TemplateLoader::replaceCDATA() - Code Metrics - Inspection of "TemplateLoader: added support for HTML 5 entities..." - s9e/TextFormatter - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 982d14...1bd195 )

by Josh

created 2020-01-10 00:20 UTC

TemplateLoader::replaceCDATA() A

↳ Parent: TemplateLoader

Complexity

Conditions	1
Paths	1

Size

Total Lines	9
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	5
CRAP Score	1

Importance

Changes

Metric	Value
eloc	5
dl	0
loc	9
ccs	5
cts	5
cp	1
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	1
crap	1

<?php

/**
* @package   s9e\TextFormatter
* @copyright Copyright (c) 2010-2020 The s9e authors
* @license   http://www.opensource.org/licenses/mit-license.php The MIT License
*/
namespace s9e\TextFormatter\Configurator\Helpers;

use DOMDocument;
use DOMElement;
use DOMXPath;
use RuntimeException;

abstract class TemplateLoader
{
	/**
	* XSL namespace
	*/
	const XMLNS_XSL = 'http://www.w3.org/1999/XSL/Transform';

	/**
	* Get the XML content of an element
	*
	* @private
	*
	* @param  DOMElement $element
	* @return string
	*/
	public static function innerXML(DOMElement $element)
	{
		// Serialize the XML then remove the outer element
		$xml = $element->ownerDocument->saveXML($element);
		$pos = 1 + strpos($xml, '>');
		$len = strrpos($xml, '<') - $pos;

		// If the template is empty, return an empty string
		return ($len < 1) ? '' : substr($xml, $pos, $len);
	}

	/**
	* Load a template as an xsl:template node
	*
	* Will attempt to load it as XML first, then as HTML as a fallback. Either way, an xsl:template
	* node is returned
	*
	* @param  string      $template
	* @return DOMDocument
	*/
	public static function load($template)
	{
		$dom = self::loadAsXML($template) ?: self::loadAsXML(self::fixEntities($template));
		if ($dom)
		{
			return $dom;
		}

		// If the template contains an XSL element, abort now. Otherwise, try reparsing it as HTML
		if (strpos($template, '<xsl:') !== false)
		{
			$error = libxml_get_last_error();

			throw new RuntimeException('Invalid XSL: ' . $error->message);
		}

		return self::loadAsHTML($template);
	}

	/**
	* Serialize a loaded template back into a string
	*
	* NOTE: removes the root node created by load()
	*
	* @param  DOMDocument $dom
	* @return string
	*/
	public static function save(DOMDocument $dom)
	{
		$xml = self::innerXML($dom->documentElement);
		if (strpos($xml, 'xmlns:xsl') !== false)
		{
			$xml = preg_replace('((<[^>]+?) xmlns:xsl="' . self::XMLNS_XSL . '")', '$1', $xml);
		}

		return $xml;
	}

	/**
	* Replace HTML entities and unescaped ampersands in given template
	*
	* @param  string $template
	* @return string
	*/
	protected static function fixEntities($template)
	{
		$template = self::replaceEntities($template);
		$template = preg_replace('(&(?!quot;|amp;|apos;|lt;|gt;|#\\d+;|#x[A-Fa-f0-9]+;))', '&amp;', $template);

		return $template;
	}

	/**
	* Load given HTML template in a DOM document
	*
	* @param  string      $template Original template
	* @return DOMDocument
	*/
	protected static function loadAsHTML($template)
	{
		$template = self::replaceCDATA($template);
		$template = self::replaceEntities($template);

		$dom  = new DOMDocument;
		$html = '<?xml version="1.0" encoding="utf-8" ?><html><body><div>' . $template . '</div></body></html>';

		$useErrors = libxml_use_internal_errors(true);
		$dom->loadHTML($html, LIBXML_NSCLEAN);
		self::removeInvalidAttributes($dom);
		libxml_use_internal_errors($useErrors);

		// Now dump the thing as XML then reload it with the proper root element
		$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . self::innerXML($dom->documentElement->firstChild->firstChild) . '</xsl:template>';

		$useErrors = libxml_use_internal_errors(true);
		$dom->loadXML($xml, LIBXML_NSCLEAN);
		libxml_use_internal_errors($useErrors);

		return $dom;
	}

	/**
	* Load given XSL template in a DOM document
	*
	* @param  string           $template Original template
	* @return bool|DOMDocument           DOMDocument on success, FALSE otherwise
	*/
	protected static function loadAsXML($template)
	{
		$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . $template . '</xsl:template>';

