def testInferFeatureSchema(self):
columns = {
'a': api._InputColumn(tf.placeholder(tf.float32, (None, )), None),
'b': api._InputColumn(tf.placeholder(tf.string, (1, 2, 3)), None),
'c': api._InputColumn(tf.placeholder(tf.int64, None), None)
}
schema = impl_helper.infer_feature_schema(columns)
expected_schema = sch.Schema(
column_schemas={
'a':
sch.ColumnSchema(
sch.LogicalColumnSchema(sch.dtype_to_domain(tf.float32),
sch.LogicalShape([])),
sch.FixedColumnRepresentation()),
'b':
sch.ColumnSchema(
sch.LogicalColumnSchema(
sch.dtype_to_domain(tf.string),
sch.LogicalShape([sch.Axis(2),
sch.Axis(3)])),
sch.FixedColumnRepresentation()),
'c':
sch.ColumnSchema(
sch.LogicalColumnSchema(sch.dtype_to_domain(tf.int64),
sch.LogicalShape(None)),
sch.FixedColumnRepresentation())
})
self.assertEqual(schema, expected_schema)
def testInferFeatureSchema(self):
d = tf.placeholder(tf.int64, None)
tensors = {
'a': tf.placeholder(tf.float32, (None, )),
'b': tf.placeholder(tf.string, (1, 2, 3)),
'c': tf.placeholder(tf.int64, None),
'd': d
}
d_column_schema = sch.ColumnSchema(tf.int64, [1, 2, 3],
sch.FixedColumnRepresentation())
api.set_column_schema(d, d_column_schema)
schema = impl_helper.infer_feature_schema(tensors)
expected_schema = sch.Schema(
column_schemas={
'a':
sch.ColumnSchema(tf.float32, [],
sch.FixedColumnRepresentation()),
'b':
sch.ColumnSchema(tf.string, [2, 3],
sch.FixedColumnRepresentation()),
'c':
sch.ColumnSchema(tf.int64, None,
sch.FixedColumnRepresentation()),
'd':
sch.ColumnSchema(tf.int64, [1, 2, 3],
sch.FixedColumnRepresentation())
})
self.assertEqual(schema, expected_schema)
def testInferFeatureSchemaBadRank(self):
tensors = {
'a': tf.placeholder(tf.float32, ()),
}
with self.assertRaises(ValueError):
_ = impl_helper.infer_feature_schema(tf.get_default_graph(),
tensors)
def testInferFeatureSchema(self):
columns = {
'a': api._InputColumn(tf.placeholder(tf.float32, (None,)), None),
'b': api._InputColumn(tf.placeholder(tf.string, (1, 2, 3)), None),
'c': api._InputColumn(tf.placeholder(tf.int64, None), None)
}
schema = impl_helper.infer_feature_schema(columns)
expected_schema = sch.Schema(column_schemas={
'a': sch.ColumnSchema(tf.float32, [],
sch.FixedColumnRepresentation()),
'b': sch.ColumnSchema(tf.string, [2, 3],
sch.FixedColumnRepresentation()),
'c': sch.ColumnSchema(tf.int64, None,
sch.FixedColumnRepresentation())
})
self.assertEqual(schema, expected_schema)
def make_transform_graph(output_dir, schema, features):
"""Writes a tft transform fn, and metadata files.
Args:
output_dir: output folder
schema: schema list
features: features dict
"""
tft_input_schema = make_tft_input_schema(schema, os.path.join(output_dir,
STATS_FILE))
tft_input_metadata = dataset_metadata.DatasetMetadata(schema=tft_input_schema)
preprocessing_fn = make_preprocessing_fn(output_dir, features)
# copy from /tft/beam/impl
inputs, outputs = impl_helper.run_preprocessing_fn(
preprocessing_fn=preprocessing_fn,
schema=tft_input_schema)
output_metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(outputs))
transform_fn_dir = os.path.join(output_dir, TRANSFORM_FN_DIR)
# This writes the SavedModel
impl_helper.make_transform_fn_def(
schema=tft_input_schema,
inputs=inputs,
outputs=outputs,
saved_model_dir=transform_fn_dir)
metadata_io.write_metadata(
metadata=output_metadata,
path=os.path.join(output_dir, TRANSFORMED_METADATA_DIR))
metadata_io.write_metadata(
metadata=tft_input_metadata,
path=os.path.join(output_dir, RAW_METADATA_DIR))
def expand(self, dataset):
"""Analyze the dataset.
