azure.eventhub.aio package

class azure.eventhub.aio.EventHubConsumerClient(fully_qualified_namespace: str, eventhub_name: str, consumer_group: str, credential: TokenCredential, **kwargs)

The EventHubConsumerClient class defines a high level interface for receiving events from the Azure Event Hubs service.

The main goal of EventHubConsumerClient is to receive events from all partitions of an EventHub with load-balancing and checkpointing.

When multiple EventHubConsumerClient instances are running against the same event hub, consumer group and checkpointing location, the partitions will be evenly distributed among them.

To enable load-balancing and persisted checkpoints, checkpoint_store must be set when creating the EventHubConsumerClient. If a checkpoint store is not provided, the checkpoint will be maintained internally in memory.

An EventHubConsumerClient can also receive from a specific partition when you call its method receive() or receive_batch() and specify the partition_id. Load-balancing won’t work in single-partition mode. But users can still save checkpoints if the checkpoint_store is set.

Parameters

• fully_qualified_namespace (str) – The fully qualified host name for the Event Hubs namespace. The namespace format is: <yournamespace>.servicebus.windows.net.

• eventhub_name (str) – The path of the specific Event Hub to connect the client to.

• consumer_group (str) – Receive events from the event hub for this consumer group.

• credential (TokenCredential) – The credential object used for authentication which implements a particular interface for getting tokens. It accepts EventHubSharedKeyCredential, or credential objects generated by the azure-identity library and objects that implement the get_token(self, *scopes) method.

Keyword Arguments

• logging_enable (bool) – Whether to output network trace logs to the logger. Default is False.

• auth_timeout (float) – The time in seconds to wait for a token to be authorized by the service. The default value is 60 seconds. If set to 0, no timeout will be enforced from the client.

• user_agent (str) – The user agent that should be appended to the built-in user agent string.

• retry_total (int) – The total number of attempts to redo a failed operation when an error occurs. Default value is 3. The context of retry_total in receiving is special: The receive method is implemented by a while-loop calling internal receive method in each iteration. In the receive case, retry_total specifies the numbers of retry after error raised by internal receive method in the while-loop. If retry attempts are exhausted, the on_error callback will be called (if provided) with the error information. The failed internal partition consumer will be closed (on_partition_close will be called if provided) and new internal partition consumer will be created (on_partition_initialize will be called if provided) to resume receiving.

• idle_timeout (float) – Timeout, in seconds, after which this client will close the underlying connection if there is no further activity. By default the value is None, meaning that the client will not shutdown due to inactivity unless initiated by the service.

• transport_type (TransportType) – The type of transport protocol that will be used for communicating with the Event Hubs service. Default is TransportType.Amqp.

• http_proxy (dict) – HTTP proxy settings. This must be a dictionary with the following keys: ‘proxy_hostname’ (str value) and ‘proxy_port’ (int value). Additionally the following keys may also be present: ‘username’, ‘password’.

• checkpoint_store (CheckpointStore) – A manager that stores the partition load-balancing and checkpoint data when receiving events. The checkpoint store will be used in both cases of receiving from all partitions or a single partition. In the latter case load-balancing does not apply. If a checkpoint store is not provided, the checkpoint will be maintained internally in memory, and the EventHubConsumerClient instance will receive events without load-balancing.

• load_balancing_interval (float) – When load-balancing kicks in. This is the interval, in seconds, between two load-balancing evaluations. Default is 10 seconds.

• partition_ownership_expiration_interval (float) – A partition ownership will expire after this number of seconds. Every load-balancing evaluation will automatically extend the ownership expiration time. Default is 6 * load_balancing_interval, i.e. 60 seconds when using the default load_balancing_interval of 10 seconds.

• load_balancing_strategy (str or LoadBalancingStrategy) – When load-balancing kicks in, it will use this strategy to claim and balance the partition ownership. Use “greedy” or LoadBalancingStrategy.GREEDY for the greedy strategy, which, for every load-balancing evaluation, will grab as many unclaimed partitions required to balance the load. Use “balanced” or LoadBalancingStrategy.BALANCED for the balanced strategy, which, for every load-balancing evaluation, claims only one partition that is not claimed by other EventHubConsumerClient. If all partitions of an EventHub are claimed by other EventHubConsumerClient and this client has claimed too few partitions, this client will steal one partition from other clients for every load-balancing evaluation regardless of the load balancing strategy. Greedy strategy is used by default.

• custom_endpoint_address (str) – The custom endpoint address to use for establishing a connection to the Event Hubs service, allowing network requests to be routed through any application gateways or other paths needed for the host environment. Default is None. The format would be like “sb://<custom_endpoint_hostname>:<custom_endpoint_port>”. If port is not specified in the custom_endpoint_address, by default port 443 will be used.

• connection_verify (str) – Path to the custom CA_BUNDLE file of the SSL certificate which is used to authenticate the identity of the connection endpoint. Default is None in which case certifi.where() will be used.

