AbstractDomParser - Code Metrics - Inspection of "[+]: check result of "html5FallbackForScriptTags()..." - voku/simple_html_dom - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( fefe4d...d45b97 )

by Lars

created 2019-07-19 00:00 UTC

AbstractDomParser A

↳ Parent: Project

Complexity

Total Complexity

Size/Duplication

Total Lines	470
Duplicated Lines	25.53 %

Coupling/Cohesion

Components	2
Dependencies	0

Test Coverage

Coverage

79.31%

Importance

Changes

Metric	Value
wmc	37
lcom	2
cbo	0
dl	120
loc	470
ccs	92
cts	116
cp	0.7931
rs	9.44
c	0
b	0
f	0

25 Methods

Rating	Name	Duplication	Size	Complexity
A	__call()	0	10	2
	__callStatic()	0	1	?
A	__clone()	0	4	1
	__get()	0	1	?
	__toString()	0	1	?
A	clear()	0	4	1
	createDOMDocument()	0	1	?
A	decodeHtmlEntity()	0	35	5
	find()	0	1	?
	findMulti()	0	1	?
	findOne()	0	1	?
A	getDocument()	0	4	1
	html()	0	1	?
A	innerHtml()	13	13	3
A	innerXml()	13	13	3
	loadHtml()	0	1	?
	loadHtmlFile()	0	1	?
A	save()	0	9	2
A	set_callback()	0	4	1
A	text()	0	4	1
A	xml()	22	22	3
A	getEncoding()	0	4	1
A	html5FallbackForScriptTags()	0	21	3
A	putReplacedBackToPreserveHtmlEntities()	37	37	4
B	replaceToPreserveHtmlEntities()	35	35	6

How to fix Duplicated Code

<?php

declare(strict_types=1);

namespace voku\helper;

abstract class AbstractDomParser implements DomParserInterface
{
    /**
     * @var string
     */
    protected static $domHtmlWrapperHelper = '____simple_html_dom__voku__html_wrapper____';

    /**
     * @var string
     */
    protected static $domHtmlSpecialScriptHelper = '____simple_html_dom__voku__html_special_sctipt____';

    /**
     * @var array
     */
    protected static $domBrokenReplaceHelper = [];

    /**
     * @var string[][]
     */
    protected static $domLinkReplaceHelper = [
        'orig' => ['[', ']', '{', '}'],
        'tmp'  => [
            '____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_LEFT____',
            '____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_RIGHT____',
            '____SIMPLE_HTML_DOM__VOKU__BRACKET_LEFT____',
            '____SIMPLE_HTML_DOM__VOKU__BRACKET_RIGHT____',
        ],
    ];

    /**
     * @var string[][]
     */
    protected static $domReplaceHelper = [
        'orig' => ['&', '|', '+', '%', '@', '<html ⚡'],
        'tmp'  => [
            '____SIMPLE_HTML_DOM__VOKU__AMP____',
            '____SIMPLE_HTML_DOM__VOKU__PIPE____',
            '____SIMPLE_HTML_DOM__VOKU__PLUS____',
            '____SIMPLE_HTML_DOM__VOKU__PERCENT____',
            '____SIMPLE_HTML_DOM__VOKU__AT____',
            '<html ____SIMPLE_HTML_DOM__VOKU__GOOGLE_AMP____="true"',
        ],
    ];

    /**
     * @var callable
     */
    protected static $callback;

    /**
     * @var string[]
     */
    protected static $functionAliases = [];

    /**
     * @var \DOMDocument
     */
    protected $document;

    /**
     * @var string
     */
    protected $encoding = 'UTF-8';

    /**
     * @param string $name
     * @param array  $arguments
     *
     * @return bool|mixed
     */
    public function __call($name, $arguments)
    {
        $name = \strtolower($name);

        if (isset(self::$functionAliases[$name])) {
            return \call_user_func_array([$this, self::$functionAliases[$name]], $arguments);
        }

        throw new \BadMethodCallException('Method does not exist: ' . $name);
    }

    /**
     * @param string $name
     * @param array  $arguments
     *
     * @throws \BadMethodCallException
     * @throws \RuntimeException
     *
     * @return static
     */
    abstract public static function __callStatic($name, $arguments);

    public function __clone()
    {
        $this->document = clone $this->document;
    }

