Selectors

Abstract

CSS (Cascading Style Sheets) is a language for describing the rendering of HTML and XML documents on screen, on paper, in speech, etc. To bind style properties to elements in the document, CSS uses selectors, which are patterns that match one or more elements. This document describes the selectors that are proposed for CSS level 3. It includes and extends the selectors of CSS level 2.

Status of this document

This document is one of the "modules" of the upcoming CSS3 specification. It not only describes the selectors that already exist in CSS1 and CSS2, but also proposes new selectors for CSS3 as well as for other languages that may need them. The CSS Working Group doesn't expect that all implementations of CSS3 will have to implement all selectors. Instead, there will probably be a small number of variants of CSS3, so-called "profiles". For example, it may be that only a profile for non-interactive user agents will include all of the proposed selectors.

This specification is being put forth as a Candidate Recommendation by the CSS Working Group. This document is a revision of the Working Draft dated 2001 January 26, and has incorporated suggestions received during last call review, comments, and further deliberations of the W3C CSS Working Group.

The duration of Candidate Recommendation is expected to last approximately six months (ending May, 2002). All persons are encouraged to review and implement this specification and return comments to the (archived) public mailing list www-style (see instructions). W3C Members can also send comments directly to the CSS Working Group.

Should this specification prove impossible to implement, the Working Group will return the document to Working Draft status and make necessary changes. Otherwise, the Working Group anticipates asking the W3C Director to advance this document to Proposed Recommendation.

This is still a draft document and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to cite a W3C Candidate Recommendation as other than "work in progress." A list of current W3C working drafts can be found at http://www.w3.org/TR.

This document may be available in translation. The English version of this specification is the only normative version.

Dependencies with other CSS3 Modules

• General Syntax
• Value Assignment, Cascade and Inheritance
• Generated Content / Markers
• User Interface

1. Context

Members of the CSS+FP Working Group proposed during the Clamart meeting to modularize the CSS specification.

This modularization, and the externalization of the general syntax of CSS will reduce the size of the specification and allow new specifications to use selectors and/or CSS general syntax. For instance, behaviors or tree transformations.

This specification contains its own test cases, one test per concept introduced in this document. These tests are not full conformance tests but are intended to provide users with a way to check if a part of this specification is implemented ad minima or is not implemented at all.

1.1 Changes from CSS2

The main differences between the selectors in CSS2 and those in Selectors are:

• the list of basic definitions (selector, group of selectors, simple selector, etc.) has been clarified
• an optional namespace component is now allowed in type element selectors, the universal selector and attribute selectors
• a new combinator
• new simple selectors including substring matching attribute selectors, and new pseudo-classes
• new pseudo-elements, and introduction of the "::" convention for pseudo-elements
• a rewriting of the selectors grammar
• profiles to be added to specifications integrating Selectors and defining the set of selectors which is actually supported by each specification
• Selectors are now a CSS3 Module and an independent specification. Other specifications can now refer to this document independently of CSS
• the specification now contains its own test suite.

2. Selectors

A Selector represents a structure. This structure can be used as a condition (e.g. in a CSS rule) that determines which elements a selector matches in the document tree, or as a flat description of the HTML or XML fragment corresponding to that structure.

Selectors may range from simple element names to rich contextual representations.

The following table summarizes Selector syntax:

The meaning of each selector is derived from the table above by prepending "matches" to the contents of each cell of the "Meaning" column.

3. Case sensitivity

The case-sensitivity of document language element names in selectors depends on the document language. For example, in HTML, element names are case-insensitive, but in XML they are case-sensitive.

The case-sensitivity of attribute names and attribute values in attribute selectors also depends on the document language.

4. Selector syntax

A selector is a chain of one or more sequences of simple selectors separated by combinators.

A sequence of simple selectors is a chain of simple selectors that are not separated by a combinator. It always begins with a type selector or a universal selector. No other type selector or universal selector is allowed in the sequence.

A simple selector is either a type selector, universal selector, attribute selector, ID selector, content selector, or pseudo-class. One pseudo-element may be appended to the last sequence of simple selectors.

Combinators are: white space, "greater-than sign" (>), "plus sign" (+) and "tilde" (~). White space may appear between a combinator and the simple selectors around it. Only the characters "space" (Unicode code 32), "tab" (9), "line feed" (10), "carriage return" (13), and "form feed" (12) can occur in white space. Other space-like characters, such as "em-space" (8195) and "ideographic space" (12288), are never part of white space.

The elements of the document tree represented by a selector are called subjects of the selector. A selector consisting of a single sequence of simple selectors represents any element satisfying its requirements. Prepending another sequence of simple selectors and a combinator to a sequence imposes additional matching constraints, so the subjects of a selector are always a subset of the elements represented by the rightmost sequence of simple selectors.

Note: an empty selector, containing no sequence of simple selectors and no combinator, is an invalid selector.

5. Groups of selectors

When several selectors share the same declarations, they may be grouped into a comma-separated list.

h1 { font-family: sans-serif }
h2 { font-family: sans-serif }
h3 { font-family: sans-serif }

h1, h2, h3 { font-family: sans-serif }

Warning: the equivalence is true in this example because all selectors are valid selectors. If just one of these selectors is invalid, the entire group of selectors is invalid thus invalidating the rule for all three heading elements, whereas only one of the three individual heading rules would be invalid.

6. Simple selectors

6.1 Type selector

A type selector is the name of a document language element type. A type selector represents an instance of the element type in the document tree.

h1

6.1.1 Type selectors and Namespaces

Type selectors allow an optional namespace ([XML-NAMES]) component. A namespace prefix that has been previously declared may be prepended to the element name separated by the namespace separator "vertical bar" (|). The namespace component may be left empty to indicate that the selector is only to represent elements with no declared namespace. Furthermore, an asterisk may be used for the namespace prefix, indicating that the selector represents elements in any namespace (including elements with no namespace). Element type selectors that have no namespace component (no namespace separator), represent elements without regard to the element's namespace (equivalent to "*|") unless a default namespace has been declared. In that case, the selector will represent only elements in the default namespace.