		$useErrors = libxml_use_internal_errors(true);
		$dom       = new DOMDocument;
		$success   = $dom->loadXML($xml, LIBXML_NOCDATA | LIBXML_NSCLEAN);
		self::removeInvalidAttributes($dom);
		libxml_use_internal_errors($useErrors);

		return ($success) ? $dom : false;
	}

	/**
	* Remove attributes with an invalid name from given DOM document
	*
	* @param  DOMDocument $dom
	* @return void
	*/
	protected static function removeInvalidAttributes(DOMDocument $dom)
	{
		$xpath = new DOMXPath($dom);
		foreach ($xpath->query('//@*') as $attribute)
		{
			if (!preg_match('(^(?:[-\\w]+:)?(?!\\d)[-\\w]+$)D', $attribute->nodeName))
			{
				$attribute->parentNode->removeAttributeNode($attribute);
			}
		}
	}

	/**
	* Replace CDATA sections in given template
	*
	* @param  string $template Original template
	* @return string           Modified template
	*/
	protected static function replaceCDATA($template)
	{
		return preg_replace_callback(
			'(<!\\[CDATA\\[(.*?)\\]\\]>)',
			function ($m)
			{
				return htmlspecialchars($m[1]);
			},
			$template
		);
	}

	/**
	* Replace known HTML entities
	*
	* @param  string $template
	* @return string
	*/
	protected static function replaceEntities(string $template): string
	{
		return preg_replace_callback(
			'(&(?!quot;|amp;|apos;|lt;|gt;)\\w+;)',
			function ($m)
			{
				return html_entity_decode($m[0], ENT_HTML5 | ENT_NOQUOTES, 'UTF-8');
			},
			$template
		);
	}
}

