fiftyone.utils.beam¶

Apache Beam utilities.

Copyright 2017-2022, Voxel51, Inc.
voxel51.com

Classes:

`ExportBatch`(dataset_name[, view_stages, …])
`ImportBatch`(dataset_name[, parse_fcn, …])
`MergeBatch`(dataset_name[, parse_fcn, …])

Functions:

beam_export(sample_collection, num_shards[, …])

Exports the given sample collection in the specified number shards via Apache Beam.

beam_import(dataset, samples[, parse_fcn, …])

Imports the given samples into the dataset via Apache Beam.

beam_merge(dataset, samples[, parse_fcn, …])

Merges the given samples into the dataset via Apache Beam.

fiftyone.utils.beam.beam_import(dataset, samples, parse_fcn=None, expand_schema=True, validate=True, options=None, verbose=False)¶

Imports the given samples into the dataset via Apache Beam.

This function is a parallelized alternative to fiftyone.core.dataset.Dataset.add_samples().

Note

The insertion order of the samples is not guaranteed.

Example:

import fiftyone as fo
import fiftyone.utils.beam as foub

samples = range(10000)

def make_sample(idx):
    return fo.Sample(filepath="image%d.png" % idx, uuid=idx)

#
# Option 1: build the samples on the workers
#

dataset = fo.Dataset()

foub.beam_import(dataset, samples, parse_fcn=make_sample)
print(dataset)

#
# Option 2: build the samples in the main thread
#
# This is generally not preferred but may be necessary if your
# ``parse_fcn`` is not serializable
#

dataset = fo.Dataset()

samples = map(make_sample, samples)

foub.beam_import(dataset, samples)
print(dataset)

Parameters

dataset – a fiftyone.core.dataset.Dataset
samples – an iterable of samples. If no parse_fcn is provided, these must be fiftyone.core.sample.Sample instances. If a parse_fcn is provided, these are passed to it for parsing
parse_fcn (None) – an optional function that converts elements of samples to fiftyone.core.sample.Sample instances
expand_schema (True) – whether to dynamically add new sample fields encountered to the dataset schema. If False, an error is raised if a sample’s schema is not a subset of the dataset schema
validate (True) – whether to validate that the fields of each sample are compliant with the dataset schema before adding it
options (None) – a apache_beam.options.pipeline_options.PipelineOptions that configures how to run the pipeline. By default, the pipeline will be run via Beam’s direct runner using multiprocessing.cpu_count() threads
verbose (False) – whether to log the Beam pipeline’s messages

fiftyone.utils.beam.beam_merge(dataset, samples, parse_fcn=None, options=None, verbose=False, **kwargs)¶

Merges the given samples into the dataset via Apache Beam.

This function is a parallelized alternative to fiftyone.core.dataset.Dataset.merge_samples().

Note

This function is only useful for merging in-memory samples into a dataset. If you are merging a sample collection, simply call fiftyone.core.dataset.Dataset.merge_samples().

Example:

import fiftyone as fo
import fiftyone.utils.beam as foub
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart").clone()

samples = iter(dataset.select_fields("predictions"))

foub.beam_merge(dataset, samples, fields={"predictions": "predictions2"})

print(dataset.count("predictions.detections"))
print(dataset.count("predictions2.detections"))

Parameters

dataset – a fiftyone.core.dataset.Dataset
samples – an iterable of samples. If no parse_fcn is provided, these must be fiftyone.core.sample.Sample instances. If a parse_fcn is provided, these are passed to it for parsing
parse_fcn (None) – an optional function that converts elements of samples to fiftyone.core.sample.Sample instances
options (None) – a apache_beam.options.pipeline_options.PipelineOptions that configures how to run the pipeline. By default, the pipeline will be run via Beam’s direct runner using multiprocessing.cpu_count() threads
verbose (False) – whether to log the Beam pipeline’s messages
**kwargs – keyword arguments for fiftyone.core.dataset.Dataset.merge_samples()

fiftyone.utils.beam.beam_export(sample_collection, num_shards, options=None, verbose=False, render_kwargs=None, **kwargs)¶

Exports the given sample collection in the specified number shards via Apache Beam.