Note : a type selector containing a namespace prefix that has not been previously declared is an invalid selector. The mechanism for declaring a namespace prefix is left up to the language implementing Selectors. In CSS, such a mechanism is defined in the General Syntax module.

It should be noted that if a namespace prefix used in a selector has not been previously declared, then the selector must be considered invalid and the entire style rule will be ignored in accordance with the standard error handling rules.

It should further be noted that in a namespace aware client, element type selectors will only match against the local part of the element's qualified name. See below for notes about matching behaviors in down-level clients.

In summary:

ns|E

elements with name E in namespace ns

*|E

elements with name E in any namespace, including those without any declared namespace

|E

elements with name E without any declared namespace

E

if no default namespace has been specified, this is equivalent to *|E. Otherwise it is equivalent to ns|E where ns is the default namespace.

@namespace foo url(http://www.example.com);

foo|h1 { color: blue }

foo|* { color: yellow }

|h1 { color: red }

*|h1 { color: green }

h1 { color: green }

6.2 Universal selector

The universal selector, written "asterisk" (*), represents the qualified name of any element type. It represents then any single element in the document tree in any namespace (including those without any declared namespace) if no default namespace has been specified. If a default namespace has been specified, see Universal selector and Namespaces below.

If the universal selector is not the only component of a sequence of simple selectors, the * may be omitted. For example:

Note: it is recommended that the *, representing the universal selector, not be omitted.

6.2.1 Universal selector and Namespaces

The universal selector allows an optional namespace component.

ns|*

all elements in namespace ns

*|*

all elements

|*

all elements without any declared namespace

*

if no default namespace has been specified, this is equivalent to *|*. Otherwise it is equivalent to ns|* where ns is the default namespace.

Note: a universal selector containing a namespace prefix that has not been previously declared is an invalid selector. The mechanism for declaring a namespace prefix is left up to the language implementing Selectors. In CSS, such a mechanism is defined in the General Syntax module.

6.3 Attribute selectors

Selectors allow the representation of an element's attributes.

6.3.1 Attribute presence and values selectors

CSS2 introduced four attribute selectors:

[att]

Represents the att attribute, whatever the value of the attribute.

[att=val]

Represents the att attribute with value exactly "val".

[att~=val]

Represents the att attribute whose value is a space-separated list of words, one of which is exactly "val". If this selector is used, the words in the value must not contain spaces (since they are separated by spaces).

[att|=val]

Represents the att attribute, its value either being exactly "val" or beginning with "val" immediately followed by "-". This is primarily intended to allow language subcode matches (e.g., the hreflang attribute on the link element in HTML) as described in RFC 3066 ([RFC3066]). Note: for lang (or xml:lang) language subcode matching, please see the :lang pseudo-class.

Attribute values must be identifiers or strings. The case-sensitivity of attribute names and values in selectors depends on the document language.

h1[title]

span[class=example]

span[hello="Cleveland"][goodbye="Columbus"]

a[rel~="copyright"]
a[href="http://www.w3.org/"]

link[hreflang=fr]

link[hreflang|="en"]

DIALOGUE[character=romeo] 

DIALOGUE[character=juliet]

6.3.2 Substring matching attribute selectors

Three additional attribute selectors are provided for matching substrings in the value of an attribute:

[att^=val]

Represents the att attribute whose value begins with the prefix "val"

[att$=val]

Represents the att attribute whose value ends with the suffix "val"

[att*=val]

Represents the att attribute whose value contains at least one instance of the substring "val"

Attribute values must be identifiers or strings. The case-sensitivity of attribute names in selectors depends on the document language.

Example:

The following selector represents an HTML object, referencing an image:

object[type^="image/"]

The following selector represents an HTML anchor a with an href attribute whose value ends with ".html".

a[href$=".html"]

The following selector represents a HTML paragraph with a title attribute whose value contains the substring "hello"

p[title*="hello"] 

6.3.3 Attribute selectors and Namespaces

Attribute selectors allow an optional namespace component to the attribute name. A namespace prefix that has been previously declared may be prepended to the attribute name separated by the namespace separator "vertical bar" (|). In keeping with the Namespaces in the XML recommendation, default namespaces do not apply to attributes, therefore attribute selectors without a namespace component apply only to attributes that have no declared namespace (equivalent to "|attr"). An asterisk may be used for the namespace prefix indicating that the selector is to match all attribute names without regard to the attribute's namespace.

Note : an attribute selector with an attribute name containing a namespace prefix that has not been previously declared is an invalid selector. The mechanism for declaring a namespace prefix is left up to the language implementing Selectors. In CSS, such a mechanism is defined in the General Syntax module.

CSS examples:

@namespace foo "http://www.example.com";

[foo|att=val] { color: blue }

[*|att] { color: yellow }

[|att] { color: green }

[att] { color: green }

6.3.4 Default attribute values in DTDs

Attribute selectors represent explicitly set attribute values in the document tree. Default attribute values may be defined in a DTD or elsewhere. Selectors should be designed so that they work even if the default values are not included in the document tree.

<!ATTLIST EXAMPLE notation (decimal,octal) "decimal">

EXAMPLE[notation=decimal]
EXAMPLE[notation=octal]

EXAMPLE:not([notation])

6.4 Class selectors

Working with HTML, authors may use the period (.) notation as an alternative to the ~= notation when representing the class attribute. Thus, for HTML, div.value and div[class~=value] have the same meaning. The attribute value must immediately follow the "period" (.). Note: UAs may apply selectors using the period (.) notation in XML documents if the UA has namespace specific knowledge that allows it to determine which attribute is the "class" attribute for the respective namespace. One such example of namespace specific knowledge is the prose in the specification for a particular namespace (e.g. SVG 1.0 [SVG] describes the SVG "class" attribute and how a UA should interpret it, and similarly MathML 1.01 [MATH] describes the MathML "class" attribute.)

*.pastoral

.pastoral

h1.pastoral

p.pastoral.marine

p.marine.pastoral

6.5 ID selectors

Document languages may contain attributes that are declared to be of type ID. What makes attributes of type ID special is that no two such attributes can have the same value in a document, regardless of the type of the elements that carry them; whatever the document language, an ID typed attribute can be used to uniquely identify its element. In HTML all ID attributes are named "id"; XML applications may name ID attributes differently, but the same restriction applies.

