Streaming Iterators

When calling GoogleAdsService.search_stream, a streaming response iterator is returned. This iterator should remain in the same scope as the GoogleAdsService client while being used in order to avoid broken streams or segmentation faults. This is because the gRPC Channel object is garbage-collected once the open GoogleAdsService object goes out of scope. If the GoogleAdsService object is no longer in scope by the time the iteration occurs on the result of search_stream, the Channel object may already be destroyed, causing undefined behavior when the iterator attempts to retrieve the next value.

The following code demonstrates incorrect usage of streaming iterators:

def stream_response(client, customer_id, query):
    return client.get_service("GoogleAdsService", version="v16").search_stream(customer_id, query=query)

def main(client, customer_id):
    query = "SELECT campaign.name FROM campaign LIMIT 10"
    response = stream_response(client, customer_id, query=query)
    # Access the iterator in a different scope from where the service object was created.
    try:
        for batch in response:
            # Iterate through response, expect undefined behavior.

In the above code, the GoogleAdsService object is created within a different scope from where the iterator is accessed. As a result, the Channel object may be destroyed before the iterator consumes the entire response.

Instead, the streaming iterator should remain in the same scope as the GoogleAdsService client for as long as it is being used:

def main(client, customer_id):
    ga_service = client.get_service("GoogleAdsService", version="v16")
    query = "SELECT campaign.name FROM campaign LIMIT 10"
    response = ga_service.search_stream(customer_id=customer_id, query=query)
    # Access the iterator in the same scope as where the service object was created.
    try:
        for batch in response:
            # Successfully iterate through response.