apache-airflow-providers-amazon 8.22.0__py3-none-any.whl → 8.23.0__py3-none-any.whl

airflow/providers/amazon/__init__.py

@@ -29,7 +29,7 @@ from airflow import __version__ as airflow_version
 
 __all__ = ["__version__"]
 
-__version__ = "8.22.0"
+__version__ = "8.23.0"
 
 if packaging.version.parse(packaging.version.parse(airflow_version).base_version) < packaging.version.parse(
     "2.7.0"

airflow/providers/amazon/aws/executors/batch/batch_executor.py

@@ -19,10 +19,11 @@
 
 from __future__ import annotations
 
+import contextlib
 import time
 from collections import defaultdict, deque
 from copy import deepcopy
-from typing import TYPE_CHECKING, Any, Dict, List
+from typing import TYPE_CHECKING, Any, Dict, List, Sequence
 
 from botocore.exceptions import ClientError, NoCredentialsError
 
@@ -34,11 +35,12 @@ from airflow.providers.amazon.aws.executors.utils.exponential_backoff_retry impo
     exponential_backoff_retry,
 )
 from airflow.providers.amazon.aws.hooks.batch_client import BatchClientHook
+from airflow.stats import Stats
 from airflow.utils import timezone
 from airflow.utils.helpers import merge_dicts
 
 if TYPE_CHECKING:
-    from airflow.models.taskinstance import TaskInstanceKey
+    from airflow.models.taskinstance import TaskInstance, TaskInstanceKey
 from airflow.providers.amazon.aws.executors.batch.boto_schema import (
     BatchDescribeJobsResponseSchema,
     BatchSubmitJobResponseSchema,
@@ -306,14 +308,20 @@ class AwsBatchExecutor(BaseExecutor):
                     self.pending_jobs.append(batch_job)
             else:
                 # Success case
+                job_id = submit_job_response["job_id"]
                 self.active_workers.add_job(
-                    job_id=submit_job_response["job_id"],
+                    job_id=job_id,
                     airflow_task_key=key,
                     airflow_cmd=cmd,
                     queue=queue,
                     exec_config=exec_config,
                     attempt_number=attempt_number,
                 )
+                with contextlib.suppress(AttributeError):
+                    # TODO: Remove this when min_airflow_version is 2.10.0 or higher in Amazon provider.
+                    # running_state is added in Airflow 2.10 and only needed to support task adoption
+                    # (an optional executor feature).
+                    self.running_state(key, job_id)
         if failure_reasons:
             self.log.error(
                 "Pending Batch jobs failed to launch for the following reasons: %s. Retrying later.",
@@ -418,3 +426,39 @@ class AwsBatchExecutor(BaseExecutor):
                 " and value should be NULL or empty."
             )
         return submit_kwargs
+
+    def try_adopt_task_instances(self, tis: Sequence[TaskInstance]) -> Sequence[TaskInstance]:
+        """
+        Adopt task instances which have an external_executor_id (the Batch job ID).
+
+        Anything that is not adopted will be cleared by the scheduler and becomes eligible for re-scheduling.
+        """
+        with Stats.timer("batch_executor.adopt_task_instances.duration"):
+            adopted_tis: list[TaskInstance] = []
+
+            if job_ids := [ti.external_executor_id for ti in tis if ti.external_executor_id]:
+                batch_jobs = self._describe_jobs(job_ids)
+
+                for batch_job in batch_jobs:
+                    ti = next(ti for ti in tis if ti.external_executor_id == batch_job.job_id)
+                    self.active_workers.add_job(
+                        job_id=batch_job.job_id,
+                        airflow_task_key=ti.key,
+                        airflow_cmd=ti.command_as_list(),
+                        queue=ti.queue,
+                        exec_config=ti.executor_config,
+                        attempt_number=ti.prev_attempted_tries,
+                    )
+                    adopted_tis.append(ti)
+
+            if adopted_tis:
+                tasks = [f"{task} in state {task.state}" for task in adopted_tis]
+                task_instance_str = "\n\t".join(tasks)
+                self.log.info(
+                    "Adopted the following %d tasks from a dead executor:\n\t%s",
+                    len(adopted_tis),
+                    task_instance_str,
+                )
+
+            not_adopted_tis = [ti for ti in tis if ti not in adopted_tis]
+            return not_adopted_tis

