Replace non-working videos with something less frustrating
Opened this issue · 5 comments
We know that support for videos is very limited in warc2zim (only Youtube player mostly).
It would be cool to replace all video players (at least all
Ideally I would like to replace the
Maybe this is too much.
Alternatives could be:
- describe which images (based on URL regex) we want to watermark, with which text (and keep the video player as-is) => this is going to be a bit frustrating for the user because he can still start the video but at least this ensure we can easily cover all kind of video players
- instead of replacing with an image of the original video poster, just replace with a text saying there was a video which is not available in the ZIM (but then we need to be able to localize this message in proper languages).
Can we maybe add use cases to this issue? I imagine all simple players based on a <video /> tag with a single file source work as expected.
If so this issue would be for non-youtube smart players that stream outside of a single file.
First use case I have in mind is the kind of video we have at the bottom of this page: https://www.nhs.uk/conditions/periods/
It is indeed not a simple player.
It might be worth revisiting solutions explored previously by @mgautierfr force-replacing JS players with simple HTML5 <video> tags referencing the scraped mp4 / webm. This made at least Vimeo videos work fine in Zimit2 ZIMs (using test site mesquartierschinois), albeit not as "elegantly" as with the original JS player.
This is a generic solution where the video is actually scraped as a file. I don't think we have yet come across cases where video is streamed and stored in a ZIM as a set of packets as opposed to a file, though it's theoretically possible.
Above assumes that proper support for Vimeo is likely to take a long time (?). It would be better to have rudimentary support for the video rather than substituting a still image. Of course if only a few seconds of a video get scraped, then such a solution wouldn't work, though it would be worth seeing if there's a simple fix for that in the crawler.
In most cases where the video doesn't work, the problem is both that player is not working properly AND only first seconds of video have been crawled. At least this was the case with Vimeo in my former experiments, and I never achieved to find how to fix this in the crawler. Maybe it has been fixed since then.
This is why I consider this is going to take time to fix non-working videos, and why I consider this issue.
How much resources should we involve in this "temporary" workaround. OK for me only if this is robust and easy to do.