Args:
dataset: A dataset.
Returns:
A TransformFn containing the deferred transform function.
Raises:
ValueError: If preprocessing_fn has no outputs.
"""
input_values, input_metadata = dataset
input_schema = input_metadata.schema
base_temp_dir = Context.create_base_temp_dir()
graph = tf.Graph()
with graph.as_default():
with tf.name_scope('inputs'):
inputs = input_schema.as_batched_placeholders()
# In order to avoid a bug where import_graph_def fails when the input_map
# and return_elements of an imported graph are the same (b/34288791), we
# avoid using the placeholder of an input column as an output of a graph.
# We do this by applying tf.identity to all inputs of the
# preprocessing_fn. Note this applies at the level of raw tensors.
outputs = self._preprocessing_fn(impl_helper.copy_tensors(inputs))
# At this point we check that the preprocessing_fn has at least one
# output. This is because if we allowed the output of preprocessing_fn to
# be empty, we wouldn't be able to determine how many instances to
# "unbatch" the output into.
if not outputs:
raise ValueError('The preprocessing function returned an empty dict')
if graph.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES):
raise ValueError(
'The preprocessing function contained trainable variables '
'{}'.format(
graph.get_collection_ref(tf.GraphKeys.TRAINABLE_VARIABLES)))
# NOTE: it's important that create_phases is called directly after
# preprocessing_fn, because we later mutate the graph's TABLE_INITIALIZERS
# collection which would break the logic in create_phases.
phases = impl_helper.create_phases()
# Iterate through levels. tensor_pcoll_mapping is a mapping from tensor
# names to singleton PCollections containing a _TensorValue. We compute
# tensor_pcoll_mapping in phases, where at each phase we compute the
# analyzers that are ready to run and update tensor_pcoll_mapping.
tensor_pcoll_mapping = {}
table_initializers = graph.get_collection_ref(
tf.GraphKeys.TABLE_INITIALIZERS)
original_table_initializers = list(table_initializers)
del table_initializers[:]
serialized_tf_config = (
common._DEFAULT_TENSORFLOW_CONFIG_BY_RUNNER.get( # pylint: disable=protected-access
input_values.pipeline.runner))
for level, phase in enumerate(phases):
# Create a SavedModel that describes the mapping from the input data
# to the inputs of the analyzers at this level. The colum names of the
# outputs are the tensor names of the analyzer inputs in the graph.
# This graph has the anaylzer outputs computed so far replaced with
# constants.
analyzer_inputs = {}
for analyzer in phase.analyzers:
for input_tensor in analyzer.inputs:
analyzer_inputs[input_tensor.name] = input_tensor
table_initializers.extend(phase.table_initializers)
unbound_saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, analyzer_inputs,
unbound_saved_model_dir)
saved_model_dir = (
tensor_pcoll_mapping
| 'CreateSavedModelForAnalyzerInputs[%d]' % level >>
_ReplaceTensorsWithConstants(unbound_saved_model_dir, base_temp_dir,
input_values.pipeline))
# Run this saved model on the input dataset to obtain the inputs to the
# analyzers.
analyzer_input_values = (
input_values
| 'BatchAnalyzerInputs[%d]' % level >> _BatchElements()
| 'ComputeAnalyzerInputs[%d]' % level >> beam.ParDo(
_RunMetaGraphDoFn(
input_schema,
serialized_tf_config,
shared_graph_state_handle=shared.Shared(),
passthrough_keys=Context.get_passthrough_keys()),
saved_model_dir=beam.pvalue.AsSingleton(saved_model_dir)))
