AbstractDomParser::createDOMDocument()

Completed

Push — master ( 75b23e...4c12d1 )

by Lars

created 2020-03-23 07:36 UTC

↳ Parent: AbstractDomParser

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	1
ccs	0
cts	0
cp	0
c	0
b	0
f	0
nc	1

<?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 $domHtmlBrokenHtmlHelper = '____simple_html_dom__voku__broken_html____';

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

    /**
     * @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|null
     */
    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 nodes with a CSS selector or false, if no element is found.
     *
     * @param string $selector
     */
    abstract public function findMultiOrFalse(string $selector);

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

    /**
     * Find one node with a CSS selector or false, if no element is found.
     *
     * @param string $selector
     */
    abstract public function findOneOrFalse(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 = $this->html();
        if ($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, $options);
        if ($xml === false) {
            return '';
        }

        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
        ) {
            $html = \str_ireplace(self::$domBrokenReplaceHelper['tmp'], self::$domBrokenReplaceHelper['orig'], $html);
        }

        return \str_ireplace($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 $domHtmlBrokenHtmlHelper = '____simple_html_dom__voku__broken_html____';
18
19			/**
20			* @var string
21			*/
22			protected static $domHtmlSpecialScriptHelper = '____simple_html_dom__voku__html_special_script____';
23
24			/**
25			* @var array
26			*/
27			protected static $domBrokenReplaceHelper = [];
28
29			/**
30			* @var string[][]
31			*/
32			protected static $domLinkReplaceHelper = [
33			'orig' => ['[', ']', '{', '}'],
34			'tmp' => [
35			'____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_LEFT____',
36			'____SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_RIGHT____',
37			'____SIMPLE_HTML_DOM__VOKU__BRACKET_LEFT____',
38			'____SIMPLE_HTML_DOM__VOKU__BRACKET_RIGHT____',
39			],
40			];
41
42			/**
43			* @var string[][]
44			*/
45			protected static $domReplaceHelper = [
46			'orig' => ['&', '\|', '+', '%', '@', '<html ⚡'],
47			'tmp' => [
48			'____SIMPLE_HTML_DOM__VOKU__AMP____',
49			'____SIMPLE_HTML_DOM__VOKU__PIPE____',
50			'____SIMPLE_HTML_DOM__VOKU__PLUS____',
51			'____SIMPLE_HTML_DOM__VOKU__PERCENT____',
52			'____SIMPLE_HTML_DOM__VOKU__AT____',
53			'<html ____SIMPLE_HTML_DOM__VOKU__GOOGLE_AMP____="true"',
54			],
55			];
56
57			/**
58			* @var callable\|null
59			*/
60			protected static $callback;
61
62			/**
63			* @var string[]
64			*/
65			protected static $functionAliases = [];
66
67			/**
68			* @var \DOMDocument
69			*/
70			protected $document;
71
72			/**
73			* @var string
74			*/
75			protected $encoding = 'UTF-8';
76
77			/**
78			* @param string $name
79			* @param array $arguments
80			*
81			* @return bool\|mixed
82			*/
83			public function __call($name, $arguments)
84			{
85			$name = \strtolower($name);
86
87			if (isset(self::$functionAliases[$name])) {
88			return \call_user_func_array([$this, self::$functionAliases[$name]], $arguments);
89			}
90
91			throw new \BadMethodCallException('Method does not exist: ' . $name);
92			}
93
94			/**
95			* @param string $name
96			* @param array $arguments
97			*
98			* @throws \BadMethodCallException
99			* @throws \RuntimeException
100			*
101			* @return static
102			*/
103			abstract public static function __callStatic($name, $arguments);
104
105			public function __clone()
106			{
