refactor(logo): use google logo endpoint

microlinkhq · Nov 14, 2022 · 9fb6a71 · 9fb6a71
1 parent d783e68
commit 9fb6a71
Show file tree

Hide file tree

Showing 5 changed files with 22 additions and 33 deletions.
diff --git a/packages/metascraper-logo-favicon/index.js b/packages/metascraper-logo-favicon/index.js
@@ -7,7 +7,6 @@ const memoize = require('@keyvhq/memoize')
 const {
   logo,
   parseUrl,
-  memoizeOne,
   normalizeUrl,
   toRule,
   url: urlFn
@@ -106,19 +105,23 @@ const pickBiggerSize = sizes => {
 pickBiggerSize.sortBySize = collection =>
   orderBy(collection, ['size.priority'], ['desc'])
 
-const getFaviconUrl = memoizeOne(url => logo('/favicon.ico', { url }))
-
 const createGetLogo = ({ gotOpts, keyvOpts }) => {
   const getLogo = async url => {
-    const faviconUrl = getFaviconUrl(url)
+    const faviconUrl = logo('/favicon.ico', { url })
     if (!faviconUrl) return
 
-    const response = await reachableUrl(faviconUrl, gotOpts)
-    return reachableUrl.isReachable(response) ? faviconUrl : undefined
+    let response = await reachableUrl(faviconUrl, gotOpts)
+    if (reachableUrl.isReachable(response)) return faviconUrl
+
+    response = await reachableUrl(
+      `https://www.google.com/s2/favicons?domain_url=${url}&sz=128`,
+      gotOpts
+    )
+
+    return reachableUrl.isReachable(response) ? response.url : undefined
   }
 
   return memoize(getLogo, keyvOpts, {
-    key: getFaviconUrl,
     value: value => (value === undefined ? null : value)
   })
 }
@@ -145,5 +148,3 @@ module.exports = ({ gotOpts, keyvOpts, pickFn = pickBiggerSize } = {}) => {
     ]
   }
 }
-
-module.exports.createGetLogo = createGetLogo
diff --git a/packages/metascraper-logo-favicon/test/index.js b/packages/metascraper-logo-favicon/test/index.js
@@ -6,8 +6,6 @@ const test = require('ava')
 
 const createMetascraper = opts => require('metascraper')([require('..')(opts)])
 
-const { createGetLogo } = require('..')
-
 const createHtml = meta =>
   `<!DOCTYPE html>
 <html lang="en">
@@ -54,18 +52,6 @@ test('`pickFn` gets the bigger size by default', async t => {
   )
 })
 
-test('.createGetLogo dedupes key from input', async t => {
-  const cache = new Map()
-  const getLogo = createGetLogo({ keyvOpts: { store: cache } })
-
-  await getLogo('https://help.kinopio.club/about/')
-  await getLogo('https://help.kinopio.club/legal/')
-  await getLogo('https://help.kinopio.club/')
-  await getLogo('https://kinopio.club/')
-
-  t.is(cache.size, 2)
-})
-
 test('create an absolute favicon url if the logo is not present', async t => {
   const url = 'https://www.nytimes.com'
   const metascraper = createMetascraper()
@@ -231,17 +217,19 @@ test('use non square logo as in the worst scenario', async t => {
   t.is(metadata.logo, 'https://s.yimg.com/kw/assets/eng-e-558x270.png')
 })
 
-test('returns null if logo is not detected', async t => {
-  const url = 'https://liu.edu/Brooklyn'
-  const html = await readFile(resolve(__dirname, 'fixtures/liuedu.html'))
+test('resolve logo using favicon associated with the domain', async t => {
+  const url = 'https://cdn.teslahunt.io/foo/bar'
   const metascraper = createMetascraper()
-  const metadata = await metascraper({ url, html })
-  t.is(metadata.logo, null)
+  const metadata = await metascraper({ url })
+  t.is(metadata.logo, 'https://teslahunt.io/favicon.ico')
 })
 
-test('resolve domain favicon', async t => {
-  const url = 'https://cdn.teslahunt.io/foo/bar'
+test('resolve logo using from google associated with the domain', async t => {
+  const url = 'https://escritopor.elenatorro.com'
   const metascraper = createMetascraper()
   const metadata = await metascraper({ url })
-  t.is(metadata.logo, 'https://teslahunt.io/favicon.ico')
+  t.is(
+    metadata.logo,
+    'https://t2.gstatic.com/faviconV2?client=SOCIAL&type=FAVICON&fallback_opts=TYPE,SIZE,URL&url=https://elenatorro.com/&size=128'
+  )
 })
diff --git a/packages/metascraper/test/integration/silicon-beat/index.js b/packages/metascraper/test/integration/silicon-beat/index.js
@@ -25,6 +25,7 @@ const url = 'http://www.siliconbeat.com/2016/04/19/time-come-woman'
 
 test('silicon-beat', async t => {
   const html = await readFile(resolve(__dirname, 'input.html'))
-  const metadata = await metascraper({ html, url })
+  const { logo, ...metadata } = await metascraper({ html, url })
+  t.is(typeof logo, 'string')
   t.snapshot(metadata)
 })
diff --git a/packages/metascraper/test/integration/silicon-beat/snapshots/index.js.md b/packages/metascraper/test/integration/silicon-beat/snapshots/index.js.md
@@ -15,7 +15,6 @@ Generated by [AVA](https://avajs.dev).
       description: 'It’s no secret that few women hold positions of power at venture capital firms. A new study quantifies what many in the industry have reported anecdotally. According to CrunchBase data released Tuesday on …',
       image: 'http://www.siliconbeat.com/wp-content/uploads/2016/04/Panel.jpg',
       lang: 'en',
-      logo: null,
       publisher: 'SiliconBeat',
       title: 'VC: Time to ‘come out as a woman’',
       url: 'http://www.siliconbeat.com/2016/04/19/time-come-woman/',

diff --git a/packages/metascraper/test/integration/silicon-beat/snapshots/index.js.snap b/packages/metascraper/test/integration/silicon-beat/snapshots/index.js.snap