# Compute the analyzers from their inputs. `analyzer_outputs_dict` is a
# map from tensor names to singleton PCollections of `_TensorValue`s.
analyzer_outputs_dict = (
analyzer_input_values
| 'ComputeAnalyzerOutputs[%d]' % level >> _ComputeAnalyzerOutputs(
phase.analyzers, base_temp_dir))
# Update the mapping for all analyzers.
tensor_pcoll_mapping.update(analyzer_outputs_dict)
del table_initializers[:]
table_initializers.extend(original_table_initializers)
saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, outputs, saved_model_dir)
transform_fn = (
tensor_pcoll_mapping
| 'ReplaceTensorsWithConstants' >> _ReplaceTensorsWithConstants(
saved_model_dir, base_temp_dir, input_values.pipeline))
# Infer metadata. The metadata may contain Futures that refer to the
# values of tensors in the graph. In that case, the tensors must be
# "constant" in that they don't depend on input data. The tensors can
# depend on analyzer outputs though. This allows us to set metadata that
# depends on analyzer outputs.
#
# We first extract the names of the tensors that are referenced by the
# Futures, and then compute them by calling _ComputeScalarConstants with
# the tensor-PCollection mapping representing the analyzer outputs.
metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(outputs))
deferred_metadata_tensor_names = {
future.name
for column_schema in metadata.schema.column_schemas.values()
for future in column_schema.substitute_futures({})
}
name_pcoll_dict = (
tensor_pcoll_mapping
| 'ComputeTensorValues' >>
_ComputeTensorValues(deferred_metadata_tensor_names, saved_model_dir,
input_values.pipeline))
full_metadata = beam_metadata_io.BeamDatasetMetadata(
metadata, name_pcoll_dict)
_clear_shared_state_after_barrier(input_values.pipeline, transform_fn)
return transform_fn, full_metadata
def expand(self, dataset):
"""Analyze the dataset.
Args:
dataset: A dataset.
Returns:
A TransformFn containing the deferred transform function.
"""
input_values, input_metadata = dataset
input_schema = input_metadata.schema
base_temp_dir = Context.create_base_temp_dir()
# NOTE: it's important that create_phases is called directly after
# run_preprocessing_fn, because we later mutate the graph's
# TABLE_INITIALIZERS collection which would break the logic in
# create_phases.
graph, inputs, outputs = impl_helper.run_preprocessing_fn(
self._preprocessing_fn, input_schema)
phases = impl_helper.create_phases(graph)
# Iterate through levels. tensor_pcoll_mapping is a mapping from tensor
# names to singleton PCollections containing a _TensorValue. We compute
# tensor_pcoll_mapping in phases, where at each phase we compute the
# analyzers that are ready to run and update tensor_pcoll_mapping.
tensor_pcoll_mapping = {}
table_initializers = graph.get_collection_ref(
tf.GraphKeys.TABLE_INITIALIZERS)
original_table_initializers = list(table_initializers)
del table_initializers[:]
serialized_tf_config = (
analyzer_impls._DEFAULT_TENSORFLOW_CONFIG_BY_RUNNER.get( # pylint: disable=protected-access
input_values.pipeline.runner))
for level, phase in enumerate(phases):
# Create a SavedModel that describes the mapping from the input data
# to the inputs of the analyzers at this level. The colum names of the
# outputs are the tensor names of the analyzer inputs in the graph. This
# graph has the anaylzer outputs computed so far replaced with constants.
analyzer_inputs = {}
for analyzer in phase.analyzers:
for input_tensor in analyzer.inputs:
analyzer_inputs[input_tensor.name] = input_tensor
table_initializers.extend(phase.table_initializers)
unbound_saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(
graph, inputs, analyzer_inputs, unbound_saved_model_dir)
saved_model_dir = (
tensor_pcoll_mapping
| 'CreateSavedModelForAnaylzerInputs[%d]' % level
>> _ReplaceTensorsWithConstants(
unbound_saved_model_dir, base_temp_dir, input_values.pipeline))
