AO_Minify_HTML::_removeDataURICB() - Code Metrics - Inspection of "version 2.8.3" - futtta/autoptimize - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 8538b3...935035 )

by frank

created 2021-04-05 14:03 UTC

AO_Minify_HTML::_removeDataURICB() A

↳ Parent: AO_Minify_HTML

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
nc	1
nop	1
dl	0
loc	4
rs	10
c	0
b	0
f	0

<?php
/**
 * Class Minify_HTML
 * @package Minify
 */

/**
 * Compress HTML
 *
 * This is a heavy regex-based removal of whitespace, unnecessary comments and
 * tokens. IE conditional comments are preserved. There are also options to have
 * STYLE and SCRIPT blocks compressed by callback functions.
 *
 * A test suite is available.
 *
 * @package Minify
 * @author Stephen Clay <[email protected]>
 */
class AO_Minify_HTML {

    /** @var string */
    private $_html;

    /**
     * "Minify" an HTML page
     *
     * @param string $html
     *
     * @param array $options
     *
     * 'cssMinifier' : (optional) callback function to process content of STYLE
     * elements.
     *
     * 'jsMinifier' : (optional) callback function to process content of SCRIPT
     * elements. Note: the type attribute is ignored.
     *
     * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
     * unset, minify will sniff for an XHTML doctype.
     *
     * 'keepComments' : (optional boolean) should the HTML comments be kept
     * in the HTML Code?
     *
     * @return string
     */
    public static function minify($html, $options = array()) {
        $min = new AO_Minify_HTML($html, $options);
        return $min->process();
    }


    /**
     * Create a minifier object
     *
     * @param string $html
     *
     * @param array $options
     *
     * 'cssMinifier' : (optional) callback function to process content of STYLE
     * elements.
     *
     * 'jsMinifier' : (optional) callback function to process content of SCRIPT
     * elements. Note: the type attribute is ignored.
     *
     * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
     * unset, minify will sniff for an XHTML doctype.
     *
     * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
     * unset, minify will sniff for an XHTML doctype.
     *
     * @return null

     */
    public function __construct($html, $options = array())
    {
        $this->_html = str_replace("\r\n", "\n", trim($html));
        if (isset($options['xhtml'])) {
            $this->_isXhtml = (bool)$options['xhtml'];
        }
        if (isset($options['cssMinifier'])) {
            $this->_cssMinifier = $options['cssMinifier'];
        }
        if (isset($options['jsMinifier'])) {
            $this->_jsMinifier = $options['jsMinifier'];
        }
        if (isset($options['keepComments'])) {
            $this->_keepComments = $options['keepComments'];
        }
    }


    /**
     * Minify the markeup given in the constructor
     *
     * @return string
     */
    public function process()
    {
        if ($this->_isXhtml === null) {
            $this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'));
        }

        $this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
        $this->_placeholders = array();

        // replace SCRIPTs (and minify) with placeholders
        $this->_html = preg_replace_callback(
            '/(\\s*)(<script\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i'
            ,array($this, '_removeScriptCB')
            ,$this->_html);

        // replace STYLEs (and minify) with placeholders
        $this->_html = preg_replace_callback(
            '/\\s*(<style\\b[^>]*?>)([\\s\\S]*?)<\\/style>\\s*/i'
            ,array($this, '_removeStyleCB')
            ,$this->_html);

        // remove HTML comments (not containing IE conditional comments).
        if  ($this->_keepComments == false) {

            $this->_html = preg_replace_callback(
                '/<!--([\\s\\S]*?)-->/'
                ,array($this, '_commentCB')
                ,$this->_html);
        }

        // replace PREs with placeholders
        $this->_html = preg_replace_callback('/\\s*(<pre\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i'
            ,array($this, '_removePreCB')
            ,$this->_html);

        // replace TEXTAREAs with placeholders
        $this->_html = preg_replace_callback(
            '/\\s*(<textarea\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i'
            ,array($this, '_removeTextareaCB')
            ,$this->_html);

        // replace data: URIs with placeholders
        $this->_html = preg_replace_callback(
            '/(=("|\')data:.*\\2)/Ui'
            ,array($this, '_removeDataURICB')
            ,$this->_html);

        // trim each line.
        // replace by space instead of '' to avoid newline after opening tag getting zapped
        $this->_html = preg_replace('/^\s+|\s+$/m', ' ', $this->_html);

        // remove ws around block/undisplayed elements
        $this->_html = preg_replace('/\\s+(<\\/?(?:area|article|aside|base(?:font)?|blockquote|body'
            .'|canvas|caption|center|col(?:group)?|dd|dir|div|dl|dt|fieldset|figcaption|figure|footer|form'
            .'|frame(?:set)?|h[1-6]|head|header|hgroup|hr|html|legend|li|link|main|map|menu|meta|nav'
            .'|ol|opt(?:group|ion)|output|p|param|section|t(?:able|body|head|d|h||r|foot|itle)'
            .'|ul|video)\\b[^>]*>)/i', '$1', $this->_html);

