feat(cache): Cache resources based on ETag headers

After parsing a response's HTML, save the parsed subresources into
lru-cache with the `Etag` header as the key. Subsequent responses can
lookup and reuse the already parsed resources.

README.md

@@ -16,13 +16,24 @@ express()
   .listen(1337);
 ```
 
-## Node.js 0.10 Requirements
+## ⚠️ Node.js 0.10 Requirements ⚠️
 
 This module utilizes the [`posthtml`][posthtml] module to find images, scripts, and stylesheets in your HTML response.
 PostHTML requires a native Promise implementation or shim, so users of Node.js 0.10 will need to ensure a Promise shim has been configured.
 
+## Options
+
+* **images**, **scripts**, and **styles**: `Boolean`:
+
+    If `true` the corresponding subresources are parsed and added as a Preload Link headers.
+
+* **cache**: `Object`:
+
+    Object passed straight to [`lru-cache`][lru-cache]. It is highly recommended to set `cache.max` to an integer.
+
 ## License
 [![MIT License](https://img.shields.io/badge/license-MIT-blue.svg?style=flat-square)](https://www.tldrlegal.com/l/mit) see `LICENSE.md`.
 
 [preload]: https://www.w3.org/TR/preload/
-[posthtml]: https://github.com/posthtml/posthtml
+[posthtml]: https://github.com/posthtml/posthtml#readme
+[lru-cache]: https://github.com/isaacs/node-lru-cache#readme

index.js

@@ -7,13 +7,18 @@ var defaults = require('lodash.defaults');
 var hijackresponse = require('hijackresponse');
 var bl = require('bl');
 
+var LRU = require('lru-cache');
+
 module.exports = function netjet(options) {
   options = defaults(options, {
     images: true,
     scripts: true,
-    styles: true
+    styles: true,
+    cache: {}
   });
 
+  var cache = new LRU(options.cache);
+
   return function (req, res, next) {
     function appendHeader(field, value) {
       var prev = res.getHeader(field);
@@ -25,50 +30,63 @@ module.exports = function netjet(options) {
       res.setHeader(field, value);
     }
 
-    function insertLinks(urls, asType) {
-      urls.forEach(function (url) {
+    function insertLinkArray(entries) {
+      entries.forEach(function (entry) {
+        var url = entry[0];
+        var asType = entry[1];
+
         appendHeader('Link', '<' + unescape(url) + '>; rel=preload; as=' + asType);
       });
     }
 
     function processBody(body) {
-      var found = {};
-
-      posthtml().use(posthtmlPreload(options, found)).process(body);
+      var foundEntries = [];
 
-      if (options.images) {
-        insertLinks(found.images, 'image');
-      }
-
-      if (options.scripts) {
-        insertLinks(found.scripts, 'script');
-      }
+      posthtml().use(posthtmlPreload(options, foundEntries)).process(body);
 
-      if (options.styles) {
-        insertLinks(found.styles, 'style');
-      }
+      return foundEntries;
     }
 
     hijackresponse(res, function (err, res) {
       /* istanbul ignore next */
       // `err` from hijackresponse is currently hardcoded to "null"
       if (err) {
         res.unhijack();
-        return next(err);
+        next(err);
+        return;
       }
 
       // Only hijack HTML responses
       if (!/^text\/html(?:;|\s|$)/.test(res.getHeader('Content-Type'))) {
-        return res.unhijack();
+        res.unhijack();
+        return;
+      }
+
+      var etag = res.getHeader('etag');
+      var entries;
+
+      // reuse previous parse if the etag already exists in cache
+      if (etag) {
+        entries = cache.get(etag);
+
+        if (entries) {
+          insertLinkArray(entries);
+          res.pipe(res);
+          return;
+        }
       }
 
       res.pipe(bl(function (err, data) {
         if (err) {
           res.unhijack();
-          return next(err);
+          next(err);
+          return;
         }
 
-        processBody(data.toString());
+        entries = processBody(data.toString());
+
+        insertLinkArray(entries);
+        cache.set(etag, entries);
 
         res.end(data);
       }));

package.json

@@ -30,9 +30,9 @@
   "dependencies": {
     "bl": "^1.0.1",
     "hijackresponse": "^1.0.2",
-    "lodash.assign": "^4.0.1",
     "lodash.defaults": "^4.0.0",
     "lodash.unescape": "^3.1.1",
+    "lru-cache": "^4.0.0",
     "posthtml": "^0.8.1"
   },
   "devDependencies": {
@@ -43,6 +43,8 @@
     "istanbul": "^0.4.2",
     "mocha": "^2.4.5",
     "supertest": "^1.1.0",
+    "supertest-as-promised": "^2.0.2",
+    "testdouble": "^0.7.3",
     "xo": "^0.12.1"
   },
   "xo": {

posthtml-preload.js

@@ -1,14 +1,7 @@
 'use strict';
-var assign = require('lodash.assign');
 
-module.exports = function (options, found) {
+module.exports = function (options, foundEntries) {
   return function (tree) {
-    assign(found, {
-      images: [],
-      scripts: [],
-      styles: []
-    });
-
     var matchers = [];
 
     if (options.images) {
@@ -27,20 +20,20 @@ module.exports = function (options, found) {
       tree.match(matchers, function (node) {
         switch (node.tag) {
           case 'img':
-            found.images.push(node.attrs.src);
+            foundEntries.push([node.attrs.src, 'image']);
             break;
           case 'script':
-            found.scripts.push(node.attrs.src);
+            foundEntries.push([node.attrs.src, 'script']);
             break;
           case 'link':
-            found.styles.push(node.attrs.href);
+            foundEntries.push([node.attrs.href, 'style']);
             break;
           // no default
         }
         return node;
       });
     }
 
-    return found;
+    return foundEntries;
   };
 };