The `SUPPORTED_MEDIA_DOMAINS` regex whitelist, used to check if media link comes from trusted hosts is invalid. It does not expose a security risk or I couldn't find an example for such as of now, but if someone would add a subdomain host to it using the same pattern, it would.
A counter example below:
```js
const SUPPORTED_MEDIA_DOMAINS = /^([^.]+\.)*(ytimg.com|cdninstagram.com|redd.it|imgur.com|fbcdn.net|pinimg.com)$/i;
console.log('Testing redd.it: ' + SUPPORTED_MEDIA_DOMAINS.test('redd.it'));
console.log('Testing reddjit: ' + SUPPORTED_MEDIA_DOMAINS.test('reddjit'));
```
Output:
```
$ node example.js
Testing redd.it: true
Testing reddjit: true
```
---
To be more clear, if someone would extend the regex in the future with e.g. `media.redd.it`, an attacker would be able to create a `mediaXredd.it` domain and bypass the whitelist.
---
A visualisation of the incorrect regex can be found on https://regexper.com/#%5E%28%5B%5E.%5D%2B%5C.%29*%28ytimg.com%7Ccdninstagram.com%7Credd.it%7Cimgur.com%7Cfbcdn.net%7Cpinimg.com%29%24
The issue has been found with LGTM: b626ef0b64/files/js/modules/link_previews.js (xdabadfc2bf20f0c3):1
fa4b2d412f