Trim all but 1 whitespace jinja condition

[GuillaumeGomez]

@vallentin suggested here that we could extend the "trim_all_whitespace" feature to be able to trim all whitespace characters but one.

So currently, we have:

<div>    {%- if foo -%} blabla {%- endif -%} </div>

Which will remove all whitespace characters before and after (by using -) and we have:

<div>    {%+ if foo +%} blabla {%+ endif +%} </div>

Which means that the whitespace should be saved (only useful when the "trim all whitespace" feature is enabled).

The idea would now to add a new character to strip all whitespace characters but one. I propose ~:

<div>    {%~ if foo %} blabla {% endif %} </div>

[djc]

I'm on board with all that, including the proposed operator.

[GuillaumeGomez]

Ok, then I'll make a PR soon. :)

[Kijewski]

I like that! For readability I'd like it best if a run of whitespaces becomes a single '\n' if it contains an '\n', or a space otherwise.

[GuillaumeGomez]

I'm fine with implementing it this way. Anyone else has an opinion maybe?

[djc]

Yeah, I think that approach makes sense.

[vallentin]

I'd love, if it is also configurable like #664. So you can configure it through the config file or part of the #[template(...)]. As minifying HTML is part of my rendering pipeline, so if it was possible to enable it globally, without needing to sprinkle ~ everywhere, then that would be lovely :)

[GuillaumeGomez]

Sure. What should we name the config parameter then?

[GuillaumeGomez]

Also, if both "trim_all" and "trim_all_but_one" are enabled, what should we do?

[vallentin]

Maybe instead of having them be individual. Maybe it might be possible to do whitespace = "trim_all" and whitespace = "trim_all_but_one"

[GuillaumeGomez]

Maybe @djc was right: I should have gone with a string argument for "trim_whitespace" so then we wouldn't have this problem.

EDIT: @vallentin Ah we reached the same conclusion haha.

[djc]

Maybe @djc was right: I should have gone with a string argument for "trim_whitespace" so then we wouldn't have this problem.

Yup!

[GuillaumeGomez]

Ok, then I'll change the parameter as well. :)

[vallentin]

Maybe @djc was right: I should have gone with a string argument for "trim_whitespace" so then we wouldn't have this problem.

@vallentin Ah we reached the same conclusion haha.

Well, there hasn't been a new release yet. So it's not to late to play with it :)

[djc]

I think the names should be suppress, preserve, and then maybe minimize?

[GuillaumeGomez]

I think a mix between @djc's and @vallentin's propositions would work better. What do you think of:

trim_whitespace = "all" | "all_but_one" | "preserve"

?

[djc]

I still dislike anything "trim" because that's not, conceptually, what it is. We're either suppressing the whitespace from the template or preserving it, and this option is somewhere in between. To my eyes, all_but_one seems pretty ugly, and it kind of begs the question about which one.

I do like the idea of just calling it whitespace by default, to keep things simple and concise.

[GuillaumeGomez]

That makes sense. Then let's go with whitespace and suppress|preserve|minimize.