puppet-canvas is a Puppeteer backed implementation of HTML Canvas API for NodeJS.
$ npm install puppet-canvas --save
Following example draws a house and grabs the data url of the image.
import { createCanvas, close } from 'puppet-canvas';
// OR const { createCanvas, close } = require('puppet-canvas')
async(() => {
const canvas = await createCanvas(400, 400);
const ctx = await canvas.getContext('2d');
// Draw a house
ctx.lineWidth = 10;
ctx.strokeRect(75, 140, 150, 110);
ctx.fillRect(130, 190, 40, 60);
ctx.moveTo(50, 140);
ctx.lineTo(150, 60);
ctx.lineTo(250, 140);
ctx.closePath();
ctx.stroke();
// Get the image as a data url
const dataUrl = await canvas.toDataURL();
// Release
await close();
})();
To use external images in your canvas, first load the image using the loadImage
method.
import { createCanvas, loadImage } from 'puppet-canvas';
const canvas = await createCanvas(400, 400);
const ctx = await canvas.getContext('2d');
const image = await loadImage('https://....', canvas);
await ctx.drawImage(image, 100, 100);
Any non-serializable object returned by puppeteer is automatically proxied as well. So one can use/pass more complex objects like a gradient:
const gradient = await ctx.createLinearGradient(20, 0, 220, 0);
gradient.addColorStop(0, 'green');
gradient.addColorStop(.5, 'cyan');
gradient.addColorStop(1, 'green');
ctx.fillStyle = gradient;
ctx.fillRect(20, 20, 200, 100);
When using a custom font for rendering text, you can ensure the font is loaded by calling the loadFont
methods.
import { createCanvas, loadFont } from 'puppet-canvas';
const canvas = await createCanvas(400, 400);
const ctx = await canvas.getContext('2d');
await loadFont(
'Bangers',
'https://fonts.gstatic.com/s/bangers/v12/FeVQS0BTqb0h60ACH55Q2J5hm24.woff2',
canvas
);
ctx.font = `bold 48px Bangers`;
ctx.fillText('Hello world', 50, 100);
puppet-canvas creates a canvas on a puppeteer instance, and exposes the API via a JavaScript Proxy.
A side effect of this is that all calls, essentially become async
. For normal drawing, one doesn't need to await
each command unless a value is being returned.
A proxied solution is somewhat better than alternate ones because, firstly, the rendering is exactly what the browser would have rendered (Chrome). Secondly, this is mostly future-proof since all methods are just proxied to the actual instance. So, any new API added to the Canvas, should automagically work.
Though a proxied implementation is simpler, smaller, and more flexible, it can be slower than alternative native implementations. For example, manipulating lots of pixels, one pixel at a time, may be slow and not recommended.
The other shortcoming is from a dev experience, everything is an async
call. So it creates more code for getting child object and properties. The following code:
const imageDataLength = ctx.createImageData(10, 10).data.length;
has to be refactored as
const imageData = await ctx.createImageData(10, 10);
const imageDataLength = await (await imageData.data).length;
Yes, you can and for some cases it may even be a better dev experience.
puppet-canvas provides a simple abstraction layer so you don't have to jump in and out of Puppeteer's execution context all the time. You treat the canvas very similar to canvas on the web, and not worry much about anything else.
Creates a canvas instance with the specified width and height (in pixels)
Say, you want to use this with an existing instance of puppeteer, you can pass in the ElementHandle of the canvase in your page.
Close associated puppeteer instance. Usually called at the end.
Release the canvas instance, if you do not want puppet-canvas to proxy it anymore, but still want to keep the canvas instance around
screenshotCanvas(canvas: HTMLCanvasElement, options?: ScreenshotOptions) => Promise<string | Buffer>
Take a screenshot of the canvas. The method optionally takes in ScreenshotOptions
which are the same options as described in Puppeteer screenshot method
Load a font for the canvas element to use. name is the font-family
name of the font. url is the url to the font file (like a woff
file)
Load an image from a URL that could be used by the canvas, e.g. for drawing the image on the canvas.