1		<?php
2
3		/**
4		* @package s9e\TextFormatter
5		* @copyright Copyright (c) 2010-2020 The s9e authors
6		* @license http://www.opensource.org/licenses/mit-license.php The MIT License
7		*/
8		namespace s9e\TextFormatter\Configurator\Helpers;
9
10		use DOMDocument;
11		use DOMElement;
12		use DOMXPath;
13		use RuntimeException;
14
15		abstract class TemplateLoader
16		{
17		/**
18		* XSL namespace
19		*/
20		const XMLNS_XSL = 'http://www.w3.org/1999/XSL/Transform';
21
22		/**
23		* Get the XML content of an element
24		*
25		* @private
26		*
27		* @param DOMElement $element
28		* @return string
29		*/
30	15	public static function innerXML(DOMElement $element)
31		{
32		// Serialize the XML then remove the outer element
33	15	$xml = $element->ownerDocument->saveXML($element);
34	15	$pos = 1 + strpos($xml, '>');
35	15	$len = strrpos($xml, '<') - $pos;
36
37		// If the template is empty, return an empty string
38	15	return ($len < 1) ? '' : substr($xml, $pos, $len);
39		}
40
41		/**
42		* Load a template as an xsl:template node
43		*
44		* Will attempt to load it as XML first, then as HTML as a fallback. Either way, an xsl:template
45		* node is returned
46		*
47		* @param string $template
48		* @return DOMDocument
49		*/
50	24	public static function load($template)
51		{
52	24	$dom = self::loadAsXML($template) ?: self::loadAsXML(self::fixEntities($template));
53	24	if ($dom)
54		{
55	13	return $dom;
56		}
57
58		// If the template contains an XSL element, abort now. Otherwise, try reparsing it as HTML
59	11	if (strpos($template, '<xsl:') !== false)
60		{
61	1	$error = libxml_get_last_error();
62
63	1	throw new RuntimeException('Invalid XSL: ' . $error->message);
64		}
65
66	10	return self::loadAsHTML($template);
67		}
68
69		/**
70		* Serialize a loaded template back into a string
71		*
72		* NOTE: removes the root node created by load()
73		*
74		* @param DOMDocument $dom
75		* @return string
76		*/
77	6	public static function save(DOMDocument $dom)
78		{
79	6	$xml = self::innerXML($dom->documentElement);
80	6	if (strpos($xml, 'xmlns:xsl') !== false)
81		{
82	1	$xml = preg_replace('((<[^>]+?) xmlns:xsl="' . self::XMLNS_XSL . '")', '$1', $xml);
83		}
84
85	6	return $xml;
86		}
87
88		/**
89		* Replace HTML entities and unescaped ampersands in given template
90		*
91		* @param string $template
92		* @return string
93		*/
94	15	protected static function fixEntities($template)
95		{
96	15	$template = self::replaceEntities($template);
97	15	$template = preg_replace('(&(?!quot;\|amp;\|apos;\|lt;\|gt;\|#\\d+;\|#x[A-Fa-f0-9]+;))', '&', $template);
98
99	15	return $template;
100		}
101
102		/**
103		* Load given HTML template in a DOM document
104		*
105		* @param string $template Original template
106		* @return DOMDocument
107		*/
108	10	protected static function loadAsHTML($template)
109		{
110	10	$template = self::replaceCDATA($template);
111	10	$template = self::replaceEntities($template);
112
113	10	$dom = new DOMDocument;
114	10	$html = '<?xml version="1.0" encoding="utf-8" ?><html><body><div>' . $template . '</div></body></html>';
115
116	10	$useErrors = libxml_use_internal_errors(true);
117	10	$dom->loadHTML($html, LIBXML_NSCLEAN);
118	10	self::removeInvalidAttributes($dom);
119	10	libxml_use_internal_errors($useErrors);
120
121		// Now dump the thing as XML then reload it with the proper root element
122	10	$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . self::innerXML($dom->documentElement->firstChild->firstChild) . '</xsl:template>';
123
124	10	$useErrors = libxml_use_internal_errors(true);
125	10	$dom->loadXML($xml, LIBXML_NSCLEAN);
126	10	libxml_use_internal_errors($useErrors);
127
128	10	return $dom;
129		}
130
131		/**
132		* Load given XSL template in a DOM document
133		*
134		* @param string $template Original template
135		* @return bool\|DOMDocument DOMDocument on success, FALSE otherwise
136		*/
137	24	protected static function loadAsXML($template)
138		{
139	24	$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . $template . '</xsl:template>';
140
141	24	$useErrors = libxml_use_internal_errors(true);
142	24	$dom = new DOMDocument;
143	24	$success = $dom->loadXML($xml, LIBXML_NOCDATA \| LIBXML_NSCLEAN);
144	24	self::removeInvalidAttributes($dom);
145	24	libxml_use_internal_errors($useErrors);
146
147	24	return ($success) ? $dom : false;
148		}
149
150		/**
151		* Remove attributes with an invalid name from given DOM document
152		*
153		* @param DOMDocument $dom
154		* @return void
155		*/
156	24	protected static function removeInvalidAttributes(DOMDocument $dom)
157		{
158	24	$xpath = new DOMXPath($dom);
159	24	foreach ($xpath->query('//@*') as $attribute)
160		{
161	12	if (!preg_match('(^(?:[-\\w]+:)?(?!\\d)[-\\w]+$)D', $attribute->nodeName))
162		{
163	3	$attribute->parentNode->removeAttributeNode($attribute);
164		}
165		}
166		}
167
168		/**
169		* Replace CDATA sections in given template
170		*
171		* @param string $template Original template
172		* @return string Modified template
173		*/
174	10	protected static function replaceCDATA($template)
175		{
176	10	return preg_replace_callback(
177	10	'(<!\\[CDATA\\[(.*?)\\]\\]>)',
178		function ($m)
179		{
180	1	return htmlspecialchars($m[1]);
181	10	},
182		$template
183		);
184		}
185
186		/**
187		* Replace known HTML entities
188		*
189		* @param string $template
190		* @return string
191		*/
192	15	protected static function replaceEntities(string $template): string
193		{
194	15	return preg_replace_callback(
195	15	'(&(?!quot;\|amp;\|apos;\|lt;\|gt;)\\w+;)',
196		function ($m)
197		{
198	4	return html_entity_decode($m[0], ENT_HTML5 \| ENT_NOQUOTES, 'UTF-8');
199	15	},
200		$template
201		);
202		}
203		}

s9e / TextFormatter

Push — master ( 982d14...1bd195 )

TemplateLoader::replaceCDATA() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like