.KB: 10**3,
ByteUnit.KIB: 1 << 10,
ByteUnit.MB: 10**6,
ByteUnit.MIB: 1 << 20,
ByteUnit.GB: 10**9,
ByteUnit.GIB: 1 << 30,
ByteUnit.TB: 10**12,
ByteUnit.TIB: 1 << 40,
ByteUnit.PB: 10**15,
ByteUnit.PIB: 1 << 50,
}[self]
ResourceSettings (BaseSettings)
pydantic-model
Hardware resource settings.
Attributes:
| Name | Type | Description |
|---|---|---|
cpu_count |
Optional[pydantic.types.PositiveFloat] |
The amount of CPU cores that should be configured. |
gpu_count |
Optional[pydantic.types.NonNegativeInt] |
The amount of GPUs that should be configured. |
memory |
Optional[str] |
The amount of memory that should be configured. |
Source code in zenml/config/resource_settings.py
class ResourceSettings(BaseSettings):
"""Hardware resource settings.
Attributes:
cpu_count: The amount of CPU cores that should be configured.
gpu_count: The amount of GPUs that should be configured.
memory: The amount of memory that should be configured.
"""
cpu_count: Optional[PositiveFloat] = None
gpu_count: Optional[NonNegativeInt] = None
memory: Optional[str] = Field(regex=MEMORY_REGEX)
@property
def empty(self) -> bool:
"""Returns if this object is "empty" (=no values configured) or not.
Returns:
`True` if no values were configured, `False` otherwise.
"""
# To detect whether this config is empty (= no values specified), we
# check if there are any attributes which are explicitly set to any
# value other than `None`.
return len(self.dict(exclude_unset=True, exclude_none=True)) == 0
def get_memory(
self, unit: Union[str, ByteUnit] = ByteUnit.GB
) -> Optional[float]:
"""Gets the memory configuration in a specific unit.
Args:
unit: The unit to which the memory should be converted.
Raises:
ValueError: If the memory string is invalid.
Returns:
The memory configuration converted to the requested unit, or None
if no memory was configured.
"""
if not self.memory:
return None
if isinstance(unit, str):
unit = ByteUnit(unit)
memory = self.memory
for memory_unit in ByteUnit:
if memory.endswith(memory_unit.value):
memory_value = int(memory[: -len(memory_unit.value)])
return memory_value * memory_unit.byte_value / unit.byte_value
else:
# Should never happen due to the regex validation
raise ValueError(f"Unable to parse memory unit from '{memory}'.")
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# prevent extra attributes during model initialization
extra = Extra.forbid
empty: bool
property
readonly
Returns if this object is "empty" (=no values configured) or not.
Returns:
| Type | Description |
|---|---|
bool |
|
Config
Pydantic configuration class.
Source code in zenml/config/resource_settings.py
class Config:
"""Pydantic configuration class."""
# public attributes are immutable
allow_mutation = False
# prevent extra attributes during model initialization
extra = Extra.forbid
get_memory(self, unit=<ByteUnit.GB: 'GB'>)
Gets the memory configuration in a specific unit.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
unit |
Union[str, zenml.config.resource_settings.ByteUnit] |
The unit to which the memory should be converted. |
<ByteUnit.GB: 'GB'> |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If the memory string is invalid. |
Returns:
| Type | Description |
|---|---|
Optional[float] |
The memory configuration converted to the requested unit, or None if no memory was configured. |
Source code in zenml/config/resource_settings.py
def get_memory(
self, unit: Union[str, ByteUnit] = ByteUnit.GB
) -> Optional[float]:
"""Gets the memory configuration in a specific unit.
Args:
unit: The unit to which the memory should be converted.
Raises:
ValueError: If the memory string is invalid.
Returns:
The memory configuration converted to the requested unit, or None
if no memory was configured.
"""
if not self.memory:
return None
if isinstance(unit, str):
unit = ByteUnit(unit)
memory = self.memory
for memory_unit in ByteUnit:
if memory.endswith(memory_unit.value):
memory_value = int(memory[: -len(memory_unit.value)])
return memory_value * memory_unit.byte_value / unit.byte_value
else:
# Should never happen due to the regex validation
raise ValueError(f"Unable to parse memory unit from '{memory}'.")
schedule
Class for defining a pipeline schedule.
Schedule (BaseModel)
pydantic-model
Class for defining a pipeline schedule.
Attributes:
| Name | Type | Description |
|---|---|---|
name |
Optional[str] |
Optional name to give to the schedule. If not set, a default name will be generated based on the pipeline name and the current date and time. |
cron_expression |
Optional[str] |
Cron expression for the pipeline schedule. If a value for this is set it takes precedence over the start time + interval. |
start_time |
Optional[datetime.datetime] |
datetime object to indicate when to start the schedule. |
end_time |
Optional[datetime.datetime] |
datetime object to indicate when to end the schedule. |
interval_second |
Optional[datetime.timedelta] |
datetime timedelta indicating the seconds between two recurring runs for a periodic schedule. |
catchup |
bool |
Whether the recurring run should catch up if behind schedule. For example, if the recurring run is paused for a while and re-enabled afterwards. If catchup=True, the scheduler will catch up on (backfill) each missed interval. Otherwise, it only schedules the latest interval if more than one interval is ready to be scheduled. Usually, if your pipeline handles backfill internally, you should turn catchup off to avoid duplicate backfill. |
Source code in zenml/config/schedule.py
class Schedule(BaseModel):
"""Class for defining a pipeline schedule.
Attributes:
name: Optional name to give to the schedule. If not set, a default name
will be generated based on the pipeline name and the current date
and time.
cron_expression: Cron expression for the pipeline schedule. If a value
for this is set it takes precedence over the start time + interval.
start_time: datetime object to indicate when to start the schedule.
end_time: datetime object to indicate when to end the schedule.
interval_second: datetime timedelta indicating the seconds between two
recurring runs for a periodic schedule.
catchup: Whether the recurring run should catch up if behind schedule.
For example, if the recurring run is paused for a while and
re-enabled afterwards. If catchup=True, the scheduler will catch
up on (backfill) each missed interval. Otherwise, it only
schedules the latest interval if more than one interval is ready to
be scheduled. Usually, if your pipeline handles backfill
internally, you should turn catchup off to avoid duplicate backfill.