This function is a parallelized alternative to fiftyone.core.collections.SampleCollection.export() that effectively performs the following sharded export in parallel:

for idx, (first, last) in enumerate(shards, 1):
    _kwargs = render_kwargs(kwargs, idx)
    sample_collection[first:last].export(**_kwargs)

Example:

from apache_beam.options.pipeline_options import PipelineOptions

import fiftyone as fo
import fiftyone.utils.beam as foub
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

# Use multithreading instead of the default multiprocessing
options = PipelineOptions(
    runner="direct",
    direct_num_workers=10,
    direct_running_mode="multi_threading",
)

foub.beam_export(
    dataset,
    num_shards=20,
    options=options,
    dataset_type=fo.types.TFObjectDetectionDataset,
    label_field="ground_truth",
    tf_records_path="/tmp/beam/tf.records-%05d-of-00020",
)

Parameters

sample_collection – a fiftyone.core.collections.SampleCollection
num_shards – the number of shards to write
options (None) – a apache_beam.options.pipeline_options.PipelineOptions that configures how to run the pipeline. By default, the pipeline will be run via Beam’s direct runner using min(num_shards, multiprocessing.cpu_count()) processes
verbose (False) – whether to log the Beam pipeline’s messages
render_kwargs (None) – a function that renders kwargs for the current shard. The function should have signature def render_kwargs(kwargs, idx) -> kwargs, where idx in [1, num_shards] is the shard index. By default, any string-valued arguments that contain format patterns like %05d will be rendered via value % idx
**kwargs – keyword arguments for fiftyone.core.collections.SampleCollection.export()

class fiftyone.utils.beam.ImportBatch(dataset_name, parse_fcn=None, expand_schema=True, validate=True)¶

Bases: apache_beam.transforms.core.DoFn

Classes:

`BundleFinalizerParam`	alias of `apache_beam.transforms.core._BundleFinalizerParam`
`RestrictionParam`	alias of `apache_beam.transforms.core._RestrictionDoFnParam`
`StateParam`	alias of `apache_beam.transforms.core._StateDoFnParam`
`TimerParam`	alias of `apache_beam.transforms.core._TimerDoFnParam`
`WatermarkEstimatorParam`	alias of `apache_beam.transforms.core._WatermarkEstimatorParam`

Attributes:

`DoFnProcessParams`
`DynamicTimerTagParam`
`ElementParam`
`KeyParam`
`PaneInfoParam`
`SideInputParam`
`TimestampParam`
`WindowParam`

Methods:

`default_label`()
`default_type_hints`()
`display_data`()	Returns the display data associated to a pipeline component.
`finish_bundle`()	Called after a bundle of elements is processed on a worker.
`from_callable`(fn)
`from_runner_api`(fn_proto, context)	Converts from an FunctionSpec to a Fn object.
`get_function_arguments`(func)
`get_input_batch_type`(input_element_type)	Determine the batch type expected as input to process_batch.
`get_output_batch_type`(input_element_type)	Determine the batch type produced by this DoFn’s `process_batch` implementation and/or its `process` implementation with `@yields_batch`.
`get_type_hints`()	Gets and/or initializes type hints for this object.
`infer_output_type`(input_type)
`process`(sample)	Method to use for processing elements.
`process_batch`(batch, args, *kwargs)
`register_pickle_urn`(pickle_urn)	Registers and implements the given urn via pickling.
`register_urn`(urn, parameter_type[, fn])	Registers a urn with a constructor.
`setup`()	Called to prepare an instance for processing bundles of elements.
`start_bundle`()	Called before a bundle of elements is processed on a worker.
`teardown`()	Called to use to clean up this instance before it is discarded.
`to_runner_api`(context)	Returns an FunctionSpec encoding this Fn.
`to_runner_api_parameter`(context)
`unbounded_per_element`()	A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.
`with_input_types`(arg_hints, *kwarg_hints)
`with_output_types`(arg_hints, *kwarg_hints)
`yields_batches`(fn)	A decorator to apply to `process` indicating it yields batches.
`yields_elements`(fn)	A decorator to apply to `process_batch` indicating it yields elements.

setup()¶

Called to prepare an instance for processing bundles of elements.

This is a good place to initialize transient in-memory resources, such as network connections. The resources can then be disposed in DoFn.teardown.

start_bundle()¶

Called before a bundle of elements is processed on a worker.

