refactor: use regex in entities.js

patrickhulce · Aug 14, 2020 · 107e6dd · 107e6dd
1 parent d8154dc
commit 107e6dd
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 16 deletions.
diff --git a/bin/shared/utils.js b/bin/shared/utils.js
@@ -1,7 +1,7 @@
 module.exports = {
   stringifyEntities(entities) {
     return JSON.stringify(entities, (key, value) =>
-      value instanceof RegExp ? value.source : value
+      value instanceof RegExp ? `REGEXP:${value.source}` : value
     )
   },
 }
diff --git a/data/entities.js b/data/entities.js
@@ -75,7 +75,7 @@ module.exports = [
     products: [
       {
         name: 'Facebook Messenger Customer Chat',
-        urlPatterns: ['.*connect\\.facebook\\.net/.*/sdk/xfbml\\.customerchat\\.js'],
+        urlPatterns: [/connect\.facebook\.net\/.*\/sdk\/xfbml\.customerchat\.js/],
         facades: [
           {
             name: 'React Live Chat Loader',
@@ -217,7 +217,7 @@ module.exports = [
     products: [
       {
         name: 'YouTube Embedded Player',
-        urlPatterns: ['.*\\.youtube\\.com/embed/.*'],
+        urlPatterns: ['youtube.com/embed/'],
         facades: [
           {
             name: 'Lite YouTube',
@@ -566,7 +566,7 @@ module.exports = [
     products: [
       {
         name: 'Help Scout Beacon',
-        urlPatterns: ['.*beacon-v2\\.helpscout\\.net'],
+        urlPatterns: ['beacon-v2.helpscout.net'],
         facades: [
           {
             name: 'React Live Chat Loader',
@@ -700,7 +700,7 @@ module.exports = [
     products: [
       {
         name: 'Vimeo Embedded Player',
-        urlPatterns: ['.*player\\.vimeo\\.com/video/.*'],
+        urlPatterns: ['player.vimeo.com/video/'],
         facades: [
           {
             name: 'Lite Vimeo',
@@ -827,7 +827,7 @@ module.exports = [
     products: [
       {
         name: 'Drift Live Chat',
-        urlPatterns: ['.*js\\.driftt\\.com/include/.*/.*\\.js'],
+        urlPatterns: [/js\.driftt\.com\/include\/.*\/.*\.js/],
         facades: [
           {
             name: 'React Live Chat Loader',
@@ -1495,7 +1495,7 @@ module.exports = [
     products: [
       {
         name: 'Intercom Widget',
-        urlPatterns: ['.*widget\\.intercom\\.io.*', '.*js\\.intercomcdn\\.com/shim\\.latest\\.js'],
+        urlPatterns: ['widget.intercom.io', 'js.intercomcdn.com/shim.latest.js'],
         facades: [
           {
             name: 'React Live Chat Loader',

diff --git a/lib/create-entity-finder-api.js b/lib/create-entity-finder-api.js
@@ -32,19 +32,32 @@ function getProductInDataset(entityByDomain, entityByRootDomain, originOrURL) {
   const entity = getEntityInDataset(entityByDomain, entityByRootDomain, originOrURL)
   const products = entity && entity.products
   if (!products) return undefined
+  if (typeof originOrURL !== 'string') return undefined
+
   for (const product of products) {
     for (const pattern of product.urlPatterns) {
-      const regex = new RegExp(pattern)
-      if (regex.test(originOrURL)) {
-        return product
-      }
+      if (pattern instanceof RegExp && pattern.test(originOrURL)) return product
+      if (typeof pattern === 'string' && originOrURL.includes(pattern)) return product
     }
   }
   return undefined
 }
 
+function cloneEntities(entities) {
+  return entities.map(entity => ({
+    ...entity,
+    products: (entity.products || []).map(product => ({
+      facades: [],
+      ...product,
+      urlPatterns: product.urlPatterns.map(s =>
+        s.startsWith('REGEXP:') ? new RegExp(s.slice('REGEXP:'.length)) : s
+      ),
+    })),
+  }))
+}
+
 function createAPIFromDataset(entities_) {
-  const entities = entities_.map(e => ({...e}))
+  const entities = cloneEntities(entities_)
   const entityByDomain = new Map()
   const entityByRootDomain = new Map()
 

diff --git a/lib/index.test.js b/lib/index.test.js
@@ -93,7 +93,7 @@ describe('getEntity', () => {
             ],
             "name": "Facebook Messenger Customer Chat",
             "urlPatterns": Array [
-              ".*connect\\\\.facebook\\\\.net/.*/sdk/xfbml\\\\.customerchat\\\\.js",
+              /connect\\\\\\.facebook\\\\\\.net\\\\/\\.\\*\\\\/sdk\\\\/xfbml\\\\\\.customerchat\\\\\\.js/,
             ],
           },
         ],
@@ -121,6 +121,7 @@ describe('getEntity', () => {
         ],
         "homepage": "https://fonts.adobe.com/",
         "name": "Adobe TypeKit",
+        "products": Array [],
         "totalExecutionTime": 1722921,
         "totalOccurrences": 18274,
       }
@@ -149,18 +150,35 @@ describe('getEntity', () => {
 
 describe('getProduct', () => {
   it('works on basic url', () => {
-    expect(getProduct('https://www.youtube.com/embed/alGcULGtiv8')).toEqual({
+    expect(getProduct('https://www.youtube.com/embed/alGcULGtiv8')).toMatchObject({
+      name: 'YouTube Embedded Player',
       facades: [
         {
           name: 'Lite YouTube',
           repo: 'https://github.com/paulirish/lite-youtube-embed',
         },
       ],
-      name: 'YouTube Embedded Player',
-      urlPatterns: ['.*\\.youtube\\.com/embed/.*'],
     })
   })
 
+  it('works on regex based', () => {
+    expect(
+      getProduct('https://connect.facebook.net/en_US/sdk/xfbml.customerchat.js')
+    ).toMatchObject({
+      name: 'Facebook Messenger Customer Chat',
+      facades: [
+        {
+          name: 'React Live Chat Loader',
+          repo: 'https://github.com/calibreapp/react-live-chat-loader',
+        },
+      ],
+    })
+  })
+
+  it('returns undefined when product does not match', () => {
+    expect(getProduct('https://js.connect.facebook.net/lib.js')).toEqual(undefined)
+  })
+
   it('returns undefined with no products', () => {
     expect(getProduct('https://unknown.typekit.net/fonts.css')).toEqual(undefined)
   })