"""
name: Optional[str] = None
cron_expression: Optional[str] = None
start_time: Optional[datetime.datetime] = None
end_time: Optional[datetime.datetime] = None
interval_second: Optional[datetime.timedelta] = None
catchup: bool = False
@root_validator
def _ensure_cron_or_periodic_schedule_configured(
cls, values: Dict[str, Any]
) -> Dict[str, Any]:
"""Ensures that the cron expression or start time + interval are set.
Args:
values: All attributes of the schedule.
Returns:
All schedule attributes.
Raises:
ValueError: If no cron expression or start time + interval were
provided.
"""
cron_expression = values.get("cron_expression")
periodic_schedule = values.get("start_time") and values.get(
"interval_second"
)
if cron_expression and periodic_schedule:
logger.warning(
"This schedule was created with a cron expression as well as "
"values for `start_time` and `interval_seconds`. The resulting "
"behavior depends on the concrete orchestrator implementation "
"but will usually ignore the interval and use the cron "
"expression."
)
return values
elif cron_expression or periodic_schedule:
return values
else:
raise ValueError(
"Either a cron expression or start time and interval seconds "
"need to be set for a valid schedule."
)
@property
def utc_start_time(self) -> Optional[str]:
"""Optional ISO-formatted string of the UTC start time.
Returns:
Optional ISO-formatted string of the UTC start time.
"""
if not self.start_time:
return None
return self.start_time.astimezone(datetime.timezone.utc).isoformat()
@property
def utc_end_time(self) -> Optional[str]:
"""Optional ISO-formatted string of the UTC end time.
Returns:
Optional ISO-formatted string of the UTC end time.
"""
if not self.end_time:
return None
return self.end_time.astimezone(datetime.timezone.utc).isoformat()
utc_end_time: Optional[str]
property
readonly
Optional ISO-formatted string of the UTC end time.
Returns:
| Type | Description |
|---|---|
Optional[str] |
Optional ISO-formatted string of the UTC end time. |
utc_start_time: Optional[str]
property
readonly
Optional ISO-formatted string of the UTC start time.
Returns:
| Type | Description |
|---|---|
Optional[str] |
Optional ISO-formatted string of the UTC start time. |
secret_reference_mixin
Secret reference mixin implementation.
SecretReferenceMixin (BaseModel)
pydantic-model
Mixin class for secret references in pydantic model attributes.
Source code in zenml/config/secret_reference_mixin.py
class SecretReferenceMixin(BaseModel):
"""Mixin class for secret references in pydantic model attributes."""
def __init__(
self, warn_about_plain_text_secrets: bool = False, **kwargs: Any
) -> None:
"""Ensures that secret references are only passed for valid fields.
This method ensures that secret references are not passed for fields
that explicitly prevent them or require pydantic validation.
Args:
warn_about_plain_text_secrets: If true, then warns about using plain-text secrets.
**kwargs: Arguments to initialize this object.
Raises:
ValueError: If an attribute that requires custom pydantic validation
or an attribute which explicitly disallows secret references is
is passed as a secret reference.
"""
for key, value in kwargs.items():
try:
field = self.__class__.__fields__[key]
except KeyError:
# Value for a private attribute or non-existing field, this
# will fail during the upcoming pydantic validation
continue
if value is None:
continue
if not secret_utils.is_secret_reference(value):
if (
secret_utils.is_secret_field(field)
and warn_about_plain_text_secrets
):
logger.warning(
"You specified a plain-text value for the sensitive "
f"attribute `{key}`. This is currently only a warning, "
"but future versions of ZenML will require you to pass "
"in sensitive information as secrets. Check out the "
"documentation on how to configure values with secrets "
"here: https://docs.zenml.io/platform-guide/set-up-your-mlops-platform/use-the-secret-store"
)
continue
if secret_utils.is_clear_text_field(field):
raise ValueError(
f"Passing the `{key}` attribute as a secret reference is "
"not allowed."
)
requires_validation = field.pre_validators or field.post_validators
if requires_validation:
raise ValueError(
f"Passing the attribute `{key}` as a secret reference is "
"not allowed as additional validation is required for "
"this attribute."
)
super().__init__(**kwargs)
def __custom_getattribute__(self, key: str) -> Any:
"""Returns the (potentially resolved) attribute value for the given key.
An attribute value may be either specified directly, or as a secret
reference. In case of a secret reference, this method resolves the
reference and returns the secret value instead.
Args:
key: The key for which to get the attribute value.
Raises:
RuntimeError: If the active stack is missing a secrets manager.
KeyError: If the secret or secret key don't exist.
Returns:
The (potentially resolved) attribute value.
"""
value = super().__getattribute__(key)
if not secret_utils.is_secret_reference(value):
return value
from zenml.client import Client
secret_ref = secret_utils.parse_secret_reference(value)
# Try to resolve the secret using the secret store first
try:
store_secret = Client().get_secret_by_name_and_scope(
name=secret_ref.name,
)
except (KeyError, NotImplementedError):
pass
else:
if secret_ref.key in store_secret.values:
return store_secret.secret_values[secret_ref.key]
else:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value "
f"for key {secret_ref.key}. Available keys: "
f"{set(store_secret.values)}."
)
secrets_manager = Client().active_stack.secrets_manager
if not secrets_manager:
raise RuntimeError(
f"Failed to resolve secret reference for attribute {key}: "
"The active stack does not have a secrets manager."
)
try:
secret = secrets_manager.get_secret(secret_ref.name)
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not exist."
)
try:
secret_value = secret.content[secret_ref.key]
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value for key "
f"{secret_ref.key}. Available keys: {set(secret.content)}."
)
return str(secret_value)
if not TYPE_CHECKING:
# When defining __getattribute__, mypy allows accessing non-existent
# attributes without failing
# (see https://github.com/python/mypy/issues/13319).
__getattribute__ = __custom_getattribute__
@property
def required_secrets(self) -> Set[secret_utils.SecretReference]:
"""All required secrets for this object.
Returns:
The required secrets of this object.
"""
return {
secret_utils.parse_secret_reference(v)
for v in self.dict().values()
if secret_utils.is_secret_reference(v)
}
required_secrets: Set[zenml.utils.secret_utils.SecretReference]
property
readonly
All required secrets for this object.