airflow/providers/amazon/aws/executors/ecs/ecs_executor.py

@@ -405,6 +405,7 @@ class AwsEcsExecutor(BaseExecutor):
                 except AttributeError:
                     # running_state is newly added, and only needed to support task adoption (an optional
                     # executor feature).
+                    # TODO: remove when min airflow version >= 2.9.2
                     pass
         if failure_reasons:
             self.log.error(

airflow/providers/amazon/aws/hooks/bedrock.py

@@ -77,3 +77,23 @@ class BedrockAgentHook(AwsBaseHook):
     def __init__(self, *args, **kwargs) -> None:
         kwargs["client_type"] = self.client_type
         super().__init__(*args, **kwargs)
+
+
+class BedrockAgentRuntimeHook(AwsBaseHook):
+    """
+    Interact with the Amazon Agents for Bedrock API.
+
+    Provide thin wrapper around :external+boto3:py:class:`boto3.client("bedrock-agent-runtime") <AgentsforBedrockRuntime.Client>`.
+
+    Additional arguments (such as ``aws_conn_id``) may be specified and
+    are passed down to the underlying AwsBaseHook.
+
+    .. seealso::
+        - :class:`airflow.providers.amazon.aws.hooks.base_aws.AwsBaseHook`
+    """
+
+    client_type = "bedrock-agent-runtime"
+
+    def __init__(self, *args, **kwargs) -> None:
+        kwargs["client_type"] = self.client_type
+        super().__init__(*args, **kwargs)

airflow/providers/amazon/aws/hooks/comprehend.py

@@ -0,0 +1,37 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+from __future__ import annotations
+
+from airflow.providers.amazon.aws.hooks.base_aws import AwsBaseHook
+
+
+class ComprehendHook(AwsBaseHook):
+    """
+    Interact with AWS Comprehend.
+
+    Provide thin wrapper around :external+boto3:py:class:`boto3.client("comprehend") <Comprehend.Client>`.
+
+    Additional arguments (such as ``aws_conn_id``) may be specified and
+    are passed down to the underlying AwsBaseHook.
+
+    .. seealso::
+        - :class:`airflow.providers.amazon.aws.hooks.base_aws.AwsBaseHook`
+    """
+
+    def __init__(self, *args, **kwargs) -> None:
+        kwargs["client_type"] = "comprehend"
+        super().__init__(*args, **kwargs)

airflow/providers/amazon/aws/hooks/neptune.py

@@ -34,6 +34,12 @@ class NeptuneHook(AwsBaseHook):
 
     AVAILABLE_STATES = ["available"]
     STOPPED_STATES = ["stopped"]
+    ERROR_STATES = [
+        "cloning-failed",
+        "inaccessible-encryption-credentials",
+        "inaccessible-encryption-credentials-recoverable",
+        "migration-failed",
+    ]
 