Elements to be processed are split into bundles and distributed to workers. Before a worker calls process() on the first element of its bundle, it calls this method.

process(sample)¶

Method to use for processing elements.

This is invoked by DoFnRunner for each element of a input PCollection.

The following parameters can be used as default values on process arguments to indicate that a DoFn accepts the corresponding parameters. For example, a DoFn might accept the element and its timestamp with the following signature:

def process(element=DoFn.ElementParam, timestamp=DoFn.TimestampParam):
  ...

The full set of parameters is:

DoFn.ElementParam: element to be processed, should not be mutated.
DoFn.SideInputParam: a side input that may be used when processing.
DoFn.TimestampParam: timestamp of the input element.
DoFn.WindowParam: Window the input element belongs to.
DoFn.TimerParam: a userstate.RuntimeTimer object defined by the spec of the parameter.
DoFn.StateParam: a userstate.RuntimeState object defined by the spec of the parameter.
DoFn.KeyParam: key associated with the element.
DoFn.RestrictionParam: an iobase.RestrictionTracker will be provided here to allow treatment as a Splittable DoFn. The restriction tracker will be derived from the restriction provider in the parameter.
DoFn.WatermarkEstimatorParam: a function that can be used to track output watermark of Splittable DoFn implementations.

Parameters

element – The element to be processed
*args – side inputs
**kwargs – other keyword arguments.

Returns

An Iterable of output elements or None.

finish_bundle()¶: Called after a bundle of elements is processed on a worker.

BundleFinalizerParam¶

alias of apache_beam.transforms.core._BundleFinalizerParam Methods:

`finalize_bundle`()
`has_callbacks`()
`register`(callback)
`reset`()

DoFnProcessParams = [ElementParam, SideInputParam, TimestampParam, WindowParam, <class 'apache_beam.transforms.core._WatermarkEstimatorParam'>, PaneInfoParam, <class 'apache_beam.transforms.core._BundleFinalizerParam'>, KeyParam, <class 'apache_beam.transforms.core._StateDoFnParam'>, <class 'apache_beam.transforms.core._TimerDoFnParam'>]¶

DynamicTimerTagParam = DynamicTimerTagParam¶

ElementParam = ElementParam¶

KeyParam = KeyParam¶

PaneInfoParam = PaneInfoParam¶

RestrictionParam¶: alias of apache_beam.transforms.core._RestrictionDoFnParam

SideInputParam = SideInputParam¶

StateParam¶: alias of apache_beam.transforms.core._StateDoFnParam

TimerParam¶: alias of apache_beam.transforms.core._TimerDoFnParam

TimestampParam = TimestampParam¶

WatermarkEstimatorParam¶: alias of apache_beam.transforms.core._WatermarkEstimatorParam

WindowParam = WindowParam¶

default_label()¶

default_type_hints()¶

display_data() → dict¶

Returns the display data associated to a pipeline component.

It should be reimplemented in pipeline components that wish to have static display data.

Returns

A dictionary containing key:value pairs. The value might be an integer, float or string value; a DisplayDataItem for values that have more data (e.g. short value, label, url); or a HasDisplayData instance that has more display data that should be picked up. For example:

{
  'key1': 'string_value',
  'key2': 1234,
  'key3': 3.14159265,
  'key4': DisplayDataItem('apache.org', url='http://apache.org'),
  'key5': subComponent
}

Return type

Dict[str, Any]

static from_callable(fn)¶

classmethod from_runner_api(fn_proto: Type[RunnerApiFnT], context: beam_runner_api_pb2.FunctionSpec) → RunnerApiFnT¶

Converts from an FunctionSpec to a Fn object.

Prefer registering a urn with its parameter type and constructor.

get_function_arguments(func)¶

get_input_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type expected as input to process_batch.

The default implementation of get_input_batch_type simply observes the input typehint for the first parameter of process_batch. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn cannot accept batches, else a Beam typehint or a native Python typehint.

get_output_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type produced by this DoFn’s process_batch implementation and/or its process implementation with @yields_batch.

The default implementation of this method observes the return type annotations on process_batch and/or process. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn will never yield batches, else a Beam typehint or a native Python typehint.

get_type_hints()¶

Gets and/or initializes type hints for this object.