Example:

Create a new instance of the EventHubConsumerClient.

import os
from azure.eventhub.aio import EventHubConsumerClient, EventHubSharedKeyCredential

fully_qualified_namespace = os.environ['EVENT_HUB_HOSTNAME']
eventhub_name = os.environ['EVENT_HUB_NAME']
shared_access_policy = os.environ['EVENT_HUB_SAS_POLICY']
shared_access_key = os.environ['EVENT_HUB_SAS_KEY']

consumer = EventHubConsumerClient(fully_qualified_namespace=fully_qualified_namespace,
                                  consumer_group='$Default',
                                  eventhub_name=eventhub_name,
                                  credential=EventHubSharedKeyCredential(shared_access_policy, shared_access_key))

async close() → None

Stop retrieving events from the Event Hub and close the underlying AMQP connection and links.

Return type

None

Example:

Close down the client.

import os

event_hub_connection_str = os.environ['EVENT_HUB_CONN_STR']
eventhub_name = os.environ['EVENT_HUB_NAME']

from azure.eventhub.aio import EventHubConsumerClient
consumer = EventHubConsumerClient.from_connection_string(
    conn_str=event_hub_connection_str,
    consumer_group='$Default',
    eventhub_name=eventhub_name  # EventHub name should be specified if it doesn't show up in connection string.
)

logger = logging.getLogger("azure.eventhub")

async def on_event(partition_context, event):
    # Put your code here.
    # If the operation is i/o intensive, async will have better performance.
    logger.info("Received event from partition: {}".format(partition_context.partition_id))

# The receive method is a coroutine which will be blocking when awaited.
# It can be executed in an async task for non-blocking behavior, and combined with the 'close' method.

recv_task = asyncio.ensure_future(consumer.receive(on_event=on_event))
await asyncio.sleep(3)  # keep receiving for 3 seconds
recv_task.cancel()  # stop receiving

# Close down the consumer handler explicitly.
await consumer.close()

classmethod from_connection_string(conn_str: str, consumer_group: str, *, eventhub_name: Optional[str] = None, logging_enable: bool = False, http_proxy: Optional[Dict[str, Union[str, int]]] = None, auth_timeout: float = 60, user_agent: Optional[str] = None, retry_total: int = 3, transport_type: Optional[TransportType] = None, checkpoint_store: Optional[CheckpointStore] = None, load_balancing_interval: float = 10, **kwargs: Any) → EventHubConsumerClient

Create an EventHubConsumerClient from a connection string.

Parameters

• conn_str (str) – The connection string of an Event Hub.

• consumer_group (str) – Receive events from the Event Hub for this consumer group.

Keyword Arguments

• eventhub_name (str) – The path of the specific Event Hub to connect the client to.

• logging_enable (bool) – Whether to output network trace logs to the logger. Default is False.

• http_proxy (dict) – HTTP proxy settings. This must be a dictionary with the following keys: ‘proxy_hostname’ (str value) and ‘proxy_port’ (int value). Additionally the following keys may also be present: ‘username’, ‘password’.

• auth_timeout (float) – The time in seconds to wait for a token to be authorized by the service. The default value is 60 seconds. If set to 0, no timeout will be enforced from the client.

• user_agent (str) – The user agent that should be appended to the built-in user agent string.

• retry_total (int) – The total number of attempts to redo a failed operation when an error occurs. Default value is 3. The context of retry_total in receiving is special: The receive method is implemented by a while-loop calling internal receive method in each iteration. In the receive case, retry_total specifies the numbers of retry after error raised by internal receive method in the while-loop. If retry attempts are exhausted, the on_error callback will be called (if provided) with the error information. The failed internal partition consumer will be closed (on_partition_close will be called if provided) and new internal partition consumer will be created (on_partition_initialize will be called if provided) to resume receiving.

• idle_timeout (float) – Timeout, in seconds, after which this client will close the underlying connection if there is no further activity. By default the value is None, meaning that the client will not shutdown due to inactivity unless initiated by the service.

• transport_type (TransportType) – The type of transport protocol that will be used for communicating with the Event Hubs service. Default is TransportType.Amqp.

• checkpoint_store (CheckpointStore) – A manager that stores the partition load-balancing and checkpoint data when receiving events. The checkpoint store will be used in both cases of receiving from all partitions or a single partition. In the latter case load-balancing does not apply. If a checkpoint store is not provided, the checkpoint will be maintained internally in memory, and the EventHubConsumerClient instance will receive events without load-balancing.

• load_balancing_interval (float) – When load-balancing kicks in. This is the interval, in seconds, between two load-balancing evaluations. Default is 10 seconds.

• partition_ownership_expiration_interval (float) – A partition ownership will expire after this number of seconds. Every load-balancing evaluation will automatically extend the ownership expiration time. Default is 6 * load_balancing_interval, i.e. 60 seconds when using the default load_balancing_interval of 10 seconds.