        // remove ws outside of all elements
        $this->_html = preg_replace_callback(
            '/>([^<]+)</'
            ,array($this, '_outsideTagCB')
            ,$this->_html);

        // use newlines before 1st attribute in open tags (to limit line lengths)
        //$this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html);

        // reverse order while preserving keys to ensure the last replacement is done first, etc ...
        $this->_placeholders = array_reverse( $this->_placeholders, true );

        // fill placeholders
        $this->_html = str_replace(
            array_keys($this->_placeholders)
            ,array_values($this->_placeholders)
            ,$this->_html
        );
        return $this->_html;
    }

    protected function _commentCB($m)
    {
        return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<!['))
            ? $m[0]
            : '';
    }

    protected function _reservePlace($content)
    {
        $placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%';
        $this->_placeholders[$placeholder] = $content;
        return $placeholder;
    }

    protected $_isXhtml = null;
    protected $_replacementHash = null;
    protected $_placeholders = array();
    protected $_cssMinifier = null;
    protected $_jsMinifier = null;
    protected $_keepComments = false;

    protected function _outsideTagCB($m)
    {
        return '>' . preg_replace('/^\\s+|\\s+$/', ' ', $m[1]) . '<';
    }

    protected function _removePreCB($m)
    {
        return $this->_reservePlace($m[1]);
    }

    protected function _removeTextareaCB($m)
    {
        return $this->_reservePlace($m[1]);
    }

    protected function _removeDataURICB($m)
    {
        return $this->_reservePlace($m[1]);
    }

    protected function _removeStyleCB($m)
    {
        $openStyle = $m[1];
        $css = $m[2];
        // remove HTML comments
        $css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/', '', $css);

        // remove CDATA section markers
        $css = $this->_removeCdata($css);

        // minify
        $minifier = $this->_cssMinifier
            ? $this->_cssMinifier
            : 'trim';
        $css = call_user_func($minifier, $css);

        return $this->_reservePlace($this->_needsCdata($css)
            ? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>"
            : "{$openStyle}{$css}</style>"
        );
    }

    protected function _removeScriptCB($m)
    {
        $openScript = $m[2];
        $js = $m[3];

        // whitespace surrounding? preserve at least one space
        $ws1 = ($m[1] === '') ? '' : ' ';
        $ws2 = ($m[4] === '') ? '' : ' ';

        if ($this->_keepComments == false) {

            // remove HTML comments (and ending "//" if present)
            $js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js);

            // remove CDATA section markers
            $js = $this->_removeCdata($js);
        }

        // minify
        $minifier = $this->_jsMinifier
            ? $this->_jsMinifier
            : 'trim';
        $js = call_user_func($minifier, $js);

        return $this->_reservePlace($this->_needsCdata($js)
            ? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}"
            : "{$ws1}{$openScript}{$js}</script>{$ws2}"
        );
    }

    protected function _removeCdata($str)
    {
        return (false !== strpos($str, '<![CDATA['))
            ? str_replace(array('/* <![CDATA[ */','/* ]]> */','/*<![CDATA[*/','/*]]>*/','<![CDATA[', ']]>'), '', $str)
            : $str;
    }

    protected function _needsCdata($str)
    {
        return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str));
    }
}


