Collecting output from Apache Beam pipeline and displaying it to console

Question

I have been working on Apache Beam for a couple of days. I wanted to quickly iterate on the application I am working and make sure the pipeline I am building is error free. In spark we can use sc.parallelise and when we apply some action we get the value that we can inspect.

Similarly when I was reading about Apache Beam, I found that we can create a PCollection and work with it using following syntax

with beam.Pipeline() as pipeline:
    lines = pipeline | beam.Create(["this is test", "this is another test"])
    word_count = (lines 
                  | "Word" >> beam.ParDo(lambda line: line.split(" "))
                  | "Pair of One" >> beam.Map(lambda w: (w, 1))
                  | "Group" >> beam.GroupByKey()
                  | "Count" >> beam.Map(lambda (w, o): (w, sum(o))))
    result = pipeline.run()

I actually wanted to print the result to console. But I couldn't find any documentation around it.

Is there a way to print the result to console instead of saving it to a file each time?

user_5299171 · Answer

After exploring furthermore and understanding how I can write testcases for my application I figure out the way to print the result to console. Please not that I am right now running everything to a single node machine and trying to understand functionality provided by apache beam and how can I adopt it without compromising industry best practices.

So, here is my solution. At the very last stage of our pipeline we can introduce a map function that will print result to the console or accumulate the result in a variable later we can print the variable to see the value

import apache_beam as beam

# lets have a sample string
data = ["this is sample data", "this is yet another sample data"]

# create a pipeline
pipeline = beam.Pipeline()
counts = (pipeline | "create" >> beam.Create(data)
    | "split" >> beam.ParDo(lambda row: row.split(" "))
    | "pair" >> beam.Map(lambda w: (w, 1))
    | "group" >> beam.CombinePerKey(sum))

# lets collect our result with a map transformation into output array
output = []
def collect(row):
    output.append(row)
    return True

counts | "print" >> beam.Map(collect)

# Run the pipeline
result = pipeline.run()

# lets wait until result a available
result.wait_until_finish()

# print the output
print output

user_6586216 · Answer

Follow an example from pycharm Edu

import apache_beam as beam

class LogElements(beam.PTransform):
    class _LoggingFn(beam.DoFn):

        def __init__(self, prefix=''):
            super(LogElements._LoggingFn, self).__init__()
            self.prefix = prefix

        def process(self, element, **kwargs):
            print self.prefix + str(element)
            yield element

    def __init__(self, label=None, prefix=''):
        super(LogElements, self).__init__(label)
        self.prefix = prefix

    def expand(self, input):
        input | beam.ParDo(self._LoggingFn(self.prefix))

class MultiplyByTenDoFn(beam.DoFn):

    def process(self, element):
        yield element * 10

p = beam.Pipeline()

(p | beam.Create([1, 2, 3, 4, 5])
   | beam.ParDo(MultiplyByTenDoFn())
   | LogElements())

p.run()

Output

10
20
30
40
50
Out[10]:

user_10808155 · Answer

Maybe logging info instead of print?

def _logging(elem):
    logging.info(elem)
    return elem

P | "logging info" >> beam.Map(_logging)

user_6199292 · Answer

I know it isn't what you asked for but why don't you store it to a text file? It's always better than printing it via stdout and it isn't volatile

Collecting output from Apache Beam pipeline and displaying it to console

5 Answers