• load_balancing_strategy (str or LoadBalancingStrategy) – When load-balancing kicks in, it will use this strategy to claim and balance the partition ownership. Use “greedy” or LoadBalancingStrategy.GREEDY for the greedy strategy, which, for every load-balancing evaluation, will grab as many unclaimed partitions required to balance the load. Use “balanced” or LoadBalancingStrategy.BALANCED for the balanced strategy, which, for every load-balancing evaluation, claims only one partition that is not claimed by other EventHubConsumerClient. If all partitions of an EventHub are claimed by other EventHubConsumerClient and this client has claimed too few partitions, this client will steal one partition from other clients for every load-balancing evaluation regardless of the load balancing strategy. Greedy strategy is used by default.

• custom_endpoint_address (str) – The custom endpoint address to use for establishing a connection to the Event Hubs service, allowing network requests to be routed through any application gateways or other paths needed for the host environment. Default is None. The format would be like “sb://<custom_endpoint_hostname>:<custom_endpoint_port>”. If port is not specified in the custom_endpoint_address, by default port 443 will be used.

• connection_verify (str) – Path to the custom CA_BUNDLE file of the SSL certificate which is used to authenticate the identity of the connection endpoint. Default is None in which case certifi.where() will be used.

Return type

EventHubConsumerClient

Example:

Create a new instance of the EventHubConsumerClient from connection string.

import os
from azure.eventhub.aio import EventHubConsumerClient
event_hub_connection_str = os.environ['EVENT_HUB_CONN_STR']
eventhub_name = os.environ['EVENT_HUB_NAME']
consumer = EventHubConsumerClient.from_connection_string(
    conn_str=event_hub_connection_str,
    consumer_group='$Default',
    eventhub_name=eventhub_name  # EventHub name should be specified if it doesn't show up in connection string.
)

async get_eventhub_properties() → Dict[str, Any]

Get properties of the Event Hub.

Keys in the returned dictionary include:

• eventhub_name (str)

• created_at (UTC datetime.datetime)

• partition_ids (list[str])

Return type

dict

Raises

EventHubError

async get_partition_ids() → List[str]

Get partition IDs of the Event Hub.

Return type

list[str]

Raises

EventHubError

async get_partition_properties(partition_id: str) → Dict[str, Any]

Get properties of the specified partition.

Keys in the properties dictionary include:

• eventhub_name (str)

• id (str)

• beginning_sequence_number (int)

• last_enqueued_sequence_number (int)

• last_enqueued_offset (str)

• last_enqueued_time_utc (UTC datetime.datetime)

• is_empty (bool)

Parameters

partition_id (str) – The target partition ID.

Return type

dict

Raises

EventHubError

async receive(on_event: Callable[[PartitionContext, Optional[EventData]], Awaitable[None]], *, max_wait_time: Optional[float] = None, partition_id: Optional[str] = None, owner_level: Optional[int] = None, prefetch: int = 300, track_last_enqueued_event_properties: bool = False, starting_position: Union[str, int, datetime.datetime, Dict[str, Any], None] = None, starting_position_inclusive: Union[bool, Dict[str, bool]] = False, on_error: Optional[Callable[[PartitionContext, Exception], Awaitable[None]]] = None, on_partition_initialize: Optional[Callable[PartitionContext, Awaitable[None]]] = None, on_partition_close: Optional[Callable[[PartitionContext, CloseReason], Awaitable[None]]] = None) → None

Receive events from partition(s), with optional load-balancing and checkpointing.

Parameters

on_event (Callable[PartitionContext, Optional[EventData]]) – The callback function for handling a received event. The callback takes two parameters: partition_context which contains partition context and event which is the received event. The callback function should be defined like: on_event(partition_context, event). For detailed partition context information, please refer to PartitionContext.

Keyword Arguments

• max_wait_time (float) – The maximum interval in seconds that the event processor will wait before calling the callback. If no events are received within this interval, the on_event callback will be called with None. If this value is set to None or 0 (the default), the callback will not be called until an event is received.

• partition_id (str) – If specified, the client will receive from this partition only. Otherwise the client will receive from all partitions.

• owner_level (int) – The priority for an exclusive consumer. An exclusive consumer will be created if owner_level is set. A consumer with a higher owner_level has higher exclusive priority. The owner level is also know as the ‘epoch value’ of the consumer.

• prefetch (int) – The number of events to prefetch from the service for processing. Default is 300.

• track_last_enqueued_event_properties (bool) – Indicates whether the consumer should request information on the last-enqueued event on its associated partition, and track that information as events are received. When information about the partitions last-enqueued event is being tracked, each event received from the Event Hubs service will carry metadata about the partition. This results in a small amount of additional network bandwidth consumption that is generally a favorable trade-off when considered against periodically making requests for partition properties using the Event Hub client. It is set to False by default.

• starting_position (str, int, datetime.datetime or dict[str,Any]) – Start receiving from this event position if there is no checkpoint data for a partition. Checkpoint data will be used if available. This can be a a dict with partition ID as the key and position as the value for individual partitions, or a single value for all partitions. The value type can be str, int or datetime.datetime. Also supported are the values “-1” for receiving from the beginning of the stream, and “@latest” for receiving only new events.