1			<?php
2			/**
3			* Class Minify_HTML
4			* @package Minify
5			*/
6
7			/**
8			* Compress HTML
9			*
10			* This is a heavy regex-based removal of whitespace, unnecessary comments and
11			* tokens. IE conditional comments are preserved. There are also options to have
12			* STYLE and SCRIPT blocks compressed by callback functions.
13			*
14			* A test suite is available.
15			*
16			* @package Minify
17			* @author Stephen Clay <[email protected]>
18			*/
19			class AO_Minify_HTML {
20
21			/** @var string */
22			private $_html;
23
24			/**
25			* "Minify" an HTML page
26			*
27			* @param string $html
28			*
29			* @param array $options
30			*
31			* 'cssMinifier' : (optional) callback function to process content of STYLE
32			* elements.
33			*
34			* 'jsMinifier' : (optional) callback function to process content of SCRIPT
35			* elements. Note: the type attribute is ignored.
36			*
37			* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
38			* unset, minify will sniff for an XHTML doctype.
39			*
40			* 'keepComments' : (optional boolean) should the HTML comments be kept
41			* in the HTML Code?
42			*
43			* @return string
44			*/
45			public static function minify($html, $options = array()) {
46			$min = new AO_Minify_HTML($html, $options);
47			return $min->process();
48			}
49
50
51			/**
52			* Create a minifier object
53			*
54			* @param string $html
55			*
56			* @param array $options
57			*
58			* 'cssMinifier' : (optional) callback function to process content of STYLE
59			* elements.
60			*
61			* 'jsMinifier' : (optional) callback function to process content of SCRIPT
62			* elements. Note: the type attribute is ignored.
63			*
64			* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
65			* unset, minify will sniff for an XHTML doctype.
66			*
67			* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
68			* unset, minify will sniff for an XHTML doctype.
69			*
70			* @return null
			0 ignored issues – show Comprehensibility Best Practice introduced 2016-11-16 14:04 UTC by Report Bug Copy Issue Report Adding a `@return` annotation to constructors is generally not recommended as a constructor does not have a meaningful return value. Adding a `@return` annotation to a constructor is not recommended, since a constructor does not have a meaningful return value. Please refer to the PHP core documentation on constructors. Loading history...
71			*/
72			public function __construct($html, $options = array())
73			{
74			$this->_html = str_replace("\r\n", "\n", trim($html));
75			if (isset($options['xhtml'])) {
76			$this->_isXhtml = (bool)$options['xhtml'];
77			}
78			if (isset($options['cssMinifier'])) {
79			$this->_cssMinifier = $options['cssMinifier'];
80			}
81			if (isset($options['jsMinifier'])) {
82			$this->_jsMinifier = $options['jsMinifier'];
83			}
84			if (isset($options['keepComments'])) {
85			$this->_keepComments = $options['keepComments'];
86			}
87			}
88
89
90			/**
91			* Minify the markeup given in the constructor
92			*
93			* @return string
94			*/
95			public function process()
96			{
97			if ($this->_isXhtml === null) {
98			$this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'));
99			}
100
101			$this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
102			$this->_placeholders = array();
103
104			// replace SCRIPTs (and minify) with placeholders
105			$this->_html = preg_replace_callback(
106			'/(\\s)(<script\\b[^>]?>)([\\s\\S]?)<\\/script>(\\s)/i'
107			,array($this, '_removeScriptCB')
108			,$this->_html);
109
110			// replace STYLEs (and minify) with placeholders
111			$this->_html = preg_replace_callback(
112			'/\\s(<style\\b[^>]?>)([\\s\\S]?)<\\/style>\\s/i'
113			,array($this, '_removeStyleCB')
114			,$this->_html);
115
116			// remove HTML comments (not containing IE conditional comments).
117			if ($this->_keepComments == false) {
			0 ignored issues – show Coding Style Best Practice introduced 2018-10-07 12:00 UTC by Report Bug Copy Issue Report It seems like you are loosely comparing two booleans. Considering using the strict comparison `===` instead. When comparing two booleans, it is generally considered safer to use the strict comparison operator. Loading history...
118			$this->_html = preg_replace_callback(
119			'/<!--([\\s\\S]*?)-->/'
120			,array($this, '_commentCB')
121			,$this->_html);
122			}
123
124			// replace PREs with placeholders
125			$this->_html = preg_replace_callback('/\\s(<pre\\b[^>]?>[\\s\\S]?<\\/pre>)\\s/i'
126			,array($this, '_removePreCB')
127			,$this->_html);
128
129			// replace TEXTAREAs with placeholders
130			$this->_html = preg_replace_callback(
131			'/\\s(<textarea\\b[^>]?>[\\s\\S]?<\\/textarea>)\\s/i'
132			,array($this, '_removeTextareaCB')
133			,$this->_html);
134
135			// replace data: URIs with placeholders
136			$this->_html = preg_replace_callback(
137			'/(=("\|\')data:.*\\2)/Ui'
138			,array($this, '_removeDataURICB')