Returns:
| Type | Description |
|---|---|
Set[zenml.utils.secret_utils.SecretReference] |
The required secrets of this object. |
__custom_getattribute__(self, key)
special
Returns the (potentially resolved) attribute value for the given key.
An attribute value may be either specified directly, or as a secret reference. In case of a secret reference, this method resolves the reference and returns the secret value instead.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The key for which to get the attribute value. |
required |
Exceptions:
| Type | Description |
|---|---|
RuntimeError |
If the active stack is missing a secrets manager. |
KeyError |
If the secret or secret key don't exist. |
Returns:
| Type | Description |
|---|---|
Any |
The (potentially resolved) attribute value. |
Source code in zenml/config/secret_reference_mixin.py
def __custom_getattribute__(self, key: str) -> Any:
"""Returns the (potentially resolved) attribute value for the given key.
An attribute value may be either specified directly, or as a secret
reference. In case of a secret reference, this method resolves the
reference and returns the secret value instead.
Args:
key: The key for which to get the attribute value.
Raises:
RuntimeError: If the active stack is missing a secrets manager.
KeyError: If the secret or secret key don't exist.
Returns:
The (potentially resolved) attribute value.
"""
value = super().__getattribute__(key)
if not secret_utils.is_secret_reference(value):
return value
from zenml.client import Client
secret_ref = secret_utils.parse_secret_reference(value)
# Try to resolve the secret using the secret store first
try:
store_secret = Client().get_secret_by_name_and_scope(
name=secret_ref.name,
)
except (KeyError, NotImplementedError):
pass
else:
if secret_ref.key in store_secret.values:
return store_secret.secret_values[secret_ref.key]
else:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value "
f"for key {secret_ref.key}. Available keys: "
f"{set(store_secret.values)}."
)
secrets_manager = Client().active_stack.secrets_manager
if not secrets_manager:
raise RuntimeError(
f"Failed to resolve secret reference for attribute {key}: "
"The active stack does not have a secrets manager."
)
try:
secret = secrets_manager.get_secret(secret_ref.name)
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not exist."
)
try:
secret_value = secret.content[secret_ref.key]
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value for key "
f"{secret_ref.key}. Available keys: {set(secret.content)}."
)
return str(secret_value)
__getattribute__(self, key)
special
Returns the (potentially resolved) attribute value for the given key.
An attribute value may be either specified directly, or as a secret reference. In case of a secret reference, this method resolves the reference and returns the secret value instead.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The key for which to get the attribute value. |
required |
Exceptions:
| Type | Description |
|---|---|
RuntimeError |
If the active stack is missing a secrets manager. |
KeyError |
If the secret or secret key don't exist. |
Returns:
| Type | Description |
|---|---|
Any |
The (potentially resolved) attribute value. |
Source code in zenml/config/secret_reference_mixin.py
def __custom_getattribute__(self, key: str) -> Any:
"""Returns the (potentially resolved) attribute value for the given key.
An attribute value may be either specified directly, or as a secret
reference. In case of a secret reference, this method resolves the
reference and returns the secret value instead.
Args:
key: The key for which to get the attribute value.
Raises:
RuntimeError: If the active stack is missing a secrets manager.
KeyError: If the secret or secret key don't exist.
Returns:
The (potentially resolved) attribute value.
"""
value = super().__getattribute__(key)
if not secret_utils.is_secret_reference(value):
return value
from zenml.client import Client
secret_ref = secret_utils.parse_secret_reference(value)
# Try to resolve the secret using the secret store first
try:
store_secret = Client().get_secret_by_name_and_scope(
name=secret_ref.name,
)
except (KeyError, NotImplementedError):
pass
else:
if secret_ref.key in store_secret.values:
return store_secret.secret_values[secret_ref.key]
else:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value "
f"for key {secret_ref.key}. Available keys: "
f"{set(store_secret.values)}."
)
secrets_manager = Client().active_stack.secrets_manager
if not secrets_manager:
raise RuntimeError(
f"Failed to resolve secret reference for attribute {key}: "
"The active stack does not have a secrets manager."
)
try:
secret = secrets_manager.get_secret(secret_ref.name)
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not exist."
)
try:
secret_value = secret.content[secret_ref.key]
except KeyError:
raise KeyError(
f"Failed to resolve secret reference for attribute {key}: "
f"The secret {secret_ref.name} does not contain a value for key "
f"{secret_ref.key}. Available keys: {set(secret.content)}."
)
return str(secret_value)
__init__(self, warn_about_plain_text_secrets=False, **kwargs)
special
Ensures that secret references are only passed for valid fields.
This method ensures that secret references are not passed for fields that explicitly prevent them or require pydantic validation.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
warn_about_plain_text_secrets |
bool |
If true, then warns about using plain-text secrets. |
False |
**kwargs |
Any |
Arguments to initialize this object. |
{} |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If an attribute that requires custom pydantic validation or an attribute which explicitly disallows secret references is is passed as a secret reference. |
Source code in zenml/config/secret_reference_mixin.py
def __init__(
self, warn_about_plain_text_secrets: bool = False, **kwargs: Any
) -> None:
"""Ensures that secret references are only passed for valid fields.
This method ensures that secret references are not passed for fields
that explicitly prevent them or require pydantic validation.
Args:
warn_about_plain_text_secrets: If true, then warns about using plain-text secrets.
**kwargs: Arguments to initialize this object.
Raises:
ValueError: If an attribute that requires custom pydantic validation
or an attribute which explicitly disallows secret references is
is passed as a secret reference.
"""
for key, value in kwargs.items():
try:
field = self.__class__.__fields__[key]
except KeyError:
# Value for a private attribute or non-existing field, this
# will fail during the upcoming pydantic validation
continue
if value is None:
continue
if not secret_utils.is_secret_reference(value):
if (
secret_utils.is_secret_field(field)
and warn_about_plain_text_secrets
):
logger.warning(
"You specified a plain-text value for the sensitive "
f"attribute `{key}`. This is currently only a warning, "
"but future versions of ZenML will require you to pass "
"in sensitive information as secrets. Check out the "
"documentation on how to configure values with secrets "
"here: https://docs.zenml.io/platform-guide/set-up-your-mlops-platform/use-the-secret-store"
)
continue
if secret_utils.is_clear_text_field(field):
raise ValueError(
f"Passing the `{key}` attribute as a secret reference is "
"not allowed."