# Run this saved model on the input dataset to obtain the inputs to the
# analyzers.
analyzer_input_values = (
input_values
| 'ComputeAnalyzerInputs[%d]' % level >> beam.ParDo(
_RunMetaGraphDoFn(
input_schema,
serialized_tf_config,
shared_graph_state_handle=shared.Shared()),
saved_model_dir=beam.pvalue.AsSingleton(saved_model_dir)))
# Compute the analyzers from their inputs. `analyzer_outputs_dict` is a
# map from tensor names to singleton PCollections of `_TensorValue`s.
analyzer_outputs_dict = (
analyzer_input_values
| 'ComputeAnalyzerOutputs[%d]' % level
>> _ComputeAnalyzerOutputs(phase.analyzers, base_temp_dir))
# Update the mapping for all analyzers.
tensor_pcoll_mapping.update(analyzer_outputs_dict)
del table_initializers[:]
table_initializers.extend(original_table_initializers)
saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, outputs, saved_model_dir)
transform_fn = (
tensor_pcoll_mapping
| 'ReplaceTensorsWithConstants'
>> _ReplaceTensorsWithConstants(
saved_model_dir, base_temp_dir, input_values.pipeline))
# Infer metadata. The metadata may contain Futures that refer to the values
# of tensors in the graph. In that case, the tensors must be "constant" in
# that they don't depend on input data. The tensors can depend on analyzer
# outputs though. This allows us to set metadata that depends on analyzer
# outputs.
#
# We first extract the names of the tensors that are referenced by the
# Futures, and then compute them by calling _ComputeScalarConstants with the
# tensor-PCollection mapping representing the analyzer outputs.
metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(graph, outputs))
deferred_metadata_tensor_names = [
future.name
for column_schema in tft_api.get_column_schemas(graph).values()
for future in column_schema.substitute_futures({})]
name_pcoll_dict = (
tensor_pcoll_mapping
| 'ComputeTensorValues' >>
_ComputeTensorValues(
deferred_metadata_tensor_names, saved_model_dir,
input_values.pipeline))
full_metadata = beam_metadata_io.BeamDatasetMetadata(
metadata, name_pcoll_dict)
_clear_shared_state_after_barrier(input_values.pipeline, transform_fn)
return transform_fn, full_metadata
def testInferFeatureSchemaBadRank(self):
columns = {
'a': api._InputColumn(tf.placeholder(tf.float32, ()), None),
}
with self.assertRaises(ValueError):
_ = impl_helper.infer_feature_schema(columns)
def expand(self, dataset):
"""Analyze the dataset.
Args:
dataset: A dataset.
Returns:
A TransformFn containing the deferred transform function.
"""
input_values, input_metadata = dataset
input_schema = input_metadata.schema
input_batches = input_values | 'BatchInstances' >> beam.ParDo(
_BatchDoFn())
class _CreateTransformFn(beam.PTransform):
"""Create a TransformFnDef, binding statistics in a deferred manner.
This function constructs a tensorflow graph eagerly and then (in a
deferred manner) fills in analyzer outputs with their actual computed
values. We construct the tensorflow graph up front because that implies
serializing MetaGraphDef protos rather than pickling the user-defined TITO
functions. The graph contains placeholders for `_AnalyzerOutput`s which
are then replaced with their actual values (as constant tensors) in a
deferred manner.
Args:
input_columns: A map from column names to `Column`s.
output_columns: A map from column names to `Column`s.
temp_dir: Temp dir to store `SavedModel`s.
"""
def __init__(self, input_columns, output_columns, temp_dir):
# Generally the pipeline is inferred from its inputs, however we need
# to know the pipeline for beam.Create.
self.pipeline = input_values.pipeline
self._input_columns = input_columns
self._output_columns = output_columns
self._temp_dir = temp_dir
def expand(self, analyzer_outputs_to_pcoll):
"""Converts a dict of statistics to a transform function.