139			,$this->_html);
140
141			// trim each line.
142			// replace by space instead of '' to avoid newline after opening tag getting zapped
143			$this->_html = preg_replace('/^\s+\|\s+$/m', ' ', $this->_html);
144
145			// remove ws around block/undisplayed elements
146			$this->_html = preg_replace('/\\s+(<\\/?(?:area\|article\|aside\|base(?:font)?\|blockquote\|body'
147			.'\|canvas\|caption\|center\|col(?:group)?\|dd\|dir\|div\|dl\|dt\|fieldset\|figcaption\|figure\|footer\|form'
148			.'\|frame(?:set)?\|h[1-6]\|head\|header\|hgroup\|hr\|html\|legend\|li\|link\|main\|map\|menu\|meta\|nav'
149			.'\|ol\|opt(?:group\|ion)\|output\|p\|param\|section\|t(?:able\|body\|head\|d\|h\|\|r\|foot\|itle)'
150			.'\|ul\|video)\\b[^>]*>)/i', '$1', $this->_html);
151
152			// remove ws outside of all elements
153			$this->_html = preg_replace_callback(
154			'/>([^<]+)</'
155			,array($this, '_outsideTagCB')
156			,$this->_html);
157
158			// use newlines before 1st attribute in open tags (to limit line lengths)
159			//$this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html);
160
161			// reverse order while preserving keys to ensure the last replacement is done first, etc ...
162			$this->_placeholders = array_reverse( $this->_placeholders, true );
163
164			// fill placeholders
165			$this->_html = str_replace(
166			array_keys($this->_placeholders)
167			,array_values($this->_placeholders)
168			,$this->_html
169			);
170			return $this->_html;
171			}
172
173			protected function _commentCB($m)
174			{
175			return (0 === strpos($m[1], '[') \|\| false !== strpos($m[1], '<!['))
176			? $m[0]
177			: '';
178			}
179
180			protected function _reservePlace($content)
181			{
182			$placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%';
183			$this->_placeholders[$placeholder] = $content;
184			return $placeholder;
185			}
186
187			protected $_isXhtml = null;
188			protected $_replacementHash = null;
189			protected $_placeholders = array();
190			protected $_cssMinifier = null;
191			protected $_jsMinifier = null;
192			protected $_keepComments = false;
193
194			protected function _outsideTagCB($m)
195			{
196			return '>' . preg_replace('/^\\s+\|\\s+$/', ' ', $m[1]) . '<';
197			}
198
199			protected function _removePreCB($m)
200			{
201			return $this->_reservePlace($m[1]);
202			}
203
204			protected function _removeTextareaCB($m)
205			{
206			return $this->_reservePlace($m[1]);
207			}
208
209			protected function _removeDataURICB($m)
210			{
211			return $this->_reservePlace($m[1]);
212			}
213
214			protected function _removeStyleCB($m)
215			{
216			$openStyle = $m[1];
217			$css = $m[2];
218			// remove HTML comments
219			$css = preg_replace('/(?:^\\s<!--\|-->\\s$)/', '', $css);
220
221			// remove CDATA section markers
222			$css = $this->_removeCdata($css);
223
224			// minify
225			$minifier = $this->_cssMinifier
226			? $this->_cssMinifier
227			: 'trim';
228			$css = call_user_func($minifier, $css);
229
230			return $this->_reservePlace($this->_needsCdata($css)
231			? "{$openStyle}/<![CDATA[/{$css}/]]>/</style>"
232			: "{$openStyle}{$css}</style>"
233			);
234			}
235
236			protected function _removeScriptCB($m)
237			{
238			$openScript = $m[2];
239			$js = $m[3];
240
241			// whitespace surrounding? preserve at least one space
242			$ws1 = ($m[1] === '') ? '' : ' ';
243			$ws2 = ($m[4] === '') ? '' : ' ';
244
245			if ($this->_keepComments == false) {
			0 ignored issues – show Coding Style Best Practice introduced 2018-10-07 12:00 UTC by Report Bug Copy Issue Report It seems like you are loosely comparing two booleans. Considering using the strict comparison `===` instead. When comparing two booleans, it is generally considered safer to use the strict comparison operator. Loading history...
246			// remove HTML comments (and ending "//" if present)
247			$js = preg_replace('/(?:^\\s<!--\\s\|\\s(?:\\/\\/)?\\s-->\\s*$)/', '', $js);
248
249			// remove CDATA section markers
250			$js = $this->_removeCdata($js);
251			}
252
253			// minify
254			$minifier = $this->_jsMinifier
255			? $this->_jsMinifier
256			: 'trim';
257			$js = call_user_func($minifier, $js);
258
259			return $this->_reservePlace($this->_needsCdata($js)
260			? "{$ws1}{$openScript}/<![CDATA[/{$js}/]]>/</script>{$ws2}"
261			: "{$ws1}{$openScript}{$js}</script>{$ws2}"
262			);
263			}
264
265			protected function _removeCdata($str)
266			{
267			return (false !== strpos($str, '<![CDATA['))
268			? str_replace(array('/* <![CDATA[ /','/ ]]> /','/<![CDATA[/','/]]>*/','<![CDATA[', ']]>'), '', $str)
269			: $str;
270			}
271
272			protected function _needsCdata($str)
273			{
274			return ($this->_isXhtml && preg_match('/(?:[<&]\|\\-\\-\|\\]\\]>)/', $str));
275			}
276			}
277

futtta / autoptimize

Push — master ( 8538b3...935035 )

AO_Minify_HTML::_removeDataURICB() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like