Spiderable middleware
Search engines and social networks — such as Google, Facebook, Twitter, Yahoo, Bing, and many others — are constantly crawling your website to index its content. However, if your site is built using a modern JavaScript framework (for example, React, Preact, Vue, Svelte, Angular, Backbone, Ember, Meteor, Blaze), it serve only a basic HTML skeleton with script tags instead of fully rendered page content. This limitation negatively impacts website's SEO score, performance, and user experience when links shared by users around the web. The mission of spiderable-middleware
and ostr.io
is to enhance SEO and performance for modern JavaScript websites.
Why Pre-render?
- 🕸 Execute JavaScript, — get rendered HTML page and its content;
- 🏃♂️ Boost response rate and decrease response time with caching;
- 🚀 Optimized HTML markup for the best SEO score;
- 🖥 Support PWAs and SPAs;
- 📱 Support mobile-like crawlers;
- 💅 Support
styled-components
; - ⚡️ Support AMP (Accelerated Mobile Pages);
- 🤓 Works with
Content-Security-Policy
and other complex front-end security rules; - 📦 This package shipped with types and TS examples;
- ❤️ Search engines and social network crawlers love straightforward and pre-rendered pages;
- 📱 Consistent link previews in messaging apps, like iMessage, Messages, Facebook, Slack, Telegram, WhatsApp, Viber, VK, Twitter, and other apps;
- 💻 Image, title, and description previews for posted links at social networks, like Facebook, Twitter, Instagram, and other social networks.
ToC
- Installation
- Basic usage
- Meteor.js usage
- Return genuine status code
- Speed-up rendering
- Detect request from Prerendering engine during runtime
- JavaScript redirects
- AMP Support
- Rendering Endpoints
- API
- Debugging
- Running Tests
About Package
This package acts as middleware and intercepts requests to a Node.js application from web crawlers. All requests proxy passed to the Prerendering Service, which returns static, rendered HTML.
This is SERVER only package. For NPM make sure it's imported only in Node.js.
We made this package with developers in mind. It's well written in a very simple way, hackable, and easily tunable to meet all projects needs, it can be used to turn dynamic pages into rendered, cached, and lightweight static pages, just set botsUA
to ['.*']
. This is the best option to offload servers unless a website gets updated more often than once in 4 hours.
- Note: This package proxies real HTTP headers and response code, to reduce overwhelming requests, try to avoid HTTP-redirect headers, like
Location
. Read how to return genuine status code and handle JS-redirects - Note: This is server only package. This package should be imported/initialized only within server code base
This middleware was tested and works like a charm with:
- express:
example.js
,example.ts
- connect:
example.js
,example.ts
- vanilla http(s) server:
example.js
,example.ts
- Nginx: example
- Apache: example
- See all examples
All other frameworks which follow Node's middleware convention - will work too.
This package was originally developed for ostr.io service. But it's not limited to, and can proxy-pass requests to any other rendering-endpoint.
Installation
Install spiderable-middleware
package from NPM:
# using npmjs npm install spiderable-middleware --save # using yarn yarn add spiderable-middleware
Usage
Get ready in a few lines of code
Basic usage
See usage examples in .js
and .ts
for quick copy-paste experience.
Start with adding fragment
meta-tag to HTML template, head, or page:
1<html> 2 <head> 3 <meta name="fragment" content="!"> 4 <!-- ... --> 5 </head> 6 <body> 7 <!-- ... --> 8 </body> 9</html>
Import or require spiderable-middleware
package:
1// ES6 import 2import Spiderable from 'spiderable-middleware'; 3 4// or CommonJS require 5const Spiderable = require('spiderable-middleware');
Register middleware handle:
1import express from 'express'; 2import Spiderable from 'spiderable-middleware'; 3 4const spiderable = new Spiderable({ 5 rootURL: 'http://example.com', 6 auth: 'test:test', 7}); 8 9const app = express(); 10// ensure this is the most top registered handle 11// to reduce response time and server load 12app.use(spiderable.handle).get('/', (req, res) => { 13 res.send('Hello World'); 14}); 15 16app.listen(3000);
We provide various options for serviceURL
as "Rendering Endpoints", each endpoint has its own features to fit every project needs.