)
requires_validation = field.pre_validators or field.post_validators
if requires_validation:
raise ValueError(
f"Passing the attribute `{key}` as a secret reference is "
"not allowed as additional validation is required for "
"this attribute."
)
super().__init__(**kwargs)
secrets_store_config
Functionality to support ZenML secrets store configurations.
SecretsStoreConfiguration (BaseModel)
pydantic-model
Generic secrets store configuration.
The store configurations of concrete secrets store implementations must inherit from this class and validate any extra attributes that are configured in addition to those defined in this class.
Attributes:
| Name | Type | Description |
|---|---|---|
type |
SecretsStoreType |
The type of store backend. |
class_path |
Optional[str] |
The Python class path of the store backend. Should point to
a subclass of |
Source code in zenml/config/secrets_store_config.py
class SecretsStoreConfiguration(BaseModel):
"""Generic secrets store configuration.
The store configurations of concrete secrets store implementations must
inherit from this class and validate any extra attributes that are
configured in addition to those defined in this class.
Attributes:
type: The type of store backend.
class_path: The Python class path of the store backend. Should point to
a subclass of `BaseSecretsStore`. This is optional and only
required if the store backend is not one of the built-in
implementations.
"""
type: SecretsStoreType
class_path: Optional[str] = None
@root_validator
def validate_custom(cls, values: Dict[str, Any]) -> Dict[str, Any]:
"""Validate that class_path is set for custom secrets stores.
Args:
values: Dict representing user-specified runtime settings.
Returns:
Validated settings.
Raises:
ValueError: If class_path is not set when using an custom secrets
store.
"""
if not values.get("type"):
return values
if values["type"] == SecretsStoreType.CUSTOM:
if values["class_path"] is None:
raise ValueError(
"A class_path must be set when using a custom secrets "
"store implementation."
)
elif values["class_path"] is not None:
raise ValueError(
f"The class_path attribute is not supported for the "
f"{values['type']} secrets store type."
)
return values
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes to be set in the base class. The concrete
# classes are responsible for validating the attributes.
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
Config
Pydantic configuration class.
Source code in zenml/config/secrets_store_config.py
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes to be set in the base class. The concrete
# classes are responsible for validating the attributes.
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
validate_custom(values)
classmethod
Validate that class_path is set for custom secrets stores.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
values |
Dict[str, Any] |
Dict representing user-specified runtime settings. |
required |
Returns:
| Type | Description |
|---|---|
Dict[str, Any] |
Validated settings. |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If class_path is not set when using an custom secrets store. |
Source code in zenml/config/secrets_store_config.py
@root_validator
def validate_custom(cls, values: Dict[str, Any]) -> Dict[str, Any]:
"""Validate that class_path is set for custom secrets stores.
Args:
values: Dict representing user-specified runtime settings.
Returns:
Validated settings.
Raises:
ValueError: If class_path is not set when using an custom secrets
store.
"""
if not values.get("type"):
return values
if values["type"] == SecretsStoreType.CUSTOM:
if values["class_path"] is None:
raise ValueError(
"A class_path must be set when using a custom secrets "
"store implementation."
)
elif values["class_path"] is not None:
raise ValueError(
f"The class_path attribute is not supported for the "
f"{values['type']} secrets store type."
)
return values
settings_resolver
Class for resolving settings.
SettingsResolver
Class for resolving settings.
This class converts a BaseSettings instance to the correct subclass
depending on the key for which these settings were specified.
Source code in zenml/config/settings_resolver.py
class SettingsResolver:
"""Class for resolving settings.
This class converts a `BaseSettings` instance to the correct subclass
depending on the key for which these settings were specified.
"""
def __init__(self, key: str, settings: "BaseSettings"):
"""Checks if the settings key is valid.
Args:
key: Settings key.
settings: The settings.
Raises:
ValueError: If the settings key is invalid.
"""
if not settings_utils.is_valid_setting_key(key):
raise ValueError(
f"Invalid setting key `{key}`. Setting keys can either refer "
"to general settings (available keys: "
f"{set(settings_utils.get_general_settings())}) or stack "
"component specific settings. Stack component specific keys "
"are of the format "
"`<STACK_COMPONENT_TYPE>.<STACK_COMPONENT_FLAVOR>`."
)
self._key = key
self._settings = settings
def resolve(self, stack: "Stack") -> "BaseSettings":
"""Resolves settings for the given stack.
Args:
stack: The stack for which to resolve the settings.
Returns:
The resolved settings.
"""
if settings_utils.is_general_setting_key(self._key):
target_class = self._resolve_general_settings_class()
else:
target_class = self._resolve_stack_component_setting_class(
stack=stack
)
return self._convert_settings(target_class=target_class)
def _resolve_general_settings_class(
self,
) -> Type["BaseSettings"]:
"""Resolves general settings.
Returns:
The resolved settings.
"""
return settings_utils.get_general_settings()[self._key]
def _resolve_stack_component_setting_class(
self, stack: "Stack"
) -> Type["BaseSettings"]:
"""Resolves stack component settings with the given stack.
Args:
stack: The stack to use for resolving.
Raises:
KeyError: If the stack contains no settings for the key.
Returns:
The resolved settings.
"""
settings_class = stack.setting_classes.get(self._key)
if not settings_class:
raise KeyError(
f"Failed to resolve settings for key {self._key}: "
"No settings for this key exist in the stack. "
"Available settings: "
f"{set(stack.setting_classes)}"
)
return settings_class
def _convert_settings(self, target_class: Type["T"]) -> "T":
"""Converts the settings to their correct class.
Args:
target_class: The correct settings class.
Raises:
SettingsResolvingError: If the conversion failed.
Returns:
The converted settings.
"""
settings_dict = self._settings.dict()
try:
return target_class(**settings_dict)
except ValidationError:
raise SettingsResolvingError(
f"Failed to convert settings `{settings_dict}` to expected "
f"class {target_class}."
)
__init__(self, key, settings)
special
Checks if the settings key is valid.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
Settings key. |
required |
settings |
BaseSettings |
The settings. |
required |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If the settings key is invalid. |
Source code in zenml/config/settings_resolver.py
def __init__(self, key: str, settings: "BaseSettings"):
"""Checks if the settings key is valid.
Args:
key: Settings key.
settings: The settings.
Raises:
ValueError: If the settings key is invalid.
