fix(audio): only consider http URLs (#678)

* refactor(media-provider): remove unnecessary dependency * fix(audio): only consider http URLs
microlinkhq · Dec 20, 2023 · 3f03005 · 3f03005
1 parent eca5a38
commit 3f03005
Show file tree

Hide file tree

Showing 5 changed files with 65 additions and 6 deletions.
diff --git a/packages/metascraper-audio/package.json b/packages/metascraper-audio/package.json
@@ -28,6 +28,7 @@
     "p-reflect": "~2.1.0"
   },
   "devDependencies": {
+    "async-listen": "latest",
     "ava": "5"
   },
   "engines": {

diff --git a/packages/metascraper-audio/src/index.js b/packages/metascraper-audio/src/index.js
@@ -79,7 +79,7 @@ module.exports = ({ getIframe = _getIframe } = {}) => {
     audio: audioRules.concat(
       async ({ htmlDom: $, url }) => {
         const srcs = [
-          ...new $('iframe')
+          ...new $('iframe[src^="http"], iframe[src^="/"]')
             .map((_, element) => $(element).attr('src'))
             .get()
             .map(src => normalizeUrl(url, src))

diff --git a/packages/metascraper-audio/test/iframe.js b/packages/metascraper-audio/test/iframe.js
@@ -0,0 +1,57 @@
+'use strict'
+
+const { default: listen } = require('async-listen')
+const { createServer } = require('http')
+const { promisify } = require('util')
+const test = require('ava')
+
+const closeServer = server => promisify(server.close)
+
+const createMetascraper = (...args) =>
+  require('metascraper')([require('../src')(...args)])
+
+test('absolute http', async t => {
+  const server = createServer((_, res) => {
+    res.setHeader('Content-Type', 'text/html')
+    res.end(
+      '<meta property="og:audio" content="https://cdn.microlink.io/file-examples/sample.mp3">'
+    )
+  })
+
+  t.teardown(() => closeServer(server))
+  const url = (await listen(server, { port: 0, host: '0.0.0.0' })).toString()
+  const html = `<iframe src="${url}">`
+  const metascraper = createMetascraper()
+  const metadata = await metascraper({ html, url })
+  t.is(metadata.audio, 'https://cdn.microlink.io/file-examples/sample.mp3')
+})
+
+test('relative http', async t => {
+  const server = createServer((_, res) => {
+    res.setHeader('Content-Type', 'text/html')
+    res.end('<meta property="og:audio" content="/file-examples/sample.mp3">')
+  })
+
+  t.teardown(() => closeServer(server))
+  const url = (await listen(server, { port: 0, host: '0.0.0.0' })).toString()
+  const html = '<iframe src="/">'
+  const metascraper = createMetascraper()
+  const metadata = await metascraper({ html, url })
+  t.is(metadata.audio, url + 'file-examples/sample.mp3')
+})
+
+test('ignore non http urls', async t => {
+  const server = createServer((_, res) => {
+    res.setHeader('Content-Type', 'text/html')
+    res.end(
+      '<meta property="og:audio" content="tg://join?invite=n3gS0R7pjFJhMWM0">'
+    )
+  })
+
+  t.teardown(() => closeServer(server))
+  const url = (await listen(server, { port: 0, host: '0.0.0.0' })).toString()
+  const html = `<iframe src="${url}">`
+  const metascraper = createMetascraper()
+  const metadata = await metascraper({ html, url })
+  t.is(metadata.audio, null)
+})
diff --git a/packages/metascraper-media-provider/package.json b/packages/metascraper-media-provider/package.json
@@ -29,7 +29,6 @@
     "debug-logfmt": "~1.0.4",
     "got": "~11.8.6",
     "lodash": "~4.17.21",
-    "p-do-whilst": "~1.1.0",
     "p-reflect": "~2.1.0",
     "p-retry": "~4.6.1",
     "p-timeout": "~4.1.0",

diff --git a/packages/metascraper-media-provider/src/get-media/provider/generic.js b/packages/metascraper-media-provider/src/get-media/provider/generic.js
@@ -6,7 +6,6 @@ const debug = require('debug-logfmt')(
 const { serializeError } = require('serialize-error')
 const youtubedl = require('youtube-dl-exec')
 const { get, constant } = require('lodash')
-const pDoWhilst = require('p-do-whilst')
 const pTimeout = require('p-timeout')
 
 const RE_UNSUPORTED_URL = /Unsupported URL/
@@ -53,16 +52,19 @@ module.exports = ({
     const userAgent = get(gotOpts, 'headers.user-agent')
 
     const task = async () => {
-      await pDoWhilst(async () => {
+      do {
         try {
           const proxy = getProxy({ url, retryCount: retryCount++ })
           const flags = getFlags({ url, proxy, userAgent, cacheDir })
           data = await youtubedl(url, flags, { timeout, ...props })
         } catch (error) {
-          if (condition()) { debug('getInfo:error', { retryCount }, serializeError(error)) }
+          if (condition()) {
+            debug('getInfo:error', { retryCount }, serializeError(error))
+          }
           isSupportedURL = !RE_UNSUPORTED_URL.test(error.stderr)
+          console.log('catch', { isSupportedURL })
         }
-      }, condition)
+      } while (condition())
 
       return data
     }