Download the PHP package sabberworm/php-css-parser without Composer

PHP CSS Parser

A Parser for CSS Files written in PHP. Allows extraction of CSS files into a data structure, manipulation of said structure and output as (optimized) CSS.

Usage

Installation using Composer

Extraction

To use the CSS Parser, create a new instance. The constructor takes the following form:

To read a file, for example, you’d do the following:

The resulting CSS document structure can be manipulated prior to being output.

Options

Charset

The charset option will only be used if the CSS file does not contain an @charset declaration. UTF-8 is the default, so you won’t have to create a settings object at all if you don’t intend to change that.

Strict parsing

To have the parser throw an exception when encountering invalid/unknown constructs (as opposed to trying to ignore them and carry on parsing), supply a thusly configured \Sabberworm\CSS\Settings object:

Note that this will also disable a workaround for parsing the unquoted variant of the legacy IE-specific filter rule.

Disable multibyte functions

To achieve faster parsing, you can choose to have PHP-CSS-Parser use regular string functions instead of mb_* functions. This should work fine in most cases, even for UTF-8 files, as all the multibyte characters are in string literals. Still it’s not recommended using this with input you have no control over as it’s not thoroughly covered by test cases.

Manipulation

The resulting data structure consists mainly of five basic types: CSSList, RuleSet, Rule, Selector and Value. There are two additional types used: Import and Charset, which you won’t use often.

CSSList

CSSList represents a generic CSS container, most likely containing declaration blocks (rule sets with a selector), but it may also contain at-rules, charset declarations, etc.

To access the items stored in a CSSList – like the document you got back when calling $parser->parse() –, use getContents(), then iterate over that collection and use instanceof to check whether you’re dealing with another CSSList, a RuleSet, a Import or a Charset.

To append a new item (selector, media query, etc.) to an existing CSSList, construct it using the constructor for this class and use the append($oItem) method.

RuleSet

RuleSet is a container for individual rules. The most common form of a rule set is one constrained by a selector. The following concrete subtypes exist:

• AtRuleSet – for generic at-rules for generic at-rules which are not covered by specific classes, i.e., not @import, @charset or @media. A common example for this is @font-face.
• DeclarationBlock – a RuleSet constrained by a Selector; contains an array of selector objects (comma-separated in the CSS) as well as the rules to be applied to the matching elements.

Note: A CSSList can contain other CSSLists (and Imports as well as a Charset), while a RuleSet can only contain Rules.

If you want to manipulate a RuleSet, use the methods addRule(Rule $rule), getRules() and removeRule($rule) (which accepts either a Rule or a rule name; optionally suffixed by a dash to remove all related rules).

Rule

Rules just have a string key (the rule) and a Value.

Value

Value is an abstract class that only defines the render method. The concrete subclasses for atomic value types are:

• Size – consists of a numeric size value and a unit.
• Color – colors can be input in the form #rrggbb, #rgb or schema(val1, val2, …) but are always stored as an array of ('s' => val1, 'c' => val2, 'h' => val3, …) and output in the second form.
• CSSString – this is just a wrapper for quoted strings to distinguish them from keywords; always output with double quotes.
• URL – URLs in CSS; always output in URL("") notation.

There is another abstract subclass of Value, ValueList: A ValueList represents a lists of Values, separated by some separation character (mostly ,, whitespace, or /).

There are two types of ValueLists:

• RuleValueList – The default type, used to represent all multivalued rules like font: bold 12px/3 Helvetica, Verdana, sans-serif; (where the value would be a whitespace-separated list of the primitive value bold, a slash-separated list and a comma-separated list).
• CSSFunction – A special kind of value that also contains a function name and where the values are the function’s arguments. Also handles equals-sign-separated argument lists like filter: alpha(opacity=90);.

Convenience methods

There are a few convenience methods on Document to ease finding, manipulating and deleting rules:

• getAllDeclarationBlocks() – does what it says; no matter how deeply nested the selectors are. Aliased as getAllSelectors().
• getAllRuleSets() – does what it says; no matter how deeply nested the rule sets are.
• getAllValues() – finds all Value objects inside Rules.

To-Do

• More convenience methods (like selectorsWithElement($sId/Class/TagName), attributesOfType($type), removeAttributesOfType($type))
• Real multibyte support. Currently, only multibyte charsets whose first 255 code points take up only one byte and are identical with ASCII are supported (yes, UTF-8 fits this description).
• Named color support (using Color instead of an anonymous string literal)

Use cases

Use Parser to prepend an ID to all selectors

Shrink all absolute sizes to half

Remove unwanted rules

Output

To output the entire CSS document into a variable, just use ->render():

If you want to format the output, pass an instance of type \Sabberworm\CSS\OutputFormat:

Or use one of the predefined formats:

To see what you can do with output formatting, look at the tests in tests/OutputFormatTest.php.

Examples

Example 1 (At-Rules)

Input

Output (render())

Example 2 (Values)

Input

Output (render())

Class diagram

Contributors/Thanks to

• oliverklee for lots of refactorings, code modernizations and CI integrations
• raxbg for contributions to parse calc, grid lines, and various bugfixes.
• westonruter for bugfixes and improvements.
• FMCorz for many patches and suggestions, for being able to parse comments and IE hacks (in lenient mode).
• Lullabot for a patch that allows to know the line number for each parsed token.
• ju1ius for the specificity parsing code and the ability to expand/compact shorthand properties.
• ossinkine for a 150 time performance boost.
• GaryJones for lots of input and https://css-specificity.info/.
• docteurklein for output formatting and CSSList->remove() inspiration.
• nicolopignatelli for PSR-0 compatibility.
• diegoembarcadero for keyframe at-rule parsing.
• goetas for @namespace at-rule support.
• View full list

Misc

Legacy Support

The latest pre-PSR-0 version of this project can be checked with the 0.9.0 tag.

Running Tests

To run all continuous integration (CI) checks for this project (including unit tests),

• run composer install to install the development dependencies managed with Composer;
• run phive install to install the development dependencies managed with PHIVE;
  • Installation of PHIVE
• run composer ci to run all static and dynamic CI checks.

Details of other Composer scripts available (e.g. to run one specific CI check) are provided with composer list.