bertha-client

bertha-client

A client library for fetching data from Bertha.

Why use this instead of fetching Bertha URLs with fetch or axios?

• takes care of the URL syntax
• gives more helpful error messages when things go wrong
• provides an easy way to transform sheets into objects

Installation

• yarn add bertha-client or npm install bertha-client

Usage

import * as bertha from 'bertha-client'; // or const bertha = require('bertha-client');

bertha.get(spreadsheetKey, ['someSheet', 'anotherSheet|object']).then((data) => {
  console.log(data);
  // { someSheet: [...], anotherSheet: [...] }
});

API

bertha.get(spreadsheetKey, sheetNames, [options])

Fetches the sheet and returns a promise for the response data.

spreadsheetKey

String (required). A valid Google spreadsheet key, or a full Google Spreadsheet URL or Bertha URL.

sheetNames

Array of strings (required). The names of the sheets you want to get. A sheet name may be appended with |object to apply the "object" transformation.

// example
const sheetNames = [
  'polls',
  'authors',
  'copy|object', // applies the "object" transformation
]

options

Plain object (optional).

• republish (default: false) – set to true if you want Bertha to trigger a republish.
• query (default: undefined) – an optional object of {[name]: value} to append to the URL as query parameters.

bertha.createURL(spreadsheetKey, sheetNames, [options])

Identical API to bertha.get(), but simply returns a URL string.

new bertha.Poller(spreadsheetKey, sheetNames, [options])

Inspired by ft-poller (but with a simpler API), this constructs a poller that fetches the data once a minute. When you call poller.getData(), it resolves immediately with the latest data, even if that data is stale. An internal timer refreshes the data periodically so it's never too old.

const poller = new bertha.Poller(spreadsheetKey, ['someSheet', 'anotherSheet']);

await poller.start();

// poller is now primed, and will continue to refresh approx. every minute

poller.getData(); // returns the latest data

Constructor takes the same arguments as bertha.get(), except that options may include the following additional property:

• refreshAfter (default: 60000) – the number of milliseconds to leave between refreshes. Minimum 1000. Note this is a timeout (not a regular interval) that begins after each attempted refresh is complete. So with the default settings, the real 'interval' will be more like 61 seconds due to the duration of each refresh attempt. This is to avoid possible problems caused by very slow requests overlapping.
• attempts (default: 3) – each fetch will be tried up to this many times before the refresh is considered failed.
• retryAfter (default: 2000) – the number of milliseconds to leave before retrying a failed fetch.

poller.start()

Returns a promise that resolves (with undefined) when the first fetch has succeeded and the poller is primed with data. It will reject if the first refresh (including retries) fails.

poller.getData()

Returns the latest available data immediately. Throws an error if the poller has not finished starting up.

poller.stop()

Stops any new fetches from occurring (but any in-progress fetch will finish).

bertha.parseKey(url, [silent])

Takes any valid Google Sheets URL or Bertha URL and returns a plain spreadsheet key. If the string you pass is already a plain key, it returns it unchanged.

If you set silent to true, invalid input will result in the function returning null instead of throwing an error.

bertha.domains

An array of known Bertha domains.

Response data

The data from bertha.get() is always returned as a plain JavaScript object (even if there is only one sheet). The key names correspond with the sheet names.

The |object transformation

If you append a sheet name with |object, that sheet will be transformed into a plain object (instead of an array), using the sheet's name and value columns as key paths and values, respectively. Any other columns are discarded. If the name or value columns are missing, an error is thrown.

Example spreadsheet:

name	value
foo	hiya
bar	123
someone.age	50
someone.name.first	Bob
someone.name.last	Hoskins

// result of "|object" transform
{
  foo: 'hiya',
  bar: 123,
  someone: {
    age: 50,
    name: {
      first: 'Bob',
      last: 'Hoskins',
    }
  },
}

NB. the Bertha server automatically converts certain values ("y" or "yes" becomes true etc.) – see Bertha's docs. This conversion is not controlled by bertha-client – you'll need to specify a non-default column transform such as ..str if you want to control it.

Development

Local setup

Recommended approach:

• Clone this repo and run npm install.
• Run npm run build -- --watch – this will compile files continually from src to dist using Babel.
• In another terminal, run npm test -- --watch to run Jest continually against files in dist.

Publishing to npm

This module is automatically published to npm via CircleCI whenever the master branch contains a higher version string than the latest published version.

To publish a new version:

• npm version patch (or replace patch with minor or major as appropriate)
• git push && git push --tags

Changelog

[4.0.1] 2017-09-14

Changed

• Now requires Node 8 or above. This makes the compiled code smaller and easier to debug, as it has real async functions and doesn't need runtime dependencies.
• Flow libdef now declares bertha-client as an ES6 module instead of a CommonJS one.

Added

• Poller class.