    /** @noinspection MagicMethodsValidityInspection */

    /**
     * @param string $name
     *
     * @return string|null
     */
    abstract public function __get($name);

    /**
     * @return string
     */
    abstract public function __toString();

    /**
     * does nothing (only for api-compatibility-reasons)
     *
     * @return bool
     *
     * @deprecated
     */
    public function clear(): bool
    {
        return true;
    }

    /**
     * Create DOMDocument from HTML.
     *
     * @param string   $html
     * @param int|null $libXMLExtraOptions
     *
     * @return \DOMDocument
     */
    abstract protected function createDOMDocument(string $html, $libXMLExtraOptions = null): \DOMDocument;

    /**
     * @param string $content
     * @param bool   $multiDecodeNewHtmlEntity
     *
     * @return string
     */
    protected function decodeHtmlEntity(string $content, bool $multiDecodeNewHtmlEntity): string
    {
        if ($multiDecodeNewHtmlEntity) {
            if (\class_exists('\voku\helper\UTF8')) {
                /** @noinspection PhpUndefinedClassInspection */
                $content = UTF8::rawurldecode($content, true);
            } else {
                do {
                    $content_compare = $content;

                    $content = \rawurldecode(
                        \html_entity_decode(
                            $content,
                            \ENT_QUOTES | \ENT_HTML5
                        )
                    );
                } while ($content_compare !== $content);
            }
        } else {
            /** @noinspection NestedPositiveIfStatementsInspection */
            if (\class_exists('\voku\helper\UTF8')) {
                /** @noinspection PhpUndefinedClassInspection */
                $content = UTF8::rawurldecode($content, false);
            } else {
                $content = \rawurldecode(
                    \html_entity_decode(
                        $content,
                        \ENT_QUOTES | \ENT_HTML5
                    )
                );
            }
        }

        return $content;
    }

    /**
     * Find list of nodes with a CSS selector.
     *
     * @param string   $selector
     * @param int|null $idx
     */
    abstract public function find(string $selector, $idx = null);

    /**
     * Find nodes with a CSS selector.
     *
     * @param string $selector
     */
    abstract public function findMulti(string $selector);

    /**
     * Find one node with a CSS selector.
     *
     * @param string $selector
     */
    abstract public function findOne(string $selector);

    /**
     * @return \DOMDocument
     */
    public function getDocument(): \DOMDocument
    {
        return $this->document;
    }

    /**
     * Get dom node's outer html.
     *
     * @param bool $multiDecodeNewHtmlEntity
     *
     * @return string
     */
    abstract public function html(bool $multiDecodeNewHtmlEntity = false): string;

    /**
     * Get dom node's inner html.
     *
     * @param bool $multiDecodeNewHtmlEntity
     *
     * @return string
     */
    public function innerHtml(bool $multiDecodeNewHtmlEntity = false): string

    {
        // init
        $text = '';

        if ($this->document->documentElement) {
            foreach ($this->document->documentElement->childNodes as $node) {
                $text .= $this->document->saveHTML($node);
            }
        }

        return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
    }

    /**
     * Get dom node's inner html.
     *
     * @param bool $multiDecodeNewHtmlEntity
     *
     * @return string
     */
    public function innerXml(bool $multiDecodeNewHtmlEntity = false): string

    {
        // init
        $text = '';

        if ($this->document->documentElement) {
            foreach ($this->document->documentElement->childNodes as $node) {
                $text .= $this->document->saveXML($node);
            }
        }

        return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
    }

    /**
     * Load HTML from string.
     *
     * @param string   $html
     * @param int|null $libXMLExtraOptions
     *
     * @return DomParserInterface
     */
    abstract public function loadHtml(string $html, $libXMLExtraOptions = null): DomParserInterface;

    /**
     * Load HTML from file.
     *
     * @param string   $filePath
     * @param int|null $libXMLExtraOptions
     *
     * @throws \RuntimeException
     *
     * @return DomParserInterface
     */
    abstract public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface;

    /**
     * Save the html-dom as string.
     *
     * @param string $filepath
     *
     * @return string
     */
    public function save(string $filepath = ''): string
    {
        $string = $this->innerHtml();
        if ($filepath !== '') {
            \file_put_contents($filepath, $string, \LOCK_EX);
        }

        return $string;
    }

    /**
     * @param callable $functionName
     */
    public function set_callback($functionName)
    {
        static::$callback = $functionName;
    }

