Doesn't always grab images

[danmed]

I'm just testing this and the archive feature doesn't appear to always grab images...

For example, when archiving the below link, none of the images are captured in the archive, but a thumbnail is generated.

https://imgur.com/a/W5wZxHT

I'm running the docker image btw..

[RadhiFadlillah]

@danmed yep, unfortunately the archival still doesn't work properly with page that uses a lot of Javascript like imgur and Reddit (the old Reddit is fine though), and unfortunately I don't have any idea where to start to solve this.

[danmed]

Ah ok, will it still does a pretty good job.. so thank you..

On Sun, 22 Sep 2019, 09:22 Radhi, [email protected] wrote:

@danmed https://github.com/danmed yep, unfortunately the archival still doesn't work properly with page that uses a lot of Javascript like imgur and Reddit (the old Reddit is fine though), and unfortunately I don't have any idea where to start to solve this.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/go-shiori/shiori/issues/160?email_source=notifications&email_token=AA5S4WVLZ7DP2EOK7WVX3J3QK4TOBA5CNFSM4IZBRP5KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7JBAXA#issuecomment-533860444, or mute the thread https://github.com/notifications/unsubscribe-auth/AA5S4WQCQWT7NSAONEDR54TQK4TOBANCNFSM4IZBRP5A .

[deanishe]

and unfortunately I don't have any idea where to start to solve this.

I was thinking about something like a set of URL rewriting rules, which might help Shiori handle certain sites (like Reddit). For example, a rule might rewrite www.reddit.com/... to old.reddit.com/... to get a more easily parseable version of a page.

For other sites, the mobile or print version (if they exist) might be much easier to parse.

For example, Shiori does a good job with this article from the Spiegel, but it also grabs a bunch of the dumb user comments that aren't in the print version.

[linwaytin]

I also have the same problem. I think there are two possible ways to deal with this problem.

First, let users enter the link manually. Second, as @deanishe mentioned, rules can help Shiori to determine where the image is.

At least, I like to be able to enter the link manually. For now, if the image is not correct, there is nothing I can do to fix it.

Anyway, thank you for this great project.

[fmartingr]

This will get worked on with #353. After the switch, we can start fixing this kind of things directly on obelisk.