     def __init__(self, *args, **kwargs):
         kwargs["client_type"] = "neptune"
@@ -82,4 +88,33 @@ class NeptuneHook(AwsBaseHook):
         :param cluster_id: The ID of the cluster to get the status of.
         :return: The status of the cluster.
         """
-        return self.get_conn().describe_db_clusters(DBClusterIdentifier=cluster_id)["DBClusters"][0]["Status"]
+        return self.conn.describe_db_clusters(DBClusterIdentifier=cluster_id)["DBClusters"][0]["Status"]
+
+    def get_db_instance_status(self, instance_id: str) -> str:
+        """
+        Get the status of a Neptune instance.
+
+        :param instance_id: The ID of the instance to get the status of.
+        :return: The status of the instance.
+        """
+        return self.conn.describe_db_instances(DBInstanceIdentifier=instance_id)["DBInstances"][0][
+            "DBInstanceStatus"
+        ]
+
+    def wait_for_cluster_instance_availability(
+        self, cluster_id: str, delay: int = 30, max_attempts: int = 60
+    ) -> None:
+        """
+        Wait for Neptune instances in a cluster to be available.
+
+        :param cluster_id: The cluster ID of the instances to wait for.
+        :param delay: Time in seconds to delay between polls.
+        :param max_attempts: Maximum number of attempts to poll for completion.
+        :return: The status of the instances.
+        """
+        filters = [{"Name": "db-cluster-id", "Values": [cluster_id]}]
+        self.log.info("Waiting for instances in cluster %s.", cluster_id)
+        self.get_waiter("db_instance_available").wait(
+            Filters=filters, WaiterConfig={"Delay": delay, "MaxAttempts": max_attempts}
+        )
+        self.log.info("Finished waiting for instances in cluster %s.", cluster_id)

airflow/providers/amazon/aws/operators/athena.py

@@ -266,7 +266,7 @@ class AthenaOperator(AwsBaseOperator[AthenaHook]):
 
         if self.output_location:
             parsed = urlparse(self.output_location)
-            outputs.append(Dataset(namespace=f"{parsed.scheme}://{parsed.netloc}", name=parsed.path))
+            outputs.append(Dataset(namespace=f"{parsed.scheme}://{parsed.netloc}", name=parsed.path or "/"))
 
         return OperatorLineage(job_facets=job_facets, run_facets=run_facets, inputs=inputs, outputs=outputs)
 

airflow/providers/amazon/aws/operators/batch.py

@@ -206,9 +206,7 @@ class BatchOperator(BaseOperator):
         self.scheduling_priority_override = scheduling_priority_override
         self.array_properties = array_properties
         self.parameters = parameters or {}
-        self.retry_strategy = retry_strategy or {}
-        if not self.retry_strategy.get("attempts", None):
-            self.retry_strategy["attempts"] = 1
+        self.retry_strategy = retry_strategy
         self.waiters = waiters
         self.tags = tags or {}
         self.wait_for_completion = wait_for_completion

airflow/providers/amazon/aws/operators/bedrock.py

@@ -20,11 +20,17 @@ import json
 from time import sleep
 from typing import TYPE_CHECKING, Any, Sequence
 
+import botocore
 from botocore.exceptions import ClientError
 
 from airflow.configuration import conf
 from airflow.exceptions import AirflowException
-from airflow.providers.amazon.aws.hooks.bedrock import BedrockAgentHook, BedrockHook, BedrockRuntimeHook
+from airflow.providers.amazon.aws.hooks.bedrock import (
+    BedrockAgentHook,
+    BedrockAgentRuntimeHook,
+    BedrockHook,
+    BedrockRuntimeHook,
+)
 from airflow.providers.amazon.aws.operators.base_aws import AwsBaseOperator
 from airflow.providers.amazon.aws.triggers.bedrock import (
     BedrockCustomizeModelCompletedTrigger,
@@ -32,7 +38,7 @@ from airflow.providers.amazon.aws.triggers.bedrock import (
     BedrockKnowledgeBaseActiveTrigger,
     BedrockProvisionModelThroughputCompletedTrigger,
 )
-from airflow.providers.amazon.aws.utils import validate_execute_complete_event
+from airflow.providers.amazon.aws.utils import get_botocore_version, validate_execute_complete_event
 from airflow.providers.amazon.aws.utils.mixins import aws_template_fields
 from airflow.utils.helpers import prune_dict
 from airflow.utils.timezone import utcnow
@@ -664,3 +670,213 @@ class BedrockIngestDataOperator(AwsBaseOperator[BedrockAgentHook]):
             )
 