Return genuine status code
To pass expected status code of a response from front-end JavaScript framework to browser/crawlers use specially formatted HTML-comment. This comment can be placed in any part of HTML-page. head
or body
tag is the best place for it.
Format
html:
1<!-- response:status-code=404 -->
jade:
// response:status-code=404
This package support any standard and custom status codes:
201
-<!-- response:status-code=201 -->
401
-<!-- response:status-code=401 -->
403
-<!-- response:status-code=403 -->
500
-<!-- response:status-code=500 -->
514
-<!-- response:status-code=514 -->
(non-standard)
Speed-up rendering
To speed-up rendering, JS-runtime should tell to the Spiderable engine when the page is ready. Set window.IS_RENDERED
to false
, and once the page is ready set this variable to true
. Example:
1<html> 2 <head> 3 <meta name="fragment" content="!"> 4 <script> 5 window.IS_RENDERED = false; 6 </script> 7 </head> 8 <body> 9 <!-- ... --> 10 <script type="text/javascript"> 11 //Somewhere deep in app-code: 12 window.IS_RENDERED = true; 13 </script> 14 </body> 15</html>
Detect request from Pre-rendering engine during runtime
Pre-rendering engine will set window.IS_PRERENDERING
global variable to true
. Detecting requests from pre-rendering engine are as easy as:
1if (window.IS_PRERENDERING) { 2 // This request is coming from Pre-rendering engine 3}
Note: window.IS_PRERENDERING
can be undefined
on initial page load, and may change during runtime. That's why we recommend to pre-define a setter for IS_PRERENDERING
:
1let isPrerendering = false; 2Object.defineProperty(window, 'IS_PRERENDERING', { 3 set(val) { 4 isPrerendering = val; 5 if (isPrerendering === true) { 6 // This request is coming from Pre-rendering engine 7 } 8 }, 9 get() { 10 return isPrerendering; 11 } 12});
JavaScript redirects
Redirect browser/crawler inside application when needed while a page is loading (imitate navigation), use any of classic JS-redirects can be used, including framework's navigation, or History.pushState()
1window.location.href = 'http://example.com/another/page'; 2window.location.replace('http://example.com/another/page'); 3 4Router.go('/another/page'); // framework's navigation !pseudo code
Note: Only 4 redirects are allowed during one request after 4 redirects session will be terminated.
API
Create new instance and pass middleware to server's routes chain;
Constructor
1new Spiderable(opts?: SpiderableOptions);
opts
{Object} - Configuration optionsopts.serviceURL
{String} - Valid URL to Spiderable endpoint (local or foreign). Default:https://render.ostr.io
. Can be set via environment variables:SPIDERABLE_SERVICE_URL
orPRERENDER_SERVICE_URL
opts.rootURL
{String} - Valid root URL of a website. Can be set via an environment variable:ROOT_URL
opts.auth
{String} - [Optional] Auth string in next format:user:pass
. Can be set via an environment variables:SPIDERABLE_SERVICE_AUTH
orPRERENDER_SERVICE_AUTH
. Defaultnull
opts.sanitizeUrls
{Boolean} - [Optional] Sanitize URLs in order to "fix" badly composed URLs. Defaultfalse
opts.botsUA
{[String]} - [Optional] An array of strings (case insensitive) with additional User-Agent names of crawlers that needs to get intercepted. See default bot's names. Set to['.*']
to match all browsers and robots, to serve static pages to all users/visitorsopts.ignoredHeaders
{[String]} - [Optional] An array of strings (case insensitive) with HTTP header names to exclude from response. See default list of ignored headers. Set to['.*']
to ignore all headersopts.ignore
{[String]} - [Optional] An array of strings (case sensitive) with ignored routes. Note: it's based on first match, so route/users
will cause ignoring of/part/users/part
,/users/_id
and/list/of/users
, but not/user/_id
