Google Cloud Dataflow, BigQueryIO and NullPointerException on TableRow.get

0
votes

I'm new to GC Dataflow and didn't find a relevant answer here. Apologies if I should have found this already answered.

I'm trying to create a simple pipeline using the v2.0 SDK and am having trouble reading data into my PCollection using BigQueryIO. I am using the .withQuery method and I have tested the query in the BigQuery interface and it seems to be working fine. The initial PCollection seems to get created without any issues, but when I think setup a simple ParDo function to convert the values from the TableRow into a PCollection I am getting a NullPointerException on the line of code that does the .get on the TableRow object.

Here is my code. (I'm probably missing something simple. I'm a total newbie at Pipeline programming. Any input would be most appreciated.)

public class ClientAutocompletePipeline {
    private static final Logger LOG = LoggerFactory.getLogger(ClientAutocompletePipeline.class);


    public static void main(String[] args) {
        //  create the pipeline  
        Pipeline p = Pipeline.create(
                PipelineOptionsFactory.fromArgs(args).withValidation().create());

        // A step to read in the product names from a BigQuery table
        p.apply(BigQueryIO.read().fromQuery("SELECT name FROM [beaming-team-169321:Products.raw_product_data]"))

        .apply("ExtractProductNames", ParDo.of(new DoFn<TableRow, String>() {
            @ProcessElement
            public void processElement(ProcessContext c) {
                // Grab a row from the BigQuery Results
                TableRow row = c.element();

                // Get the value of the "name" column from the table row.
                //NOTE: This is the line that is giving me the NullPointerException 
                String productName = row.get("name").toString();

                // Make sure it isn't empty
                if (!productName.isEmpty()) {
                    c.output(productName);
                }
            }
        }))

The query definitely works in the BigQuery UI and the column called "name" is returned when I test the query. Why am I getting a NullPointerException on this line:

String productName = row.get("name").toString();

Any ideas?

3
google-bigquerygoogle-cloud-dataflow
Are all of the values in the name column guaranteed not to be null? - Elliott Brossard
If you run SELECT name FROM [beaming-team-169321:Products.raw_product_data] where name is null in BigQuery you'll see that there are null values. So, you need to take this into consideration in your pipeline. - Graham Polley
Well, now that you mention it, that does make perfect sense. I guess I was under the incorrect impression that something was causing them all to be null because I had some error in my code, but I guess that might not have been the case. Thanks for replying! - MrSimmonsSr

3 Answers

0
votes

This is a common problem when working with BigQuery and Dataflow (most likely the field is indeed null). If you are ok with using Scala, you could take a look at Scio (which is a Scala DSL for Dataflow) and its BigQuery IO.

0
votes

Just make your code null safe. Replace this:

String productName = row.get("name").toString();

With something like this:

String productName = String.valueOf(row.get("name"));

0
votes

I think I'm late for this but you can do something like if(row.containsKey("column-name")). This will basically tell you if the field is null or not. In BigQuery what happens is, while reading data, if a column value is null, it is not available as a part of that particular TableRow. Hence, you are getting that error. You can also do something like if(null == row.get("column-name")) to check if the field is null or not.