If type hints have not been set, attempts to initialize type hints in this order: - Using self.default_type_hints(). - Using self.__class__ type hints.

infer_output_type(input_type)¶

process_batch(batch, *args, **kwargs)¶

classmethod register_pickle_urn(pickle_urn)¶: Registers and implements the given urn via pickling.

classmethod register_urn(urn, parameter_type, fn=None)¶

Registers a urn with a constructor.

For example, if ‘beam:fn:foo’ had parameter type FooPayload, one could write RunnerApiFn.register_urn(‘bean:fn:foo’, FooPayload, foo_from_proto) where foo_from_proto took as arguments a FooPayload and a PipelineContext. This function can also be used as a decorator rather than passing the callable in as the final parameter.

A corresponding to_runner_api_parameter method would be expected that returns the tuple (‘beam:fn:foo’, FooPayload)

teardown()¶

Called to use to clean up this instance before it is discarded.

A runner will do its best to call this method on any given instance to prevent leaks of transient resources, however, there may be situations where this is impossible (e.g. process crash, hardware failure, etc.) or unnecessary (e.g. the pipeline is shutting down and the process is about to be killed anyway, so all transient resources will be released automatically by the OS). In these cases, the call may not happen. It will also not be retried, because in such situations the DoFn instance no longer exists, so there’s no instance to retry it on.

Thus, all work that depends on input elements, and all externally important side effects, must be performed in DoFn.process or DoFn.finish_bundle.

to_runner_api(context: PipelineContext) → beam_runner_api_pb2.FunctionSpec¶

Returns an FunctionSpec encoding this Fn.

Prefer overriding self.to_runner_api_parameter.

to_runner_api_parameter(context)¶

static unbounded_per_element()¶: A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.

with_input_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

with_output_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

static yields_batches(fn)¶

A decorator to apply to process indicating it yields batches.

By default process is assumed to both consume and produce individual elements at a time. This decorator indicates that process produces “batches”, which are collections of multiple logical Beam elements.

static yields_elements(fn)¶

A decorator to apply to process_batch indicating it yields elements.

By default process_batch is assumed to both consume and produce “batches”, which are collections of multiple logical Beam elements. This decorator indicates that process_batch produces individual elements at a time. process_batch is always expected to consume batches.

class fiftyone.utils.beam.MergeBatch(dataset_name, parse_fcn=None, key_field='filepath', key_fcn=None, expand_schema=True, **kwargs)¶

Bases: apache_beam.transforms.core.DoFn

Classes:

`BundleFinalizerParam`	alias of `apache_beam.transforms.core._BundleFinalizerParam`
`RestrictionParam`	alias of `apache_beam.transforms.core._RestrictionDoFnParam`
`StateParam`	alias of `apache_beam.transforms.core._StateDoFnParam`
`TimerParam`	alias of `apache_beam.transforms.core._TimerDoFnParam`
`WatermarkEstimatorParam`	alias of `apache_beam.transforms.core._WatermarkEstimatorParam`

Attributes:

`DoFnProcessParams`
`DynamicTimerTagParam`
`ElementParam`
`KeyParam`
`PaneInfoParam`
`SideInputParam`
`TimestampParam`
`WindowParam`

Methods:

`default_label`()
`default_type_hints`()
`display_data`()	Returns the display data associated to a pipeline component.
`finish_bundle`()	Called after a bundle of elements is processed on a worker.
`from_callable`(fn)
`from_runner_api`(fn_proto, context)	Converts from an FunctionSpec to a Fn object.
`get_function_arguments`(func)
`get_input_batch_type`(input_element_type)	Determine the batch type expected as input to process_batch.
`get_output_batch_type`(input_element_type)	Determine the batch type produced by this DoFn’s `process_batch` implementation and/or its `process` implementation with `@yields_batch`.
`get_type_hints`()	Gets and/or initializes type hints for this object.
`infer_output_type`(input_type)
`process`(sample)	Method to use for processing elements.
`process_batch`(batch, args, *kwargs)
`register_pickle_urn`(pickle_urn)	Registers and implements the given urn via pickling.
`register_urn`(urn, parameter_type[, fn])	Registers a urn with a constructor.
`setup`()	Called to prepare an instance for processing bundles of elements.
`start_bundle`()	Called before a bundle of elements is processed on a worker.
`teardown`()	Called to use to clean up this instance before it is discarded.
`to_runner_api`(context)	Returns an FunctionSpec encoding this Fn.
`to_runner_api_parameter`(context)
`unbounded_per_element`()	A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.
`with_input_types`(arg_hints, *kwarg_hints)
`with_output_types`(arg_hints, *kwarg_hints)
`yields_batches`(fn)	A decorator to apply to `process` indicating it yields batches.
`yields_elements`(fn)	A decorator to apply to `process_batch` indicating it yields elements.

setup()¶

Called to prepare an instance for processing bundles of elements.