• starting_position_inclusive (bool or dict[str,bool]) – Determine whether the given starting_position is inclusive(>=) or not (>). True for inclusive and False for exclusive. This can be a dict with partition ID as the key and bool as the value indicating whether the starting_position for a specific partition is inclusive or not. This can also be a single bool value for all starting_position. The default value is False.

• on_error (Callable[[PartitionContext, Exception]]) – The callback function that will be called when an error is raised during receiving after retry attempts are exhausted, or during the process of load-balancing. The callback takes two parameters: partition_context which contains partition information and error being the exception. partition_context could be None if the error is raised during the process of load-balance. The callback should be defined like: on_error(partition_context, error). The on_error callback will also be called if an unhandled exception is raised during the on_event callback.

• on_partition_initialize (Callable[[PartitionContext]]) – The callback function that will be called after a consumer for a certain partition finishes initialization. It would also be called when a new internal partition consumer is created to take over the receiving process for a failed and closed internal partition consumer. The callback takes a single parameter: partition_context which contains the partition information. The callback should be defined like: on_partition_initialize(partition_context).

• on_partition_close (Callable[[PartitionContext, CloseReason]]) – The callback function that will be called after a consumer for a certain partition is closed. It would be also called when error is raised during receiving after retry attempts are exhausted. The callback takes two parameters: partition_context which contains partition information and reason for the close. The callback should be defined like: on_partition_close(partition_context, reason). Please refer to CloseReason for the various closing reasons.

Return type

None

Example:

Receive events from the EventHub.

    logger = logging.getLogger("azure.eventhub")

    async def on_event(partition_context, event):
        # Put your code here.
        # If the operation is i/o intensive, async will have better performance.
        logger.info("Received event from partition: {}".format(partition_context.partition_id))

    async with consumer:
        await consumer.receive(
            on_event=on_event,
            starting_position="-1",  # "-1" is from the beginning of the partition.
        )

async receive_batch(on_event_batch: Callable[[PartitionContext, List[EventData]], Awaitable[None]], *, max_batch_size: int = 300, max_wait_time: Optional[float] = None, partition_id: Optional[str] = None, owner_level: Optional[int] = None, prefetch: int = 300, track_last_enqueued_event_properties: bool = False, starting_position: Union[str, int, datetime.datetime, Dict[str, Any], None] = None, starting_position_inclusive: Union[bool, Dict[str, bool]] = False, on_error: Optional[Callable[[PartitionContext, Exception], Awaitable[None]]] = None, on_partition_initialize: Optional[Callable[PartitionContext, Awaitable[None]]] = None, on_partition_close: Optional[Callable[[PartitionContext, CloseReason], Awaitable[None]]] = None) → None

Receive events from partition(s) in batches, with optional load-balancing and checkpointing.

Parameters

on_event_batch (Callable[PartitionContext, List[EventData]]) – The callback function for handling a batch of received events. The callback takes two parameters: partition_context which contains partition context and event_batch, which is the received events. The callback function should be defined like: on_event_batch(partition_context, event_batch). event_batch could be an empty list if max_wait_time is not None nor 0 and no event is received after max_wait_time. For detailed partition context information, please refer to PartitionContext.

Keyword Arguments

• max_batch_size (int) – The maximum number of events in a batch passed to callback on_event_batch. If the actual received number of events is larger than max_batch_size, the received events are divided into batches and call the callback for each batch with up to max_batch_size events.

• max_wait_time (float) – The maximum interval in seconds that the event processor will wait before calling the callback. If no events are received within this interval, the on_event_batch callback will be called with an empty list. If this value is set to None or 0 (the default), the callback will not be called until events are received.

• partition_id (str) – If specified, the client will receive from this partition only. Otherwise the client will receive from all partitions.

• owner_level (int) – The priority for an exclusive consumer. An exclusive consumer will be created if owner_level is set. A consumer with a higher owner_level has higher exclusive priority. The owner level is also know as the ‘epoch value’ of the consumer.

• prefetch (int) – The number of events to prefetch from the service for processing. Default is 300.

• track_last_enqueued_event_properties (bool) – Indicates whether the consumer should request information on the last-enqueued event on its associated partition, and track that information as events are received. When information about the partitions last-enqueued event is being tracked, each event received from the Event Hubs service will carry metadata about the partition. This results in a small amount of additional network bandwidth consumption that is generally a favorable trade-off when considered against periodically making requests for partition properties using the Event Hub client. It is set to False by default.

• starting_position (str, int, datetime.datetime or dict[str,Any]) – Start receiving from this event position if there is no checkpoint data for a partition. Checkpoint data will be used if available. This can be a a dict with partition ID as the key and position as the value for individual partitions, or a single value for all partitions. The value type can be str, int or datetime.datetime. Also supported are the values “-1” for receiving from the beginning of the stream, and “@latest” for receiving only new events.

• starting_position_inclusive (bool or dict[str,bool]) – Determine whether the given starting_position is inclusive(>=) or not (>). True for inclusive and False for exclusive. This can be a dict with partition ID as the key and bool as the value indicating whether the starting_position for a specific partition is inclusive or not. This can also be a single bool value for all starting_position. The default value is False.

