Ceci est une ancienne révision du document !
I’m going to start this month with yet another complaint about the state of font support in web browsers. Once again, a single browser vendor has effectively scuppered a format that is supported by their peers, leaving us users and developers without a practical alternative.
I’ve written in the past about how Firefox’s lack of support for SVG fonts effectively killed the format. This was a real blow: the ability for a web developer to dynamically modify fonts on-the-fly using JavaScript could have been truly revolutionary. More recently, it looked like SVG in OpenType would provide some new capabilities in terms of color fonts and bitmap support, but Google has closed the relevant issue on the Chromium bug tracker (#306078) with a status of WONTFIX, indicating that they won’t be adding support for this format to the browser. Given Chrome’s dominant market share, this effectively neuters some of the new font support in Inkscape 1.0 that I covered in part 108 of this series.
Variable fonts, as described in that article, already work and will continue to do so. It’s the SVG in OpenType format that is now effectively dead, despite being supported in Firefox, Safari and even older versions of Edge. My own experiments suggested some level of support for the fonts I was using (e.g. the Gilbert Color font was rendered in Chrome, but only as a monochrome font). I now suspect that was due to the fonts themselves providing non-SVG fallback content, rather than Chrome providing any support at all for this format.
What’s most disheartening about this development is that Google’s primary reason for not supporting SVG in OpenType appears to be that they are developing their own format for color fonts (again, embedded into OpenType). If you were in any doubt that Google might use their market dominance to force the direction of web standards, perhaps this will make you realise that a single vendor having so much control is not really in the best interest of the users. The need for multiple, independent rendering engines is the reason I continue to use Firefox as my daily driver. The battle against IE may have been won, but the war for a truly open and independent web still wages.
Time will tell whether or not Google’s new format will win out, or font foundries will just embed both the Google and the SVG versions of color glyphs into ever-more-bloated font files. For now, however, despite Inkscape’s support for color fonts – and its partial support for bitmap fonts – you might want to think twice about using these in your projects.
CSS in SVG
The main topic for this month was supposed to be the Selectors and CSS dialog that was added experimentally in Inkscape v1.0, and then promoted to non-experimental in v1.0.1. But in order to understand what this dialog does, it’s essential to first have a decent grounding in how CSS works in a stand-alone SVG file. This month’s article will put in that ground work – if you’re already a CSS aficionado, then you may want to skip this one and come back next month for the details of the dialog.
Let’s begin with some history. The SVG format has always been a little confused about its identity. It was created during the great XML uprising of the late 1990s, when the World Wide Web Consortium (W3C) were pushing for a world in which XML formed a common basis for many file formats, allowing for tools and workflows that could easily combine and convert different types of data. SVG was a stand-alone vector format, not yet directly integrated into any web browsers, but the intention was clear that it should conform to, and work with, existing web standards. This left it with something of a dichotomy: on the one hand, it needed a way to directly define fonts, colors and dimensions in order to be used as a generic vector format with no dependency on a browser engine; on the other hand, it had to work well with the growing power of CSS to define styles for those users who did want to work in a more web-focused way.
And so we have ended up with a profusion, and confusion, of ways to style SVG content. First there are the classic “presentation attributes” from the SVG format. These are attributes that can be applied directly to SVG elements in order to style them individually. For example, a red rectangle with a thick black border might be defined like this:
<rect x=“20” y=“20”
width="100" height="100" fill="red" stroke="black" stroke-width="10"
/>
This approach requires only the consuming program to understand XML and SVG, and doesn’t impose the need for a complete CSS parser (though things like color values and most units are still taken from the CSS spec).
In the early days of HTML, presentation attributes were common there, too. Veteran web developers might remember the “border”, “color” and “bgcolor” attributes, amongst others, but those capabilities were rapidly subsumed by the growing scope of CSS. In order to override CSS styles on a per-element basis, the relevant CSS rules had to be combined into a single “style” attribute. This method also works with SVG, meaning that our thick-stroked red rectangle could also be written like this:
<rect x=“20” y=“20”
width="100" height="100" style="fill: red; stroke: black; stroke-width: 10;"
/>
If CSS only offered a way to put all the style information into a single attribute, it wouldn’t be terribly useful. But as well as setting styles on each element individually, CSS also offers a mechanism for applying styles across an entire document (or even across multiple documents, but that doesn’t work with Inkscape). It makes it trivial to give all your rectangles the same fill color; or to divide your objects into common classes that share similar styles; or to add override rules that make, for example, the third rectangle in a group have a different stroke width to the others. These tricks are achieved by moving the style information out of the style attributes, and into a common stylesheet in a <style> element. The style attribute still applies, and will override the stylesheet, as shown in this abridged example where both the <rect> elements share the same fill color, stroke width, and their width and height, but have a different stroke color:
<style>
rect {
fill: red;
stroke-width: 10;
width: 100;
height: 100;
}
</style>
<rect x=“20” y=“20”
style="stroke: black;"
/> <rect x=“60” y=“60”
style="stroke: darkred;"
/>
Notice that the syntax of the stylesheet is quite different to that of the SVG content. I won’t go into details about how to write CSS here, but suffice to say that it’s made up of multiple rules (delimited by curly braces: { and } ), with each rule starting with a “selector” that determines what elements the rule will apply to. In this case, the selector is just the string “rect” which will make this rule apply to all the <rect> elements in the document.