An ID typed attribute of a document language allows authors to assign an identifier to one element instance in the document tree. W3C ID selectors represent an element instance based on its identifier. An ID selector contains a "number sign" (#) immediately followed by the ID value.

h1#chapter1

#chapter1

*#z98y

6.6 Pseudo-classes

The pseudo-class concept is introduced to permit selection based on information that lies outside of the document tree or that cannot be expressed using the other simple selectors.

A pseudo-class always contains a "colon" (:) followed by the name of the pseudo-class and optionally by a value between parentheses.

Pseudo-classes are allowed in all sequences of simple selectors contained in a selector. Pseudo-classes are allowed anywhere in sequences of simple selectors, after the leading type selector or universal selector (possibly omitted). Pseudo-class names are case-insensitive. Some pseudo-classes are mutually exclusive, while others can be applied simultaneously to the same element. Pseudo-classes may be dynamic, in the sense that an element may acquire or lose a pseudo-class while a user interacts with the document.

6.6.1 Dynamic pseudo-classes

Dynamic pseudo-classes classify elements on characteristics other than their name, attributes or content, in principle characteristics that cannot be deduced from the document tree.

Dynamic pseudo-classes do not appear in the document source or document tree.

The link pseudo-classes: :link and :visited

User agents commonly display unvisited links differently from previously visited ones. Selectors provides the pseudo-classes :link and :visited to distinguish them:

• The :link pseudo-class applies for links that have not yet been visited.
• The :visited pseudo-class applies once the link has been visited by the user.

The two states are mutually exclusive.

a.external:visited

The user action pseudo-classes :hover, :active, and :focus

Interactive user agents sometimes change the rendering in response to user actions. Selectors provides three pseudo-classes for the selection of an element the user is acting on.

• The :hover pseudo-class applies while the user designates an element (with some pointing device), but does not activate it. For example, a visual user agent could apply this pseudo-class when the cursor (mouse pointer) hovers over a box generated by the element. User agents not supporting interactive media do not have to support this pseudo-class. Some conforming user agents supporting interactive media may not be able to support this pseudo-class (e.g., a pen device).
• The :active pseudo-class applies while an element is being activated by the user. For example, between the times the user presses the mouse button and releases it.
• The :focus pseudo-class applies while an element has the focus (accepts keyboard or mouse events, or other forms of input).

There may be document language or implementation specific limits on which elements can become :active or acquire :focus.

These pseudo-classes are not mutually exclusive. An element may match several of them at the same time.

a:link    /* unvisited links */
a:visited /* visited links   */
a:hover   /* user hovers     */
a:active  /* active links    */

a:focus
a:focus:hover

6.6.2 The target pseudo-class :target

Some URIs refer to a location within a resource. This kind of URI ends with a "number sign" (#) followed by an anchor identifier (called the fragment identifier).

URIs with fragment identifiers link to a certain element within the document, known as the target element. For instance, here is a URI pointing to an anchor named section_2 in a HTML document:

http://example.com/html/top.html#section_2

A target element can be represented by the :target pseudo-class:

p.note:target

represents a p of class note that is the target element of the referring URI.

*:target { color : red }

*:target::before { content : url(target.png) }

6.6.3 The language pseudo-class :lang

If the document language specifies how the human language of an element is determined, it is possible to write selectors that represent an element based on its language. For example, in HTML [HTML4.01], the language is determined by a combination of the lang attribute, the meta element, and possibly by information from the protocol (such as HTTP headers). XML uses an attribute called xml:lang, and there may be other document language-specific methods for determining the language.

The pseudo-class :lang(C) represents an element that is in language C. Here C is a language code as specified in HTML 4.01 [HTML4.01] and RFC 3066 [RFC3066].

html:lang(fr-be)
html:lang(de)
:lang(fr-be) > q
:lang(de) > q

6.6.4 The UI element states pseudo-classes

The :enabled and :disabled pseudo-classes

The purpose of the :enabled pseudo-class is to allow authors to customize the look of user interface elements which are enabled - which the user can select/activate in some fashion (e.g. clicking on a button with a mouse). There is a need for such a pseudo-class because there is no way to programmatically specify the default appearance of say, an enabled input element without also specifying what it would look like when it was disabled.

Similar to :enabled, :disabled allows the author to specify precisely how a disabled or inactive user interface element should look.

It should be noted that most elements will be neither enabled nor disabled. An element is enabled if the user can either activate it or transfer the focus to it. An element is disabled if it could be enabled, but the user cannot presently activate it or transfer focus to it.

The :checked pseudo-class

Radio and checkbox elements can be toggled by the user. Some menu items are "checked" when the user selects them. When such elements are toggled "on" the :checked pseudo-class applies. The :checked pseudo-class initially applies to such elements that have the HTML4 selected attribute as described in Section 17.2.1 of HTML4, but of course the user can toggle "off" such elements in which case the :checked pseudo-class would no longer apply. While the :checked pseudo-class is dynamic in nature, and is altered by user action, since it can also be based on the presence of the semantic HTML4 selected attribute, it applies to all media.

The :indeterminate pseudo-class

Radio and checkbox elements can be toggled by the user, but are sometimes in an indeterminate state, neither checked nor unchecked. This can be due to an element attribute, or DOM manipulation. The :indeterminate pseudo-class applies to such elements. While the :indeterminate pseudo-class is dynamic in nature, and is altered by user action, since it can also be based on the presence of an element attribute, it applies to all media.

Components of a radio-group initialized with no pre-selected choice are an example of :indeterminate state.

6.6.5 Structural pseudo-classes

Selectors introduces the concept of structural pseudo-classes to permit selection based on extra information that lies in the document tree but cannot be represented by other simple selectors or combinators.

Note that standalone PCDATA are not counted when calculating the position of an element in the list of children of its parent. When calculating the position of an element in the list of children of its parent, the index numbering starts at 1.

:root pseudo-class

The :root pseudo-class represents an element that is the root of the document. In HTML 4, this is the HTML element. In XML, it is whatever is appropriate for the DTD or schema and namespace for that XML document.