         return ingestion_job_id
+
+
+class BedrockRaGOperator(AwsBaseOperator[BedrockAgentRuntimeHook]):
+    """
+    Query a knowledge base and generate responses based on the retrieved results with sources citations.
+
+    NOTE:  Support for EXTERNAL SOURCES was added in botocore 1.34.90
+
+    .. seealso::
+        For more information on how to use this operator, take a look at the guide:
+        :ref:`howto/operator:BedrockRaGOperator`
+
+    :param input: The query to be made to the knowledge base. (templated)
+    :param source_type: The type of resource that is queried by the request. (templated)
+        Must be one of 'KNOWLEDGE_BASE' or 'EXTERNAL_SOURCES', and the appropriate config values must also be provided.
+        If set to 'KNOWLEDGE_BASE' then `knowledge_base_id` must be provided, and `vector_search_config` may be.
+        If set to `EXTERNAL_SOURCES` then `sources` must also be provided.
+        NOTE:  Support for EXTERNAL SOURCES was added in botocore 1.34.90
+    :param model_arn: The ARN of the foundation model used to generate a response. (templated)
+    :param prompt_template: The template for the prompt that's sent to the model for response generation.
+        You can include prompt placeholders, which are replaced before the prompt is sent to the model
+        to provide instructions and context to the model. In addition, you can include XML tags to delineate
+        meaningful sections of the prompt template. (templated)
+    :param knowledge_base_id: The unique identifier of the knowledge base that is queried. (templated)
+            Can only be specified if source_type='KNOWLEDGE_BASE'.
+    :param vector_search_config: How the results from the vector search should be returned. (templated)
+        Can only be specified if source_type='KNOWLEDGE_BASE'.
+        For more information, see https://docs.aws.amazon.com/bedrock/latest/userguide/kb-test-config.html.
+    :param sources: The documents used as reference for the response. (templated)
+        Can only be specified if source_type='EXTERNAL_SOURCES'
+        NOTE:  Support for EXTERNAL SOURCES was added in botocore 1.34.90
+    :param rag_kwargs: Additional keyword arguments to pass to the  API call. (templated)
+    """
+
+    aws_hook_class = BedrockAgentRuntimeHook
+    template_fields: Sequence[str] = aws_template_fields(
+        "input",
+        "source_type",
+        "model_arn",
+        "prompt_template",
+        "knowledge_base_id",
+        "vector_search_config",
+        "sources",
+        "rag_kwargs",
+    )
+
+    def __init__(
+        self,
+        input: str,
+        source_type: str,
+        model_arn: str,
+        prompt_template: str | None = None,
+        knowledge_base_id: str | None = None,
+        vector_search_config: dict[str, Any] | None = None,
+        sources: list[dict[str, Any]] | None = None,
+        rag_kwargs: dict[str, Any] | None = None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.input = input
+        self.prompt_template = prompt_template
+        self.source_type = source_type.upper()
+        self.knowledge_base_id = knowledge_base_id
+        self.model_arn = model_arn
+        self.vector_search_config = vector_search_config
+        self.sources = sources
+        self.rag_kwargs = rag_kwargs or {}
+
+    def validate_inputs(self):
+        if self.source_type == "KNOWLEDGE_BASE":
+            if self.knowledge_base_id is None:
+                raise AttributeError(
+                    "If `source_type` is set to 'KNOWLEDGE_BASE' then `knowledge_base_id` must be provided."
+                )
+            if self.sources is not None:
+                raise AttributeError(
+                    "`sources` can not be used when `source_type` is set to 'KNOWLEDGE_BASE'."
+                )
+        elif self.source_type == "EXTERNAL_SOURCES":
+            if not self.sources is not None:
+                raise AttributeError(
+                    "If `source_type` is set to `EXTERNAL_SOURCES` then `sources` must also be provided."
+                )
+            if self.vector_search_config or self.knowledge_base_id:
+                raise AttributeError(
+                    "`vector_search_config` and `knowledge_base_id` can not be used "
+                    "when `source_type` is set to `EXTERNAL_SOURCES`"
+                )
+        else:
+            raise AttributeError(
+                "`source_type` must be one of 'KNOWLEDGE_BASE' or 'EXTERNAL_SOURCES', "
+                "and the appropriate config values must also be provided."