    /**
     * Get dom node's plain text.
     *
     * @param bool $multiDecodeNewHtmlEntity
     *
     * @return string
     */
    public function text(bool $multiDecodeNewHtmlEntity = false): string
    {
        return $this->fixHtmlOutput($this->document->textContent, $multiDecodeNewHtmlEntity);
    }

    /**
     * Get the HTML as XML or plain XML if needed.
     *
     * @param bool $multiDecodeNewHtmlEntity
     * @param bool $htmlToXml
     * @param bool $removeXmlHeader
     * @param int  $options
     *
     * @return string
     */
    public function xml(

        bool $multiDecodeNewHtmlEntity = false,
        bool $htmlToXml = true,
        bool $removeXmlHeader = true,
        int $options = \LIBXML_NOEMPTYTAG
    ): string {
        $xml = $this->document->saveXML(null, $options);

        if ($removeXmlHeader) {
            $xml = \ltrim((string) \preg_replace('/<\?xml.*\?>/', '', $xml));
        }

        if ($htmlToXml) {
            $return = $this->fixHtmlOutput($xml, $multiDecodeNewHtmlEntity);
        } else {
            $xml = $this->decodeHtmlEntity($xml, $multiDecodeNewHtmlEntity);

            $return = self::putReplacedBackToPreserveHtmlEntities($xml);
        }

        return $return;
    }

    /**
     * Get the encoding to use.
     *
     * @return string
     */
    protected function getEncoding(): string
    {
        return $this->encoding;
    }

    /**
     * workaround for bug: https://bugs.php.net/bug.php?id=74628
     *
     * @param string $html
     */
    protected function html5FallbackForScriptTags(string &$html)
    {
        // regEx for e.g.: [<script id="elements-image-2">...<script>]
        /** @noinspection HtmlDeprecatedTag */
        $regExSpecialScript = '/<(script)(?<attr>[^>]*)>(?<content>.*)<\/\1>/isU';
        $htmlTmp = \preg_replace_callback(
            $regExSpecialScript,
            static function ($scripts) {
                if (empty($scripts['content'])) {
                    return $scripts[0];
                }

                return '<script' . $scripts['attr'] . '>' . str_replace('</', '<\/', $scripts['content']) . '</script>';
            },
            $html
        );

        if ($htmlTmp !== null) {
            $html = $htmlTmp;
        }
    }

    /**
     * @param string $html
     *
     * @return string
     */
    public static function putReplacedBackToPreserveHtmlEntities(string $html): string

    {
        static $DOM_REPLACE__HELPER_CACHE = null;

        if ($DOM_REPLACE__HELPER_CACHE === null) {
            $DOM_REPLACE__HELPER_CACHE['tmp'] = \array_merge(
                self::$domLinkReplaceHelper['tmp'],
                self::$domReplaceHelper['tmp']
            );
            $DOM_REPLACE__HELPER_CACHE['orig'] = \array_merge(
                self::$domLinkReplaceHelper['orig'],
                self::$domReplaceHelper['orig']
            );

            $DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__start'] = '<' . self::$domHtmlWrapperHelper . '>';
            $DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__end'] = '</' . self::$domHtmlWrapperHelper . '>';

            $DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__start'] = '';
            $DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__end'] = '';

            $DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__start'] = '<' . self::$domHtmlSpecialScriptHelper;
            $DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__end'] = '</' . self::$domHtmlSpecialScriptHelper . '>';

            $DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__start'] = '<script';
            $DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__end'] = '</script>';
        }

        if (
            isset(self::$domBrokenReplaceHelper['tmp'])
            &&
            \count(self::$domBrokenReplaceHelper['tmp']) > 0
        ) {
            $html = \str_replace(self::$domBrokenReplaceHelper['tmp'], self::$domBrokenReplaceHelper['orig'], $html);
        }

        return \str_replace($DOM_REPLACE__HELPER_CACHE['tmp'], $DOM_REPLACE__HELPER_CACHE['orig'], $html);
    }

    /**
     * @param string $html
     *
     * @return string
     */
    public static function replaceToPreserveHtmlEntities(string $html): string