:nth-child() pseudo-class

The :nth-child(an+b) pseudo-class notation represents an element that has an+b-1 siblings before it in the document tree, for a given positive integer or zero value of n. In other words, this matches the bth child of an element after all the children have been split into groups of a elements each. For example, this allows the selectors to address every other row in a table, and could be used, for example, to alternate the color of paragraph text in a cycle of four. The a and b values must be zero, negative integers or positive integers. The index of the first child of an element is 1.

In addition to this, :nth-child() can take 'odd' and 'even' for argument. 'odd' has the same signification as 2n+1, and 'even' has the same signification as 2n.

tr:nth-child(2n+1) /* represents every odd row of a HTML table */
tr:nth-child(odd)  /* same */
tr:nth-child(2n)   /* represents every even row of a HTML table */
tr:nth-child(even) /* same */

/* Alternate paragraph colours in CSS */
p:nth-child(4n+1) { color: navy; }
p:nth-child(4n+2) { color: green; }
p:nth-child(4n+3) { color: maroon; }
p:nth-child(4n+4) { color: purple; }

When a=0, no repeating is used, so for example :nth-child(0n+5) matches only the fifth child. When a=0, the a part need not be included, so the syntax simplifies to :nth-child(b) and the last example simplifies to :nth-child(5).

foo:nth-child(0n+1)   /* represents an element foo, first child of its parent element */
foo:nth-child(1)      /* same */

When a=1, the number may be omitted from the rule, so the following examples are equivalent:

bar:nth-child(1n+0)   /* represents all bar elements, specificity (0,1,1) */
bar:nth-child(n+0)    /* same */
bar:nth-child(n)      /* same */
bar                   /* same but lower specificity (0,0,1) */

If b=0, then every a-th element is picked:

tr:nth-child(2n) /* represents every even row of a HTML table */

If both a and b are equal to zero, the pseudo-class represents no element in the document tree.

The value a can be negative, but only the positive values of an+b, for n>= 0, may represent an element in the document tree, of course:

html|tr:nth-child(-n+6)  /* represents the 6 first rows of XHTML tables */

:nth-last-child() pseudo-class

The :nth-last-child(an+b) pseudo-class notation represents an element that has an+b-1 siblings after it in the document tree, for a given positive integer or zero value of n. See :nth-child() pseudo-class for the syntax of its argument. It also accepts the 'even' and 'odd' values for argument.

tr:nth-last-child(-n+2)    /* represents the two last rows of a HTML table */

foo:nth-last-child(odd)    /* represents all odd foo elements in their parent element,
                              counting from the last one */

:nth-of-type() pseudo-class

The :nth-of-type(an+b) pseudo-class notation represents an element that has an+b-1 siblings with the same element name before it in the document tree, for a given zero or positive integer value of n. In other words, this matches the bth child of that type after all the children of that type have been split into groups of a elements each. See :nth-child() pseudo-class for the syntax of its argument. It also accepts the 'even' and 'odd' values for argument.

img:nth-of-type(2n+1) { float: right; }
img:nth-of-type(2n) { float: left; }

:nth-last-of-type() pseudo-class

The :nth-last-of-type(an+b) pseudo-class notation represents an element that has an+b-1 siblings with the same element name after it in the document tree, for a given zero or positive integer value of n. See :nth-child() pseudo-class for the syntax of its argument. It also accepts the 'even' and 'odd' values for argument.

body > h2:nth-of-type(n+2):nth-last-of-type(n+2)

body > h2:not(:first-of-type):not(:last-of-type) 

:first-child pseudo-class

Same as :nth-child(1). The :first-child pseudo-class represents an element that is the first child of some other element.

div > p:first-child

<p> The last P before the note.</p>
<div class="note">
   <p> The first P inside the note.</p>
</div>

<p> The last P before the note.</p>
<div class="note">
   <h2>Note</h2>
   <p> The first P inside the note.</p>
</div>

* > a:first-child   /* a is first child of any element */
a:first-child       /* Same */

:last-child pseudo-class

Same as :nth-last-child(1).The :last-child pseudo-class represents an element that is the last child of some other element.

The following selector represents a list item li that is the last child of an ordered list ol.

ol > li:last-child

:first-of-type pseudo-class

Same as :nth-of-type(1).The :first-of-type pseudo-class represents an element that is the first sibling of its type in the list of children of its parent element.

dl dt:first-of-type

<dl><dt>gigogne</dt>
        <dd><dl><dt>fus&eacute;e</dt>
                    <dd>multistage rocket</dd>
                <dt>table</dt>
                    <dd>nest of tables</dd>
            </dl></dd>
</dl>

:last-of-type pseudo-class

Same as :nth-last-of-type(1).The :last-of-type pseudo-class represents an element that is the last sibling of its type in the list of children of its parent element.

tr > td:last-of-type

:only-child pseudo-class

Represents an element that has no siblings. Same as :first-child:last-child or :nth-child(1):nth-last-child(1), but with a lower specificity.

:only-of-type pseudo-class

Represents an element that has no siblings with the same element name. Same as :first-of-type:last-of-type or :nth-of-type(1):nth-last-of-type(1), but with a lower specificity.

:empty pseudo-class

The :empty pseudo-class represents an element that has no children at all, including possibly empty text nodes, from a DOM point of view.

<p></p>

<foo>bar</foo>

<foo><bar>bla</bar></foo>

<foo>this is not <bar>:empty</bar></foo>

6.6.6 Content pseudo-class

The :contains("foo") pseudo-class notation represents an element whose textual contents contain the given substring. The argument of this pseudo-class can be a string (surrounded by double quotes) or a keyword.

Usage of the content pseudo-class is restricted to static media types (see [CSS2]).

The textual contents of a given element is determined by the concatenation of all PCDATA contained in the element and sub-elements.

p:contains("Markup")

<p><strong>H</strong>yper<strong>t</strong>ext
   <strong>M</strong><em>arkup</em>
   <strong>L</strong>anguage</p>

Special characters can be inserted in the argument of a content pseudo-class using the escape mechanism for Unicode characters and carriage returns.