This is a good place to initialize transient in-memory resources, such as network connections. The resources can then be disposed in DoFn.teardown.

start_bundle()¶

Called before a bundle of elements is processed on a worker.

Elements to be processed are split into bundles and distributed to workers. Before a worker calls process() on the first element of its bundle, it calls this method.

process(sample)¶

Method to use for processing elements.

This is invoked by DoFnRunner for each element of a input PCollection.

The following parameters can be used as default values on process arguments to indicate that a DoFn accepts the corresponding parameters. For example, a DoFn might accept the element and its timestamp with the following signature:

def process(element=DoFn.ElementParam, timestamp=DoFn.TimestampParam):
  ...

The full set of parameters is:

DoFn.ElementParam: element to be processed, should not be mutated.
DoFn.SideInputParam: a side input that may be used when processing.
DoFn.TimestampParam: timestamp of the input element.
DoFn.WindowParam: Window the input element belongs to.
DoFn.TimerParam: a userstate.RuntimeTimer object defined by the spec of the parameter.
DoFn.StateParam: a userstate.RuntimeState object defined by the spec of the parameter.
DoFn.KeyParam: key associated with the element.
DoFn.RestrictionParam: an iobase.RestrictionTracker will be provided here to allow treatment as a Splittable DoFn. The restriction tracker will be derived from the restriction provider in the parameter.
DoFn.WatermarkEstimatorParam: a function that can be used to track output watermark of Splittable DoFn implementations.

Parameters

element – The element to be processed
*args – side inputs
**kwargs – other keyword arguments.

Returns

An Iterable of output elements or None.

finish_bundle()¶: Called after a bundle of elements is processed on a worker.

BundleFinalizerParam¶

alias of apache_beam.transforms.core._BundleFinalizerParam Methods:

`finalize_bundle`()
`has_callbacks`()
`register`(callback)
`reset`()

DoFnProcessParams = [ElementParam, SideInputParam, TimestampParam, WindowParam, <class 'apache_beam.transforms.core._WatermarkEstimatorParam'>, PaneInfoParam, <class 'apache_beam.transforms.core._BundleFinalizerParam'>, KeyParam, <class 'apache_beam.transforms.core._StateDoFnParam'>, <class 'apache_beam.transforms.core._TimerDoFnParam'>]¶

DynamicTimerTagParam = DynamicTimerTagParam¶

ElementParam = ElementParam¶

KeyParam = KeyParam¶

PaneInfoParam = PaneInfoParam¶

RestrictionParam¶: alias of apache_beam.transforms.core._RestrictionDoFnParam

SideInputParam = SideInputParam¶

StateParam¶: alias of apache_beam.transforms.core._StateDoFnParam

TimerParam¶: alias of apache_beam.transforms.core._TimerDoFnParam

TimestampParam = TimestampParam¶

WatermarkEstimatorParam¶: alias of apache_beam.transforms.core._WatermarkEstimatorParam

WindowParam = WindowParam¶

default_label()¶

default_type_hints()¶

display_data() → dict¶

Returns the display data associated to a pipeline component.

It should be reimplemented in pipeline components that wish to have static display data.

Returns

A dictionary containing key:value pairs. The value might be an integer, float or string value; a DisplayDataItem for values that have more data (e.g. short value, label, url); or a HasDisplayData instance that has more display data that should be picked up. For example:

{
  'key1': 'string_value',
  'key2': 1234,
  'key3': 3.14159265,
  'key4': DisplayDataItem('apache.org', url='http://apache.org'),
  'key5': subComponent
}

Return type

Dict[str, Any]

static from_callable(fn)¶

classmethod from_runner_api(fn_proto: Type[RunnerApiFnT], context: beam_runner_api_pb2.FunctionSpec) → RunnerApiFnT¶

Converts from an FunctionSpec to a Fn object.