• on_error (Callable[[PartitionContext, Exception]]) – The callback function that will be called when an error is raised during receiving after retry attempts are exhausted, or during the process of load-balancing. The callback takes two parameters: partition_context which contains partition information and error being the exception. partition_context could be None if the error is raised during the process of load-balance. The callback should be defined like: on_error(partition_context, error). The on_error callback will also be called if an unhandled exception is raised during the on_event callback.

• on_partition_initialize (Callable[[PartitionContext]]) – The callback function that will be called after a consumer for a certain partition finishes initialization. It would also be called when a new internal partition consumer is created to take over the receiving process for a failed and closed internal partition consumer. The callback takes a single parameter: partition_context which contains the partition information. The callback should be defined like: on_partition_initialize(partition_context).

• on_partition_close (Callable[[PartitionContext, CloseReason]]) – The callback function that will be called after a consumer for a certain partition is closed. It would be also called when error is raised during receiving after retry attempts are exhausted. The callback takes two parameters: partition_context which contains partition information and reason for the close. The callback should be defined like: on_partition_close(partition_context, reason). Please refer to CloseReason for the various closing reasons.

Return type

None

Example:

Receive events in batches from the EventHub.

    logger = logging.getLogger("azure.eventhub")

    async def on_event_batch(partition_context, event_batch):
        # Put your code here.
        # If the operation is i/o intensive, async will have better performance.
        logger.info(
            "{} events received from partition: {}".format(len(event_batch), partition_context.partition_id)
        )

    async with consumer:
        await consumer.receive_batch(
            on_event_batch=on_event_batch,
            starting_position="-1",  # "-1" is from the beginning of the partition.
        )

class azure.eventhub.aio.EventHubProducerClient(fully_qualified_namespace: str, eventhub_name: str, credential: TokenCredential, **kwargs)

The EventHubProducerClient class defines a high level interface for sending events to the Azure Event Hubs service.

Parameters

• fully_qualified_namespace (str) – The fully qualified host name for the Event Hubs namespace. This is likely to be similar to <yournamespace>.servicebus.windows.net

• eventhub_name (str) – The path of the specific Event Hub to connect the client to.

• credential (TokenCredential) – The credential object used for authentication which implements a particular interface for getting tokens. It accepts EventHubSharedKeyCredential, or credential objects generated by the azure-identity library and objects that implement the get_token(self, *scopes) method.

Keyword Arguments

• logging_enable (bool) – Whether to output network trace logs to the logger. Default is False.

• auth_timeout (float) – The time in seconds to wait for a token to be authorized by the service. The default value is 60 seconds. If set to 0, no timeout will be enforced from the client.

• user_agent (str) – The user agent that should be appended to the built-in user agent string.

• retry_total (int) – The total number of attempts to redo a failed operation when an error occurs. Default value is 3.

• idle_timeout (float) – Timeout, in seconds, after which this client will close the underlying connection if there is no activity. By default the value is None, meaning that the client will not shutdown due to inactivity unless initiated by the service.

• transport_type (TransportType) – The type of transport protocol that will be used for communicating with the Event Hubs service. Default is TransportType.Amqp.

• http_proxy (dict) – HTTP proxy settings. This must be a dictionary with the following keys: ‘proxy_hostname’ (str value) and ‘proxy_port’ (int value). Additionally the following keys may also be present: ‘username’, ‘password’.

• custom_endpoint_address (str) – The custom endpoint address to use for establishing a connection to the Event Hubs service, allowing network requests to be routed through any application gateways or other paths needed for the host environment. Default is None. The format would be like “sb://<custom_endpoint_hostname>:<custom_endpoint_port>”. If port is not specified in the custom_endpoint_address, by default port 443 will be used.

• connection_verify (str) – Path to the custom CA_BUNDLE file of the SSL certificate which is used to authenticate the identity of the connection endpoint. Default is None in which case certifi.where() will be used.

• enable_idempotent_partitions (bool) – Indicates whether or not the producer should enable idempotent publishing to the Event Hub partitions. If enabled, the producer will only be able to publish directly to partitions; it will not be able to publish to the Event Hubs gateway for automatic partition routing nor will it be able to use a partition key. Default is False.

• partition_config (dict[str, Union[PartitionPublishingConfiguration, dict]]) – The set of configurations that can be specified to influence publishing behavior specific to the configured Event Hub partition. These configurations are not necessary in the majority of scenarios and are intended for use with specialized scenarios, such as when recovering the state used for idempotent publishing. It is highly recommended that these configurations only be specified if there is a proven need to do so; Incorrectly configuring these values may result in an EventHubProducerClient instance that is unable to publish to the Event Hubs. These configurations are ignored when publishing to the Event Hubs gateway for automatic routing or when using a partition key. If a dict is provided as the value instead of a PartitionPublishingConfiguration object, it may contain the optional keys: ‘producer_group_id’ (int value), ‘owner_level’ (int value) and ‘starting_sequence_number’ (int value).