"""
if not settings_utils.is_valid_setting_key(key):
raise ValueError(
f"Invalid setting key `{key}`. Setting keys can either refer "
"to general settings (available keys: "
f"{set(settings_utils.get_general_settings())}) or stack "
"component specific settings. Stack component specific keys "
"are of the format "
"`<STACK_COMPONENT_TYPE>.<STACK_COMPONENT_FLAVOR>`."
)
self._key = key
self._settings = settings
resolve(self, stack)
Resolves settings for the given stack.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
stack |
Stack |
The stack for which to resolve the settings. |
required |
Returns:
| Type | Description |
|---|---|
BaseSettings |
The resolved settings. |
Source code in zenml/config/settings_resolver.py
def resolve(self, stack: "Stack") -> "BaseSettings":
"""Resolves settings for the given stack.
Args:
stack: The stack for which to resolve the settings.
Returns:
The resolved settings.
"""
if settings_utils.is_general_setting_key(self._key):
target_class = self._resolve_general_settings_class()
else:
target_class = self._resolve_stack_component_setting_class(
stack=stack
)
return self._convert_settings(target_class=target_class)
source
Source classes.
CodeRepositorySource (Source)
pydantic-model
Source representing an object from a code repository.
Attributes:
| Name | Type | Description |
|---|---|---|
repository_id |
UUID |
The code repository ID. |
commit |
str |
The commit. |
subdirectory |
str |
The subdirectory of the source root inside the code repository. |
Source code in zenml/config/source.py
class CodeRepositorySource(Source):
"""Source representing an object from a code repository.
Attributes:
repository_id: The code repository ID.
commit: The commit.
subdirectory: The subdirectory of the source root inside the code
repository.
"""
repository_id: UUID
commit: str
subdirectory: str
type: SourceType = SourceType.CODE_REPOSITORY
@validator("type")
def _validate_type(cls, value: SourceType) -> SourceType:
"""Validate the source type.
Args:
value: The source type.
Raises:
ValueError: If the source type is not `CODE_REPOSITORY`.
Returns:
The source type.
"""
if value != SourceType.CODE_REPOSITORY:
raise ValueError("Invalid source type.")
return value
DistributionPackageSource (Source)
pydantic-model
Source representing an object from a distribution package.
Attributes:
| Name | Type | Description |
|---|---|---|
package_name |
str |
Name of the package. |
version |
Optional[str] |
The package version. |
Source code in zenml/config/source.py
class DistributionPackageSource(Source):
"""Source representing an object from a distribution package.
Attributes:
package_name: Name of the package.
version: The package version.
"""
package_name: str
version: Optional[str] = None
type: SourceType = SourceType.DISTRIBUTION_PACKAGE
@validator("type")
def _validate_type(cls, value: SourceType) -> SourceType:
"""Validate the source type.
Args:
value: The source type.
Raises:
ValueError: If the source type is not `DISTRIBUTION_PACKAGE`.
Returns:
The source type.
"""
if value != SourceType.DISTRIBUTION_PACKAGE:
raise ValueError("Invalid source type.")
return value
Source (BaseModel)
pydantic-model
Source specification.
A source specifies a module name as well as an optional attribute of that module. These values can be used to import the module and get the value of the attribute inside the module.
Examples:
The source Source(module="zenml.config.source", attribute="Source")
references the class that this docstring is describing. This class is
defined in the zenml.config.source module and the name of the
attribute is the class name Source.
Attributes:
| Name | Type | Description |
|---|---|---|
module |
str |
The module name. |
attribute |
Optional[str] |
Optional name of the attribute inside the module. |
type |
SourceType |
The type of the source. |
Source code in zenml/config/source.py
class Source(BaseModel):
"""Source specification.
A source specifies a module name as well as an optional attribute of that
module. These values can be used to import the module and get the value
of the attribute inside the module.
Example:
The source `Source(module="zenml.config.source", attribute="Source")`
references the class that this docstring is describing. This class is
defined in the `zenml.config.source` module and the name of the
attribute is the class name `Source`.
Attributes:
module: The module name.
attribute: Optional name of the attribute inside the module.
type: The type of the source.
"""
module: str
attribute: Optional[str] = None
type: SourceType
@classmethod
def from_import_path(
cls, import_path: str, is_module_path: bool = False
) -> "Source":
"""Creates a source from an import path.
Args:
import_path: The import path.
is_module_path: If the import path points to a module or not.
Raises:
ValueError: If the import path is empty.
Returns:
The source.
"""
if not import_path:
raise ValueError(
"Invalid empty import path. The import path needs to refer "
"to a Python module and an optional attribute of that module."
)
# Remove internal version pins for backwards compatibility
if "@" in import_path:
import_path = import_path.split("@", 1)[0]
if is_module_path or "." not in import_path:
module = import_path
attribute = None
else:
module, attribute = import_path.rsplit(".", maxsplit=1)
return Source(
module=module, attribute=attribute, type=SourceType.UNKNOWN
)
@property
def import_path(self) -> str:
"""The import path of the source.
Returns:
The import path of the source.
"""
if self.attribute:
return f"{self.module}.{self.attribute}"
else:
return self.module
@property
def is_internal(self) -> bool:
"""If the source is internal (=from the zenml package).
Returns:
True if the source is internal, False otherwise
"""
if self.type not in {SourceType.UNKNOWN, SourceType.INTERNAL}:
return False
return self.module.split(".", maxsplit=1)[0] == "zenml"
@property
def is_module_source(self) -> bool:
"""If the source is a module source.
Returns:
If the source is a module source.
"""
return self.attribute is None
class Config:
"""Pydantic config class."""
extra = Extra.allow
import_path: str
property
readonly
The import path of the source.
Returns:
| Type | Description |
|---|---|
str |
The import path of the source. |
is_internal: bool
property
readonly
If the source is internal (=from the zenml package).
Returns:
| Type | Description |
|---|---|
bool |
True if the source is internal, False otherwise |
is_module_source: bool
property
readonly
If the source is a module source.
Returns:
| Type | Description |
|---|---|
bool |
If the source is a module source. |
Config
Pydantic config class.
