content() function captures text after text-transform

[mdekstrand]

With the following effective rule:

h1 {
    text-transform: uppercase;
    -weasy-string-set: section content();
}

Weasyprint sets the section register/variable to the uppercased version of the <h1>'s content, instead of the original lowercase version.

I am not entirely sure if I am right about the expected behavior, but the CSS Text Transformation spec says this about text-transform:

This property transforms text for styling purposes. (It has no effect on the underlying content.)

This suggests to me that content() should return the untransformed text. The definition of content() does not shed light on this matter.

[liZe]

Thanks for reporting! This issue is a duplicate of #137.