+            )
+
+    def build_rag_config(self) -> dict[str, Any]:
+        result: dict[str, Any] = {}
+        base_config: dict[str, Any] = {
+            "modelArn": self.model_arn,
+        }
+
+        if self.prompt_template:
+            base_config["generationConfiguration"] = {
+                "promptTemplate": {"textPromptTemplate": self.prompt_template}
+            }
+
+        if self.source_type == "KNOWLEDGE_BASE":
+            if self.vector_search_config:
+                base_config["retrievalConfiguration"] = {
+                    "vectorSearchConfiguration": self.vector_search_config
+                }
+
+            result = {
+                "type": self.source_type,
+                "knowledgeBaseConfiguration": {
+                    **base_config,
+                    "knowledgeBaseId": self.knowledge_base_id,
+                },
+            }
+
+        if self.source_type == "EXTERNAL_SOURCES":
+            result = {
+                "type": self.source_type,
+                "externalSourcesConfiguration": {**base_config, "sources": self.sources},
+            }
+        return result
+
+    def execute(self, context: Context) -> Any:
+        self.validate_inputs()
+
+        try:
+            result = self.hook.conn.retrieve_and_generate(
+                input={"text": self.input},
+                retrieveAndGenerateConfiguration=self.build_rag_config(),
+                **self.rag_kwargs,
+            )
+        except botocore.exceptions.ParamValidationError as error:
+            if (
+                'Unknown parameter in retrieveAndGenerateConfiguration: "externalSourcesConfiguration"'
+                in str(error)
+            ) and (self.source_type == "EXTERNAL_SOURCES"):
+                self.log.error(
+                    "You are attempting to use External Sources and the BOTO API returned an "
+                    "error message which may indicate the need to update botocore to do this.  \n"
+                    "Support for External Sources was added in botocore 1.34.90 and you are using botocore %s",
+                    ".".join(map(str, get_botocore_version())),
+                )
+            raise
+
+        self.log.info(
+            "\nPrompt: %s\nResponse: %s\nCitations: %s",
+            self.input,
+            result["output"]["text"],
+            result["citations"],
+        )
+        return result
+
+
+class BedrockRetrieveOperator(AwsBaseOperator[BedrockAgentRuntimeHook]):
+    """
+    Query a knowledge base and retrieve results with source citations.
+
+    .. seealso::
+        For more information on how to use this operator, take a look at the guide:
+        :ref:`howto/operator:BedrockRetrieveOperator`
+
+    :param retrieval_query: The query to be made to the knowledge base. (templated)
+    :param knowledge_base_id: The unique identifier of the knowledge base that is queried. (templated)
+    :param vector_search_config: How the results from the vector search should be returned. (templated)
+        For more information, see https://docs.aws.amazon.com/bedrock/latest/userguide/kb-test-config.html.
+    :param retrieve_kwargs: Additional keyword arguments to pass to the  API call. (templated)
+    """
+
+    aws_hook_class = BedrockAgentRuntimeHook
+    template_fields: Sequence[str] = aws_template_fields(
+        "retrieval_query",
+        "knowledge_base_id",
+        "vector_search_config",
+        "retrieve_kwargs",
+    )
+
+    def __init__(
+        self,
+        retrieval_query: str,
+        knowledge_base_id: str,
+        vector_search_config: dict[str, Any] | None = None,
+        retrieve_kwargs: dict[str, Any] | None = None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.retrieval_query = retrieval_query
+        self.knowledge_base_id = knowledge_base_id
+        self.vector_search_config = vector_search_config
+        self.retrieve_kwargs = retrieve_kwargs or {}
+
+    def execute(self, context: Context) -> Any:
+        retrieval_configuration = (
+            {"retrievalConfiguration": {"vectorSearchConfiguration": self.vector_search_config}}
+            if self.vector_search_config
+            else {}
+        )
+
+        result = self.hook.conn.retrieve(
+            retrievalQuery={"text": self.retrieval_query},
+            knowledgeBaseId=self.knowledge_base_id,
+            **retrieval_configuration,
+            **self.retrieve_kwargs,
+        )
+
+        self.log.info("\nQuery: %s\nRetrieved: %s", self.retrieval_query, result["retrievalResults"])
+        return result