Source code in zenml/config/source.py
class Config:
"""Pydantic config class."""
extra = Extra.allow
from_import_path(import_path, is_module_path=False)
classmethod
Creates a source from an import path.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
import_path |
str |
The import path. |
required |
is_module_path |
bool |
If the import path points to a module or not. |
False |
Exceptions:
| Type | Description |
|---|---|
ValueError |
If the import path is empty. |
Returns:
| Type | Description |
|---|---|
Source |
The source. |
Source code in zenml/config/source.py
@classmethod
def from_import_path(
cls, import_path: str, is_module_path: bool = False
) -> "Source":
"""Creates a source from an import path.
Args:
import_path: The import path.
is_module_path: If the import path points to a module or not.
Raises:
ValueError: If the import path is empty.
Returns:
The source.
"""
if not import_path:
raise ValueError(
"Invalid empty import path. The import path needs to refer "
"to a Python module and an optional attribute of that module."
)
# Remove internal version pins for backwards compatibility
if "@" in import_path:
import_path = import_path.split("@", 1)[0]
if is_module_path or "." not in import_path:
module = import_path
attribute = None
else:
module, attribute = import_path.rsplit(".", maxsplit=1)
return Source(
module=module, attribute=attribute, type=SourceType.UNKNOWN
)
SourceType (Enum)
Enum representing different types of sources.
Source code in zenml/config/source.py
class SourceType(Enum):
"""Enum representing different types of sources."""
USER = "user"
BUILTIN = "builtin"
INTERNAL = "internal"
DISTRIBUTION_PACKAGE = "distribution_package"
CODE_REPOSITORY = "code_repository"
UNKNOWN = "unknown"
convert_source_validator(*attributes)
Function to convert pydantic fields containing legacy class paths.
In older versions, sources (sometimes also called class paths) like
zenml.materializers.BuiltInMaterializer were stored as strings in our
configuration classes. These strings got replaced by a separate class, and
this function returns a validator to convert those old strings to the new
classes.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
*attributes |
str |
List of attributes to convert. |
() |
Returns:
| Type | Description |
|---|---|
AnyClassMethod |
Pydantic validator class method to be used on BaseModel subclasses to convert source fields. |
Source code in zenml/config/source.py
def convert_source_validator(*attributes: str) -> "AnyClassMethod":
"""Function to convert pydantic fields containing legacy class paths.
In older versions, sources (sometimes also called class paths) like
`zenml.materializers.BuiltInMaterializer` were stored as strings in our
configuration classes. These strings got replaced by a separate class, and
this function returns a validator to convert those old strings to the new
classes.
Args:
*attributes: List of attributes to convert.
Returns:
Pydantic validator class method to be used on BaseModel subclasses
to convert source fields.
"""
@validator(*attributes, pre=True, allow_reuse=True)
def _convert_source(
cls: Type[BaseModel], value: Union[Source, str, None]
) -> Optional[Source]:
"""Converts an old source string to a source object.
Args:
cls: The class on which the attributes are defined.
value: Source string or object.
Returns:
The converted source.
"""
if isinstance(value, str):
value = Source.from_import_path(value)
return value
return _convert_source
step_configurations
Pipeline configuration classes.
ArtifactConfiguration (PartialArtifactConfiguration)
pydantic-model
Class representing a complete input/output artifact configuration.
Source code in zenml/config/step_configurations.py
class ArtifactConfiguration(PartialArtifactConfiguration):
"""Class representing a complete input/output artifact configuration."""
materializer_source: Tuple[Source, ...]
@validator("materializer_source", pre=True)
def _convert_source(
cls, value: Union[Source, Dict[str, Any], str, Tuple[Source, ...]]
) -> Tuple[Source, ...]:
"""Converts old source strings to tuples of source objects.
Args:
value: Source string or object.
Returns:
The converted source.
"""
if isinstance(value, str):
value = (Source.from_import_path(value),)
elif isinstance(value, dict):
value = (Source.parse_obj(value),)
elif isinstance(value, Source):
value = (value,)
return value
InputSpec (StrictBaseModel)
pydantic-model
Step input specification.
Source code in zenml/config/step_configurations.py
class InputSpec(StrictBaseModel):
"""Step input specification."""
step_name: str
output_name: str
PartialArtifactConfiguration (StrictBaseModel)
pydantic-model
Class representing a partial input/output artifact configuration.
Source code in zenml/config/step_configurations.py
class PartialArtifactConfiguration(StrictBaseModel):
"""Class representing a partial input/output artifact configuration."""
materializer_source: Optional[Tuple[Source, ...]] = None
# TODO: This could be moved to the `PipelineDeployment` as it's the same
# for all steps/outputs
default_materializer_source: Optional[Source] = None
@root_validator(pre=True)
def _remove_deprecated_attributes(
cls, values: Dict[str, Any]
) -> Dict[str, Any]:
"""Removes deprecated attributes from the values dict.
Args:
values: The values dict used to instantiate the model.
Returns:
The values dict without deprecated attributes.
"""
deprecated_attributes = ["artifact_source"]
for deprecated_attribute in deprecated_attributes:
if deprecated_attribute in values:
values.pop(deprecated_attribute)
return values
@validator("materializer_source", pre=True)
def _convert_source(
cls,
value: Union[None, Source, Dict[str, Any], str, Tuple[Source, ...]],
) -> Optional[Tuple[Source, ...]]:
"""Converts old source strings to tuples of source objects.
Args:
value: Source string or object.
Returns:
The converted source.
"""
if isinstance(value, str):
value = (Source.from_import_path(value),)
elif isinstance(value, dict):
value = (Source.parse_obj(value),)
elif isinstance(value, Source):
value = (value,)
return value
PartialStepConfiguration (StepConfigurationUpdate)
pydantic-model
Class representing a partial step configuration.
Source code in zenml/config/step_configurations.py
class PartialStepConfiguration(StepConfigurationUpdate):
"""Class representing a partial step configuration."""
name: str
caching_parameters: Mapping[str, Any] = {}
external_input_artifacts: Mapping[str, UUID] = {}
outputs: Mapping[str, PartialArtifactConfiguration] = {}
# Override the deprecation validator as we do not want to deprecate the
# `name`` attribute on this class.
_deprecation_validator = deprecation_utils.deprecate_pydantic_attributes()
@root_validator(pre=True)
def _remove_deprecated_attributes(
cls, values: Dict[str, Any]
) -> Dict[str, Any]:
"""Removes deprecated attributes from the values dict.
