This page documents the detailed steps to load CSV file from GCS into BigQuery using Dataflow to demo a simple data flow creation using Dataflow Tools for Eclipse. However it doesn’t necessarily mean this is the right use case for DataFlow.

Alternatively bq command line or programming APIs (Java, .NET, Python, Ruby, GO, etc.) can be used to ingest data into BigQuery; files in GCS can also be mapped directly as external table in BigQuery.


  • Eclipse 4.6 +
  • JDK 1.8+
  • GCP account
    • GCS and BigQuery enabled
    • Service account

For my environment, I am using Eclipse Photon Release (4.8.0). Make sure the Eclipse installation path has not space, otherwise you may encounter the following error:

Eclipse Photon - Eclipse Marketplace not launching


Detailed steps

Install Cloud SDK

Follow this page to install Cloud SDK.

Install Cloud Tools for Eclipse

Install Cloud Tools for Eclipse by following this page:

*Cloud developer tools are also available for other IDEs, incl. InteliJ, Visual Studio, ISE, PowerShell, etc.

Create a DataFlow project

Create a new project through New Project wizard.

Select Google Cloud Dataflow Java Project wizard. Click Next to continue.


Input the details for this project:


Setup account details:


Click Finish to complete the wizard.

Build the project

Run Maven Install to install the dependencies. You can do this through Run Configurations or Maven command line interfaces.


Once build is successful, you can create a run configuraiton for Dataflow pipeline:


For Pipeline Arguments tab, choose Direct Runner for now.


The output of the result will show in Console:


Create a pipeline to load CSV file in GCS to BigQuery

Create a sample file named sample.csv with the following content:


Upload the file into a bucket. For example, dataflow_examples/sample.csv

Copy the StarterPipeline class to create a new one named CsvToBQPipeline.

Refer to the sample data flows:

For this demo purpose, create the class with the following content:

  * Licensed to the Apache Software Foundation (ASF) under one
  * or more contributor license agreements.  See the NOTICE file
  * distributed with this work for additional information
  * regarding copyright ownership.  The ASF licenses this file
  * to you under the Apache License, Version 2.0 (the
  * "License"); you may not use this file except in compliance
  * with the License.  You may obtain a copy of the License at
  * Unless required by applicable law or agreed to in writing, software
  * distributed under the License is distributed on an "AS IS" BASIS,
  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  * See the License for the specific language governing permissions and
  * limitations under the License.
package tech.kontext.gcp;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

import org.apache.beam.runners.dataflow.DataflowRunner;
import org.apache.beam.runners.dataflow.options.DataflowPipelineOptions;
import org.apache.beam.sdk.Pipeline;
import org.apache.beam.sdk.options.PipelineOptions;
import org.apache.beam.sdk.options.PipelineOptionsFactory;
import org.apache.beam.sdk.transforms.Create;
import org.apache.beam.sdk.transforms.DoFn;
import org.apache.beam.sdk.transforms.MapElements;
import org.apache.beam.sdk.transforms.ParDo;
import org.apache.beam.sdk.transforms.SimpleFunction;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;


  * A starter example for writing Beam programs.
  * <p>
  * The example takes two strings, converts them to their upper-case
  * representation and logs them.
  * <p>
  * To run this starter example locally using DirectRunner, just execute it
  * without any additional parameters from your favorite development environment.
  * <p>
  * To run this starter example using managed resource in Google Cloud Platform,
  * you should specify the following command-line options:
  * --project=<YOUR_PROJECT_ID>
  * --stagingLocation=<STAGING_LOCATION_IN_CLOUD_STORAGE> --runner=DataflowRunner
public class CsvToBQPipeline {
     private static final Logger LOG = LoggerFactory.getLogger(CsvToBQPipeline.class);
     private static String HEADERS = "ID,Code,Value,Date";

    public static class FormatForBigquery extends DoFn<String, TableRow> {

        private String[] columnNames = HEADERS.split(",");

         public void processElement(ProcessContext c) {
             TableRow row = new TableRow();
             String[] parts = c.element().split(",");

            if (!c.element().contains(HEADERS)) {
                 for (int i = 0; i < parts.length; i++) {
                     // No typy conversion at the moment.
                     row.set(columnNames[i], parts[i]);

