From Spreadsheet to Test Fixtures: CSV to JSON for CI/CD Pipelines

Your QA team has two hundred test cases in a spreadsheet. They update it constantly. Your test suite needs JSON fixtures, but nobody wants to retype two hundred rows every time someone fixes a typo. A CSV to JSON converter turns the spreadsheet into a JSON fixture on every build, the test suite reads the JSON, and the QA team keeps working in Excel. Everyone stops fighting about file types and gets back to work.

The Spreadsheet-to-Fixture Pipeline

The pattern is simple: the QA team owns a CSV file in version control, the build pipeline runs CSV to JSON conversion, and the test suite consumes the JSON output. No human touches the JSON. No human retypes data. When QA adds a row to the spreadsheet, it's automatically in the next test run. The counter-intuitive part is that spreadsheets are better than JSON files for non-developers, even though JSON is better for tests. The fix isn't to make everyone use JSON — it's to make the conversion automatic so nobody has to.

How to Make It Reliable

Three rules. First, keep the CSV header row as your schema — that's where the JSON keys come from. Second, commit the converted JSON to a build artifact directory, not to the source tree, so it's always fresh. Third, validate the JSON with a JSON formatter as part of the conversion step, so a typo in the CSV doesn't quietly produce malformed test data. For testing edge cases — what happens when a cell is empty, what happens with a quote in a quote — a regex tester can help verify the parsing patterns against the CSV before they hit the pipeline. The whole thing runs in CI on every PR, takes seconds, and your QA team never has to touch JSON.

Convert at the Edge, Not in the Code

We covered legacy migrations in our guide to CSV to JSON for legacy systems; the CI/CD version is the same idea applied to living data. Let the humans use the tool they like, convert it at the boundary, and the pipeline stays clean. The cost is one tiny conversion script in CI, the benefit is that QA and engineering stop arguing about file formats and start testing the actual feature. Most teams that try it never go back, because the alternative — retype everything by hand every time — is exactly the kind of manual work that breaks down in production.