TLDR

Eleventy has built-in support for custom data formats. To read data from a CSV file:

• Add it to the _data folder
• Install csv-parse using npm install --save-dev csv-parse
• Add the following to your Eleventy config:

.eleventy.js

import { parse } from 'csv-parse/sync'

export default function (eleventyConfig) {
+	eleventyConfig.addDataExtension('csv', (contents) => {
+		return parse(contents, { columns: true, skip_empty_lines: true })
+	})
}

This will add the data to Eleventy's data cascade using the CSV's filename as the key.

Background + alternate approach

I like to use Google Sheets as a CMS for one-off website projects. This means I author content in a sheet, export to a CSV file, and throw that into a static site generator to produce the HTML I need.

Eleventy doesn't have a built-in way to do that. It has a concept of global data files, but those only support json files out of the box.

However, you can roll your own support for CSV or any other data format using Javascript Data Files. The idea is to put a Javascript file into the _data folder that exports whatever data we need. Eleventy runs that file and adds the output to its global data object, making it available in your templates.

For example, if you have a CSV file called foo.csv, you can put the following Javascript file next to it:

_data/bar.js

import { parse } from 'csv-parse/lib/sync'
import { fs } from 'fs'

export default function () {
	const input = fs.readFileSync('./values.csv')
	const records = parse(input, { columns: true, skip_empty_lines: true })
	return records
}

This will make the data from the CSV availble under the bar key in Eleventy's data object, so template code like this will work as expected:

index.liquid

{% for row in bar %}
    {{ row.title }}
{% endfor %}

We could also use Eleventy's pagination feature to turn our data into individual pages.

The idea that you can run arbitrary Javascript pipe the results right into Eleventy's data object is pretty powerful. The documentation gives examples of fetching data from an API and exposing environment variables, but you could also do calculations, parse data in proprietary formats, or anything else you might need to do.