Warning: the selector ul:contains("chief") will match the list <ul><li>... the greek letter chi</li><li>effective</li></ul>

P:contains("Markup") { background-color : red }

6.6.7 The negation pseudo-class

The negation pseudo-class is a functional notation taking a simple selector (excluding the negation pseudo-class itself and pseudo-elements) as an argument. It represents an element that is not represented by the argument.

button:not([DISABLED])

*:not(FOO)

html|*:not(:link):not(:visited)

Note: the :not() pseudo allows useless selectors to be written. For instance :not(*|*), which represents no element at all, or foo:not(bar), which is equivalent to foo but with a higher specificity.

7. Pseudo-elements

Pseudo-elements create abstractions about the document tree beyond those specified by the document language. For instance, document languages do not offer mechanisms to access the first letter or first line of an element's content. Pseudo-elements allow designers to refer to this otherwise inaccessible information. Pseudo-elements may also provide designers a way to refer to content that does not exist in the source document (e.g., the ::before and ::after pseudo-elements give access to generated content).

A pseudo-element is made of two colons (::) followed by the name of the pseudo-element.

Note: this :: notation is introduced by the current document in order to establish a discrimination between pseudo-classes and pseudo-elements. For compatibility with existing style sheets, user agents must also accept the previous one-colon notation for pseudo-elements introduced in CSS levels 1 and 2. This compatibility is not allowed for the new pseudo-elements introduced in CSS level 3.

Pseudo-elements may only appear once in the sequence of simple selectors that represents the subjects of the selector.

7.1 The ::first-line pseudo-element

The ::first-line pseudo-element describes the first formatted line of an element.

For instance in CSS:

p::first-line { text-transform: uppercase }

The above rule means "change the letters of the first line of every paragraph to uppercase". However, the selector p::first-line does not match any real HTML element. It does match a pseudo-element that conforming user agents will insert at the beginning of every paragraph.

Note that the length of the first line depends on a number of factors, including the width of the page, the font size, etc. Thus, an ordinary HTML paragraph such as:

<p>This is a somewhat long HTML
paragraph that will be broken into several
lines. The first line will be identified
by a fictional tag sequence. The other lines
will be treated as ordinary lines in the
paragraph.</p>

the lines of which happen to be rendered as follows if the style rule above applies:

THIS IS A SOMEWHAT LONG HTML PARAGRAPH THAT
will be broken into several lines. The first
line will be identified by a fictional tag
sequence. The other lines will be treated as
ordinary lines in the paragraph.

might be "rewritten" by user agents to include the fictional tag sequence for ::first-line. This fictional tag sequence helps to show how properties are inherited.

<p><p::first-line> This is a somewhat long HTML
paragraph that</p::first-line> will be broken into several
lines. The first line will be identified
by a fictional tag sequence. The other lines
will be treated as ordinary lines in the
paragraph.</p>

If a pseudo-element breaks up a real element, the desired effect can be described by closing and then re-opening the fictional tag sequence. Thus, if we mark up the previous paragraph with a span element:

<p><span class="test"> This is a somewhat</span> long HTML
paragraph that will be broken into several
lines. The first line will be identified
by a fictional tag sequence. The other lines
will be treated as ordinary lines in the
paragraph.</p>

the user agent could generate the appropriate start and end tags for the fictional tag sequence for ::first-line.

<p><span class="test"><p::first-line> This is a
somewhat</p::first-line></span><p::first-line>
long HTML paragraph that</p::first-line> will be broken into
several lines. The first line will be identified
by a fictional tag sequence. The other lines
will be treated as ordinary lines in the
paragraph.</p>

The ::first-line pseudo-element can only be attached to a block-level element.

The ::first-line pseudo-element is similar to an inline-level element, but with certain restrictions, depending on usage. Only the following properties apply to a ::first-line pseudo-element: font properties, color properties, background properties, 'word-spacing', 'letter-spacing', 'text-decoration', 'vertical-align', 'text-transform', 'line-height', 'text-shadow', and 'clear'.

7.2 The ::first-letter pseudo-element

The ::first-letter pseudo-element describes the first formatted letter of an element.

The ::first-letter pseudo-element can be attached to all elements.

The ::first-letter pseudo-element may be used for "initial caps" and "drop caps", which are common typographical effects. This type of initial letter is similar to an inline-level element if its CSS 'float' property is 'none', but with certain restrictions, depending on usage. Otherwise it is similar to a floated element.

These are the CSS properties that apply to ::first-letter pseudo-elements: font properties, color properties, background properties, 'text-decoration', 'vertical-align' (only if 'float' is 'none'), 'text-transform', 'line-height', margin properties, padding properties, border properties, 'float', 'text-shadow', and 'clear'.

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN">
<HTML>
 <HEAD>
  <TITLE>Drop cap initial letter</TITLE>
  <STYLE type="text/css">
   P               { font-size: 12pt; line-height: 12pt }
   P::first-letter { font-size: 200%; font-style: italic;
                     font-weight: bold; float: left }
   SPAN            { text-transform: uppercase }
  </STYLE>
 </HEAD>
 <BODY>
  <P><SPAN>The first</SPAN> few words of an article
    in The Economist.</P>
 </BODY>
</HTML>

<P>
<SPAN>
<P::first-letter>
T
</P::first-letter>he first
</SPAN>
few words of an article in the Economist.
</P>

In order to achieve traditional drop caps formatting, user agents may approximate font sizes, for example to align baselines. Also, the glyph outline may be taken into account when formatting.

Punctuation (i.e, characters defined in Unicode [UNICODE] in the "open" (Ps), "close" (Pe), and "other" (Po) punctuation classes), that precedes the first letter should be included, as in:

The ::first-letter pseudo-element matches parts of elements only.

Some languages may have specific rules about how to treat certain letter combinations. In Dutch, for example, if the letter combination "ij" appears at the beginning of a word, both letters should be considered within the ::first-letter pseudo-element. The ::first-letter pseudo-element should select select from beginning of element up to the first non-opening-punctuation character cluster.