Args:
analyzer_outputs_to_pcoll: A dictionary mapping `_AnalyzerOutput`s
to the values of these statistics as a PCollection.
Returns:
A single-element PCollection containing the directory name with the
SavedModel.
"""
# Create a transform_fn with unbound values.
unbound_transform_fn_dir = os.path.join(
self._temp_dir, 'unbound_transform_fn')
input_columns_to_statistics = impl_helper.make_transform_fn_def(
input_schema, self._input_columns, self._output_columns,
unbound_transform_fn_dir)
transform_fn = (self.pipeline | 'CreateTransformFn' >>
beam.Create([unbound_transform_fn_dir]))
if not analyzer_outputs_to_pcoll:
return transform_fn
# Convert the statistics dict into a DictPCollectionView so it can be
# passed as a side input to the beam Map below.
tagged_statistics = []
for tag, statistic in input_columns_to_statistics.items():
pcoll = analyzer_outputs_to_pcoll[statistic]
tagged_statistics.append(
pcoll
| 'AddTag[%s]' % tag >> beam.Map(lambda x, tag=tag:
(tag, x)))
statistics_side_input = beam.pvalue.AsDict(
tagged_statistics | 'MergeStatistics' >> beam.Flatten())
# Run a mapper that inserts statistic values into the graph.
return (transform_fn
| 'ReplaceTensorsWithConstantValues' >> beam.Map(
impl_helper.replace_tensors_with_constant_values,
bound_saved_model_dir=os.path.join(
self._temp_dir, 'transform_fn'),
input_value_mapping=statistics_side_input))
inputs, outputs = impl_helper.run_preprocessing_fn(
self._preprocessing_fn, input_schema)
# Get a list of lists, containing analyzers (i.e. _AnalyzerOutput objects)
# by level in the DAG of Columns/Statistics. Analyzers at level n are ready
# to run once all analyzers at level n - 1 are complete.
analyzers_by_level = self._analyzers_by_level(outputs)
# Iterate through levels, keeping track of analyzer outputs (i.e.
# statistics) via a mapping of `_AnalyzerOutput` -> single element
# PCollection.
analyzer_outputs_to_pcoll = {}
for level, analyzer_outputs in enumerate(analyzers_by_level):
# Create a TransformFnDef representing the graph needed to generate
# all the inputs required by the analyzer_outputs at this level. We
# assign arbitrary names to the outputs of this TransformFnDef.
analyzer_input_columns = {}
for idx, analyzer_output in enumerate(analyzer_outputs):
if len(analyzer_output.inputs) != 1:
raise NotImplementedError(
'Analyzers must have exactly one input')
analyzer_input_key = 'analyzer_%d_input' % idx
analyzer_input_columns[
analyzer_input_key] = analyzer_output.inputs[0]
transform_fn = (
analyzer_outputs_to_pcoll
| 'CreateTransformFn_%d' % level >> _CreateTransformFn(
inputs, analyzer_input_columns,
os.path.join(self._output_dir, 'tmp', 'level_%s' % level)))
analyzer_input_schema = impl_helper.infer_feature_schema(
analyzer_input_columns)
# Run the TransformFnDef in a mapper.
analysis_inputs = (
input_batches
| 'ComputeAnalyzerInputs_%d' % level >> beam.ParDo(
_RunMetaGraphDoFn(input_schema, analyzer_input_schema),
saved_model_dir=beam.pvalue.AsSingleton(transform_fn)))
# For each analyzer output, look up its input values (by tensor name)
# and run the analyzer in these values.
for idx, analyzer_output in enumerate(analyzer_outputs):
analyzer_input_key = 'analyzer_%d_input' % idx
analyzer_outputs_to_pcoll[analyzer_output] = (
analysis_inputs
| 'Extract_%d_%d' % (level, idx) >> beam.Map(
# pylint: disable=cell-var-from-loop
# This lint warning is prone to false positives, and it's not
# clear why the warning is required here.
lambda x, key=analyzer_input_key:
[inst[key] for inst in x])
| 'Analyze_%d_%d' %
(level, idx) >> self._Analyze(analyzer_output))
output_metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(outputs))
transform_fn = (analyzer_outputs_to_pcoll
| 'CreateTransformFn' >> _CreateTransformFn(
inputs, outputs, self._output_dir))
return transform_fn, output_metadata
def expand(self, dataset):
"""Analyze the dataset.
