convos-chat / linkembedder Goto Github PK

View Code? Open in Web Editor NEW

6.0 6.0 2.0 482 KB

Embed / expand oEmbed resources and other URL / links

Perl 92.90% Dockerfile 0.98% Raku 6.12%

mojo

linkembedder's Issues

Add support for Facebook video

Example https://www.facebook.com/watch/?v=2170684279662399

A couple of predefined links in embedder.pl do not work

https://appear.in/link-embedder-demo

appear.in is now whereby.com, and I couldn't find a embedder demo on the new site. also don't see embeddable "sare" link on a room, only invites.

http://paste.scsys.co.uk/557716

"Paste not found." , but I guess even if the link is updated it will die after a while anyway.

Also, imgur.com links don't open for me, but the image URL is correct, so maybe it's hotlinking to localhost that's being blocked..

Undeclared test (?) dependency Carp::Always

See subject. Test log:

Can't locate Carp/Always.pm in @INC (you may need to install the Carp::Always module) (@INC contains: ... .) at t/link.t line 5.
BEGIN failed--compilation aborted at t/link.t line 5.
t/link.t ............... 
Dubious, test returned 2 (wstat 512, 0x200)
No subtests run

Minimum Mojolicious version?

It seems that LinkEmbedder-1.18 needs to increase its Mojolicious version. It fails for Mojolicious versions < 9.11. Statistical analysis:

****************************************************************
Regression 'mod:Mojolicious'
****************************************************************
Name           	       Theta	      StdErr	 T-stat
[0='const']    	     -0.0000	      0.0000	 -12.63
[1='eq_8.26']  	      0.0000	      0.0000	   0.00
[2='eq_8.40']  	      0.0000	      0.0000	   0.00
[3='eq_8.41']  	      0.0000	      0.0000	  13.31
[4='eq_8.43']  	      0.0000	      0.0000	   2.10
[5='eq_9.01']  	      0.0000	      0.0000	  10.52
[6='eq_9.03']  	      0.0000	      0.0000	   4.21
[7='eq_9.11']  	      1.0000	      0.0000	57321465052731984.00
[8='eq_9.12']  	      1.0000	      0.0000	47958578471990840.00
[9='eq_9.13']  	      1.0000	      0.0000	72300277395640688.00
[10='eq_9.14'] 	      1.0000	      0.0000	72066673303892016.00

R^2= 1.000, N= 74, K= 11
****************************************************************

t/serve.t fails (with Mojolicious 8?)

See subject.

There's also a ticket on RT about failing the test suite with older Mojolicious versions: https://rt.cpan.org/Ticket/Display.html?id=120901

t/serve.t started to fail (with newest Mojolicious?)

The test suite started to fail on my smoker systems:

#   Failed test '400 Bad Request'
#   at t/serve.t line 25.
#          got: '404'
#     expected: '400'

#   Failed test 'content is similar'
#   at t/serve.t line 25.
#                   '<!DOCTYPE html>
# <!-- Request ID: jCXyAk9vwlpw -->
# <html>
#   <head>
#     <title>Page Not Found (development mode)</title>
#     <meta http-equiv="Pragma" content="no-cache">
...
#     </footer>
#   </body>
# </html>
# '
#     doesn't match '(?^u:^cb\(\{.*"title":"example page".*\}\)$)'
# Looks like you failed 10 tests of 28.
t/serve.t .............. 
Dubious, test returned 10 (wstat 2560, 0xa00)
Failed 10/28 subtests

This seems to happen with Mojolicious 9.11+. Here's the statistical analysis:

****************************************************************
Regression 'mod:Mojolicious'
****************************************************************
Name           	       Theta	      StdErr	 T-stat
[0='const']    	      1.0000	      0.0000	10754762396631854.00
[1='eq_8.12']  	      0.0000	      0.0000	   0.49
[2='eq_8.17']  	      0.0000	      0.0000	   1.06
[3='eq_8.40']  	      0.0000	      0.0000	   1.48
[4='eq_8.41']  	      0.0000	      0.0000	   0.97
[5='eq_8.43']  	      0.0000	      0.0000	   1.48
[6='eq_8.58']  	      0.0000	      0.0000	   1.06
[7='eq_8.65']  	      0.0000	      0.0000	   0.00
[8='eq_8.67']  	     -0.0000	      0.0000	  -0.28
[9='eq_8.68']  	      0.0000	      0.0000	   0.42
[10='eq_8.69'] 	      0.0000	      0.0000	   0.97
[11='eq_8.70'] 	      0.0000	      0.0000	   2.23
[12='eq_8.71'] 	      0.0000	      0.0000	   0.28
[13='eq_8.72'] 	      0.0000	      0.0000	   1.18
[14='eq_9.11'] 	     -1.0000	      0.0000	-9619351920290954.00
[15='eq_9.12'] 	     -1.0000	      0.0000	-8781226725539987.00
[16='eq_9.13'] 	     -1.0000	      0.0000	-10495573407417170.00

R^2= 1.000, N= 125, K= 17
****************************************************************

Twitter links are no longer rich media

Example: https://twitter.com/convosby/status/1280032587931463680

See how good it can get without platform.twitter.com/widgets.js

Seems like double newlines are added to some pastebins

Example https://github.com/convos-chat/convos/blob/master/assets/js/commands.js

Add support for https://youtu.be links

https://www.youtube.com/ links works, while https://youtu.be does not. Example: https://youtu.be/v4Ijkq6Myfc

add an option for the size of video thumbnail

In Convos sharing link to youtube returns an iframe with hardcoded height and width, is this configurable in this module as a parameter? some users find the video thumbnails to be too big.

Add support for perlbot

Example URL: https://perlbot.pl/p/ggqqik

Add support for hub.docker.com

Not much information is shown when linking to https://hub.docker.com/r/convos/convos

Support for <pre><code> basic pastebins

A user told me that they have a home-rolled pastebin that they use that doesn't get embedded. While we can't expect to support every little bin, I think the situation as it exists could be somewhat improved.

The fallback in basic for generic pastetbins uses oddly-specific selectors, which I'm assuming conform to some specific paste sites. Notably the fact that it uses direct descendent selectors body > pre and body > div > pre is somewhat limiting. However in the specific case I'm looking at it wouldn't be insurmountable.

The site would work with a selector body > div > pre > code. That selector is again rather specific, it would be odd to require a div tag like that, it could be body pre > code to be a bit more accepting, though it would have to be attempted before body > div > pre or it would never get there. Alternatively it could optionally look for a code tag inside the resulting selector and walk into it if present. Or it could use all_text rather than text to get the tag contents. Are any of these suggestions acceptable?

convos-chat / linkembedder Goto Github PK

linkembedder's Issues

Recommend Projects

Recommend Topics

Recommend Org