P { color: red; font-size: 12pt }
P::first-letter { color: green; font-size: 200% }
P::first-line { color: blue }

<P>Some text that ends up on two lines</P>

<P>
<P::first-line>
<P::first-letter>
S
</P::first-letter>ome text that
</P::first-line>
ends up on two lines
</P>

7.3 The UI element fragments pseudo-elements

The ::selection pseudo-element

The ::selection pseudo-element applies to the portion of a document that has been highlighted by the user. This also applies, for example, to selected text within an editable text field. This pseudo-element should not be confused with the :checked pseudo-class (which used to be named :selected)

Although the ::selection pseudo-element is dynamic in nature, and is altered by user action, it is reasonable to expect that when a UA rerenders to a static medium (such as a printed page, see [CSS2]) which was originally rendered to a dynamic medium (like screen), the UA may wish to transfer the current ::selection state to that other medium, and have all the appropriate formatting and rendering take effect as well. This is not required - UAs may omit the ::selection pseudo-element for static media.

These are the CSS properties that apply to ::selection pseudo-elements: color, cursor, background, outline. The computed value of the 'background-image' property on ::selection may be ignored.

7.4 The ::before and ::after pseudo-elements

The ::before and ::after pseudo-elements can be used to describe generated content before or after an element's content. They are explained in the Generated Content/Markers CSS3 Module.

When the ::first-letter and ::first-line pseudo-elements are combined with ::before and ::after, they apply to the first letter or line of the element including the inserted text.

8. Combinators

8.1 Descendant combinator

At times, authors may want selectors to describe an element that is the descendant of another element in the document tree (e.g., "an EM element that is contained within an H1 element"). Descendant combinators express such a relationship. A descendant combinator is a white space that separates two sequences of simple selectors. A selector of the form "A B" represents an element B that is an arbitrary descendant of some ancestor element A.

h1 em

<h1>This <span class="myclass">headline 
is <em>very</em> important</span></h1>

div * p

div p *[href]

8.2 Child combinators

A child combinator describes a childhood relationship between two elements. A child combinator is made of the "greater-than sign" (>) character and separates two sequences of simple selectors.

body > p

div ol>li p

8.3 Adjacent sibling combinators

There are two different adjacent sibling combinators: direct adjacent combinator and indirect adjacent combinator.

8.3.1 Direct adjacent combinators

Direct adjacent combinators are made of the "plus sign" (+) character that separates two sequences of simple selectors. The elements represented by the two sequences share the same parent in the document tree and the element represented by the first sequence immediately precedes the element represented by the second one.

math + p

h1.opener + h2

8.3.2 Indirect adjacent combinator

Indirect adjacent combinators are made of the "tilde" (~) character that separates two sequences of simple selectors. The elements represented by the two sequences share the same parent in the document tree and the element represented by the first sequence precedes (not necessarily immediately) the element represented by the second one.

h1 ~ pre

<h1>Definition of the function a</h1>
<p>Function a(x) has to be applied to all figures in the table.</p>
<pre>function a(x) = 12x/13.5</pre>

9. Calculating a selector's specificity

A selector's specificity is calculated as follows:

• negative selectors are counted like their simple selectors argument
• count the number of ID attributes in the selector (= a)
• count the number of other attributes and pseudo-classes in the selector (= b)
• count the number of element names in the selector (= c)
• ignore pseudo-elements.

Concatenating the three numbers a-b-c (in a number system with a large base) gives the specificity.

*               /* a=0 b=0 c=0 -> specificity =   0 */
LI              /* a=0 b=0 c=1 -> specificity =   1 */
UL LI           /* a=0 b=0 c=2 -> specificity =   2 */
UL OL+LI        /* a=0 b=0 c=3 -> specificity =   3 */
H1 + *[REL=up]  /* a=0 b=1 c=1 -> specificity =  11 */
UL OL LI.red    /* a=0 b=1 c=3 -> specificity =  13 */
LI.red.level    /* a=0 b=2 c=1 -> specificity =  21 */
#x34y           /* a=1 b=0 c=0 -> specificity = 100 */
#s12:not(FOO)   /* a=1 b=0 c=1 -> specificity = 101 */

10. The grammar of Selectors

10.1 Grammar

The grammar below defines the syntax of Selectors. It is globally LL(1) and can be locally LL(2) (but note that most UA's should not use it directly, since it doesn't express the parsing conventions). The format of the productions is optimized for human consumption and some shorthand notations beyond Yacc (see [YACC]) are used:

• *: 0 or more
• +: 1 or more
• ?: 0 or 1
• |: separates alternatives
• [ ]: grouping

The productions are:

selectors_group
  : selector [ ',' S* selector ]*
  ;

selector
  /* there is at least one sequence of simple selectors in a */
  /* selector and the pseudo-elements occur only in the last */
  /* sequence ; only pseudo-element may occur */
  : [ simple_selector_sequence combinator ]*
       simple_selector_sequence [ pseudo_element ]?
  ;

combinator
  /* combinators can be surrounded by white space */
  : S* [ '+' | '>' | '~' | /* empty */ ] S*
  ;

simple_selector_sequence
  /* the universal selector is optional */
  : [ type_selector | universal ]?
        [ HASH | class | attrib | pseudo_class | negation ]+ |
    type_selector | universal
  ;

type_selector
  : [ namespace_prefix ]? element_name
  ;

namespace_prefix
  : [ IDENT | '*' ]? '|'
  ;

element_name
  : IDENT
  ;

universal
  : [ namespace_prefix ]? '*'
  ;

class
  : '.' IDENT
  ;

attrib
  : '[' S* [ namespace_prefix ]? IDENT S*
        [ [ PREFIXMATCH |
            SUFFIXMATCH |
            SUBSTRINGMATCH |
            '=' |
            INCLUDES |
            DASHMATCH ] S* [ IDENT | STRING ] S*
        ]? ']'
  ;

pseudo_class
  /* a pseudo-class is an ident, or a function taking an */
  /* ident or a string or a number or a simple selector  */
  /* (excluding negation and pseudo-elements) */
  /* or a an+b expression for argument */
  : ':' [ IDENT | functional_pseudo ]
  ;

functional_pseudo
  : FUNCTION S* [ IDENT | STRING | NUMBER |
                  expression | negation_arg ] S* ')'
  ;

expression
  :  [ [ '-' | INTEGER ]? 'n' [ SIGNED_INTEGER ]? ] | INTEGER
  ;

negation_arg
  : type_selector | universal | HASH | class | attrib | pseudo_class
  ;

pseudo_element
  : [ ':' ]? ':' IDENT
  ;

10.2 Lexical scanner

The following is the tokenizer, written in Flex (see [FLEX]) notation. The tokenizer is case-insensitive.