    {
        // init
        $linksNew = [];
        $linksOld = [];

        if (\strpos($html, 'http') !== false) {

            // regEx for e.g.: [https://www.domain.de/foo.php?foobar=1&email=lars%40moelleken.org&guid=test1233312&{{foo}}#foo]
            $regExUrl = '/(\[?\bhttps?:\/\/[^\s<>]+(?:\([\w]+\)|[^[:punct:]\s]|\/|\}|\]))/i';
            \preg_match_all($regExUrl, $html, $linksOld);

            if (!empty($linksOld[1])) {
                $linksOld = $linksOld[1];
                foreach ((array) $linksOld as $linkKey => $linkOld) {
                    $linksNew[$linkKey] = \str_replace(
                        self::$domLinkReplaceHelper['orig'],
                        self::$domLinkReplaceHelper['tmp'],
                        $linkOld
                    );
                }
            }
        }

        $linksNewCount = \count($linksNew);
        if ($linksNewCount > 0 && \count($linksOld) === $linksNewCount) {
            $search = \array_merge($linksOld, self::$domReplaceHelper['orig']);
            $replace = \array_merge($linksNew, self::$domReplaceHelper['tmp']);
        } else {
            $search = self::$domReplaceHelper['orig'];
            $replace = self::$domReplaceHelper['tmp'];
        }