Args:
values: The values dict used to instantiate the model.
Returns:
The values dict without deprecated attributes.
"""
deprecated_attributes = ["docstring", "inputs"]
for deprecated_attribute in deprecated_attributes:
if deprecated_attribute in values:
values.pop(deprecated_attribute)
return values
Step (StrictBaseModel)
pydantic-model
Class representing a ZenML step.
Source code in zenml/config/step_configurations.py
class Step(StrictBaseModel):
"""Class representing a ZenML step."""
spec: StepSpec
config: StepConfiguration
StepConfiguration (PartialStepConfiguration)
pydantic-model
Step configuration class.
Source code in zenml/config/step_configurations.py
class StepConfiguration(PartialStepConfiguration):
"""Step configuration class."""
outputs: Mapping[str, ArtifactConfiguration] = {}
@property
def resource_settings(self) -> "ResourceSettings":
"""Resource settings of this step configuration.
Returns:
The resource settings of this step configuration.
"""
from zenml.config import ResourceSettings
model_or_dict: SettingsOrDict = self.settings.get(
RESOURCE_SETTINGS_KEY, {}
)
return ResourceSettings.parse_obj(model_or_dict)
@property
def docker_settings(self) -> "DockerSettings":
"""Docker settings of this step configuration.
Returns:
The Docker settings of this step configuration.
"""
from zenml.config import DockerSettings
model_or_dict: SettingsOrDict = self.settings.get(
DOCKER_SETTINGS_KEY, {}
)
return DockerSettings.parse_obj(model_or_dict)
docker_settings: DockerSettings
property
readonly
Docker settings of this step configuration.
Returns:
| Type | Description |
|---|---|
DockerSettings |
The Docker settings of this step configuration. |
resource_settings: ResourceSettings
property
readonly
Resource settings of this step configuration.
Returns:
| Type | Description |
|---|---|
ResourceSettings |
The resource settings of this step configuration. |
StepConfigurationUpdate (StrictBaseModel)
pydantic-model
Class for step configuration updates.
Source code in zenml/config/step_configurations.py
class StepConfigurationUpdate(StrictBaseModel):
"""Class for step configuration updates."""
name: Optional[str] = None
enable_cache: Optional[bool] = None
enable_artifact_metadata: Optional[bool] = None
enable_artifact_visualization: Optional[bool] = None
enable_step_logs: Optional[bool] = None
step_operator: Optional[str] = None
experiment_tracker: Optional[str] = None
parameters: Dict[str, Any] = {}
settings: Dict[str, BaseSettings] = {}
extra: Dict[str, Any] = {}
failure_hook_source: Optional[Source] = None
success_hook_source: Optional[Source] = None
outputs: Mapping[str, PartialArtifactConfiguration] = {}
_convert_source = convert_source_validator(
"failure_hook_source", "success_hook_source"
)
_deprecation_validator = deprecation_utils.deprecate_pydantic_attributes(
"name"
)
StepSpec (StrictBaseModel)
pydantic-model
Specification of a pipeline.
Source code in zenml/config/step_configurations.py
class StepSpec(StrictBaseModel):
"""Specification of a pipeline."""
source: Source
upstream_steps: List[str]
inputs: Dict[str, InputSpec] = {}
# The default value is to ensure compatibility with specs of version <0.2
pipeline_parameter_name: str = ""
_convert_source = convert_source_validator("source")
def __eq__(self, other: Any) -> bool:
"""Returns whether the other object is referring to the same step.
This is the case if the other objects is a `StepSpec` with the same
`upstream_steps` and a `source` that meets one of the following
conditions:
- it is the same as the `source` of this step
- it refers to the same absolute path as the `source` of this step
Args:
other: The other object to compare to.
Returns:
True if the other object is referring to the same step.
"""
if isinstance(other, StepSpec):
if self.upstream_steps != other.upstream_steps:
return False
if self.inputs != other.inputs:
return False
if self.pipeline_parameter_name != other.pipeline_parameter_name:
return False
return self.source.import_path == other.source.import_path
return NotImplemented
__eq__(self, other)
special
Returns whether the other object is referring to the same step.
This is the case if the other objects is a StepSpec with the same
upstream_steps and a source that meets one of the following
!!! conditions
- it is the same as the source of this step
- it refers to the same absolute path as the source of this step
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
other |
Any |
The other object to compare to. |
required |
Returns:
| Type | Description |
|---|---|
bool |
True if the other object is referring to the same step. |
Source code in zenml/config/step_configurations.py
def __eq__(self, other: Any) -> bool:
"""Returns whether the other object is referring to the same step.
This is the case if the other objects is a `StepSpec` with the same
`upstream_steps` and a `source` that meets one of the following
conditions:
- it is the same as the `source` of this step
- it refers to the same absolute path as the `source` of this step
Args:
other: The other object to compare to.
Returns:
True if the other object is referring to the same step.
"""
if isinstance(other, StepSpec):
if self.upstream_steps != other.upstream_steps:
return False
if self.inputs != other.inputs:
return False
if self.pipeline_parameter_name != other.pipeline_parameter_name:
return False
return self.source.import_path == other.source.import_path
return NotImplemented
step_run_info
Step run info.
StepRunInfo (StrictBaseModel)
pydantic-model
All information necessary to run a step.
Source code in zenml/config/step_run_info.py
class StepRunInfo(StrictBaseModel):
"""All information necessary to run a step."""
step_run_id: UUID
run_id: UUID
run_name: str
pipeline_step_name: str
config: StepConfiguration
pipeline: PipelineConfiguration
def get_image(self, key: str) -> str:
"""Gets the Docker image for the given key.
Args:
key: The key for which to get the image.
Raises:
RuntimeError: If the run does not have an associated build.
Returns:
The image name or digest.
"""
from zenml.client import Client
run = Client().get_pipeline_run(self.run_id)
if not run.build:
raise RuntimeError(
f"Missing build for run {run.id}. This is probably because "
"the build was manually deleted."
)
return run.build.get_image(
component_key=key, step=self.pipeline_step_name
)
get_image(self, key)
Gets the Docker image for the given key.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key |
str |
The key for which to get the image. |
required |
Exceptions:
| Type | Description |
|---|---|
RuntimeError |
If the run does not have an associated build. |
Returns:
| Type | Description |
|---|---|
str |
The image name or digest. |
Source code in zenml/config/step_run_info.py
def get_image(self, key: str) -> str:
"""Gets the Docker image for the given key.
