Reading Test Data from CSV

Inline Object[][] works for 5-10 rows. But when you have 50+ test data combinations, or when the business team needs to add test cases, you move data to external files. CSV is the simplest format — just commas and newlines. No libraries needed. Every spreadsheet app can export CSV. And it plays nicely with Git version control.

The CSV File

Create a folder src/test/resources/test-data/ in your project. This is the standard location for test data files.

src/test/resources/test-data/search-terms.csvtext

searchTerm,expectResults,minCount
laptop,true,3
headphones,true,2
smartphone,true,5
xyznoexist,false,0
,false,0
LAPTOP,true,3
"laptop case",true,1
<script>alert(1)</script>,false,0

Reading CSV in a DataProvider

ShoppingSearchCsvTest.javajava

import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import org.testng.annotations.DataProvider;
import org.testng.annotations.Test;
import org.testng.Assert;

public class ShoppingSearchCsvTest extends BaseTest {

    @DataProvider(name = "searchDataFromCsv")
    public Object[][] readSearchData() throws IOException {
        List<Object[]> data = new ArrayList<>();
        String csvFile = "src/test/resources/test-data/search-terms.csv";

        try (BufferedReader br = new BufferedReader(
                new FileReader(csvFile))) {
            String line;
            boolean isHeader = true;

            while ((line = br.readLine()) != null) {
                // Skip the header row
                if (isHeader) {
                    isHeader = false;
                    continue;
                }

                // -1 keeps trailing empty strings
                String[] values = line.split(",", -1);
                String searchTerm = values[0];
                boolean expectResults = Boolean.parseBoolean(values[1]);
                int minCount = Integer.parseInt(values[2]);

                data.add(new Object[] {
                    searchTerm, expectResults, minCount
                });
            }
        }

        return data.toArray(new Object[0][]);
    }

    @Test(dataProvider = "searchDataFromCsv")
    public void testShoppingSearch(String searchTerm,
                                    boolean expectResults,
                                    int minCount) {
        navigateTo("/shopping");
        ShopHomePage homePage = new ShopHomePage(driver);
        CatalogPage catalog = homePage.searchFor(searchTerm);

        if (expectResults) {
            Assert.assertTrue(catalog.getProductCount() >= minCount,
                "Search for '" + searchTerm
                + "' should return at least " + minCount
                + " results but got " + catalog.getProductCount());
        } else {
            Assert.assertTrue(
                catalog.isNoResultsDisplayed()
                || catalog.getProductCount() == 0,
                "Search for '" + searchTerm
                + "' should return no results");
        }
    }
}

Line-by-Line Walkthrough

BufferedReader opens the CSV file for reading line by line
The boolean flag isHeader skips the first row (column names)
line.split(",", -1) splits on commas — the -1 keeps empty trailing values
Each value is parsed into the correct type (String stays as-is, boolean/int get parsed)
data.add() collects each row as an Object array

The split(",", -1) is critical. Without -1, if your last column is empty (like an empty searchTerm), Java drops it silently. You get an ArrayIndexOutOfBoundsException and waste an hour debugging. Always use -1 with split.

When CSV Breaks Down

Data contains commas — "New York, NY" breaks column splitting. Use a CSV library like OpenCSV for complex cases.
Data contains newlines — multi-line cell values break BufferedReader.readLine().
Data is nested — JSON is better for hierarchical structures.
Business team wants formatting/colors — they want Excel, not CSV.

Q: How do you read test data from a CSV file in your framework?

A: I use BufferedReader to read the CSV file line by line. I skip the header row, split each line by comma using split(",", -1) to preserve empty values, parse the values into the correct types, and collect them into a List<Object[]>. The DataProvider method converts this List to Object[][] and returns it. For CSV files with commas inside values, I use OpenCSV library instead of manual splitting.

Keep your CSV files in src/test/resources/test-data/. This folder gets included in the classpath automatically by Maven. Name files descriptively: login-data.csv, search-terms.csv, transfer-scenarios.csv.

Key Point: CSV is the simplest external data source. Use split(",", -1) to avoid losing empty values.

Previous Up NextReading Test Data from Excel

Chapter 4: Data-Driven Testing