        return \str_replace($search, $replace, $html);
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace voku\helper;
6
7			abstract class AbstractDomParser implements DomParserInterface
8			{
9			/**
10			* @var string
11			*/
12			protected static $domHtmlWrapperHelper = '____simple_html_dom__voku__html_wrapper____';
13
14			/**
15			* @var string
16			*/
17			protected static $domHtmlSpecialScriptHelper = '____simple_html_dom__voku__html_special_sctipt____';
18
19			/**
20			* @var array
21			*/
22			protected static $domBrokenReplaceHelper = [];
23
24			/**
25			* @var string[][]
26			*/
27			protected static $domLinkReplaceHelper = [
28			'orig' => ['[', ']', '{', '}'],
29			'tmp' => [
30			'____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_LEFT____',
31			'____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_RIGHT____',
32			'____SIMPLE_HTML_DOM__VOKU__BRACKET_LEFT____',
33			'____SIMPLE_HTML_DOM__VOKU__BRACKET_RIGHT____',
34			],
35			];
36
37			/**
38			* @var string[][]
39			*/
40			protected static $domReplaceHelper = [
41			'orig' => ['&', '\|', '+', '%', '@', '<html ⚡'],
42			'tmp' => [
43			'____SIMPLE_HTML_DOM__VOKU__AMP____',
44			'____SIMPLE_HTML_DOM__VOKU__PIPE____',
45			'____SIMPLE_HTML_DOM__VOKU__PLUS____',
46			'____SIMPLE_HTML_DOM__VOKU__PERCENT____',
47			'____SIMPLE_HTML_DOM__VOKU__AT____',
48			'<html ____SIMPLE_HTML_DOM__VOKU__GOOGLE_AMP____="true"',
49			],
50			];
51
52			/**
53			* @var callable
54			*/
55			protected static $callback;
56
57			/**
58			* @var string[]
59			*/
60			protected static $functionAliases = [];
61
62			/**
63			* @var \DOMDocument
64			*/
65			protected $document;
66
67			/**
68			* @var string
69			*/
70			protected $encoding = 'UTF-8';
71
72			/**
73			* @param string $name
74			* @param array $arguments
75			*
76			* @return bool\|mixed
77			*/
78			public function __call($name, $arguments)
79			{
80			$name = \strtolower($name);
81
82			if (isset(self::$functionAliases[$name])) {
83			return \call_user_func_array([$this, self::$functionAliases[$name]], $arguments);
84			}
85
86			throw new \BadMethodCallException('Method does not exist: ' . $name);
87			}
88
89			/**
90			* @param string $name
91			* @param array $arguments
92			*
93			* @throws \BadMethodCallException
94			* @throws \RuntimeException
95			*
96			* @return static
97			*/
98			abstract public static function __callStatic($name, $arguments);
99
100			public function __clone()
101			{
102			$this->document = clone $this->document;
103			}
104
105			/** @noinspection MagicMethodsValidityInspection */
106
107			/**
108			* @param string $name
109			*
110			* @return string\|null
111			*/
112			abstract public function __get($name);
113
114			/**
115			* @return string
116			*/
117			abstract public function __toString();
118
119			/**
120			* does nothing (only for api-compatibility-reasons)
121			*
122			* @return bool
123			*
124			* @deprecated
125			*/
126			public function clear(): bool
127			{
128			return true;
129			}
130
131			/**
132			* Create DOMDocument from HTML.
133			*
134			* @param string $html
135			* @param int\|null $libXMLExtraOptions
136			*
137			* @return \DOMDocument
138			*/
139			abstract protected function createDOMDocument(string $html, $libXMLExtraOptions = null): \DOMDocument;
140
141			/**
142			* @param string $content
143			* @param bool $multiDecodeNewHtmlEntity
144			*
145			* @return string
146			*/
147	78		protected function decodeHtmlEntity(string $content, bool $multiDecodeNewHtmlEntity): string
148			{
149	78		if ($multiDecodeNewHtmlEntity) {
150	3		if (\class_exists('\voku\helper\UTF8')) {
151			/** @noinspection PhpUndefinedClassInspection */
152			$content = UTF8::rawurldecode($content, true);
153			} else {
154			do {
155	3		$content_compare = $content;
156
157	3		$content = \rawurldecode(
158	3		\html_entity_decode(
159	3		$content,
160	3		\ENT_QUOTES \| \ENT_HTML5
161			)
162			);
163	3		} while ($content_compare !== $content);
164			}
165			} else {
166			/** @noinspection NestedPositiveIfStatementsInspection */
167	77		if (\class_exists('\voku\helper\UTF8')) {
168			/** @noinspection PhpUndefinedClassInspection */
169			$content = UTF8::rawurldecode($content, false);
170			} else {
171	77		$content = \rawurldecode(
172	77		\html_entity_decode(
173	77		$content,
174	77		\ENT_QUOTES \| \ENT_HTML5
175			)
176			);
177			}
178			}
179
180	78		return $content;
181			}
182
183			/**
184			* Find list of nodes with a CSS selector.
185			*
186			* @param string $selector
187			* @param int\|null $idx
188			*/
189			abstract public function find(string $selector, $idx = null);
190
191			/**
192			* Find nodes with a CSS selector.
193			*
194			* @param string $selector
195			*/
196			abstract public function findMulti(string $selector);
197
198			/**
199			* Find one node with a CSS selector.
200			*
201			* @param string $selector
202			*/
203			abstract public function findOne(string $selector);