We can target a specific element by ensuring it has an “id” attribute in the SVG content (which Inkscape does by default), then using that ID, prefixed with a hash character (#), as our selector. In this next example, we have three rectangles which all share a common base style, but one of them has its fill color overridden in this way.
<style>
rect {
fill: red;
stroke: black;
stroke-width: 10;
width: 100;
height: 100;
}
#r2 {
fill: blue;
}
</style>
<rect
id="r1" x="20" y="20"
/> <rect
id="r2" x="60" y="60"
/> <rect
id="r3" x="100" y="100"
/> We can also target multiple items by giving them the same value in a “class” attribute and using it preceded by a dot (.) as the CSS selector. This example has four rectangles, split into two different classes. Note that an element can have more than one class, separated by spaces, which allows for a huge amount of flexibility in combining objects into overlapping sets.
<style>
rect {
stroke: black;
stroke-width: 10;
width: 100;
height: 100;
}
.important {
fill: red;
}
.warning {
fill: orange;
}
</style>
<rect
class="important" x="20" y="20"
/> <rect
class="warning" x="60" y="60"
/> <rect
class="important high" x="100" y="100"
/> <rect
class="p3 warning low" x="140" y="140"
/>
CSS in web browsers also provides attribute selectors. These give the ability to select elements based on the presence, and optionally the value, of specific attributes on the elements. This would be particularly useful when dealing with some of Inkscape’s native objects, some of which are implemented as <path> elements, with a custom attribute to tell Inkscape which sort of native objects they are. Similarly, groups and layers are both implemented as SVG <g> elements, with an Inkscape-specific attribute indicating which ones are layers, so this would offer a means to distinguish between them in CSS selectors. Unfortunately it appears that the Selectors and CSS dialog doesn’t recognise the syntax for attribute selectors, so I shan’t consider it further here.
These basic selectors can be combined to further refine the elements that will match. Using “rect.important”, for example, would only match <rect> elements with a class of “important”. It would not match <rect class=“warning”> or <path class=“important”>. In a similar manner, “#r1.high.important” would only match the element with an id of “r1” if it also had both the “important” and “high” classes assigned.
The browser provides some classes for free, in the form of “pseudo-classes” which allow you to create selectors that target things that the browser calculates at runtime. These may change dynamically, especially if your page uses JavaScript to manipulate the document. Most of them aren’t applicable to the non-dynamic world of Inkscape, but some relate to the order in which elements appear in the document model, and do work in a limited manner in the Selectors and CSS dialog. For example, a selector of “rect:first-child” would select any <rect> element that is the first child of its parent (e.g. the first element inside a group). Inkscape doesn’t quite allow this syntax, though, but there is a workaround. I’ll take a deeper dive into pseudo-elements and how to work with them in Inkscape as we progress with the Selectors and CSS dialog.
Selectors can be combined in different ways to match more elements, or to further refine the match based on the hierarchical structure of the SVG document. “#id1, #id2, .important” would match the two elements with the specified IDs, but also any element with the “important” class. “text + path”, meanwhile, would match any <path> element that is an immediate sibling of a <text> element. That one can be useful for styling any underlines you’ve drawn with your text elements, for example.
A simple space character creates a rule that matches if the second element is a descendent (possibly even a deeply nested descendent) of the first. E.g. “g.primary rect” matches a <rect> that is somewhere inside a group with a class of “primary”. On the other hand, “g.primary > rect” would only match the <rect> if it’s an immediate child of the primary group, not if it’s nested further inside other groups.
With the information from the last few paragraphs, see if you can make sense of this example file:
<style>
rect {
stroke: black;
stroke-width: 10;
width: 100;
height: 100;
}
g.primary rect {
fill: yellow;
}
g.primary > rect {
fill: green;
}
g > rect:first-child {
fill: purple;
}
</style>
<g class=“primary”>
<rect x="20" y="20" /> <rect x="60" y="60" /> <g transform="translate(200)"> <rect x="20" y="20" /> <rect x="60" y="60" /> </g>
</g>
There’s even more to CSS than the rules I’ve laid out here, but these are the ones that are most useful or relevant when working with the Selectors and CSS dialog. Even so, it’s a lot to take on board if you’re not already familiar with CSS. Next month, it will hopefully start to make a bit more sense as we begin our look at the new dialog.