PHPPowertools is a web application framework for PHP >= 5.4.
PHPPowertools/HTML5 is the fourth component of the PHPPowertools that has been released to the public.
This project provides an HTML5 parser for PHP. It originated as a fork of Masterminds/html5-php, which itself started out as a fork of html5lib/html5lib-php.
- An HTML5 serializer
- Event-based (SAX-like) parser
- DOM tree builder
namespace App;
use \PowerTools\HTML5 as HTML5;
// An example HTML document:
$html = <<< 'HERE'
<html>
<head>
<title>TEST</title>
</head>
<body id='foo'>
<h1>Hello World</h1>
<p>This is a test of the HTML5 parser.</p>
</body>
</html>
HERE;
// Parse the document. $dom is a DOMDocument.
$html5 = new HTML5();
$dom = $html5->loadHTML($html);
// Render it as HTML5:
print $html5->saveHTML($dom);
// Or save it to a file:
$html5->save($dom, 'out.html');
The $dom
created by the parser is a full DOMDocument
object. And the
save()
and saveHTML()
methods will take any DOMDocument.
It is possible to pass in an array of configuration options when loading an HTML5 document.
// An associative array of options
$options = array(
'option_name' => 'option_value',
);
// Provide the options to the constructor
$html5 = new HTML5($options);
$dom = $html5->loadHTML($html);
The following options are supported:
encode_entities
(boolean): Indicates that the serializer should aggressively encode characters as entities. Without this, it only encodes the bare minimum.disable_html_ns
(boolean): Prevents the parser from automatically assigning the HTML5 namespace to the DOM document. This is for non-namespace aware DOM tools.target_document
(\DOMDocument): A DOM document that will be used as the destination for the parsed nodes.implicit_namespaces
(array): An assoc array of namespaces that should be used by the parser. Name is tag prefix, value is NS URI.
This library provides the following low-level APIs that you can use to create more customized HTML5 tools:
- An
InputStream
abstraction that can work with different kinds of input source (not just files and strings). - A SAX-like event-based parser that you can hook into for special kinds of parsing.
- A flexible error-reporting mechanism that can be tuned to document syntax checking.
- A DOM implementation that uses PHP's built-in DOM library.
The parser is designed as follows:
- The
InputStream
portion handles direct I/O. - The
Scanner
handles scanning on behalf of the parser. - The
Tokenizer
requests data off of the scanner, parses it, clasifies it, and sends it to anEventHandler
. It is a recursive descent parser. - The
EventHandler
receives notifications and data for each specific semantic event that occurs during tokenization. - The
DOMBuilder
is anEventHandler
that listens for tokenizing events and builds a document tree (DOMDocument
) based on the events.
The serializer takes a data structure (the DOMDocument
) and transforms
it into a character representation -- an HTML5 document.
The serializer is broken into three parts:
- The
OutputRules
contain the rules to turn DOM elements into strings. The rules are an implementation of the interfaceRulesInterface
allowing for different rule sets to be used. - The
Traverser
, which is a special-purpose tree walker. It visits each node node in the tree and uses theOutputRules
to transform the node into a string. HTML5
manages theTraverser
and stores the resultant data in the correct place.
The serializer (save()
, saveHTML()
) follows the
section 8.9 of the HTML 5.0 spec.
So tags are serialized according to these rules:
- A tag with children: <foo>CHILDREN</foo>
- A tag that cannot have content: <foo> (no closing tag)
- A tag that could have content, but doesn't: <foo></foo>
- Namespaces: HTML5 only supports a selected list of namespaces
and they do not operate in the same way as XML namespaces. A
:
has no special meaning. By default the parser does not support XML style namespaces via:
; to enable the XML namespaces see the XML Namespaces section - Scripts: This parser does not contain a JavaScript or a CSS interpreter. While one may be supplied, not all features will be supported.
- Rentrance: The current parser is not re-entrant. (Thus you can't pause the parser to modify the HTML string mid-parse.)
- Validation: The current tree builder is not a validating parser.
While it will correct some HTML, it does not check that the HTML
conforms to the standard. (Should you wish, you can build a validating
parser by extending DOMTree or building your own EventHandler
implementation.)
- There is limited support for insertion modes.
- Some autocorrection is done automatically.
- Per the spec, many legacy tags are admitted and correctly handled, even though they are technically not part of HTML5.
- Attribute names and values: Due to the implementation details of the PHP implementation of DOM, attribute names that do not follow the XML 1.0 standard are not inserted into the DOM. (Effectively, they are ignored.) If you've got a clever fix for this, jump in!
- Processor Instructions: The HTML5 spec does not allow processor instructions. We do. Since this is a server-side library, we think this is useful. And that means, dear reader, that in some cases you can parse the HTML from a mixed PHP/HTML document. This, however, is an incidental feature, not a core feature.
- HTML manifests: Unsupported.
- PLAINTEXT: Unsupported.
- Adoption Agency Algorithm: Not yet implemented. (8.2.5.4.7)
To use XML style namespaces you have to configure well the main HTML5
instance.
use Masterminds\HTML5;
$html = new HTML5(array(
"xmlNamespaces" => true
));
$dom = $html->loadHTML('<t:tag xmlns:t="http://www.example.com"/>');
$dom->documentElement->namespaceURI; // http://www.example.com
You can also add some default prefixes that will not require the namespace declaration, but it's elements will be namespaced.
use Masterminds\HTML5;
$html = new HTML5(array(
"implicitNamespaces"=>array(
"t"=>"http://www.example.com"
)
));
$dom = $html->loadHTML('<t:tag/>');
$dom->documentElement->namespaceURI; // http://www.example.com
A huge debt of gratitude to the original authors of Masterminds/html5-php and html5lib/html5lib-php. Their names are credited in the code of the classes I borrowed from Masterminds/html5-php.
John slegers |