204
205			/**
206			* @return \DOMDocument
207			*/
208	39		public function getDocument(): \DOMDocument
209			{
210	39		return $this->document;
211			}
212
213			/**
214			* Get dom node's outer html.
215			*
216			* @param bool $multiDecodeNewHtmlEntity
217			*
218			* @return string
219			*/
220			abstract public function html(bool $multiDecodeNewHtmlEntity = false): string;
221
222			/**
223			* Get dom node's inner html.
224			*
225			* @param bool $multiDecodeNewHtmlEntity
226			*
227			* @return string
228			*/
229	20	View Code Duplication	public function innerHtml(bool $multiDecodeNewHtmlEntity = false): string
			0 ignored issues – show Duplication introduced 2019-07-01 08:00 UTC by Report Bug Copy Issue Report This method seems to be duplicated in your project. Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation. You can also find more detailed suggestions in the “Code” section of your repository. Loading history...
230			{
231			// init
232	20		$text = '';
233
234	20		if ($this->document->documentElement) {
235	20		foreach ($this->document->documentElement->childNodes as $node) {
236	20		$text .= $this->document->saveHTML($node);
237			}
238			}
239
240	20		return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
241			}
242
243			/**
244			* Get dom node's inner html.
245			*
246			* @param bool $multiDecodeNewHtmlEntity
247			*
248			* @return string
249			*/
250		View Code Duplication	public function innerXml(bool $multiDecodeNewHtmlEntity = false): string
			0 ignored issues – show Duplication introduced 2019-07-01 08:00 UTC by Report Bug Copy Issue Report This method seems to be duplicated in your project. Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation. You can also find more detailed suggestions in the “Code” section of your repository. Loading history...
251			{
252			// init
253			$text = '';
254
255			if ($this->document->documentElement) {
256			foreach ($this->document->documentElement->childNodes as $node) {
257			$text .= $this->document->saveXML($node);
258			}
259			}
260
261			return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
262			}
263
264			/**
265			* Load HTML from string.
266			*
267			* @param string $html
268			* @param int\|null $libXMLExtraOptions
269			*
270			* @return DomParserInterface
271			*/
272			abstract public function loadHtml(string $html, $libXMLExtraOptions = null): DomParserInterface;
273
274			/**
275			* Load HTML from file.
276			*
277			* @param string $filePath
278			* @param int\|null $libXMLExtraOptions
279			*
280			* @throws \RuntimeException
281			*
282			* @return DomParserInterface
283			*/
284			abstract public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface;
285
286			/**
287			* Save the html-dom as string.
288			*
289			* @param string $filepath
290			*
291			* @return string
292			*/
293	1		public function save(string $filepath = ''): string
294			{
295	1		$string = $this->innerHtml();
296	1		if ($filepath !== '') {
297			\file_put_contents($filepath, $string, \LOCK_EX);
298			}
299
300	1		return $string;
301			}
302
303			/**
304			* @param callable $functionName
305			*/
306			public function set_callback($functionName)
307			{
308			static::$callback = $functionName;
309			}
310
311			/**
312			* Get dom node's plain text.
313			*
314			* @param bool $multiDecodeNewHtmlEntity
315			*
316			* @return string
317			*/
318	3		public function text(bool $multiDecodeNewHtmlEntity = false): string
319			{
320	3		return $this->fixHtmlOutput($this->document->textContent, $multiDecodeNewHtmlEntity);
321			}
322
323			/**
324			* Get the HTML as XML or plain XML if needed.
325			*
326			* @param bool $multiDecodeNewHtmlEntity
327			* @param bool $htmlToXml
328			* @param bool $removeXmlHeader
329			* @param int $options
330			*
331			* @return string
332			*/
333	2	View Code Duplication	public function xml(
			0 ignored issues – show Duplication introduced 2019-07-01 08:00 UTC by Report Bug Copy Issue Report This method seems to be duplicated in your project. Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation. You can also find more detailed suggestions in the “Code” section of your repository. Loading history...
334			bool $multiDecodeNewHtmlEntity = false,
335			bool $htmlToXml = true,
336			bool $removeXmlHeader = true,
337			int $options = \LIBXML_NOEMPTYTAG
338			): string {
339	2		$xml = $this->document->saveXML(null, $options);
340
341	2		if ($removeXmlHeader) {
342	2		$xml = \ltrim((string) \preg_replace('/<\?xml.*\?>/', '', $xml));
343			}
344
345	2		if ($htmlToXml) {
346			$return = $this->fixHtmlOutput($xml, $multiDecodeNewHtmlEntity);
347			} else {
348	2		$xml = $this->decodeHtmlEntity($xml, $multiDecodeNewHtmlEntity);
349
350	2		$return = self::putReplacedBackToPreserveHtmlEntities($xml);
351			}
352
353	2		return $return;
354			}
355
356			/**
357			* Get the encoding to use.
358			*
359			* @return string
360			*/