Prefer registering a urn with its parameter type and constructor.

get_function_arguments(func)¶

get_input_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type expected as input to process_batch.

The default implementation of get_input_batch_type simply observes the input typehint for the first parameter of process_batch. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn cannot accept batches, else a Beam typehint or a native Python typehint.

get_output_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type produced by this DoFn’s process_batch implementation and/or its process implementation with @yields_batch.

The default implementation of this method observes the return type annotations on process_batch and/or process. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn will never yield batches, else a Beam typehint or a native Python typehint.

get_type_hints()¶

Gets and/or initializes type hints for this object.

If type hints have not been set, attempts to initialize type hints in this order: - Using self.default_type_hints(). - Using self.__class__ type hints.

infer_output_type(input_type)¶

process_batch(batch, *args, **kwargs)¶

classmethod register_pickle_urn(pickle_urn)¶: Registers and implements the given urn via pickling.

classmethod register_urn(urn, parameter_type, fn=None)¶

Registers a urn with a constructor.

For example, if ‘beam:fn:foo’ had parameter type FooPayload, one could write RunnerApiFn.register_urn(‘bean:fn:foo’, FooPayload, foo_from_proto) where foo_from_proto took as arguments a FooPayload and a PipelineContext. This function can also be used as a decorator rather than passing the callable in as the final parameter.

A corresponding to_runner_api_parameter method would be expected that returns the tuple (‘beam:fn:foo’, FooPayload)

teardown()¶

Called to use to clean up this instance before it is discarded.

A runner will do its best to call this method on any given instance to prevent leaks of transient resources, however, there may be situations where this is impossible (e.g. process crash, hardware failure, etc.) or unnecessary (e.g. the pipeline is shutting down and the process is about to be killed anyway, so all transient resources will be released automatically by the OS). In these cases, the call may not happen. It will also not be retried, because in such situations the DoFn instance no longer exists, so there’s no instance to retry it on.

Thus, all work that depends on input elements, and all externally important side effects, must be performed in DoFn.process or DoFn.finish_bundle.

to_runner_api(context: PipelineContext) → beam_runner_api_pb2.FunctionSpec¶

Returns an FunctionSpec encoding this Fn.

Prefer overriding self.to_runner_api_parameter.

to_runner_api_parameter(context)¶

static unbounded_per_element()¶: A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.

with_input_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

with_output_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

static yields_batches(fn)¶

A decorator to apply to process indicating it yields batches.

By default process is assumed to both consume and produce individual elements at a time. This decorator indicates that process produces “batches”, which are collections of multiple logical Beam elements.

static yields_elements(fn)¶

A decorator to apply to process_batch indicating it yields elements.

By default process_batch is assumed to both consume and produce “batches”, which are collections of multiple logical Beam elements. This decorator indicates that process_batch produces individual elements at a time. process_batch is always expected to consume batches.

class fiftyone.utils.beam.ExportBatch(dataset_name, view_stages=None, render_kwargs=None)¶

Bases: apache_beam.transforms.core.DoFn

Classes:

`BundleFinalizerParam`	alias of `apache_beam.transforms.core._BundleFinalizerParam`
`RestrictionParam`	alias of `apache_beam.transforms.core._RestrictionDoFnParam`
`StateParam`	alias of `apache_beam.transforms.core._StateDoFnParam`
`TimerParam`	alias of `apache_beam.transforms.core._TimerDoFnParam`
`WatermarkEstimatorParam`	alias of `apache_beam.transforms.core._WatermarkEstimatorParam`

Attributes:

`DoFnProcessParams`
`DynamicTimerTagParam`
`ElementParam`
`KeyParam`
`PaneInfoParam`
`SideInputParam`
`TimestampParam`
`WindowParam`

Methods:

`default_label`()
`default_render_kwargs`(kwargs, idx)
`default_type_hints`()
`display_data`()	Returns the display data associated to a pipeline component.
`finish_bundle`()	Called after a bundle of elements is processed on a worker.
`from_callable`(fn)
`from_runner_api`(fn_proto, context)	Converts from an FunctionSpec to a Fn object.
`get_function_arguments`(func)
`get_input_batch_type`(input_element_type)	Determine the batch type expected as input to process_batch.
`get_output_batch_type`(input_element_type)	Determine the batch type produced by this DoFn’s `process_batch` implementation and/or its `process` implementation with `@yields_batch`.
`get_type_hints`()	Gets and/or initializes type hints for this object.
`infer_output_type`(input_type)
`process`(element, **kwargs)	Method to use for processing elements.
`process_batch`(batch, args, *kwargs)
`register_pickle_urn`(pickle_urn)	Registers and implements the given urn via pickling.
`register_urn`(urn, parameter_type[, fn])	Registers a urn with a constructor.
`setup`()	Called to prepare an instance for processing bundles of elements.
`start_bundle`()	Called before a bundle of elements is processed on a worker.
`teardown`()	Called to use to clean up this instance before it is discarded.
`to_runner_api`(context)	Returns an FunctionSpec encoding this Fn.
`to_runner_api_parameter`(context)
`unbounded_per_element`()	A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.
`with_input_types`(arg_hints, *kwarg_hints)
`with_output_types`(arg_hints, *kwarg_hints)
`yields_batches`(fn)	A decorator to apply to `process` indicating it yields batches.
`yields_elements`(fn)	A decorator to apply to `process_batch` indicating it yields elements.

static default_render_kwargs(kwargs, idx)¶

setup()¶

Called to prepare an instance for processing bundles of elements.

This is a good place to initialize transient in-memory resources, such as network connections. The resources can then be disposed in DoFn.teardown.

process(element, **kwargs)¶

Method to use for processing elements.

This is invoked by DoFnRunner for each element of a input PCollection.

The following parameters can be used as default values on process arguments to indicate that a DoFn accepts the corresponding parameters. For example, a DoFn might accept the element and its timestamp with the following signature:

def process(element=DoFn.ElementParam, timestamp=DoFn.TimestampParam):
  ...

The full set of parameters is:

DoFn.ElementParam: element to be processed, should not be mutated.
DoFn.SideInputParam: a side input that may be used when processing.
DoFn.TimestampParam: timestamp of the input element.
DoFn.WindowParam: Window the input element belongs to.
DoFn.TimerParam: a userstate.RuntimeTimer object defined by the spec of the parameter.
DoFn.StateParam: a userstate.RuntimeState object defined by the spec of the parameter.
DoFn.KeyParam: key associated with the element.
DoFn.RestrictionParam: an iobase.RestrictionTracker will be provided here to allow treatment as a Splittable DoFn. The restriction tracker will be derived from the restriction provider in the parameter.
DoFn.WatermarkEstimatorParam: a function that can be used to track output watermark of Splittable DoFn implementations.

Parameters

element – The element to be processed
*args – side inputs
**kwargs – other keyword arguments.

Returns

An Iterable of output elements or None.

BundleFinalizerParam¶

alias of apache_beam.transforms.core._BundleFinalizerParam Methods:

`finalize_bundle`()
`has_callbacks`()
`register`(callback)
`reset`()

DoFnProcessParams = [ElementParam, SideInputParam, TimestampParam, WindowParam, <class 'apache_beam.transforms.core._WatermarkEstimatorParam'>, PaneInfoParam, <class 'apache_beam.transforms.core._BundleFinalizerParam'>, KeyParam, <class 'apache_beam.transforms.core._StateDoFnParam'>, <class 'apache_beam.transforms.core._TimerDoFnParam'>]¶

DynamicTimerTagParam = DynamicTimerTagParam¶

ElementParam = ElementParam¶

KeyParam = KeyParam¶

PaneInfoParam = PaneInfoParam¶

RestrictionParam¶: alias of apache_beam.transforms.core._RestrictionDoFnParam

SideInputParam = SideInputParam¶

StateParam¶: alias of apache_beam.transforms.core._StateDoFnParam

TimerParam¶: alias of apache_beam.transforms.core._TimerDoFnParam

TimestampParam = TimestampParam¶

WatermarkEstimatorParam¶: alias of apache_beam.transforms.core._WatermarkEstimatorParam

WindowParam = WindowParam¶

default_label()¶

default_type_hints()¶

display_data() → dict¶

Returns the display data associated to a pipeline component.

It should be reimplemented in pipeline components that wish to have static display data.

