tl;dr I have a Google Apps Script project getting a major overhaul. If you want to look at the code and contribute, it's on GitHub.
A couple years back, I published a little addon which would scan a Google Doc for linked YouTube videos and allow you to watch them in a popup or sidebar. I called it DocuTube and published without much more thought.
Since writing that app, I've learned a ton more and decided to give it a major overhaul. It was mediocre on the web store with some valid complaints about a lack of clarity and functionality.
As I added functions, I took my time to figure out better ways to structure my code. I followed Bruce McPherson's wonderful advice to add namespacing (isolating functions from one another) to help keep everything tidy. It bent my brain into pretzels, but it was so good to wrestle though. I now have an application that is more manageable and extensible because separate parts are sequestered from one another.
This month, I published an update which adds search, video previews, automatic embedding, and cleans up video playback.
This was the biggest addition (and the most frequent request) in the app listing. This version of DocuTube includes YouTube searching, which allows authors to find - and insert content - they want in the document without managing several hundred tabs.
This is used in schools - I've had several contacts of GSuite admins asking about the permissions and user data storage (minimal permissions, nothing is stored on users) before they pushed it out to their domains. Since studets are (potentially) searching for videos, all searches in DocuTube are forced to safe videos only.
I will not be making that optional.
The YouTube API is quotaed, which means every search counts against your total allowance for the day. To manage this, I cache results heavily. Each search returns 50 videos and they're each stored in the cache as you page back and forth. Embeds are pulled from the cached data and used to populate the preview portion and do the embedding work. That said, my quota is limited to 10,000 units per day. As use increases, I can see the need to apply for more search credits very soon.
Since searching was included, I wanted to be able to provide a way to actually check and make sure the video clicked was the one the user wanted to actually embed.
Clicking on a video gives a playable iframe for a preview. The user can then choose how to attach the video to the document: copy the link to the clipboard (manual paste), insert the thumbnail, or add some custom text. All of this is done with the cached resource so the API isn't hit again to get the link or thumbnail.
Google Docs doesn't include an "embed" in the traditional sense. When I use the term embed in the context of DocuTube, I mean it handles the link for you. Inserting a video as a thumbnail grabs the title image, throws it in the document, and then adds the link. You can certainly do the same thing manually with several tabs and clicks.
I think "embed" is an okay term because it leads into the other function of DocuTube: watching videos.
This hasn't changed a whole lot. The major update from v0.7 (the current version) to v1.0 is that video watching is inclusive. Prior to 1.0, you needed to choose where to pull links from: the document or the comments.
With the 1.0 update, all videos linked in the document, regardless of location, are added to the Watch sidebar. This also removes the option of watching videos in a popup because the whole point of including the video as an embedded item is not leaving the document. If videos are loaded in a popup that takes up the entire editor, I've essentially kicked you out of the document.