361	146		protected function getEncoding(): string
362			{
363	146		return $this->encoding;
364			}
365
366			/**
367			* workaround for bug: https://bugs.php.net/bug.php?id=74628
368			*
369			* @param string $html
370			*/
371	15		protected function html5FallbackForScriptTags(string &$html)
372			{
373			// regEx for e.g.: [<script id="elements-image-2">...<script>]
374			/** @noinspection HtmlDeprecatedTag */
375	15		$regExSpecialScript = '/<(script)(?<attr>[^>])>(?<content>.)<\/\1>/isU';
376	15		$htmlTmp = \preg_replace_callback(
377	15		$regExSpecialScript,
378	15		static function ($scripts) {
379	14		if (empty($scripts['content'])) {
380	6		return $scripts[0];
381			}
382
383	11		return '<script' . $scripts['attr'] . '>' . str_replace('</', '<\/', $scripts['content']) . '</script>';
384	15		},
385	15		$html
386			);
387
388	15		if ($htmlTmp !== null) {
389	15		$html = $htmlTmp;
390			}
391	15		}
392
393			/**
394			* @param string $html
395			*
396			* @return string
397			*/
398	2	View Code Duplication	public static function putReplacedBackToPreserveHtmlEntities(string $html): string
			0 ignored issues – show Duplication introduced 2019-07-01 08:00 UTC by Report Bug Copy Issue Report This method seems to be duplicated in your project. Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation. You can also find more detailed suggestions in the “Code” section of your repository. Loading history...
399			{
400	2		static $DOM_REPLACE__HELPER_CACHE = null;
401
402	2		if ($DOM_REPLACE__HELPER_CACHE === null) {
403	2		$DOM_REPLACE__HELPER_CACHE['tmp'] = \array_merge(
404	2		self::$domLinkReplaceHelper['tmp'],
405	2		self::$domReplaceHelper['tmp']
406			);
407	2		$DOM_REPLACE__HELPER_CACHE['orig'] = \array_merge(
408	2		self::$domLinkReplaceHelper['orig'],
409	2		self::$domReplaceHelper['orig']
410			);
411
412	2		$DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__start'] = '<' . self::$domHtmlWrapperHelper . '>';
413	2		$DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__end'] = '</' . self::$domHtmlWrapperHelper . '>';
414
415	2		$DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__start'] = '';
416	2		$DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__end'] = '';
417
418	2		$DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__start'] = '<' . self::$domHtmlSpecialScriptHelper;
419	2		$DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__end'] = '</' . self::$domHtmlSpecialScriptHelper . '>';
420
421	2		$DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__start'] = '<script';
422	2		$DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__end'] = '</script>';
423			}
424
425			if (
426	2		isset(self::$domBrokenReplaceHelper['tmp'])
427			&&
428	2		\count(self::$domBrokenReplaceHelper['tmp']) > 0
429			) {
430			$html = \str_replace(self::$domBrokenReplaceHelper['tmp'], self::$domBrokenReplaceHelper['orig'], $html);
431			}
432
433	2		return \str_replace($DOM_REPLACE__HELPER_CACHE['tmp'], $DOM_REPLACE__HELPER_CACHE['orig'], $html);
434			}
435
436			/**
437			* @param string $html
438			*
439			* @return string
440			*/
441	3	View Code Duplication	public static function replaceToPreserveHtmlEntities(string $html): string
			0 ignored issues – show Duplication introduced 2019-07-01 08:00 UTC by Report Bug Copy Issue Report This method seems to be duplicated in your project. Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation. You can also find more detailed suggestions in the “Code” section of your repository. Loading history...
442			{
443			// init
444	3		$linksNew = [];
445	3		$linksOld = [];
446
447	3		if (\strpos($html, 'http') !== false) {
448
449			// regEx for e.g.: [https://www.domain.de/foo.php?foobar=1&email=lars%40moelleken.org&guid=test1233312&{{foo}}#foo]
450	1		$regExUrl = '/(\[?\bhttps?:\/\/[^\s<>]+(?:\([\w]+\)\|[^[:punct:]\s]\|\/\|\}\|\]))/i';
451	1		\preg_match_all($regExUrl, $html, $linksOld);
452
453	1		if (!empty($linksOld[1])) {
454	1		$linksOld = $linksOld[1];
455	1		foreach ((array) $linksOld as $linkKey => $linkOld) {
456	1		$linksNew[$linkKey] = \str_replace(
457	1		self::$domLinkReplaceHelper['orig'],
458	1		self::$domLinkReplaceHelper['tmp'],
459	1		$linkOld
460			);
461			}
462			}
463			}
464
465	3		$linksNewCount = \count($linksNew);
466	3		if ($linksNewCount > 0 && \count($linksOld) === $linksNewCount) {
467	1		$search = \array_merge($linksOld, self::$domReplaceHelper['orig']);
468	1		$replace = \array_merge($linksNew, self::$domReplaceHelper['tmp']);
469			} else {
470	2		$search = self::$domReplaceHelper['orig'];
471	2		$replace = self::$domReplaceHelper['tmp'];
472			}
473
474	3		return \str_replace($search, $replace, $html);
475			}
476			}
477

voku / simple_html_dom

Push — master ( fefe4d...d45b97 )

AbstractDomParser A

Complexity

Size/Duplication

Coupling/Cohesion

Test Coverage

Importance

25 Methods

How to fix Duplicated Code

Duplicated Code

Duplication Side-by-Side

Filter issues like