107			$this->document = clone $this->document;
108			}
109
110			/** @noinspection MagicMethodsValidityInspection */
111
112			/**
113			* @param string $name
114			*
115			* @return string\|null
116			*/
117			abstract public function __get($name);
118
119			/**
120			* @return string
121			*/
122			abstract public function __toString();
123
124			/**
125			* does nothing (only for api-compatibility-reasons)
126			*
127			* @return bool
128			*
129			* @deprecated
130			*/
131			public function clear(): bool
132			{
133			return true;
134			}
135
136			/**
137			* Create DOMDocument from HTML.
138			*
139			* @param string $html
140			* @param int\|null $libXMLExtraOptions
141			*
142			* @return \DOMDocument
143			*/
144			abstract protected function createDOMDocument(string $html, $libXMLExtraOptions = null): \DOMDocument;
145
146			/**
147			* @param string $content
148			* @param bool $multiDecodeNewHtmlEntity
149			*
150			* @return string
151			*/
152	119		protected function decodeHtmlEntity(string $content, bool $multiDecodeNewHtmlEntity): string
153			{
154	119		if ($multiDecodeNewHtmlEntity) {
155	3		if (\class_exists('\voku\helper\UTF8')) {
156			/** @noinspection PhpUndefinedClassInspection */
157			$content = UTF8::rawurldecode($content, true);
158			} else {
159			do {
160	3		$content_compare = $content;
161
162	3		$content = \rawurldecode(
163	3		\html_entity_decode(
164	3		$content,
165	3		\ENT_QUOTES \| \ENT_HTML5
166			)
167			);
168	3		} while ($content_compare !== $content);
169			}
170			} else {
171			/** @noinspection NestedPositiveIfStatementsInspection */
172	118		if (\class_exists('\voku\helper\UTF8')) {
173			/** @noinspection PhpUndefinedClassInspection */
174			$content = UTF8::rawurldecode($content, false);
175			} else {
176	118		$content = \rawurldecode(
177	118		\html_entity_decode(
178	118		$content,
179	118		\ENT_QUOTES \| \ENT_HTML5
180			)
181			);
182			}
183			}
184
185	119		return $content;
186			}
187
188			/**
189			* Find list of nodes with a CSS selector.
190			*
191			* @param string $selector
192			* @param int\|null $idx
193			*/
194			abstract public function find(string $selector, $idx = null);
195
196			/**
197			* Find nodes with a CSS selector.
198			*
199			* @param string $selector
200			*/
201			abstract public function findMulti(string $selector);
202
203			/**
204			* Find nodes with a CSS selector or false, if no element is found.
205			*
206			* @param string $selector
207			*/
208			abstract public function findMultiOrFalse(string $selector);
209
210			/**
211			* Find one node with a CSS selector.
212			*
213			* @param string $selector
214			*/
215			abstract public function findOne(string $selector);
216
217			/**
218			* Find one node with a CSS selector or false, if no element is found.
219			*
220			* @param string $selector
221			*/
222			abstract public function findOneOrFalse(string $selector);
223
224			/**
225			* @return \DOMDocument
226			*/
227	47		public function getDocument(): \DOMDocument
228			{
229	47		return $this->document;
230			}
231
232			/**
233			* Get dom node's outer html.
234			*
235			* @param bool $multiDecodeNewHtmlEntity
236			*
237			* @return string
238			*/
239			abstract public function html(bool $multiDecodeNewHtmlEntity = false): string;
240
241			/**
242			* Get dom node's inner html.
243			*
244			* @param bool $multiDecodeNewHtmlEntity
245			*
246			* @return string
247			*/
248	28	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...
249			{
250			// init
251	28		$text = '';
252
253	28		if ($this->document->documentElement) {
254	28		foreach ($this->document->documentElement->childNodes as $node) {
255	28		$text .= $this->document->saveHTML($node);
