perf(audio,video): short-circuit iframe probing on first hit (#810)

Author: Kikobeats

packages/metascraper-audio/src/index.js

@@ -12,7 +12,6 @@ const {
 } = require('@metascraper/helpers')
 
 const { find, chain, isEqual } = require('lodash')
-const pReflect = require('p-reflect')
 
 const toAudio = toRule(audio)
 
@@ -88,16 +87,13 @@ module.exports = ({ getIframe = _getIframe } = {}) => {
             .map(src => normalizeUrl(url, src))
         ]
         if (srcs.length === 0) return
-        return pReflect(
-          Promise.any(
-            srcs.map(async src => {
-              const htmlDom = await getIframe(url, $, { src })
-              const result = await findRule(audioRules, { htmlDom, url })
-              if (!has(result)) throw TypeError('no result')
-              return result
-            })
-          )
-        ).then(({ value }) => value)
+        for (const src of srcs) {
+          try {
+            const htmlDom = await getIframe(url, $, { src })
+            const result = await findRule(audioRules, { htmlDom, url })
+            if (has(result)) return result
+          } catch (_) {}
+        }
       },
       async ({ htmlDom: $, url }) => {
         const src = $('meta[name="twitter:player"]').attr('content')

packages/metascraper-audio/test/iframe.js

@@ -1,5 +1,6 @@
 'use strict'
 
+const cheerio = require('cheerio')
 const test = require('ava')
 
 const { runServer } = require('./helpers')
@@ -43,3 +44,26 @@ test('ignore non http urls', async t => {
   const metadata = await metascraper({ html, url })
   t.is(metadata.audio, null)
 })
+
+test('stop iframe probing after first audio match', async t => {
+  const calls = []
+  const metascraper = createMetascraper({
+    getIframe: async (url, $, { src }) => {
+      calls.push(src)
+      if (src.endsWith('/ok')) {
+        return cheerio.load(
+          '<meta property="og:audio" content="https://cdn.microlink.io/file-examples/sample.mp3">'
+        )
+      }
+      throw new Error('should not be called')
+    }
+  })
+
+  const metadata = await metascraper({
+    url: 'https://example.com',
+    html: '<iframe src="/ok"></iframe><iframe src="/skip"></iframe>'
+  })
+
+  t.is(metadata.audio, 'https://cdn.microlink.io/file-examples/sample.mp3')
+  t.deepEqual(calls, ['https://example.com/ok'])
+})

packages/metascraper-video/src/index.js

@@ -11,8 +11,6 @@ const {
   video
 } = require('@metascraper/helpers')
 
-const pReflect = require('p-reflect')
-
 const { chain, find, isEqual } = require('lodash')
 
 const toUrl = toRule(urlFn)
@@ -87,16 +85,13 @@ const withIframe = (rules, getIframe) =>
           .map(src => normalizeUrl(url, src))
       ]
       if (srcs.length === 0) return
-      return pReflect(
-        Promise.any(
-          srcs.map(async src => {
-            const htmlDom = await getIframe(url, $, { src })
-            const result = await findRule(rules, { htmlDom, url })
-            if (!has(result)) throw TypeError('no result')
-            return result
-          })
-        )
-      ).then(({ value }) => value)
+      for (const src of srcs) {
+        try {
+          const htmlDom = await getIframe(url, $, { src })
+          const result = await findRule(rules, { htmlDom, url })
+          if (has(result)) return result
+        } catch (_) {}
+      }
     },
     async ({ htmlDom: $, url }) => {
       const src = $('meta[name="twitter:player"]').attr('content')

packages/metascraper-video/test/iframe.js

@@ -1,5 +1,6 @@
 'use strict'
 
+const cheerio = require('cheerio')
 const test = require('ava')
 
 const { runServer } = require('./helpers')
@@ -43,3 +44,27 @@ test('ignore non http urls', async t => {
   const metadata = await metascraper({ html, url })
   t.is(metadata.video, null)
 })
+
+test('stop iframe probing after first video match', async t => {
+  const calls = []
+  const metascraper = createMetascraper({
+    getIframe: async (url, $, { src }) => {
+      calls.push(src)
+      if (src.endsWith('/ok')) {
+        return cheerio.load(
+          '<meta property="og:video" content="https://cdn.microlink.io/file-examples/sample.mp4">'
+        )
+      }
+      throw new Error('should not be called')
+    }
+  })
+
+  const metadata = await metascraper({
+    url: 'https://example.com',
+    html: '<iframe src="/ok"></iframe><iframe src="/skip"></iframe>',
+    pickPropNames: new Set(['video'])
+  })
+
+  t.is(metadata.video, 'https://cdn.microlink.io/file-examples/sample.mp4')
+  t.deepEqual(calls, ['https://example.com/ok'])
+})