The two occurrences of "\377" represent the highest character number that current versions of Flex can deal with (decimal 255). They should be read as "\4177777" (decimal 1114111), which is the highest possible code point in Unicode/ISO-10646.

%option case-insensitive

h                       [0-9a-f]
nonascii                [\200-\377]
unicode                 \\{h}{1,6}[ \t\r\n\f]?
escape                  {unicode}|\\[ -~\200-\377]
nmstart                 [a-z_]|{nonascii}|{escape}
nmchar                  [a-z0-9-_]|{nonascii}|{escape}
string1                 \"([\t !#$%&(-~]|\\{nl}|\'|{nonascii}|{escape})*\"
string2                 \'([\t !#$%&(-~]|\\{nl}|\"|{nonascii}|{escape})*\'

ident                   {nmstart}{nmchar}*
name                    {nmchar}+
integer                 [-]?[0-9]+
signed_integer          [-+][0-9]+
num                     {integer}|[0-9]*"."[0-9]+
string                  {string1}|{string2}
nl                      \n|\r\n|\r|\f
%%

[ \t\r\n\f]+    {return S;}

\/\*[^*]*\*+([^/][^*]*\*+)*\/   /* ignore comments */

"~="                    {return INCLUDES;}
"|="                    {return DASHMATCH;}
"^="                    (return PREFIXMATCH;)
"$="                    (return SUFFIXMATCH;)
"*="                    (return SUBSTRINGMATCH;)
{string}                {return STRING;}
{ident}                 {return IDENT;}
{ident}"("              {return FUNCTION;}
{num}                   {return NUMBER;}
{signed_integer}        {return SIGNED_INTEGER;}
{integer]               {return INTEGER;}
"#"{name}               {return HASH;}

.                       {return *yytext;}

11. Namespaces and Down-Level Clients

An important issue is the interaction of CSS selectors with XML documents in web clients that were produced prior to this document. Unfortunately, due to the fact that namespaces must be matched based on the URI which identifies the namespace, not the namespace prefix, some mechanism is required to identify namespaces in CSS by their URI as well. Without such a mechanism, it is impossible to construct a CSS style sheet which will properly match selectors in all cases against a random set of XML documents. However, given complete knowledge of the XML document to which a style sheet is to be applied, and a limited use of namespaces within the XML document, it is possible to construct a style sheet in which selectors would match elements and attributes correctly.

It should be noted that a down-level CSS client will (if it properly conforms to CSS forward compatible parsing rules) ignore all @namespace at-rules, as well as all style rules that make use of namespace qualified element type or attribute selectors. The syntax of delimiting namespace prefixes in CSS was deliberately chosen so that down-level CSS clients would ignore the style rules rather than possibly match them incorrectly.

The use of default namespaces in CSS makes it possible to write element type selectors that will function in both namespace aware CSS clients as well as down-level clients. It should be noted that down-level clients may incorrectly match selectors against XML elements in other namespaces.

The following are scenarios and examples in which it is possible to construct style sheets which would function properly in web clients that do not implement this proposal.

1. The XML document does not use namespaces.
   • In this case, it is obviously not necessary to declare or use namespaces in the style sheet. Standard CSS element type and attribute selectors will function adequately in a down-level client.
   • In a CSS namespace aware client, the default behavior of element selectors matching without regard to namespace will function properly against all elements, since no namespaces are present. However, the use of specific element type selectors that match only elements that have no namespace ("|name") will guarantee that selectors will match only XML elements that do not have a declared namespace.
2. The XML document defines a single, default namespace used throughout the document. No namespace prefixes are used in element names.
   • In this case, a down-level client will function as if namespaces were not used in the XML document at all. Standard CSS element type and attribute selectors will match against all elements.
3. The XML document does not use a default namespace, all namespace prefixes used are known to the style sheet author and there is a direct mapping between namespace prefixes and namespace URIs. (A given prefix may only be mapped to one namespace URI throughout the XML document, there may be multiple prefixes mapped to the same URI).
   • In this case, the down-level client will view and match element type and attribute selectors based on their fully qualified name, not the local part as outlined in the Type selectors and Namespaces section. CSS selectors may be declared using an escaped colon "\:" to describe the fully qualified names, e.g. "html\:h1" will match <html:h1>. Selectors using the qualified name will only match XML elements that use the same prefix. Other namespace prefixes used in the XML that are mapped to the same URI will not match as expected unless additional CSS style rules are declared for them.
   • Note that selectors declared in this fashion will only match in down-level clients. A CSS namespace aware client will match element type and attribute selectors based on the name's local part. So selectors declared with the fully qualified name will not match (unless there is no namespace prefix in the fully qualified name).

In other scenarios: when the namespace prefixes used in the XML are not known in advance by the style sheet author; or a combination of elements with no namespace are used in conjunction with elements using a default namespace; or the same namespace prefix is mapped to different namespace URIs within the same document, or in different documents; it is impossible to construct a CSS style sheet that will function properly against all elements in those documents, unless, the style sheet is written using a namespace URI syntax (as outlined in this document or similar) and the document is processed by a CSS and XML namespace aware client.

12. Profiles

Each specification using Selectors must define the subset of W3C Selectors it allows and excludes, and describe the local meaning of all the components of that subset.

Non normative examples:

h1 a[name]

13. Conformance and Requirements

This section defines conformance with the present specification only.

The inability of a user agent to implement part of this specification due to the limitations of a particular device (e.g., non interactive user agents will probably not implement dynamic pseudo-classes because they make no sense without interactivity) does not imply non-conformance.

All specifications reusing Selectors must contain a Profile listing the subset of Selectors it accepts or excludes, and describing the constraints it adds to the current specification.

Invalidity is caused by a parsing error, e.g. an unrecognized token or a token which is not allowed at the current parsing point.

User agents must observe the rules for handling parsing errors:

• a simple selector containing an undeclared namespace prefix is invalid
• a selector containing an invalid simple selector, an invalid combinator or an invalid token is invalid.
• a group of selectors containing an invalid selector is invalid.