Example:

Create a new instance of the EventHubProducerClient.

import os
from azure.eventhub.aio import EventHubProducerClient, EventHubSharedKeyCredential

fully_qualified_namespace = os.environ['EVENT_HUB_HOSTNAME']
eventhub_name = os.environ['EVENT_HUB_NAME']
shared_access_policy = os.environ['EVENT_HUB_SAS_POLICY']
shared_access_key = os.environ['EVENT_HUB_SAS_KEY']

producer = EventHubProducerClient(fully_qualified_namespace=fully_qualified_namespace,
                                  eventhub_name=eventhub_name,
                                  credential=EventHubSharedKeyCredential(shared_access_policy, shared_access_key))

async close() → None

Close the Producer client underlying AMQP connection and links.

Return type

None

Example:

Close down the handler.

import os
from azure.eventhub.aio import EventHubProducerClient
from azure.eventhub import EventData

event_hub_connection_str = os.environ['EVENT_HUB_CONN_STR']
eventhub_name = os.environ['EVENT_HUB_NAME']

producer = EventHubProducerClient.from_connection_string(
    conn_str=event_hub_connection_str,
    eventhub_name=eventhub_name  # EventHub name should be specified if it doesn't show up in connection string.
)
try:
    event_data_batch = await producer.create_batch()
    while True:
        try:
            event_data_batch.add(EventData('Message inside EventBatchData'))
        except ValueError:
            # The EventDataBatch object reaches its max_size.
            # You can send the full EventDataBatch object and create a new one here.
            break
    await producer.send_batch(event_data_batch)
finally:
    # Close down the producer handler.
    await producer.close()

async create_batch(*, partition_id: Optional[str] = None, partition_key: Optional[str] = None, max_size_in_bytes: Optional[int] = None) → azure.eventhub._common.EventDataBatch

Create an EventDataBatch object with the max size of all content being constrained by max_size_in_bytes.

The max_size_in_bytes should be no greater than the max allowed message size defined by the service.

Parameters

• partition_id (str) – The specific partition ID to send to. Default is None, in which case the service will assign to all partitions using round-robin.

• partition_key (str) – With the given partition_key, event data will be sent to a particular partition of the Event Hub decided by the service. If both partition_id and partition_key are provided, the partition_id will take precedence. WARNING: Setting partition_key of non-string value on the events to be sent is discouraged as the partition_key will be ignored by the Event Hub service and events will be assigned to all partitions using round-robin. Furthermore, there are SDKs for consuming events which expect partition_key to only be string type, they might fail to parse the non-string value.

• max_size_in_bytes (int) – The maximum size of bytes data that an EventDataBatch object can hold. By default, the value is determined by your Event Hubs tier.

Return type

EventDataBatch

Example:

Create EventDataBatch object within limited size

    from azure.eventhub import EventData
    event_data_batch = await producer.create_batch()
    while True:
        try:
            event_data_batch.add(EventData('Message inside EventBatchData'))
        except ValueError:
            # The EventDataBatch object reaches its max_size.
            # You can send the full EventDataBatch object and create a new one here.
            break

classmethod from_connection_string(conn_str: str, *, eventhub_name: Optional[str] = None, logging_enable: bool = False, http_proxy: Optional[Dict[str, Union[str, int]]] = None, auth_timeout: float = 60, user_agent: Optional[str] = None, retry_total: int = 3, transport_type: Optional[TransportType] = None, **kwargs: Any) → EventHubProducerClient

Create an EventHubProducerClient from a connection string.

Parameters

conn_str (str) – The connection string of an Event Hub.

Keyword Arguments

• eventhub_name (str) – The path of the specific Event Hub to connect the client to.

• logging_enable (bool) – Whether to output network trace logs to the logger. Default is False.

• http_proxy (dict) – HTTP proxy settings. This must be a dictionary with the following keys: ‘proxy_hostname’ (str value) and ‘proxy_port’ (int value). Additionally the following keys may also be present: ‘username’, ‘password’.

• auth_timeout (float) – The time in seconds to wait for a token to be authorized by the service. The default value is 60 seconds. If set to 0, no timeout will be enforced from the client.

• user_agent (str) – The user agent that should be appended to the built-in user agent string.

• retry_total (int) – The total number of attempts to redo a failed operation when an error occurs. Default value is 3.

• idle_timeout (float) – Timeout, in seconds, after which this client will close the underlying connection if there is no activity. By default the value is None, meaning that the client will not shutdown due to inactivity unless initiated by the service.

• transport_type (TransportType) – The type of transport protocol that will be used for communicating with the Event Hubs service. Default is TransportType.Amqp.

• custom_endpoint_address (str) – The custom endpoint address to use for establishing a connection to the Event Hubs service, allowing network requests to be routed through any application gateways or other paths needed for the host environment. Default is None. The format would be like “sb://<custom_endpoint_hostname>:<custom_endpoint_port>”. If port is not specified in the custom_endpoint_address, by default port 443 will be used.

• connection_verify (str) – Path to the custom CA_BUNDLE file of the SSL certificate which is used to authenticate the identity of the connection endpoint. Default is None in which case certifi.where() will be used.