or/list/of/blocked-users
. Defaultnull
opts.only
{[String|RegExp]} - [Optional] An array of strings (case sensitive) or regular expressions (could be mixed). Define exclusive route rules for pre-rendering. Could be used withopts.onlyRE
rules. Note: To define "safe" rules as {RegExp} it should start with^
and end with$
symbols, examples:[/^\/articles\/?$/, /^\/article\/[A-z0-9]{16}\/?$/]
opts.onlyRE
{RegExp} - [Optional] Regular Expression with exclusive route rules for pre-rendering. Could be used withopts.only
rulesopts.timeout
{Number} - [Optional] Number, proxy-request timeout to rendering endpoint in milliseconds. Default:180000
opts.requestOptions
{Object} - [Optional] Options for request module (like:timeout
,lookup
,insecureHTTPParser
), for all available options seehttp
API docsopts.debug
{Boolean} - [Optional] Enable debug and extra logging, default:false
Note: Setting .onlyRE
and/or .only
rules are highly recommended. Otherwise, all routes, including randomly generated by bots will be subject of Pre-rendering and may cause unexpectedly higher usage.
1// CommonJS 2// const Spiderable = require('spiderable-middleware'); 3 4// ES6 import 5// import Spiderable from 'spiderable-middleware'; 6 7const spiderable = new Spiderable({ 8 rootURL: 'http://example.com', 9 auth: 'test:test' 10}); 11 12// More complex setup (recommended): 13const spiderable = new Spiderable({ 14 rootURL: 'http://example.com', 15 serviceURL: 'https://render.ostr.io', 16 auth: 'test:test', 17 only: [ 18 /\/?/, // Root of the website 19 /^\/posts\/?$/, // "/posts" path with(out) trailing slash 20 /^\/post\/[A-z0-9]{16}\/?$/ // "/post/:id" path with(out) trailing slash 21 ], 22 // [Optionally] force ignore for secret paths: 23 ignore: [ 24 '/account/', // Ignore all routes under "/account/*" path 25 '/billing/' // Ignore all routes under "/billing/*" path 26 ] 27});
Configuration via env.vars
Same configuration can get achieved via setting up environment variables:
ROOT_URL='http://example.com' SPIDERABLE_SERVICE_URL='https://render.ostr.io' SPIDERABLE_SERVICE_AUTH='APIUser:APIPass'
alternatively, when migrating from other pre-rendering service — keep using existing variables, we support the next ones for compatibility:
ROOT_URL='http://example.com' PRERENDER_SERVICE_URL='https://render.ostr.io' PRERENDER_SERVICE_AUTH='APIUser:APIPass'
handle
Middleware handle
1const spiderable = new Spiderable(); 2spiderable.handle(req: IncomingMessage, res: ServerResponse, next: NextFunction): void; 3 4// Alias that returns {boolean} 5// true — if prerendering takes over the request 6spiderable.handler(req: IncomingMessage, res: ServerResponse, next: NextFunction): boolean;
Example using connect
and express
package:
1import { createServer } from 'node:http'; 2import Spiderable from 'spiderable-middleware'; 3 4const app = express(); 5// const app = connect(); 6const spiderable = new Spiderable(); 7 8app.use(spiderable.handle).use((_req, res) => { 9 res.end('Hello from Connect!\n'); 10}); 11 12createServer(app).listen(3000;
Example using node.js http
server:
1import { createServer } from 'node:http'; 2import Spiderable from 'spiderable-middleware'; 3 4// HTTP(s) Server 5http.createServer((req, res) => { 6 spiderable.handle(req, res, () => { 7 // Callback, triggered if this request 8 // is not a subject of spiderable pre-rendering 9 res.writeHead(200, {'Content-Type': 'text/plain; charset=UTF-8'}); 10 res.end('Hello vanilla NodeJS!'); 11 // Or do something else ... 12 }); 13}).listen(3000);
Types
Import types right from NPM package