Args:
dataset: A dataset.
Returns:
A TransformFn containing the deferred transform function.
"""
input_values, input_metadata = dataset
input_schema = input_metadata.schema
base_temp_dir = Context.create_base_temp_dir()
class _ReplaceTensorsWithConstants(beam.PTransform):
"""Bind statistics in a deferred manner.
This transform fills in analyzer outputs with their actual computed
values.
Args:
saved_model_dir: The directory containing the SavedModel.
"""
def __init__(self, saved_model_dir):
# Generally the pipeline is inferred from its inputs, however we need
# to know the pipeline for beam.Create.
self.pipeline = input_values.pipeline
self._saved_model_dir = saved_model_dir
def expand(self, tensor_pcoll_mapping):
"""Converts a dict of statistics to a transform function.
Args:
tensor_pcoll_mapping: A dictionary mapping `Tensor`s to singleton
`PCollection`s.
Returns:
A single-element PCollection containing the directory name with the
SavedModel.
"""
transform_fn = (self.pipeline | 'CreateTransformFn' >>
beam.Create([self._saved_model_dir]))
if not tensor_pcoll_mapping:
return transform_fn
# Convert tensor_value_mapping into a DictPCollectionView so it can be
# passed as a side input to the beam Map below.
tensor_value_pairs = []
for name, pcoll in six.iteritems(tensor_pcoll_mapping):
tensor_value_pairs.append(
pcoll
| 'AddName[%s]' % name >> beam.Map(lambda x, name=name:
(name, x)))
tensor_value_mapping = beam.pvalue.AsDict(
tensor_value_pairs
| 'MergeTensorValuePairs' >> beam.Flatten())
# Run a mapper that inserts statistic values into the graph. We wrap
# replace_tensors_with_constant_values in a wrapper that also creates
# a temp dir. This makes the wrapper idempotent since any retry will
# use a different temp dir.
def replace_tensors_with_constant_values(
saved_model_dir, tensor_value_mapping,
serialized_tf_config):
tf_config = _maybe_deserialize_tf_config(
serialized_tf_config)
with tf.Session(config=tf_config) as session:
temp_dir = _make_unique_temp_dir(base_temp_dir)
input_tensors, output_tensors = (
saved_transform_io.partially_apply_saved_transform(
saved_model_dir, {}, tensor_value_mapping))
saved_transform_io.write_saved_transform_from_session(
session, input_tensors, output_tensors, temp_dir)
return temp_dir
serialized_tf_config = _DEFAULT_TENSORFLOW_CONFIG_BY_RUNNER.get(
self.pipeline.runner)
return (transform_fn | 'ReplaceTensorsWithConstantValues' >>
beam.Map(replace_tensors_with_constant_values,
tensor_value_mapping=tensor_value_mapping,
serialized_tf_config=serialized_tf_config))
class _ComputeTensorPcollMappingUpdate(beam.PTransform):
"""Create a mapping from `Tensor`s to PCollections.
Creates a mapping from `Tensor`s to PCollections for the outputs of the
new analyzers. An existing mapping will be provided as the argument
to the extend() method.
