Merge pull request #698 from microlinkhq/next

fix(logo-favicon): ensure url exists

packages/metascraper-logo-favicon/src/index.js

@@ -56,11 +56,13 @@ const getDomNodeSizes = (domNodes, attr, url) =>
     .reduce((acc, domNode) => {
       const relativeUrl = domNode.attribs[attr]
       if (!relativeUrl) return acc
+      const normalizedUrl = normalizeUrl(url, relativeUrl)
+      if (!normalizedUrl) return acc
       return [
         ...acc,
         {
           ...domNode.attribs,
-          url: normalizeUrl(url, relativeUrl),
+          url: normalizedUrl,
           size: getSize(url, domNode.attribs.sizes)
         }
       ]

packages/metascraper-logo-favicon/test/index.js

@@ -80,10 +80,7 @@ test('get the biggest icon possible', async t => {
     '<link rel="icon" type="image/png" href="/logo/favicon-128.png" sizes="128x128">'
   ])
   const metadata = await metascraper({ url, html })
-  t.is(
-    metadata.logo,
-    'https://cdn.microlink.io/logo/favicon-196x196.png'
-  )
+  t.is(metadata.logo, 'https://cdn.microlink.io/logo/favicon-196x196.png')
 })
 
 test('detect `rel="fluid-icon"`', async t => {
@@ -139,7 +136,10 @@ test('detect `rel="apple-touch-icon-precomposed"`', async t => {
     '<link rel="apple-touch-icon-precomposed" sizes="144x144" href="logo/apple-touch-icon-144x144.png">'
   ])
   const metadata = await metascraper({ url, html })
-  t.is(metadata.logo, 'https://cdn.microlink.io/logo/apple-touch-icon-144x144.png')
+  t.is(
+    metadata.logo,
+    'https://cdn.microlink.io/logo/apple-touch-icon-144x144.png'
+  )
 })
 
 test('detect `rel="apple-touch-icon"`', async t => {
@@ -149,7 +149,10 @@ test('detect `rel="apple-touch-icon"`', async t => {
     '<link rel="apple-touch-icon" sizes="144x144" href="logo/apple-touch-icon-144x144.png">'
   ])
   const metadata = await metascraper({ url, html })
-  t.is(metadata.logo, 'https://cdn.microlink.io/logo/apple-touch-icon-144x144.png')
+  t.is(
+    metadata.logo,
+    'https://cdn.microlink.io/logo/apple-touch-icon-144x144.png'
+  )
 })
 
 test('detect `rel="shortcut icon"`', async t => {
@@ -159,10 +162,7 @@ test('detect `rel="shortcut icon"`', async t => {
     '<link rel="shortcut icon" href="logo/favicon.ico">'
   ])
   const metadata = await metascraper({ url, html })
-  t.is(
-    metadata.logo,
-    'https://cdn.microlink.io/logo/favicon.ico'
-  )
+  t.is(metadata.logo, 'https://cdn.microlink.io/logo/favicon.ico')
 })
 
 test('detect `rel="mask icon"`', async t => {
@@ -236,10 +236,18 @@ test('resolve logo using from google associated with the domain', async t => {
   t.true(metadata.logo.includes('gstatic'))
 })
 
-test('avoid data URI when data length is 0', async t => {
+test('avoid empty data URI', async t => {
   const url = 'https://www.adobe.com/'
   const html = '<link rel="icon" href="data:,">'
   const metascraper = createMetascraper()
   const metadata = await metascraper({ url, html })
   t.is(metadata.logo, 'https://www.adobe.com/favicon.ico')
 })
+
+test('avoid wrong data URI', async t => {
+  const url = 'https://www.adobe.com/'
+  const html = '<link rel="icon" href="data:">'
+  const metascraper = createMetascraper()
+  const metadata = await metascraper({ url, html })
+  t.is(metadata.logo, 'https://www.adobe.com/favicon.ico')
+})

packages/metascraper/test/integration/san-francisco-chronicle/snapshots/index.js.md

@@ -15,7 +15,7 @@ Generated by [AVA](https://avajs.dev).
       description: '[...] it’s natural to think that San Francisco’s new Nasdaq Entrepreneurial Center,…',
       image: 'https://s.hdnux.com/photos/45/53/13/9876166/6/rawImage.jpg',
       lang: 'en',
-      logo: 'https://www.sfchronicle.com/sites/premiumsfgate/apple-touch-icon-152x152.png',
+      logo: 'https://www.sfchronicle.com/sites/premiumsfgate/apple-touch-icon-196x196.png',
       publisher: 'SFGATE',
       title: 'Nasdaq center in SF offers free classes for entrepreneurs',
       url: 'https://www.sfchronicle.com/business/article/Nasdaq-center-in-SF-offers-free-classes-for-7338290.php',