• enable_idempotent_partitions (bool) – Indicates whether or not the producer should enable idempotent publishing to the Event Hub partitions. If enabled, the producer will only be able to publish directly to partitions; it will not be able to publish to the Event Hubs gateway for automatic partition routing nor will it be able to use a partition key. Default is False.

• partition_config (dict[str, Union[PartitionPublishingConfiguration, dict]]) – The set of configurations that can be specified to influence publishing behavior specific to the configured Event Hub partition. These configurations are not necessary in the majority of scenarios and are intended for use with specialized scenarios, such as when recovering the state used for idempotent publishing. It is highly recommended that these configurations only be specified if there is a proven need to do so; Incorrectly configuring these values may result in an EventHubProducerClient instance that is unable to publish to the Event Hubs. These configurations are ignored when publishing to the Event Hubs gateway for automatic routing or when using a partition key. If a dict is provided as the value instead of a PartitionPublishingConfiguration object, it may contain the optional keys: ‘producer_group_id’ (int value), ‘owner_level’ (int value) and ‘starting_sequence_number’ (int value).

Return type

EventHubProducerClient

Example:

Create a new instance of the EventHubProducerClient from connection string.

import os
from azure.eventhub.aio import EventHubProducerClient
event_hub_connection_str = os.environ['EVENT_HUB_CONN_STR']
eventhub_name = os.environ['EVENT_HUB_NAME']
producer = EventHubProducerClient.from_connection_string(
    conn_str=event_hub_connection_str,
    eventhub_name=eventhub_name  # EventHub name should be specified if it doesn't show up in connection string.
)

async get_eventhub_properties() → Dict[str, Any]

Get properties of the Event Hub.

Keys in the returned dictionary include:

• eventhub_name (str)

• created_at (UTC datetime.datetime)

• partition_ids (list[str])

Return type

dict

Raises

EventHubError

async get_partition_ids() → List[str]

Get partition IDs of the Event Hub.

Return type

list[str]

Raises

EventHubError

async get_partition_properties(partition_id: str) → Dict[str, Any]

Get properties of the specified partition.

Keys in the properties dictionary include:

• eventhub_name (str)

• id (str)

• beginning_sequence_number (int)

• last_enqueued_sequence_number (int)

• last_enqueued_offset (str)

• last_enqueued_time_utc (UTC datetime.datetime)

• is_empty (bool)

Parameters

partition_id (str) – The target partition ID.

Return type

dict

Raises

EventHubError

async get_partition_publishing_properties(partition_id: str) → Dict[str, Any]

Get the information about the state of publishing for a partition as observed by the EventHubProducerClient. This data can always be read, but will only be populated with information relevant to the active features for the producer client.

• enable_idempotent_publishing (bool)

• partition_id (str)

• last_published_sequence_number (Optional[int])

• producer_group_id (Optional[int])

• owner_level (Optional[int]

Parameters

partition_id (str) – The target partition ID.

Return type

dict

async send_batch(event_data_batch: Union[azure.eventhub._common.EventDataBatch, List[azure.eventhub._common.EventData]], *, timeout: Union[int, float, None] = None, **kwargs) → None

Sends event data and blocks until acknowledgement is received or operation times out.

If you’re sending a finite list of EventData and you know it’s within the event hub frame size limit, you can send them with a send_batch call. Otherwise, use create_batch() to create EventDataBatch and add EventData into the batch one by one until the size limit, and then call this method to send out the batch.

Parameters

event_data_batch (Union[EventDataBatch, List[EventData]]) – The EventDataBatch object to be sent or a list of EventData to be sent in a batch. All EventData in the list or EventDataBatch will land on the same partition.

Keyword Arguments

• timeout (float) – The maximum wait time to send the event data. If not specified, the default wait time specified when the producer was created will be used.

• partition_id (str) – The specific partition ID to send to. Default is None, in which case the service will assign to all partitions using round-robin. A TypeError will be raised if partition_id is specified and event_data_batch is an EventDataBatch because EventDataBatch itself has partition_id.

• partition_key (str) – With the given partition_key, event data will be sent to a particular partition of the Event Hub decided by the service. A TypeError will be raised if partition_key is specified and event_data_batch is an EventDataBatch because EventDataBatch itself has partition_key. If both partition_id and partition_key are provided, the partition_id will take precedence. WARNING: Setting partition_key of non-string value on the events to be sent is discouraged as the partition_key will be ignored by the Event Hub service and events will be assigned to all partitions using round-robin. Furthermore, there are SDKs for consuming events which expect partition_key to only be string type, they might fail to parse the non-string value.

Return type

None

Raises

AuthenticationError ConnectError ConnectionLostError EventDataError EventDataSendError EventHubError ValueError TypeError

Example:

Asynchronously sends event data

    async with producer:
        event_data_batch = await producer.create_batch()
        while True:
            try:
                event_data_batch.add(EventData('Message inside EventBatchData'))
            except ValueError:
                # The EventDataBatch object reaches its max_size.
                # You can send the full EventDataBatch object and create a new one here.
                break
        await producer.send_batch(event_data_batch)

class azure.eventhub.aio.EventHubSharedKeyCredential(policy: str, key: str)

The shared access key credential used for authentication.

