| layout | permalink |
|---|---|
default |
docs/selectors.html |
Stylus is "pythonic" (i.e. indentation-based). Whitespace is significant, so we substitute { and } with an indent, and an outdent as shown below:
body
color white
Which compiles to:
body {
color: #fff;
}
If preferred, you can use colons to separate properties and values:
body
color: white
Stylus, just like CSS, allows you to define properties for several selectors at once through comma separation.
textarea, input
border 1px solid #eee
The same can be done with a newline:
textarea
input
border 1px solid #eee
Both compile to:
textarea,
input {
border: 1px solid #eee;
}
The only exception to this rule are selectors that look like properties. For example, the following foo bar baz might be a property or a selector:
foo bar baz
> input
border 1px solid
So for this reason (or simply if preferred), we may trail with a comma:
foo bar baz,
form input,
> a
border 1px solid
The & character references the parent selector(s). In the example below our two selectors (textarea and input) both alter the color on the :hover pseudo selector.
textarea
input
color #A7A7A7
&:hover
color #000
Compiles to:
textarea,
input {
color: #a7a7a7;
}
textarea:hover,
input:hover {
color: #000;
}
Below is an example providing a simple 2px border for Internet Explorer utilizing the parent reference within a mixin:
box-shadow()
-webkit-box-shadow arguments
-moz-box-shadow arguments
box-shadow arguments
html.ie8 &,
html.ie7 &,
html.ie6 &
border 2px solid arguments[length(arguments) - 1]
body
#login
box-shadow 1px 1px 3px #eee
Yielding:
body #login {
-webkit-box-shadow: 1px 1px 3px #eee;
-moz-box-shadow: 1px 1px 3px #eee;
box-shadow: 1px 1px 3px #eee;
}
html.ie8 body #login,
html.ie7 body #login,
html.ie6 body #login {
border: 2px solid #eee;
}
If you'd need to use the ampersand symbol in a selector without it behaving like a parent reference, you can just escape it:
.foo[title*='\&']
// => .foo[title*='&']
^[N] anywhere in a selector, where N can be a number, represents a partial reference.
Partial reference works similar to the parent reference, but while parent reference contains the whole selector, partial selectors contain only the first merged N levels of the nested selectors, so you could access those nesting levels individually.
The ^[0] would give you the selector from the first level, the ^[1] would give you the rendered selector from the second level and so on:
.foo
&__bar
width: 10px
^[0]:hover &
width: 20px
would be rendered as
.foo__bar {
width: 10px;
}
.foo:hover .foo__bar {
width: 20px;
}
Negative values are counting from the end, so ^[-1] would give you the last selector from the chain before &:
.foo
&__bar
&_baz
width: 10px
^[-1]:hover &
width: 20px
would be rendered as
.foo__bar_baz {
width: 10px;
}
.foo__bar:hover .foo__bar_baz {
width: 20px;
}
Negative values are especially helpful for usage inside mixins when you don't know at what nesting level you're calling it.
Note that partial reference contain the whole rendered chain of selectors until the given nesting level, not the “part” of the selector.
^[N..M] anywhere in a selector, where both N and M can be numbers, represents a partial reference.
If you'd have a case when you'd need to get the raw part of the selector, or to get the range of parts programmatically, you could use ranges inside partial reference.
If the range would start from the positive value, the result won't contain the selectors of the previous levels and you'd get the result as if the selectors of those levels were inserted at the root of the stylesheet with the combinators omitted:
.foo
& .bar
width: 10px
^[0]:hover ^[1..-1]
width: 20px
would be rendered as
.foo .bar {
width: 10px;
}
.foo:hover .bar {
width: 20px;
}
One number in the range would be the start index, the second — the end index. Note that the order of those numbers won't matter as the selectors would always render from the first levels to the last, so ^[1..-1] would be equal to the ^[-1..1].
When both numbers are equal, the result would be just one raw level of a selector, so you could replace ^[1..-1] in a previous example to ^[-1..-1], and it would be equal to the same last one raw selector, but would be more reliable if used inside mixins.
The ~/ characters at the start of a selector can be used to point at the selector at the first nesting and could be considered as a shortcut to ^[0]. The only drawback is that you can use initial reference only at the start of a selector:
.block
&__element
~/:hover &
color: red
Would be rendered as
.block:hover .block__element {
color: #f00;
}
The ../ characters at the start of a selector mark a relative reference, which points to the previous to the & compiled selector. You can nest relative reference: ../../ to get deeper levels, but note that it can be used only at the start of the selector.
.foo
.bar
width: 10px
&,
../ .baz
height: 10px
would be rendered as
.foo .bar {
width: 10px;
}
.foo .bar,
.foo .baz {
height: 10px;
}
Relative references can be considered as shortcuts to the partial references with ranges like ^[0..-(N + 1)] where the N is the number of relative references used.
The / character at the start of a selector is a root reference. It references the root context and this means the selector won't prepend the parent's selector to it (unless you would use it with &). It is helpful when you need to write some styles both to some nested selector and to another one, not in the current scope.
textarea
input
color #A7A7A7
&:hover,
/.is-hovered
color #000
Compiles to:
textarea,
input {
color: #a7a7a7;
}
textarea:hover,
input:hover,
.is-hovered {
color: #000;
}
You can use the built-in selector() to get the current compiled selector. Could be used inside mixins for checks or other clever things.
.foo
selector()
// => '.foo'
.foo
&:hover
selector()
// '.foo:hover'
This bif could also accept an optional string argument, in this case it would return the compiled selector. Note that it wouldn't prepend the selector of the current scope in case it don't have any & symbols.
.foo
selector('.bar')
// => '.bar'
.foo
selector('&:hover')
// '.foo:hover'
selector() bif can accept multiple values or a comma-separated list in order to create a nested selector structure easier.
{selector('.a', '.b', '.c, .d')}
color: red
would be equal to the
.a
.b
.c,
.d
color: red
and would be rendered as
.a .b .c,
.a .b .d {
color: #f00;
}
This bif returns a comma-separated list of nested selectors for the current level:
.a
.b
&__c
content: selectors()
would be rendered as
.a .b__c {
content: '.a', '& .b', '&__c';
}
Expressions such as margin - n could be interpreted both as a subtraction operation, as well as a property with an unary minus. To disambiguate, wrap the expression with parens:
pad(n)
margin (- n)
body
pad(5px)
Compiles to:
body {
margin: -5px;
}
However, this is only true in functions (since functions act both as mixins, or calls with return values).
For example, the following is fine (and yields the same results as above):
body
margin -5px
Have weird property values that Stylus can't process? unquote() can help you out:
filter unquote('progid:DXImageTransform.Microsoft.BasicImage(rotation=1)')
Yields:
filter progid:DXImageTransform.Microsoft.BasicImage(rotation=1)