Args:
phase: The Phase to run
"""
def __init__(self, saved_model_dir, analyzer_inputs_schema,
analyzers):
self._saved_model_dir = saved_model_dir
self._analyzer_inputs_schema = analyzer_inputs_schema
self._analyzers = analyzers
def expand(self, input_values_and_tensor_pcoll_mapping):
input_values, tensor_pcoll_mapping = (
input_values_and_tensor_pcoll_mapping)
# Create a transform_fn to produce inputs to new analyzers.
transform_fn = (
tensor_pcoll_mapping
| 'ReplaceTensorsWithConstants' >>
_ReplaceTensorsWithConstants(self._saved_model_dir))
# Run the transform_fn.
serialized_tf_config = _DEFAULT_TENSORFLOW_CONFIG_BY_RUNNER.get(
self.pipeline.runner)
analyzer_input_values = (
input_values | 'ComputeAnalyzerInputs' >> beam.ParDo(
_RunMetaGraphDoFn(input_schema,
self._analyzer_inputs_schema,
serialized_tf_config),
saved_model_dir=beam.pvalue.AsSingleton(transform_fn)))
# For each analyzer output, look up its input values (by tensor name)
# and run the analyzer on these values.
#
tensor_pcoll_mapping_update = {}
for idx, analyzer in enumerate(self._analyzers):
analyzer_impl = analyzer_impls._impl_for_analyzer(
analyzer.spec)
# pylint: enable=protected-access
assert len(analyzer.inputs) == 1
output_pcolls = (analyzer_input_values
| 'Extract_%d' % idx >> beam.Map(
lambda batch, key: batch[key],
key=analyzer.inputs[0].name)
| 'Analyze_%d' % idx >> analyzer_impl)
assert len(analyzer.outputs) == len(output_pcolls)
for tensor, pcoll in zip(analyzer.outputs, output_pcolls):
tensor_pcoll_mapping_update[tensor.name] = pcoll
return tensor_pcoll_mapping_update
# NOTE: it's important that create_phases is called directly after
# run_preprocessing_fn, because we later mutate the graph's
# TABLE_INITIALIZERS collection which would break the logic in
# create_phases.
graph, inputs, outputs = impl_helper.run_preprocessing_fn(
self._preprocessing_fn, input_schema)
phases = impl_helper.create_phases(graph)
# Iterate through levels, generating PCollections for columns that are the
# outputs of `Operations` that are not `MapOperation`s.
tensor_pcoll_mapping = {}
table_initializers = graph.get_collection_ref(
tf.GraphKeys.TABLE_INITIALIZERS)
original_table_initializers = list(table_initializers)
del table_initializers[:]
for level, phase in enumerate(phases):
analyzer_inputs = {}
for analyzer in phase.analyzers:
for input_tensor in analyzer.inputs:
analyzer_inputs[input_tensor.name] = input_tensor
analyzer_inputs_schema = impl_helper.infer_feature_schema(
analyzer_inputs)
table_initializers.extend(phase.table_initializers)
saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, analyzer_inputs,
saved_model_dir)
tensor_pcoll_mapping_update = (
(input_values, tensor_pcoll_mapping)
| 'ComputeTensorPcollMappingUpdate_%d' % level >>
_ComputeTensorPcollMappingUpdate(
saved_model_dir, analyzer_inputs_schema, phase.analyzers))
tensor_pcoll_mapping.update(tensor_pcoll_mapping_update)
output_metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(outputs))
del table_initializers[:]
table_initializers.extend(original_table_initializers)
saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, outputs, saved_model_dir)
transform_fn = (tensor_pcoll_mapping
| 'ReplaceTensorsWithConstants' >>
_ReplaceTensorsWithConstants(saved_model_dir))
return transform_fn, output_metadata
def expand(self, dataset):
"""Analyze the dataset.
Args:
dataset: A dataset.
Returns:
A TransformFn containing the deferred transform function.
Raises:
ValueError: If preprocessing_fn has no outputs.
