Add 'rem' support

[fdemmer]

I am using Foundation, that uses the 'rem' unit a lot.

I am not sure about what i am doing here is correctly implementing the spec and the implementation follows "weasyprint conventions". Is there documentation on how weasy works? sequence of operations, what happens where etc?

I read through #264 to maybe figure this out, but I don't need all the viewport related relative units, so I attempted to just solve 'rem' for now.

It's based on the 'font-size' of the root element, so I am passing on the computed style of the 'body' element to compute() and use it in length().

My pdf output now works as I'd expect it.

[SimonSapin]

.get() here should be [] indexing. We want this to fail early if the root element’s style can’t be found. This should be fine since get_all_computed_styles deals with the root element first.

But this brings the question: what about computing styles for the root element itself? Let’s look at the spec:

https://drafts.csswg.org/css-values/#rem

Equal to the computed value of font-size on the root element. When specified on the font-size property of the root element, the rem units refer to the property’s initial value.

All in all, please replace this line with:

root_style = computed_styles[root, None] if root is not element else properties.INITIAL_VALUES

[SimonSapin]

Ah, sorry, it’s more subtle than I first thought. rem refers to the initial value for font-size on the root element, but it still has its usual definition for other properties of the root element.

I’ll leave it to you to figure out how to do this exactly. Unless you’d rather I do it, let me know.

[SimonSapin]

Thanks for your contribution! This looks good overall, there are just some minor issues noted inline.

It’d be good to add tests for this as well: change weasyprint/tests/resources/doc1.html to use rem (both in font-size and other properties, both on the root element and other elements) and test_annotate_document() in weasyprint/tests/test_css.py to check computed values.

Let me know if you want to do these fixes and add tests or if you’d rather I do it.

[fdemmer]

updated the branch:

• using set()
• use actual root element, not body
• take the initial font-size as input into the style computation for the root element

in my dev environment i get a lot of failing tests, would be great and probably save me a lot of time if you could complete that. thanks!

ah and i thought you could do a merge --squash or do you want me to squash the branch...?

[SimonSapin]

I didn’t know about merge --squash, thanks! (I’ve been using rebase -i.)

I’ll look into tests later if @liZe doesn’t get to it first ;)

[liZe]

Everything is OK on Travis, even with Python 2.6, good news. I've updated the tests and it works very well, but I can't push as it's not my branch.

@SimonSapin if everything is OK for you, I can merge and commit the test after.
@fdemmer I can squash the branch and add you in the AUTHORS :).

[fdemmer]

👍 and thanks for weasyprint!

[liZe]

Thank you for your contribution!

[SimonSapin]

@liZe, wanna add this to http://weasyprint.org/docs/features/ ?

[liZe]

Will be done in #297.