Parameters

• policy (str) – The name of the shared access policy.

• key (str) – The shared access key.

async get_token(*scopes, **kwargs)

class azure.eventhub.aio.CheckpointStore

CheckpointStore deals with the interaction with the chosen storage service.

It can list and claim partition ownerships as well as list and save checkpoints.

abstract async claim_ownership(ownership_list: Iterable[Dict[str, Any]]) → Iterable[Dict[str, Any]]

Tries to claim ownership for a list of specified partitions.

Parameters

ownership_list (Iterable[Dict[str,Any]]) – Iterable of dictionaries containing all the ownerships to claim.

Return type

Iterable[Dict[str,Any]], Iterable of dictionaries containing partition ownership information:

• fully_qualified_namespace (str): The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str): The name of the specific Event Hub the checkpoint is associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str): The name of the consumer group the ownership are associated with.

• partition_id (str): The partition ID which the checkpoint is created for.

• owner_id (str): A UUID representing the owner attempting to claim this partition.

• last_modified_time (UTC datetime.datetime): The last time this ownership was claimed.

• etag (str): The Etag value for the last time this ownership was modified. Optional depending on storage implementation.

abstract async list_checkpoints(fully_qualified_namespace: str, eventhub_name: str, consumer_group: str) → Iterable[Dict[str, Any]]

List the updated checkpoints from the chosen storage service.

Parameters

• fully_qualified_namespace (str) – The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str) – The name of the specific Event Hub the checkpoints are associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str) – The name of the consumer group the checkpoints are associated with.

Return type

Iterable[Dict[str,Any]], Iterable of dictionaries containing partition checkpoint information:

• fully_qualified_namespace (str): The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str): The name of the specific Event Hub the checkpoints are associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str): The name of the consumer group the checkpoints are associated with.

• partition_id (str): The partition ID which the checkpoint is created for.

• sequence_number (int): The sequence number of the EventData.

• offset (str): The offset of the EventData.

abstract async list_ownership(fully_qualified_namespace: str, eventhub_name: str, consumer_group: str) → Iterable[Dict[str, Any]]

Retrieves a complete ownership list from the chosen storage service.

Parameters

• fully_qualified_namespace (str) – The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str) – The name of the specific Event Hub the partition ownerships are associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str) – The name of the consumer group the ownerships are associated with.

Return type

Iterable[Dict[str, Any]], Iterable of dictionaries containing partition ownership information:

• fully_qualified_namespace (str): The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str): The name of the specific Event Hub the checkpoint is associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str): The name of the consumer group the ownership are associated with.

• partition_id (str): The partition ID which the checkpoint is created for.

• owner_id (str): A UUID representing the current owner of this partition.

• last_modified_time (UTC datetime.datetime): The last time this ownership was claimed.

• etag (str): The Etag value for the last time this ownership was modified. Optional depending on storage implementation.

abstract async update_checkpoint(checkpoint: Dict[str, Union[str, int, None]]) → None

Updates the checkpoint using the given information for the offset, associated partition and consumer group in the chosen storage service.

Note: If you plan to implement a custom checkpoint store with the intention of running between cross-language EventHubs SDKs, it is recommended to persist the offset value as an integer.

Parameters

checkpoint (Dict[str,Any]) –

A dict containing checkpoint information:

• fully_qualified_namespace (str): The fully qualified namespace that the Event Hub belongs to. The format is like “<namespace>.servicebus.windows.net”.

• eventhub_name (str): The name of the specific Event Hub the checkpoint is associated with, relative to the Event Hubs namespace that contains it.

• consumer_group (str): The name of the consumer group the checkpoint is associated with.

• partition_id (str): The partition ID which the checkpoint is created for.

• sequence_number (int): The sequence number of the EventData the new checkpoint will be associated with.

• offset (str): The offset of the EventData the new checkpoint will be associated with.

Return type

None

class azure.eventhub.aio.PartitionContext(fully_qualified_namespace: str, eventhub_name: str, consumer_group: str, partition_id: str, checkpoint_store: azure.eventhub.aio._eventprocessor.checkpoint_store.CheckpointStore = None)

Contains partition related context information.

A PartitionContext instance will be passed to the event, error and initialization callbacks defined when calling EventHubConsumerClient.receive(). Users can call update_checkpoint() of this class to persist checkpoint data.

async update_checkpoint(event: Optional[EventData] = None) → None

Updates the receive checkpoint to the given events offset.

Parameters

event (EventData) – The EventData instance which contains the offset and sequence number information used for checkpoint.

Return type

None

property last_enqueued_event_properties

The latest enqueued event information.

This property will be updated each time an event is received if the receiver is created with track_last_enqueued_event_properties set to True. The properties dict includes following information of the last enqueued event:

• sequence_number (int)

• offset (str)

• enqueued_time (UTC datetime.datetime)

• retrieval_time (UTC datetime.datetime)

Return type

dict or None