Usage instructions

The examples below assume you’ve installed the streamsets library, activated the library, and are inside a Python 3.4+ interpreter.

Use of the SDK begins by importing the library. For convenience, we tend to directly import the classes we need:

>>> from streamsets.sdk import DataCollector

Next, create an instance of streamsets.sdk.DataCollector, passing in the URL of your StreamSets Data Collector instance:

>>> data_collector = DataCollector('http://localhost:18630')


If no user credentials are passed to streamsets.sdk.DataCollector when it’s being instantiated, streamsets.sdk.sdc.DEFAULT_SDC_USERNAME and streamsets.sdk.sdc.DEFAULT_SDC_PASSWORD will be used for the username and password arguments, respectively. If your Data Collector instance is registered with StreamSets Control Hub, your Control Hub credentials need to be used to instantiate an instance of streamsets.sdk.ControlHub before it’s passed as an argument to streamsets.sdk.DataCollector instead:

>>> from streamsets.sdk import ControlHub
>>> control_hub = ControlHub('',
                             username=<your username>,
                             password=<your password>)
>>> data_collector = DataCollector('http://localhost:18630', control_hub=control_hub)

Creating a pipeline

Next, you can now get an instance of streamsets.sdk.sdc_models.PipelineBuilder:

>>> builder = data_collector.get_pipeline_builder()

We get streamsets.sdk.sdc_models.Stage instances from this builder by calling streamsets.sdk.sdc_models.PipelineBuilder.add_stage().

See the API reference for this method for details on the arguments it takes.

As shown in the first example, the simplest type of pipeline directs one origin into one destination. For this example, we do this with Dev Raw Data Source origin and Trash destination, respectively:

>>> dev_raw_data_source = builder.add_stage('Dev Raw Data Source')
>>> trash = builder.add_stage('Trash')

With streamsets.sdk.sdc_models.Stage instances in hand, we can connect them by using the >> operator, and then building a streamsets.sdk.sdc_models.Pipeline instance with the method:

>>> dev_raw_data_source >> trash
>>> pipeline ='My first pipeline')

Finally, to add this pipeline to your Data Collector instance, pass it to the streamsets.sdk.DataCollector.add_pipeline() method:

>>> data_collector.add_pipeline(pipeline)

Configuring stages

In practice, it’s rare to have stages in your pipeline that haven’t had some configurations changed from their default values. When using the SDK, the names to use when referring to these configuration properties can generally be inferred from the StreamSets Data Collector UI (e.g. Data Format becomes data_format), but they can also be directly inspected in a Python interpreter using the dir() built-in function on an instance of the streamsets.sdk.sdc_models.Stage instance:

>>> dir(dev_raw_data_source)

or by using Python’s built-in help() function:

>>> help(dev_raw_data_source)

With the attribute name in hand, you can read the value of the configuration:

>>> dev_raw_data_source.max_line_length

As for setting the value of the configuration, this can be done in one of two ways depending on your use case:

Single configurations

If you only have one or two configurations to update, you can set them using attributes of the streamsets.sdk.sdc_models.Stage instance. Continuing in the vein of our example:

>>> dev_raw_data_source.data_format = 'TEXT'
>>> dev_raw_data_source.raw_data = 'hi\nhello\nhow are you?'

Multiple configurations

For readability, it’s sometimes better to set all attributes simultaneously with one call to the streamsets.sdk.sdc_models.Stage.set_attributes() method:

>>> dev_raw_data_source.set_attributes(data_format='TEXT',
                                       raw_data='hi\nhello\nhow are you?')

Connecting stages

As described above, to connect the output of one stage to the input of another, simply use the >> operator between two streamsets.sdk.sdc_models.Stage instances:

>>> dev_raw_data_source >> trash

For stages with multiple outputs, simply use >> multiple times:

>>> file_tail = builder.add_stage('File Tail')
>>> file_tail >> trash_1
>>> file_tail >> trash_2

It is also possible to connect the output of one stage to the inputs of multiple stages, as in the image below:


To do this, put the streamsets.sdk.sdc_models.Stage instances to which you’ll be connecting the same output into a list before using the >> operator:

>>> trash_1 = builder.add_stage('Trash')
>>> trash_2 = builder.add_stage('Trash')
>>> dev_raw_data_source >> [trash_1, trash_2]


To connect the event lane of one stage to another, use the >= operator:

>>> dev_data_generator >> trash_1
>>> dev_data_generator >= trash_2

Error stages

To add an error stage, use streamsets.sdk.sdc_models.PipelineBuilder.add_error_stage():

>>> discard = builder.add_error_stage('Discard')