        /** Defines the BigQuery schema used for the output. */

        static TableSchema getSchema() {
             List<TableFieldSchema> fields = new ArrayList<>();
             // Currently store all values as String
             fields.add(new TableFieldSchema().setName("ID").setType("STRING"));
             fields.add(new TableFieldSchema().setName("Code").setType("STRING"));
             fields.add(new TableFieldSchema().setName("Value").setType("STRING"));
             fields.add(new TableFieldSchema().setName("Date").setType("STRING"));

            return new TableSchema().setFields(fields);

    public static void main(String[] args) throws Throwable {
         // Currently hard-code the variables, this can be passed into as parameters
         String sourceFilePath = "gs://dataflow_examples/sample.csv";
         String tempLocationPath = "gs://poc-dataflow-temp/bq";
         boolean isStreaming = false;
         TableReference tableRef = new TableReference();
         // Replace this with your own GCP project id

        PipelineOptions options = PipelineOptionsFactory.fromArgs(args).withValidation().create();
         // This is required for BigQuery
         Pipeline p = Pipeline.create(options);

        p.apply("Read CSV File",
                 .apply("Log messages", ParDo.of(new DoFn<String, String>() {
                     public void processElement(ProcessContext c) {
               "Processing row: " + c.element());
                 })).apply("Convert to BigQuery TableRow", ParDo.of(new FormatForBigquery()))
                 .apply("Write into BigQuery",
                                 .withWriteDisposition(isStreaming ? BigQueryIO.Write.WriteDisposition.WRITE_APPEND
                                         : BigQueryIO.Write.WriteDisposition.WRITE_TRUNCATE));;


And then run the code through local DirectRunner (set through Pipeline Arguments tab).


You may get errors like the following if your temp location is different from your BigQuery location:

"errorResult" : {
       "message" : "Cannot read and write in different locations: source: asia, destination: asia-northeast1",
       "reason" : "invalid"

The logs will also be printed to the Output window of Eclipse:


View the job in Console

You can also run through DataflowRunner (set through Pipeline Arguments tab). The job will then be uploaded into Dataflow in GCP.


View the jobs status through cloud SDK shell

Run the following command in Cloud SDK Shell or Client Tools for PowerShell:

gcloud dataflow jobs list

The result will be similar as the followings:


Verify the result in BigQuery

Once data is loaded, you can run the following query to query it:

SELECT * FROM `gcp-project-id.sample_ds.sample` LIMIT 1000

Remember to change the project ID.


Error debugs

Exception in thread "main" java.lang.RuntimeException: Failed to construct instance from factory method DataflowRunner#fromOptions

To resolve this issue, set the staging location to a folder in a bucket:


info Last modified by Raymond at 2 years ago * This page is subject to Site terms.

More from Kontext

local_offer teradata local_offer python local_offer Java

visibility 396
thumb_up 0
access_time 3 months ago

Python JayDeBeApi module allows you to connect from Python to Teradata databases using Java JDBC drivers. In article Connect to Teradata database through Python , I showed ho...

open_in_new Python Programming

local_offer hadoop local_offer hive local_offer Java

visibility 477
thumb_up 1
access_time 3 months ago

When I was configuring Hive 3.0.0 in Hadoop 3.2.1 environment, I encountered the following error: Exception in thread "main" java.lang.NoSuchMethodError:;Ljava/lang/Object;)V Ro...

open_in_new Hadoop

local_offer sqlite local_offer python local_offer Java

visibility 96
thumb_up 0
access_time 4 months ago

To read data from SQLite database in Python, you can use the built-in sqlite3 package . Another approach is to use SQLite JDBC driver via  ...

open_in_new Python Programming

local_offer Java local_offer python local_offer SQL Server

visibility 1521
thumb_up 0
access_time 4 months ago

In my previous article  Connect to SQL Server via JayDeBeApi in Python , I showed examples of u...

open_in_new Python Programming

info About author

comment Comments (0)

comment Add comment

Please log in or register to comment.

account_circle Log in person_add Register

Log in with external accounts

No comments yet.

Dark theme mode

Dark theme mode is available on Kontext.

Learn more arrow_forward

Kontext Column

Created for everyone to publish data, programming and cloud related articles. Follow three steps to create your columns.

Learn more arrow_forward