256			}
257			}
258
259	28		return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
260			}
261
262			/**
263			* Get dom node's inner html.
264			*
265			* @param bool $multiDecodeNewHtmlEntity
266			*
267			* @return string
268			*/
269		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...
270			{
271			// init
272			$text = '';
273
274			if ($this->document->documentElement) {
275			foreach ($this->document->documentElement->childNodes as $node) {
276			$text .= $this->document->saveXML($node);
277			}
278			}
279
280			return $this->fixHtmlOutput($text, $multiDecodeNewHtmlEntity);
281			}
282
283			/**
284			* Load HTML from string.
285			*
286			* @param string $html
287			* @param int\|null $libXMLExtraOptions
288			*
289			* @return DomParserInterface
290			*/
291			abstract public function loadHtml(string $html, $libXMLExtraOptions = null): DomParserInterface;
292
293			/**
294			* Load HTML from file.
295			*
296			* @param string $filePath
297			* @param int\|null $libXMLExtraOptions
298			*
299			* @throws \RuntimeException
300			*
301			* @return DomParserInterface
302			*/
303			abstract public function loadHtmlFile(string $filePath, $libXMLExtraOptions = null): DomParserInterface;
304
305			/**
306			* Save the html-dom as string.
307			*
308			* @param string $filepath
309			*
310			* @return string
311			*/
312	13		public function save(string $filepath = ''): string
313			{
314	13		$string = $this->html();
315	13		if ($filepath !== '') {
316	1		\file_put_contents($filepath, $string, \LOCK_EX);
317			}
318
319	13		return $string;
320			}
321
322			/**
323			* @param callable $functionName
324			*/
325			public function set_callback($functionName)
326			{
327			static::$callback = $functionName;
328			}
329
330			/**
331			* Get dom node's plain text.
332			*
333			* @param bool $multiDecodeNewHtmlEntity
334			*
335			* @return string
336			*/
337	4		public function text(bool $multiDecodeNewHtmlEntity = false): string
338			{
339	4		return $this->fixHtmlOutput($this->document->textContent, $multiDecodeNewHtmlEntity);
340			}
341
342			/**
343			* Get the HTML as XML or plain XML if needed.
344			*
345			* @param bool $multiDecodeNewHtmlEntity
346			* @param bool $htmlToXml
347			* @param bool $removeXmlHeader
348			* @param int $options
349			*
350			* @return string
351			*/
352	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...
353			bool $multiDecodeNewHtmlEntity = false,
354			bool $htmlToXml = true,
355			bool $removeXmlHeader = true,
356			int $options = \LIBXML_NOEMPTYTAG
357			): string {
358	2		$xml = $this->document->saveXML(null, $options);
359	2		if ($xml === false) {
360			return '';
361			}
362
363	2		if ($removeXmlHeader) {
364	2		$xml = \ltrim((string) \preg_replace('/<\?xml.*\?>/', '', $xml));
365			}
366
367	2		if ($htmlToXml) {
368			$return = $this->fixHtmlOutput($xml, $multiDecodeNewHtmlEntity);
369			} else {
370	2		$xml = $this->decodeHtmlEntity($xml, $multiDecodeNewHtmlEntity);
371
372	2		$return = self::putReplacedBackToPreserveHtmlEntities($xml);
373			}
374
375	2		return $return;
376			}
377
378			/**
379			* Get the encoding to use.
380			*
381			* @return string
382			*/
383	209		protected function getEncoding(): string
384			{
385	209		return $this->encoding;
386			}
387
388			/**
389			* workaround for bug: https://bugs.php.net/bug.php?id=74628
390			*
391			* @param string $html
392			*/
393	20		protected function html5FallbackForScriptTags(string &$html)
394			{
395			// regEx for e.g.: [<script id="elements-image-2">...<script>]
396			/** @noinspection HtmlDeprecatedTag */
397	20		$regExSpecialScript = '/<(script)(?<attr>[^>])>(?<content>.)<\/\1>/isU';
