# CSV data files

**Keep CSV files clean** — a header row and data rows, nothing else. Put deck
configuration and templates in the `yaml role=tag` block's `data-config`, not in
the CSV. This keeps data portable and presentation in one place. See
`reference.md` for the element and role catalog.

```csv
label,body,name,type
1,"### 火球术卷轴

一张泛黄的羊皮纸。",火球术卷轴,消耗品
```

- `label` — unique row id (number or text).
- `body` — main content, supports markdown.
- Any other columns are custom fields.
- `{{prop}}` in `body` interpolates another column from the same row (or a
  random row when the table uses `{remix}`).
- Wrap values containing `,`, `"`, or newlines in double quotes; escape `"` as
  `""`. Lines beginning with `#` are comments.

## Front matter (avoid)

The parser also accepts YAML front matter delimited by `---`, which can declare
`fields`, a `deck:` block, and arbitrary keys merged into every row. **Prefer
not to use it** — it mixes presentation into the data file. Use the YAML tag
block instead.

```csv
---
fields:
  - name: name
    description: 卡牌名称
deck:
  size: 63x88
  grid: 5x5
game_system: D&D 5e
---
label,body,name,type
1,"### 火球术卷轴",火球术卷轴,消耗品
```