airflow/providers/amazon/aws/operators/comprehend.py

@@ -0,0 +1,192 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+from __future__ import annotations
+
+from functools import cached_property
+from typing import TYPE_CHECKING, Any, Sequence
+
+from airflow.configuration import conf
+from airflow.exceptions import AirflowException
+from airflow.providers.amazon.aws.hooks.comprehend import ComprehendHook
+from airflow.providers.amazon.aws.operators.base_aws import AwsBaseOperator
+from airflow.providers.amazon.aws.triggers.comprehend import ComprehendPiiEntitiesDetectionJobCompletedTrigger
+from airflow.providers.amazon.aws.utils import validate_execute_complete_event
+from airflow.providers.amazon.aws.utils.mixins import aws_template_fields
+from airflow.utils.timezone import utcnow
+
+if TYPE_CHECKING:
+    import boto3
+
+    from airflow.utils.context import Context
+
+
+class ComprehendBaseOperator(AwsBaseOperator[ComprehendHook]):
+    """
+    This is the base operator for Comprehend Service operators (not supposed to be used directly in DAGs).
+
+    :param input_data_config: The input properties for a PII entities detection job. (templated)
+    :param output_data_config: Provides `configuration` parameters for the output of PII entity detection
+        jobs. (templated)
+    :param data_access_role_arn: The Amazon Resource Name (ARN) of the IAM role that grants Amazon Comprehend
+        read access to your input data. (templated)
+    :param language_code: The language of the input documents. (templated)
+    """
+
+    aws_hook_class = ComprehendHook
+
+    template_fields: Sequence[str] = aws_template_fields(
+        "input_data_config", "output_data_config", "data_access_role_arn", "language_code"
+    )
+
+    template_fields_renderers: dict = {"input_data_config": "json", "output_data_config": "json"}
+
+    def __init__(
+        self,
+        input_data_config: dict,
+        output_data_config: dict,
+        data_access_role_arn: str,
+        language_code: str,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.input_data_config = input_data_config
+        self.output_data_config = output_data_config
+        self.data_access_role_arn = data_access_role_arn
+        self.language_code = language_code
+
+    @cached_property
+    def client(self) -> boto3.client:
+        """Create and return the Comprehend client."""
+        return self.hook.conn
+
+    def execute(self, context: Context):
+        """Must overwrite in child classes."""
+        raise NotImplementedError("Please implement execute() in subclass")
+
+
+class ComprehendStartPiiEntitiesDetectionJobOperator(ComprehendBaseOperator):
+    """
+    Create a comprehend pii entities detection job for a collection of documents.
+
+    .. seealso::
+        For more information on how to use this operator, take a look at the guide:
+        :ref:`howto/operator:ComprehendStartPiiEntitiesDetectionJobOperator`
+
+    :param input_data_config: The input properties for a PII entities detection job. (templated)
+    :param output_data_config: Provides `configuration` parameters for the output of PII entity detection
+        jobs. (templated)
+    :param mode: Specifies whether the output provides the locations (offsets) of PII  entities or a file in
+        which PII entities are redacted. If you set the mode parameter to ONLY_REDACTION. In that case you
+        must provide a RedactionConfig in start_pii_entities_kwargs.
+    :param data_access_role_arn: The Amazon Resource Name (ARN) of the IAM role that grants Amazon Comprehend
+        read access to your input data. (templated)
+    :param language_code: The language of the input documents. (templated)
+    :param start_pii_entities_kwargs: Any optional parameters to pass to the job. If JobName is not provided
+        in start_pii_entities_kwargs, operator will create.
+
+    :param wait_for_completion: Whether to wait for job to stop. (default: True)
+    :param waiter_delay: Time in seconds to wait between status checks. (default: 60)
+    :param waiter_max_attempts: Maximum number of attempts to check for job completion. (default: 20)
