使用python的Apache Beam/Dataflow pub/sub侧输入

Question

我是Apache Beam的新手，所以我在以下场景中有点吃力：

• 使用流模式发布/子主题
• 转换以取出CustomerID
• 使用transform/pardo的并行PCollection根据pub/sub主题中接收的“customerid”从Firestore中获取数据（使用Side Input）
• ...

class getFirestoreUsers(beam.DoFn):
    def process(self, element, customerId):

        print(f'Getting Users from Firestore, ID: {customerId}')

        # Call function to initialize Database
        db = intializeFirebase()

        """ # get customer information from the database
        doc = db.document(f'Customers/{customerId}').get()
        customer = doc.to_dict() """
        usersList = {}

        # Get Optin Users
        try:
            docs = db.collection(
                f'Customers/{customerId}/DevicesWiFi_v3').where(u'optIn', u'==', True).stream()
            usersList = {user.id: user.to_dict() for user in docs}
        except Exception as err:
            print(f"Error: couldn't retrieve OPTIN users from DevicesWiFi")
            print(err)

        return([usersList])

def run(argv=None):
    """Build and run the pipeline."""
    parser = argparse.ArgumentParser()
    parser.add_argument(
        '--topic',
        type=str,
        help='Pub/Sub topic to read from')
    parser.add_argument(
        '--output',
        help=('Output local filename'))

    args, pipeline_args = parser.parse_known_args(argv)

    options = PipelineOptions(pipeline_args)
    options.view_as(SetupOptions).save_main_session = True
    options.view_as(StandardOptions).streaming = True

    p = beam.Pipeline(options=options)

    users = (p | 'Create chars' >> beam.Create([
        {
             "clientMac": "7c:d9:5c:b8:6f:38",
             "username": "Louis"
             },
        {
            "clientMac": "48:fd:8e:b0:6f:38",
            "username": "Paul"
        }
    ]))


    # Get Dictionary from Pub/Sub
    data = (p | 'Read from PubSub' >> beam.io.ReadFromPubSub(topic=args.topic)
            | 'Parse JSON to Dict' >> beam.Map(lambda e: json.loads(e))
            )

    # Get customerId from Pub/Sub information
    PcustomerId = (data | 'get customerId from Firestore' >>
                   beam.ParDo(lambda x: [x.get('customerId')]))
    PcustomerId | 'print customerId' >> beam.Map(print)

    # Get Users from Firestore
    custUsers = (users | 'Read from Firestore' >> beam.ParDo(
        getFirestoreUsers(), customerId=beam.pvalue.AsSingleton(PcustomerId)))
    custUsers | 'print Users from Firestore' >> beam.Map(print)

Answer 1

我不清楚示例代码中如何使用usersPCollection（因为element在process定义中没有处理）。我使用窗口重新排列了代码，并使用customer_id作为主要输入。

class GetFirestoreUsers(beam.DoFn):
  def setup(self):
    # Call function to initialize Database
    self.db = intializeFirebase()

  def process(self, element):
    print(f'Getting Users from Firestore, ID: {element}')

    """ # get customer information from the database
    doc = self.db.document(f'Customers/{element}').get()
    customer = doc.to_dict() """
    usersList = {}

    # Get Optin Users
    try:
        docs = self.db.collection(
            f'Customers/{element}/DevicesWiFi_v3').where(u'optIn', u'==', True).stream()
        usersList = {user.id: user.to_dict() for user in docs}
    except Exception as err:
        print(f"Error: couldn't retrieve OPTIN users from DevicesWiFi")
        print(err)

    return([usersList])



data = (p | 'Read from PubSub' >> beam.io.ReadFromPubSub(topic=args.topic)
          | beam.WindowInto(window.FixedWindow(60))
          | 'Parse JSON to Dict' >> beam.Map(lambda e: json.loads(e)))

# Get customerId from Pub/Sub information
customer_id = (data | 'get customerId from Firestore' >>
               beam.Map(lambda x: x.get('customerId')))
customer_id | 'print customerId' >> beam.Map(print)

# Get Users from Firestore
custUsers = (cutomer_id | 'Read from Firestore' >> beam.ParDo(
    GetFirestoreUsers())
custUsers | 'print Users from Firestore' >> beam.Map(print)

从你的评论来看：

当使用来自pub/sub的原始JSON数据运行“main”PCollection时，所需的数据（首先是customerID，然后是customers数据）尚未就绪

在阅读pub/sub主题时，你的意思是firestore中的数据还没有准备好？

您总是可以在主函数中将逻辑拆分为2个管道，然后一个接一个地运行它们。