1import Spiderable from 'spiderable-middleware'; 2import type { SpiderableOptions, NextFunction } from 'spiderable-middleware'; 3 4const options: SpiderableOptions = { 5 rootURL: 'http://example.com', 6 auth: 'test:test', 7 debug: false, 8 /* ..and other options.. */ 9}; 10expectType<SpiderableOptions>(options); 11 12const spiderable = new Spiderable(options); 13expectType<Spiderable>(spiderable); 14 15const next: NextFunction = (_err?: unknown): void => {}; 16expectType<void>(spiderable.handle(req, res, next));
AMP Support
To properly serve pages for Accelerated Mobile Pages (AMP) we support following URI schemes:
# Regular URIs: https://example.com/index.html https://example.com/articles/article-title.html https://example.com/articles/article-uniq-id/article-slug # AMP optimized URIs (prefix): https://example.com/amp/index.html https://example.com/amp/articles/article-title.html https://example.com/amp/articles/article-uniq-id/article-slug # AMP optimized URIs (extension): https://example.com/amp/index.amp.html https://example.com/amp/articles/article-title.amp.html
All URLs with .amp.
extension and /amp/
prefix will be optimized for AMP.
Rendering Endpoints
- render (default) -
https://render.ostr.io
- This endpoint has "optimal" settings, and should fit 98% cases. This endpoint respects cache headers of Crawler and origin server - render-bypass (devel/debug) -
https://render-bypass.ostr.io
- This endpoint will bypass caching mechanisms. Use it when experiencing an issue, or during development, to make sure responses are not cached. It's safe to use this endpoint in production, but it may result in higher usage and response time - render-cache (under attack) -
https://render-cache.ostr.io
- This endpoint has the most aggressive caching mechanism. Use it to achieve the shortest response time, and when outdated pages (for 6-12 hours) are acceptable
To change default endpoint, grab integration examples code and replace render.ostr.io
, with endpoint from the list above. For NPM integration change value of serviceURL
option.
Note: Described differences in caching behavior related to intermediate proxy caching, Cache-Control
header will be always set to the value defined in "Cache TTL". Cached results at the "Pre-rendering Engine" end can be purged at any time.
Debugging
Pass { debug: true }
or set DEBUG=true
environment variable to enable debugging mode.
To make sure a server can reach a rendering endpoint run cURL
command or send request via Node.js to (replace example.com with your domain name):
curl -v "https://test:test@render-bypass.ostr.io/?url=http://example.com"
In this example we're using render-bypass.ostr.io
endpoint to avoid any possible cached results, read more about rendering endpoints. As API credentials we're using test:test
, this part of URL can be replaced with auth
option from Node.js example. The API credentials and instructions can be found at the very bottom of Pre-rendering Panel of a particular host, — click on Integration Guide
# cURL example: curl -v "https://test:test@render-bypass.ostr.io/?url=http://example.com"
1// Node.js example: 2const https = require('https'); 3 4https.get('https://test:test@render-bypass.ostr.io/?url=http://example.com', (resp) => { 5 let data = ''; 6 7 resp.on('data', (chunk) => { 8 data += chunk.toString('utf8'); 9 }); 10 11 resp.on('end', () => { 12 console.log(data); 13 }); 14}).on('error', (error) => { 15 console.error(error); 16});
Running Tests
- Clone this package
- In Terminal (Console) go to directory where package was cloned
- Then run:
Node.js/Mocha
# Install development NPM dependencies: npm install --save-dev # Install NPM dependencies: npm install --save # Link package to itself npm link npm link spiderable-middleware # Run tests: ROOT_URL=http://127.0.0.1:3003 npm test # Run same tests with extra-logging DEBUG=true ROOT_URL=http://127.0.0.1:3003 npm test # http://127.0.0.1:3003 can be changed to any local address, PORT is required!