HTML export with clickable urls

[jathri]

Thank you for #251 again! Have another feature proposal. Another part of "convenience package" ;) Can it be implemented sometime? Since Signal desktop app contains clickable links as well. Perhaps it can be yet another switch.

[bepaald]

Yeah sure. This was always planned, but it quickly dropped in priority on the (back then) long to-do list when I discovered how difficult it was :-). Things have slowed down enough to tak e another crack at it though.

I actually believe I have something working now (locally), it's not perfect, but I think it works the same as Signal's linkify function (it's more or less copied from Android Open Source Project).

When I find some time (I'll be busy for the next few days), I will clean my code up and then find some way to integrate it with the project. I'll get back to you when it's done.

Thanks!

[bepaald]

There is now initial support for this. This one was quite difficult, it could do with a lot of thorough testing.

Notes:

• There are known valid URL that aren't detected. I think (hope?) they are rare, and Signal does not detect them either, but still. (For example, none of the URLs mentioned here are linkified).
• There are some false hits, at least in my backup, around typos where erroneous .'s are present. Again, Signal does this too, and there is not really an issue (the tool simply can't know the intention of the user).
• When links without a protocol are present (www.example.com), these would normally be interpreted as relative paths to the current location. Normally, this can be worked around by prepending //, but that would default to the current protocol, which in this case would be file://, which would not be correct. I'm therefor always prepending protocol-less URLs with https://. This could produce dead links if the target does not support https. Again, I see no way around this.

For testing, please try to just look at as many URLs as you can (I realize manually checking each individual one in a backup could be too much, depending on the backup's size). Also, if possible pass the generated HTML through a validator (W3C). During development there were many difficulties when links were combined with text formatting (bold, italics) or multiwidth unicode characters (both around, and in the URL). This could lead to invalid HTML, or to the link tags (<a>) being at an offset from the actual link text (www.example.com).

The feature is currently behind --linkify. If it turns out to work properly, I will probably have that default true in the near future (to be disabled by --no-linkify), but for now it needs to be enabled explicitly.

Thanks!

[bepaald]

I bit the bullet and made --linkify default. Thanks again!

[jathri]

Just tried it and looking good :)

I think it works the same as Signal's linkify function (it's more or less copied from Android Open Source Project).

Consider this right approach. Thank you.