[RavSinghChandan]

The icons of Firefox set fill="context-fill" fill-opacity="context-fill-opacity" on the root <svg>, as in the icons of Firefox-UI-Fix. They fail with ValueError: unsupported color specification context-fill.

These SVG 2 paints take the fill or stroke of the context element. For an element drawn by a <use>, that is the <use>. For the SVG itself, it is the document drawing it: Firefox lets the page set it. So context-fill, context-stroke, context-fill-opacity and context-stroke-opacity are now inherited: inside a <use> they take its paint, and at the root the colors set on the FPDF, as currentColor does without a color property.

test_svg_context_fill_is_the_color_of_the_fpdf and test_svg_context_fill_is_the_paint_of_the_use fail without the change.

[andersonhc]

I noticed one case to address before merging: mapping both context-fill and context-stroke to INHERIT loses the distinction between them. For example, with , a referenced path using fill="context-stroke" should have a lime fill, but currently renders red. An intervening group’s fill can also override the color.

Could we preserve these as distinct values and resolve them against the referencing ? Regression tests for swapped fill/stroke and an intervening group would help cover this.

[andersonhc]

I found one remaining inheritance case:

<defs>
  <path id="p" d="M0 0h5v5z" fill="context-stroke"/>
</defs>
<g fill="red" stroke="lime">
  <use href="#p"/>
</g>

The path fills red instead of lime. The context lookup misses presentation attributes inherited from the parent; writing those colors in style="..." works. Could we handle inherited presentation attributes too and add regression coverage, including opacity?

One small lint fix: reassigning value in the new test triggers pylint’s redefined-loop-name. A separate variable for the converted expected color should resolve it.

[RavSinghChandan]

Thanks, fixed. context-fill and context-stroke (and their opacities) are now resolved against the referencing : an element that uses them is built again with the fill, stroke and opacities of the , so fill="context-stroke" is lime in your example, and a group in between no longer overrides it. The definition keeps its context paints for other s, and outside a they are still inherited. test_svg_context_paints_of_the_use and test_svg_context_fill_through_a_group cover swapped fill/stroke, a style property, the opacities, an intervening group and two s with different colors.

[RavSinghChandan]

Thanks, fixed. When the doesn't set a fill, stroke or opacity, the context lookup now walks up its ancestors and reads presentation attributes as well as styles, so your example fills lime. test_svg_context_paints_inherited_by_the_use covers the attributes (with stroke-opacity) and the style form, through an extra . The test now uses a separate variable for the expected color.