fix(audio): don't consider iframe referencing itself (#663)

microlinkhq · Sep 18, 2023 · 134ec60 · 134ec60
1 parent 4409000
commit 134ec60
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 29 deletions.
diff --git a/packages/metascraper-audio/index.js b/packages/metascraper-audio/index.js
@@ -78,35 +78,18 @@ module.exports = ({ getIframe = _getIframe } = {}) => {
   return {
     audio: audioRules.concat(
       async ({ htmlDom: $, url }) => {
-        const iframe = $('iframe')
-        if (iframe.length === 0) return
-
-        const srcs = []
-
-        iframe.each(function () {
-          const src = $(this).attr('src')
-          const normalizedUrl = normalizeUrl(url, src)
-          if (
-            typeof normalizedUrl === 'string' &&
-            normalizedUrl.startsWith('http') &&
-            srcs.indexOf(normalizedUrl) === -1
-          ) {
-            srcs.push(normalizedUrl)
-          }
-        })
-
-        return srcs.length > 0
-          ? 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)
-          : undefined
+        const srcs = [...new $('iframe').map((_, element) => $(element).attr('src')).get().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)
       },
       async ({ htmlDom: $, url }) => {
         const src = $twitter($, 'twitter:player')

diff --git a/packages/metascraper-helpers/test/index.js b/packages/metascraper-helpers/test/index.js
@@ -58,6 +58,12 @@ test('.normalizeUrl', t => {
   )
   t.is(normalizeUrl('https://example.com/'), 'https://example.com/')
   t.is(normalizeUrl('https://example.com'), 'https://example.com/')
+  t.is(normalizeUrl('https://www.example.com', 'https://www.example.com/foo'), 'https://www.example.com/foo')
+  t.is(normalizeUrl('https://www.example.com', '/foo'), 'https://www.example.com/foo')
+  t.is(normalizeUrl('https://www.example.com', 'file.html'), 'https://www.example.com/file.html')
+  t.is(normalizeUrl('https://www.example.com', 'data:text/html;base64,PGh0bWw+SGVsbG8sIHdvcmxkITwvaHRtbD4='), 'data:text/html;base64,PGh0bWw+SGVsbG8sIHdvcmxkITwvaHRtbD4=')
+  t.is(normalizeUrl('https://www.example.com', 'javascript:alert(\'Hello, world!\');'), undefined)
+  t.is(normalizeUrl('https://www.example.com', 'javascript:void(0)'), undefined)
 })
 
 test('.author', t => {