# DataLab sample data

Open the [visual sample catalog](index.html) for descriptions and direct download links. These files are optional test fixtures; DataLab never loads them automatically. Download a fixture and choose it explicitly in the workspace import dialog.

# DataLab sample data catalog

DataLab always opens with an empty workspace. These files are optional test fixtures; none of them is automatically loaded or copied into application state. To try one, open **DataLab → Import data → From a file** and select a fixture from this folder. The application reads the chosen file into the active tab only. Export a dataset or workspace backup if you want to keep results.

## Suggested test path

1. Import `retail-orders.csv` to see the overview, table, profile, and chart builder together. It intentionally includes missing values, an exact duplicate, and a value that does not fit its apparent numeric column.
2. Open **Profile** and check the missing-value, unique-value, and quality summaries.
3. Open **Prepare**, try **Trim whitespace**, preview its effect, apply it, then use Undo. Try duplicate removal only after inspecting the duplicate signal.
4. Open **Visualize**, choose a category and numeric measure, and export the SVG or grouped data.
5. Export the cleaned file and a JSON workspace backup. Reload the app: the workspace should be empty again until you explicitly restore that backup.

## Fixtures

| File | Format | Use it to test |
|---|---|---|
| `retail-orders.csv` | CSV | Main end-to-end workflow; mixed types, missing values, duplicate row, malformed numeric value |
| `clean-small.csv` | CSV | Small clean baseline for sorting, charts, and export comparisons |
| `edge-cases.csv` | CSV | Quoted commas, embedded line breaks, missing fields, negative/zero values, and different null tokens |
| `regions.tsv` | TSV | Tab delimiter detection and ordinary text/number/date columns |
| `mixed-records.json` | JSON array | Different object keys, explicit nulls, booleans, numbers, and nested values |
| `events.jsonl` | JSON Lines | One JSON record per line, including optional fields |
| `chart-measure-cases.csv` | CSV | A real zero, a negative measure, and a category with only missing values; confirm the chart/table keeps those meanings distinct |
| `monthly-trend.csv` | CSV | Chronological periods with rising and falling measures; confirm line chart defaults to source order rather than numeric ranking |

All records are invented test data. They do not represent real people, businesses, customers, or transactions.

## Chart-specific test path

1. Import `chart-measure-cases.csv`, then open **Visualize** and group by `category` with `amount` as the measure. Choose Sum or Average.
2. Confirm `South` is a real zero, `East` remains negative, and `No measure` is labeled as missing rather than drawn as a zero bar.
3. Import `monthly-trend.csv`, select `month` and `revenue`, then choose a line chart. Its default order should remain Jan through Dec even though values rise and dip.

## Expected behavior notes

- A value such as `unknown` in `unit_price` is deliberately malformed, not a value to silently convert.
- Empty cells and explicit null-like tokens should appear in profile checks according to DataLab's documented missing-value rules.
- A duplicate row is present in `retail-orders.csv`; do not assume all duplicate-looking business keys are exact row duplicates.
- `mixed-records.json` contains heterogeneous object fields so the importer can exercise its union-of-keys behavior.
- The sample `events.jsonl` is plain UTF-8 text; import it as a `.jsonl` file.