MensBeam
/
HTML-Parser
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
A modern, accurate HTML parser and serializer for PHP
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
305 Commits
5 Branches
9 Tags
1.5 MiB
 
Branch: ns
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'ns'
${ noResults }
J. King fcf162f8e6
Fix the last error and most remaining failures 		3 years ago
lib Fix the last error and most remaining failures 3 years ago
tests Fix up most namespace-related failures 3 years ago
vendor-bin Update dependencies 3 years ago
.gitattributes Add missing tests for charset pre-scan 3 years ago
.gitignore Return value on Parser::parse is now \DOMDocument 3 years ago
AUTHORS Added authors file and updated license 6 years ago
LICENSE Added authors file and updated license 6 years ago
README.md Support processing instructions 3 years ago
RoboFile.php Remove DOM features and related 3 years ago
composer.json Remove DOM features and related 3 years ago
composer.lock Remove DOM features and related 3 years ago
robo Basic skeleton of test suite 4 years ago
robo.bat Basic skeleton of test suite 4 years ago

README.md

HTML-Parser

A modern, accurate HTML parser for PHP.

Usage

<?php
$out = MensBeam\HTML\Parser::parse('<!DOCTYPE html><html lang="en" charset="utf-8"><head><title>Ook!</title></head><body><h1>Ook!</h1><p>Ook-ook? Oooook. Ook ook oook ook oooooook ook ooook ook.</p><p>Eek!</p></body></html>');
$document = $out->document; // the parsed document
$encoding = $out->encoding; // the canonical name of the detected or supplied encoding
$quirks = $out->quirksMode; // the quirks-mode setting of the document, needed for parsing fragments into the document later

The API is still in flux, but should be finalized soon.

Limitations

The primary aim of this library is accuracy. If the document object differs from what the specification mandates, this is probably a bug. However, we are also constrained by PHP, which imposes various limtations. These are as follows:

  • Due to PHP's DOM being designed for XML, element and attribute names which are illegal in XML are mangled as recommended by the specification
  • PHP's DOM does not allow comments to be inserted outside the root element. The parser will perform the insertions, but the comment nodes are then silently dropped
  • PHP's DOM has no special understanding of the HTML <template> element. Consequently template contents is treated no differently from the children of other elements
  • PHP's DOM treats xmlns attributes specially. Attributes which would change the namespace URI of an element or prefix to inconsistent values are thus dropped
  • Due to a PHP bug which severely degrades performance with large documents and in consideration of existing PHP software, HTML elements are placed in the null namespace rather than in the HTML namespace
  • PHP's DOM does not allow DOCTYPEs with no name (i.e. <!DOCTYPE> rather than <!DOCTYPE html>); in such cases the parser will create a DOCTYPE using a single U+0020 SPACE character as its name

Comparison with masterminds/html5

This library and masterminds/html5 serve similar purposes. Generally, we are more accurate, but they are much faster. The following table summarizes the main functional differences.

DOMDocument Masterminds MensBeam
Minimum PHP version 5.0 5.3 7.1
Extensions required dom dom, ctype, mbstring or iconv dom
Target HTML version HTML 4.01 HTML 5.0 WHATWG Living Standard
Supported encodings System-dependent System-dependent Per specification
Encoding detection BOM, http-equiv None Per specification (Steps 1-5 & 9)
Fallback encoding ISO 8859-1 UTF-8, configurable Windows-1252, configurable
Handling of invalid characters Bytes are passed through Characters are dropped Per specification
Handling of invalid XML element names Variable Name is changed to "invalid" Per specification
Handling of invalid XML attribute names Variable Attribute is dropped Per specification
Handling of misnested tags Parent end tags always close children Parent end tags always close children Per specification
Handling of data between table cells Left as-is Left as-is Per specification
Handling of omitted start tags Elements are not inserted Elements are not inserted Per specification
Handling of processing instructions Retained Retained Per specification, configurable
Handling of bogus XLink namespace* Foreign content not supported XLink attributes are lost if preceded by bogus namespace Bogus namespace is ignored
Namespace for HTML elements Null Per specification, configurable Null
Time needed to parse single-page HTML specification 0.5 seconds 2.7 seconds† 6.0 seconds‡
Peak memory needed for same 11.6 MB 38 MB 13.9 MB

* For example: <svg xmlns:xlink='http://www.w3.org/1999/xhtml' xlink:href='http://example.com/'/>. It is unclear what correct behaviour is, but we believe our behaviour to be more consistent with the intent of the specification.

† With HTML namespace disabled. With HTML namespace enabled it does not finish in a reasonable time due to a PHP bug.

‡ With parse errors suppressed. Reporting parse errors adds approximately 10% overhead.