+    :param deferrable: If True, the operator will wait asynchronously for the job to stop.
+        This implies waiting for completion. This mode requires aiobotocore module to be installed.
+        (default: False)
+    :param aws_conn_id: The Airflow connection used for AWS credentials.
+        If this is ``None`` or empty then the default boto3 behaviour is used. If
+        running Airflow in a distributed manner and aws_conn_id is None or
+        empty, then default boto3 configuration would be used (and must be
+        maintained on each worker node).
+    :param region_name: AWS region_name. If not specified then the default boto3 behaviour is used.
+    :param verify: Whether to verify SSL certificates. See:
+        https://boto3.amazonaws.com/v1/documentation/api/latest/reference/core/session.html
+    :param botocore_config: Configuration dictionary (key-values) for botocore client. See:
+        https://botocore.amazonaws.com/v1/documentation/api/latest/reference/config.html
+    """
+
+    def __init__(
+        self,
+        input_data_config: dict,
+        output_data_config: dict,
+        mode: str,
+        data_access_role_arn: str,
+        language_code: str,
+        start_pii_entities_kwargs: dict[str, Any] | None = None,
+        wait_for_completion: bool = True,
+        waiter_delay: int = 60,
+        waiter_max_attempts: int = 20,
+        deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False),
+        **kwargs,
+    ):
+        super().__init__(
+            input_data_config=input_data_config,
+            output_data_config=output_data_config,
+            data_access_role_arn=data_access_role_arn,
+            language_code=language_code,
+            **kwargs,
+        )
+        self.mode = mode
+        self.start_pii_entities_kwargs = start_pii_entities_kwargs or {}
+        self.wait_for_completion = wait_for_completion
+        self.waiter_delay = waiter_delay
+        self.waiter_max_attempts = waiter_max_attempts
+        self.deferrable = deferrable
+
+    def execute(self, context: Context) -> str:
+        if self.start_pii_entities_kwargs.get("JobName", None) is None:
+            self.start_pii_entities_kwargs["JobName"] = (
+                f"start_pii_entities_detection_job-{int(utcnow().timestamp())}"
+            )
+
+        self.log.info(
+            "Submitting start pii entities detection job '%s'.", self.start_pii_entities_kwargs["JobName"]
+        )
+        job_id = self.client.start_pii_entities_detection_job(
+            InputDataConfig=self.input_data_config,
+            OutputDataConfig=self.output_data_config,
+            Mode=self.mode,
+            DataAccessRoleArn=self.data_access_role_arn,
+            LanguageCode=self.language_code,
+            **self.start_pii_entities_kwargs,
+        )["JobId"]
+
+        message_description = f"start pii entities detection job {job_id} to complete."
+        if self.deferrable:
+            self.log.info("Deferring %s", message_description)
+            self.defer(
+                trigger=ComprehendPiiEntitiesDetectionJobCompletedTrigger(
+                    job_id=job_id,
+                    waiter_delay=self.waiter_delay,
+                    waiter_max_attempts=self.waiter_max_attempts,
+                    aws_conn_id=self.aws_conn_id,
+                ),
+                method_name="execute_complete",
+            )
+        elif self.wait_for_completion:
+            self.log.info("Waiting for %s", message_description)
+            self.hook.get_waiter("pii_entities_detection_job_complete").wait(
+                JobId=job_id,
+                WaiterConfig={"Delay": self.waiter_delay, "MaxAttempts": self.waiter_max_attempts},
+            )
+
+        return job_id
+
+    def execute_complete(self, context: Context, event: dict[str, Any] | None = None) -> str:
+        event = validate_execute_complete_event(event)
+        if event["status"] != "success":
+            raise AirflowException("Error while running job: %s", event)
+
+        self.log.info("Comprehend pii entities detection job `%s` complete.", event["job_id"])
+        return event["job_id"]