Args:
key: The key for which to get the image.
Raises:
RuntimeError: If the run does not have an associated build.
Returns:
The image name or digest.
"""
from zenml.client import Client
run = Client().get_pipeline_run(self.run_id)
if not run.build:
raise RuntimeError(
f"Missing build for run {run.id}. This is probably because "
"the build was manually deleted."
)
return run.build.get_image(
component_key=key, step=self.pipeline_step_name
)
store_config
Functionality to support ZenML store configurations.
StoreConfiguration (BaseModel)
pydantic-model
Generic store configuration.
The store configurations of concrete store implementations must inherit from this class and validate any extra attributes that are configured in addition to those defined in this class.
Attributes:
| Name | Type | Description |
|---|---|---|
type |
StoreType |
The type of store backend. |
url |
str |
The URL of the store backend. |
secrets_store |
Optional[zenml.config.secrets_store_config.SecretsStoreConfiguration] |
The configuration of the secrets store to use to store secrets. If not set, secrets management is disabled. |
Source code in zenml/config/store_config.py
class StoreConfiguration(BaseModel):
"""Generic store configuration.
The store configurations of concrete store implementations must inherit from
this class and validate any extra attributes that are configured in addition
to those defined in this class.
Attributes:
type: The type of store backend.
url: The URL of the store backend.
secrets_store: The configuration of the secrets store to use to store
secrets. If not set, secrets management is disabled.
"""
type: StoreType
url: str
secrets_store: Optional[SecretsStoreConfiguration] = None
@classmethod
def copy_configuration(
cls,
config: "StoreConfiguration",
config_path: str,
load_config_path: Optional[PurePath] = None,
) -> "StoreConfiguration":
"""Create a copy of the store config using a different configuration path.
This method is used to create a copy of the store configuration that can
be loaded using a different configuration path or in the context of a
new environment, such as a container image.
The configuration files accompanying the store configuration are also
copied to the new configuration path (e.g. certificates etc.).
Args:
config: The store configuration to copy.
config_path: new path where the configuration copy will be loaded
from.
load_config_path: absolute path that will be used to load the copied
configuration. This can be set to a value different from
`config_path` if the configuration copy will be loaded from
a different environment, e.g. when the configuration is copied
to a container image and loaded using a different absolute path.
This will be reflected in the paths and URLs encoded in the
copied configuration.
Returns:
A new store configuration object that reflects the new configuration
path.
"""
return config.copy()
@classmethod
def supports_url_scheme(cls, url: str) -> bool:
"""Check if a URL scheme is supported by this store.
Concrete store configuration classes should override this method to
check if a URL scheme is supported by the store.
Args:
url: The URL to check.
Returns:
True if the URL scheme is supported, False otherwise.
"""
return True
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes to be set in the base class. The concrete
# classes are responsible for validating the attributes.
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
Config
Pydantic configuration class.
Source code in zenml/config/store_config.py
class Config:
"""Pydantic configuration class."""
# Validate attributes when assigning them. We need to set this in order
# to have a mix of mutable and immutable attributes
validate_assignment = True
# Allow extra attributes to be set in the base class. The concrete
# classes are responsible for validating the attributes.
extra = "allow"
# all attributes with leading underscore are private and therefore
# are mutable and not included in serialization
underscore_attrs_are_private = True
copy_configuration(config, config_path, load_config_path=None)
classmethod
Create a copy of the store config using a different configuration path.
This method is used to create a copy of the store configuration that can be loaded using a different configuration path or in the context of a new environment, such as a container image.
The configuration files accompanying the store configuration are also copied to the new configuration path (e.g. certificates etc.).
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
config |
StoreConfiguration |
The store configuration to copy. |
required |
config_path |
str |
new path where the configuration copy will be loaded from. |
required |
load_config_path |
Optional[pathlib.PurePath] |
absolute path that will be used to load the copied
configuration. This can be set to a value different from
|
None |
Returns:
| Type | Description |
|---|---|
StoreConfiguration |
A new store configuration object that reflects the new configuration path. |
Source code in zenml/config/store_config.py
@classmethod
def copy_configuration(
cls,
config: "StoreConfiguration",
config_path: str,
load_config_path: Optional[PurePath] = None,
) -> "StoreConfiguration":
"""Create a copy of the store config using a different configuration path.
This method is used to create a copy of the store configuration that can
be loaded using a different configuration path or in the context of a
new environment, such as a container image.
The configuration files accompanying the store configuration are also
copied to the new configuration path (e.g. certificates etc.).
Args:
config: The store configuration to copy.
config_path: new path where the configuration copy will be loaded
from.
load_config_path: absolute path that will be used to load the copied
configuration. This can be set to a value different from
`config_path` if the configuration copy will be loaded from
a different environment, e.g. when the configuration is copied
to a container image and loaded using a different absolute path.
This will be reflected in the paths and URLs encoded in the
copied configuration.
Returns:
A new store configuration object that reflects the new configuration
path.
"""
return config.copy()
supports_url_scheme(url)
classmethod
Check if a URL scheme is supported by this store.
Concrete store configuration classes should override this method to check if a URL scheme is supported by the store.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
url |
str |
The URL to check. |
required |
Returns:
| Type | Description |
|---|---|
bool |
True if the URL scheme is supported, False otherwise. |
Source code in zenml/config/store_config.py
@classmethod
def supports_url_scheme(cls, url: str) -> bool:
"""Check if a URL scheme is supported by this store.
Concrete store configuration classes should override this method to
check if a URL scheme is supported by the store.
Args:
url: The URL to check.
Returns:
True if the URL scheme is supported, False otherwise.
"""
return True
strict_base_model
Strict immutable pydantic model.
StrictBaseModel (BaseModel)
pydantic-model
Immutable pydantic model which prevents extra attributes.
Source code in zenml/config/strict_base_model.py
class StrictBaseModel(BaseModel):
"""Immutable pydantic model which prevents extra attributes."""
class Config:
"""Pydantic config class."""
allow_mutation = False
extra = Extra.forbid
Config
Pydantic config class.
Source code in zenml/config/strict_base_model.py
class Config:
"""Pydantic config class."""
allow_mutation = False
extra = Extra.forbid