"""
input_values, input_metadata = dataset
input_schema = input_metadata.schema
base_temp_dir = Context.create_base_temp_dir()
graph = tf.Graph()
with graph.as_default():
with tf.name_scope('inputs'):
inputs = input_schema.as_batched_placeholders()
# In order to avoid a bug where import_graph_def fails when the input_map
# and return_elements of an imported graph are the same (b/34288791), we
# avoid using the placeholder of an input column as an output of a graph.
# We do this by applying tf.identity to all inputs of the
# preprocessing_fn. Note this applies at the level of raw tensors.
outputs = self._preprocessing_fn(impl_helper.copy_tensors(inputs))
# At this point we check that the preprocessing_fn has at least one
# output. This is because if we allowed the output of preprocessing_fn to
# be empty, we wouldn't be able to determine how many instances to
# "unbatch" the output into.
if not outputs:
raise ValueError(
'The preprocessing function returned an empty dict')
if graph.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES):
raise ValueError(
'The preprocessing function contained trainable variables '
'{}'.format(
graph.get_collection_ref(
tf.GraphKeys.TRAINABLE_VARIABLES)))
# NOTE: it's important that create_phases is called directly after
# preprocessing_fn, because we later mutate the graph's TABLE_INITIALIZERS
# collection which would break the logic in create_phases.
phases = impl_helper.create_phases()
# Iterate through levels. tensor_pcoll_mapping is a mapping from tensor
# names to singleton PCollections containing a _TensorValue. We compute
# tensor_pcoll_mapping in phases, where at each phase we compute the
# analyzers that are ready to run and update tensor_pcoll_mapping.
tensor_pcoll_mapping = {}
table_initializers = graph.get_collection_ref(
tf.GraphKeys.TABLE_INITIALIZERS)
original_table_initializers = list(table_initializers)
del table_initializers[:]
serialized_tf_config = (
common._DEFAULT_TENSORFLOW_CONFIG_BY_RUNNER.get( # pylint: disable=protected-access
input_values.pipeline.runner))
for level, phase in enumerate(phases):
# Create a SavedModel that describes the mapping from the input data
# to the inputs of the analyzers at this level. The colum names of the
# outputs are the tensor names of the analyzer inputs in the graph.
# This graph has the anaylzer outputs computed so far replaced with
# constants.
analyzer_inputs = {}
for analyzer in phase.analyzer_infos:
for input_tensor_name in analyzer.input_tensor_names:
analyzer_inputs[
input_tensor_name] = graph.get_tensor_by_name(
input_tensor_name)
table_initializers.extend(phase.table_initializers)
unbound_saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, analyzer_inputs,
unbound_saved_model_dir)
tensor_pcoll_mapping_update = (
(input_values, tensor_pcoll_mapping)
| 'RunPhase[{}]'.format(level) >> _RunPhase(
phase.analyzer_infos, unbound_saved_model_dir,
base_temp_dir, input_schema, serialized_tf_config))
# Update the mapping for all analyzers.
tensor_pcoll_mapping.update(tensor_pcoll_mapping_update)
del table_initializers[:]
table_initializers.extend(original_table_initializers)
saved_model_dir = _make_unique_temp_dir(base_temp_dir)
_write_saved_transform(graph, inputs, outputs, saved_model_dir)
transform_fn = (
tensor_pcoll_mapping
|
'ReplaceTensorsWithConstants' >> _ReplaceTensorsWithConstants(
saved_model_dir, base_temp_dir, input_values.pipeline))
# Infer metadata. We take the inferred metadata and apply overrides that
# refer to values of tensors in the graph. The override tensors must
# be "constant" in that they don't depend on input data. The tensors can
# depend on analyzer outputs though. This allows us to set metadata that
# depends on analyzer outputs. _augment_metadata will use the analyzer
# outputs stored in `transform_fn` to compute the metadata in a
# deferred manner, once the analyzer outputs are known.
metadata = dataset_metadata.DatasetMetadata(
schema=impl_helper.infer_feature_schema(outputs))
deferred_metadata = (transform_fn
| 'ComputeDeferredMetadata' >> beam.Map(
_augment_metadata, metadata))
full_metadata = beam_metadata_io.BeamDatasetMetadata(
metadata, deferred_metadata)
_clear_shared_state_after_barrier(input_values.pipeline,
transform_fn)
return transform_fn, full_metadata
