From d1f8bcefb6d5c363f16231032c3222a4bbb22f6c Mon Sep 17 00:00:00 2001
From: Kiko Beats <josefrancisco.verdu@gmail.com>
Date: Mon, 18 Dec 2023 15:11:13 +0100
Subject: [PATCH] fix(audio): only consider http URLs

---
 packages/metascraper-audio/package.json   |  1 +
 packages/metascraper-audio/src/index.js   |  2 +-
 packages/metascraper-audio/test/iframe.js | 57 +++++++++++++++++++++++
 3 files changed, 59 insertions(+), 1 deletion(-)
 create mode 100644 packages/metascraper-audio/test/iframe.js

diff --git a/packages/metascraper-audio/package.json b/packages/metascraper-audio/package.json
index 89a2ac77e..e1f93c7f5 100644
--- 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
index 0e501ac4c..004acbecf 100644
--- 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) => $(element).attr('src'))
             .get()
             .map(src => normalizeUrl(url, src))
diff --git a/packages/metascraper-audio/test/iframe.js b/packages/metascraper-audio/test/iframe.js
new file mode 100644
index 000000000..6e70c829f
--- /dev/null
+++ b/packages/metascraper-audio/test/iframe.js
@@ -0,0 +1,57 @@
+'use strict'
+
+const { default: listen } = require('async-listen')
+const { createServer } = require('http')
+const { promisify } = require('util')
+const test = require('ava')
+
+const closeServer = server => promisify(server.close)
+
+const createMetascraper = (...args) =>
+  require('metascraper')([require('../src')(...args)])
+
+test('absolute http', async t => {
+  const server = createServer((_, res) => {
+    res.setHeader('Content-Type', 'text/html')
+    res.end(
+      '<meta property="og:audio" content="https://cdn.microlink.io/file-examples/sample.mp3">'
+    )
+  })
+
+  t.teardown(() => closeServer(server))
+  const url = (await listen(server, { port: 0, host: '0.0.0.0' })).toString()
+  const html = `<iframe src="${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)
+})