Returns

A dictionary containing key:value pairs. The value might be an integer, float or string value; a DisplayDataItem for values that have more data (e.g. short value, label, url); or a HasDisplayData instance that has more display data that should be picked up. For example:

{
  'key1': 'string_value',
  'key2': 1234,
  'key3': 3.14159265,
  'key4': DisplayDataItem('apache.org', url='http://apache.org'),
  'key5': subComponent
}

Return type

Dict[str, Any]

finish_bundle()¶: Called after a bundle of elements is processed on a worker.

static from_callable(fn)¶

classmethod from_runner_api(fn_proto: Type[RunnerApiFnT], context: beam_runner_api_pb2.FunctionSpec) → RunnerApiFnT¶

Converts from an FunctionSpec to a Fn object.

Prefer registering a urn with its parameter type and constructor.

get_function_arguments(func)¶

get_input_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type expected as input to process_batch.

The default implementation of get_input_batch_type simply observes the input typehint for the first parameter of process_batch. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn cannot accept batches, else a Beam typehint or a native Python typehint.

get_output_batch_type(input_element_type) → Union[apache_beam.typehints.typehints.TypeConstraint, type, None]¶

Determine the batch type produced by this DoFn’s process_batch implementation and/or its process implementation with @yields_batch.

The default implementation of this method observes the return type annotations on process_batch and/or process. A Batched DoFn may override this method if a dynamic approach is required.

Parameters: input_element_type – The element type of the input PCollection this DoFn is being applied to.
Returns: None if this DoFn will never yield batches, else a Beam typehint or a native Python typehint.

get_type_hints()¶

Gets and/or initializes type hints for this object.

If type hints have not been set, attempts to initialize type hints in this order: - Using self.default_type_hints(). - Using self.__class__ type hints.

infer_output_type(input_type)¶

process_batch(batch, *args, **kwargs)¶

classmethod register_pickle_urn(pickle_urn)¶: Registers and implements the given urn via pickling.

classmethod register_urn(urn, parameter_type, fn=None)¶

Registers a urn with a constructor.

For example, if ‘beam:fn:foo’ had parameter type FooPayload, one could write RunnerApiFn.register_urn(‘bean:fn:foo’, FooPayload, foo_from_proto) where foo_from_proto took as arguments a FooPayload and a PipelineContext. This function can also be used as a decorator rather than passing the callable in as the final parameter.

A corresponding to_runner_api_parameter method would be expected that returns the tuple (‘beam:fn:foo’, FooPayload)

start_bundle()¶

Called before a bundle of elements is processed on a worker.

Elements to be processed are split into bundles and distributed to workers. Before a worker calls process() on the first element of its bundle, it calls this method.

teardown()¶

Called to use to clean up this instance before it is discarded.

A runner will do its best to call this method on any given instance to prevent leaks of transient resources, however, there may be situations where this is impossible (e.g. process crash, hardware failure, etc.) or unnecessary (e.g. the pipeline is shutting down and the process is about to be killed anyway, so all transient resources will be released automatically by the OS). In these cases, the call may not happen. It will also not be retried, because in such situations the DoFn instance no longer exists, so there’s no instance to retry it on.

Thus, all work that depends on input elements, and all externally important side effects, must be performed in DoFn.process or DoFn.finish_bundle.

to_runner_api(context: PipelineContext) → beam_runner_api_pb2.FunctionSpec¶

Returns an FunctionSpec encoding this Fn.

Prefer overriding self.to_runner_api_parameter.

to_runner_api_parameter(context)¶

static unbounded_per_element()¶: A decorator on process fn specifying that the fn performs an unbounded amount of work per input element.

with_input_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

with_output_types(*arg_hints: WithTypeHintsT, **kwarg_hints: Any) → WithTypeHintsT¶

static yields_batches(fn)¶

A decorator to apply to process indicating it yields batches.

By default process is assumed to both consume and produce individual elements at a time. This decorator indicates that process produces “batches”, which are collections of multiple logical Beam elements.

static yields_elements(fn)¶

A decorator to apply to process_batch indicating it yields elements.

By default process_batch is assumed to both consume and produce “batches”, which are collections of multiple logical Beam elements. This decorator indicates that process_batch produces individual elements at a time. process_batch is always expected to consume batches.