398	20		$htmlTmp = \preg_replace_callback(
399	20		$regExSpecialScript,
400	20		static function ($scripts) {
401	18		if (empty($scripts['content'])) {
402	6		return $scripts[0];
403			}
404
405	15		return '<script' . $scripts['attr'] . '>' . \str_replace('</', '<\/', $scripts['content']) . '</script>';
406	20		},
407	20		$html
408			);
409
410	20		if ($htmlTmp !== null) {
411	20		$html = $htmlTmp;
412			}
413	20		}
414
415			/**
416			* @param string $html
417			*
418			* @return string
419			*/
420	135		public static function putReplacedBackToPreserveHtmlEntities(string $html): string
421			{
422	135		static $DOM_REPLACE__HELPER_CACHE = null;
423
424	135		if ($DOM_REPLACE__HELPER_CACHE === null) {
425	30		$DOM_REPLACE__HELPER_CACHE['tmp'] = \array_merge(
426	30		self::$domLinkReplaceHelper['tmp'],
427	30		self::$domReplaceHelper['tmp']
428			);
429	30		$DOM_REPLACE__HELPER_CACHE['orig'] = \array_merge(
430	30		self::$domLinkReplaceHelper['orig'],
431	30		self::$domReplaceHelper['orig']
432			);
433
434	30		$DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__start'] = '<' . self::$domHtmlWrapperHelper . '>';
435	30		$DOM_REPLACE__HELPER_CACHE['tmp']['html_wrapper__end'] = '</' . self::$domHtmlWrapperHelper . '>';
436
437	30		$DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__start'] = '';
438	30		$DOM_REPLACE__HELPER_CACHE['orig']['html_wrapper__end'] = '';
439
440	30		$DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__start'] = '<' . self::$domHtmlSpecialScriptHelper;
441	30		$DOM_REPLACE__HELPER_CACHE['tmp']['html_special_script__end'] = '</' . self::$domHtmlSpecialScriptHelper . '>';
442
443	30		$DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__start'] = '<script';
444	30		$DOM_REPLACE__HELPER_CACHE['orig']['html_special_script__end'] = '</script>';
445			}
446
447			if (
448	135		isset(self::$domBrokenReplaceHelper['tmp'])
449			&&
450	135		\count(self::$domBrokenReplaceHelper['tmp']) > 0
451			) {
452	4		$html = \str_ireplace(self::$domBrokenReplaceHelper['tmp'], self::$domBrokenReplaceHelper['orig'], $html);
453			}
454
455	135		return \str_ireplace($DOM_REPLACE__HELPER_CACHE['tmp'], $DOM_REPLACE__HELPER_CACHE['orig'], $html);
456			}
457
458			/**
459			* @param string $html
460			*
461			* @return string
462			*/
463	195		public static function replaceToPreserveHtmlEntities(string $html): string
464			{
465			// init
466	195		$linksNew = [];
467	195		$linksOld = [];
468
469	195		if (\strpos($html, 'http') !== false) {
470
471			// regEx for e.g.: [https://www.domain.de/foo.php?foobar=1&email=lars%40moelleken.org&guid=test1233312&{{foo}}#foo]
472	61		$regExUrl = '/(\[?\bhttps?:\/\/[^\s<>]+(?:\([\w]+\)\|[^[:punct:]\s]\|\/\|\}\|\]))/i';
473	61		\preg_match_all($regExUrl, $html, $linksOld);
474
475	61		if (!empty($linksOld[1])) {
476	58		$linksOld = $linksOld[1];
477	58		foreach ((array) $linksOld as $linkKey => $linkOld) {
478	58		$linksNew[$linkKey] = \str_replace(
479	58		self::$domLinkReplaceHelper['orig'],
480	58		self::$domLinkReplaceHelper['tmp'],
481	58		$linkOld
482			);
483			}
484			}
485			}
486
487	195		$linksNewCount = \count($linksNew);
488	195		if ($linksNewCount > 0 && \count($linksOld) === $linksNewCount) {
489	58		$search = \array_merge($linksOld, self::$domReplaceHelper['orig']);
490	58		$replace = \array_merge($linksNew, self::$domReplaceHelper['tmp']);
491			} else {
492	142		$search = self::$domReplaceHelper['orig'];
493	142		$replace = self::$domReplaceHelper['tmp'];
494			}
495
496	195		return \str_replace($search, $replace, $html);
497			}
498			}
499

voku / simple_html_dom

Push — master ( 75b23e...4c12d1 )

AbstractDomParser::createDOMDocument()

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like