Implementations of this specification must behave as "recipients of text data" as defined by [CWWW] when parsing selectors and attempting matches. (In particular, implementations must assume the data is normalized and must not normalize it.) Normative rules for matching strings are defined in [CWWW] and [UNICODE] and apply to implementations of this specification.

14. Tests

This specification contains a test suite allowing user agents to verify their basic conformance to the specification. This test suite does not pretend to be exhaustive and does not cover all possible combined cases of Selectors.

These tests are available [link forthcoming].

15. Acknowledgements

This specification is the product of the W3C Working Group on Cascading Style Sheets and Formatting Properties. In addition to the editors of this specification, the members of the Working Group are:

• Marc Attinasi (Netscape/AOL)
• Bert Bos (W3C)
• Tantek Çelik (Microsoft Corp.)
• Don Day (IBM)
• Martin Dürst (W3C)
• Angel Diaz (IBM)
• Daniel Glazman (Netscape/AOL from November 2000, and Electricité de France until February 2000)
• Håkon W. Lie (Opera Software from April 1999, and W3C until April 1999)
• Chris Lilley (W3C)
• Dave Raggett (W3C/Openwave Systems Inc.)
• Pierre Saslawsky (Netscape/AOL)
• Robert Stevahn (Hewlett-Packard)
• Michel Suignard (Microsoft Corp.)
• Ted Wugofski (Openwave Systems Inc.)
• Steve Zilles (Adobe)

A number of invited experts to the Working Group have significantly contributed to CSS3: L. David Baron, Tim Boland (NIST), Todd Fahrner, Daniel Glazman, Ian Hickson, Eric Meyer (The OPAL Group), Jeff Veen.

Former members of the Working Group:

• Chris Brichford (Adobe)
• Troy Chevalier (Netscape/AOL)
• Dwayne Dicks (SoftQuad)
• Ian Jacobs (W3C)
• Lorin Jurow (Quark)
• Sho Kuwamoto (Macromedia)
• Peter Linss (Netscape/AOL)
• Steven Pemberton (CWI)
• Robert Pernett (Lotus)
• Douglas Rand (SGI)
• Nisheeth Ranjan (Netscape/AOL)
• Ed Tecot (Microsoft Corp.)
• Jared Sorensen (Novell)
• Mike Wexler (Adobe)
• John Williams (Quark)
• Chris Wilson (Microsoft Corp.)

We thank all of them (members, invited experts and former members) for their efforts.

Of course, this document derives from the CSS Level 1 and CSS level 2 Recommendations. We thank all CSS1 and CSS2 authors, editors and contributors.

Dr. Hasan Ali Çelik suggested the simple and powerful syntax of the argument for :nth-child() while the Working Group was considering much more complex solutions.

The discussions on www-style@w3.org have been influential in many key issues. Especially, we would like to thank Ian Graham, David Baron, Björn Höhrmann, fantasai, Jelks Cabanis and Matthew Brealey for their active and useful participation.

16. References

1. [CSS1] Bert Bos, Håkon Wium Lie; "Cascading Style Sheets, level 1", W3C Recommendation, 17 Dec 1996, revised 11 Jan 1999
   (http://www.w3.org/TR/REC-CSS1)
2. [CSS2] Bert Bos, Håkon Wium Lie, Chris Lilley, Ian Jacobs, editors; "Cascading Style Sheets, level 2", W3C Recommendation, 12 May 1998
   (http://www.w3.org/TR/REC-CSS2/)
3. [CWWW] Martin J. Dürst, François Yergeau, Misha Wolf, Asmus Freytag, Tex Texin, editors; "Character Model for the World Wide Web", W3C Working Draft, 26 January 2001
   (http://www.w3.org/TR/2001/WD-charmod-20010126)
4. [FLEX] "Flex: The Lexical Scanner Generator", Version 2.3.7, ISBN 1882114213
5. [HTML4.01] Dave Ragget, Arnaud Le Hors, Ian Jacobs, editors; "HTML 4.01 Specification", W3C Recommendation, 24 December 1999
   (http://www.w3.org/TR/html401/)
6. [MATH] Patrick Ion, Robert Miner; "Mathematical Markup Language (MathML) 1.01", W3C Recommendation, revision of 7 July 1999
   (http://www.w3.org/1999/07/REC-MathML-19990707)
   
7. [NMSP] Peter Linss, editor; "CSS Namespace Enhancements (Proposal)", W3C Working Draft, 25 June 1999
   (http://www.w3.org/1999/06/25/WD-css3-namespace-19990625/)
8. [RFC3066] H. Alvestrand; "Tags for the Identification of Languages", Request for Comments 3066, January 2001
   (http://www.ietf.org/rfc/rfc3066.txt)
9. [STTS3] Daniel Glazman ; "Simple Tree Transformation Sheets 3", Electricité de France, submission to the W3C, 11 Nov 1998
   (http://www.w3.org/TR/NOTE-STTS3)
10. [SVG] Jon Ferraiolo ed.; "Scalable Vector Graphics (SVG) 1.0 Specification", W3C Proposed Recommendation, 19 July 2001
    (http://www.w3.org/TR/2001/PR-SVG-20010719)
    
11. [UI] Tantek Çelik, editor; "User Interface for CSS3", W3C Working Draft, 16 February 2000
    (http://www.w3.org/TR/2000/WD-css3-userint-20000216)
12. [UNICODE] "The Unicode Standard: Version 3.0.1", The Unicode Consortium, Addison Wesley Longman, 2000, ISBN 0-201-61633-5.
    URL: http://www.unicode.org/unicode/standard/versions/Unicode3.0.1.html.
    The latest version of Unicode. For more information, consult the Unicode Consortium's home page at http://www.unicode.org/.
13. [XML-NAMES] Tim Bray, Dave Hollander, Andrew Layman, editors; "Namespaces in XML", W3C Recommendation, 14 January 1999
    (http://www.w3.org/TR/REC-xml-names/)
14. [YACC] "YACC - Yet another compiler compiler", S. C. Johnson, Technical Report, Murray Hill, 1975