Document::loadHTMLFile() - Code Metrics - o2system/html - Measure and Improve Code Quality continuously with Scrutinizer

Document::loadHTMLFile() A
last analyzed 2019-09-25 13:07 UTC

↳ Parent: Document

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	1
nc	1
nop	2
dl	0
loc	3
rs	10
c	0
b	0
f	0

<?php
/**
 * This file is part of the O2System Framework package.
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 *
 * @author         Steeve Andrian Salim
 * @copyright      Copyright (c) Steeve Andrian Salim
 */

// ------------------------------------------------------------------------

namespace O2System\Html;

// ------------------------------------------------------------------------

/**
 * Class Document
 *
 * @package O2System\HTML
 */
class Document extends \DOMDocument
{
    /**
     * Document Meta Nodes
     *
     * @var \O2System\Html\Dom\Lists\Meta
     */
    public $metaNodes;

    /**
     * Document Link Nodes
     *
     * @var \O2System\Html\Dom\Lists\Asset
     */
    public $linkNodes;

    /**
     * Document Style Content
     *
     * @var \O2System\Html\Dom\Style
     */
    public $styleContent;

    /**
     * Document Script Nodes
     *
     * @var \O2System\Html\Dom\Lists\Asset
     */
    public $headScriptNodes;

    /**
     * Document Script Content
     *
     * @var \O2System\Html\Dom\Script
     */
    public $headScriptContent;

    /**
     * Document Script Nodes
     *
     * @var \O2System\Html\Dom\Lists\Asset
     */
    public $bodyScriptNodes;

    /**
     * Document Script Content
     *
     * @var \O2System\Html\Dom\Script
     */
    public $bodyScriptContent;

    // ------------------------------------------------------------------------

    /**
     * Document::__construct
     *
     * @param string $version  Document version.
     * @param string $encoding Document encoding.
     *
     * @return Document
     */
    public function __construct($version = '1.0', $encoding = 'UTF-8')
    {
        language()
            ->addFilePath(__DIR__ . DIRECTORY_SEPARATOR)
            ->loadFile('html');

        parent::__construct($version, $encoding);

        $this->registerNodeClass('DOMElement', '\O2System\Html\Dom\Element');
        $this->registerNodeClass('DOMAttr', '\O2System\Html\Dom\Attr');

        $this->formatOutput = true;

        $this->metaNodes = new Dom\Lists\Meta($this);

        $this->linkNodes = new Dom\Lists\Asset($this);
        $this->linkNodes->element = 'link';

        $this->styleContent = new Dom\Style();

        $this->headScriptNodes = new Dom\Lists\Asset($this);
        $this->headScriptNodes->element = 'script';
        $this->headScriptContent = new Dom\Script();

        $this->bodyScriptNodes = new Dom\Lists\Asset($this);
        $this->bodyScriptNodes->element = 'script';
        $this->bodyScriptContent = new Dom\Script();

        $this->loadHTMLTemplate();
    }

    // ------------------------------------------------------------------------

    /**
     * Document::loadHTMLTemplate
     *
     * Load HTML template from a file.
     *
     * @return void
     */
    protected function loadHTMLTemplate()
    {
        $htmlTemplate = <<<HTML
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <title>O2System HTML</title>
</head>
<body>
</body>
</html>
HTML;

        parent::loadHTML($htmlTemplate);
    }

    // ------------------------------------------------------------------------

    /**
     * Document::__get
     *
     * @param string $tagName The document tag element.
     *
     * @return mixed The value at the specified index or false.
     */
    public function &__get($tagName)
    {
        $getDocument[ $tagName ] = null;


        if (in_array($tagName, ['html', 'head', 'body', 'title'])) {
            $getDocument[ $tagName ] = $this->getElementsByTagName($tagName)->item(0);
        }

        return $getDocument[ $tagName ];
    }

    // ------------------------------------------------------------------------

    /**
     * Document::saveHTMLFile
     *
     * Dumps the internal document into a file using HTML formatting
     *
     * @see   http://php.net/manual/domdocument.savehtmlfile.php
     *
     * @param string $filePath <p>
     *                         The path to the saved HTML document.
     *                         </p>
     *
     * @return int the number of bytes written or false if an error occurred.
     * @since 5.0
     */
    public function saveHTMLFile($filePath)
    {
        if ( ! is_string($filePath)) {

            throw new \InvalidArgumentException('The filename argument must be of type string');
        }

        if ( ! is_writable($filePath)) {
            return false;

        }

        $result = $this->saveHTML();
        file_put_contents($filePath, $result);
        $bytesWritten = filesize($filePath);

        if ($bytesWritten === strlen($result)) {
            return $bytesWritten;
        }

        return false;

    }

    // ------------------------------------------------------------------------

    /**
     * Document::saveHTML
     *
     * Dumps the internal document into a string using HTML formatting.
     *
     * @see   http://php.net/manual/domdocument.savehtml.php
     *
     * @param \DOMNode $node [optional] parameter to output a subset of the document.
     *
     * @return string the HTML, or false if an error occurred.
     * @since 5.0
     */
    public function saveHTML(\DOMNode $node = null)
    {
        $headElement = $this->getElementsByTagName('head')->item(0);

        $styleContent = trim($this->styleContent->__toString());

        if ( ! empty($styleContent)) {
            $styleElement = $this->createElement('style', $styleContent);
            $styleElement->setAttribute('type', 'text/css');
            $headElement->appendChild($styleElement);
        }

        $titleElement = $this->getElementsByTagName('title')->item(0);

        // Insert Meta
        if ($this->metaNodes->count()) {
            $metaNodes = array_reverse($this->metaNodes->getArrayCopy());

            foreach ($metaNodes as $metaNode) {
                $headElement->insertBefore($this->importNode($metaNode), $titleElement);
            }
        }

        // Insert Link
        if ($this->linkNodes->count()) {
            foreach ($this->linkNodes as $linkNode) {
                $headElement->appendChild($this->importNode($linkNode));
            }
        }

        // Insert Head Script
        if (count($this->headScriptNodes)) {
            foreach ($this->headScriptNodes as $scriptNode) {
                $headElement->appendChild($this->importNode($scriptNode));
            }
        }

        $headScriptContent = trim($this->headScriptContent->__toString());

        if ( ! empty($headScriptContent)) {
            $scriptElement = $this->createElement('script', $headScriptContent);
            $scriptElement->setAttribute('type', 'text/javascript');
            $headElement->appendChild($scriptElement);
        }

        $bodyElement = $this->getElementsByTagName('body')->item(0);

        // Insert Body Script
        if (count($this->bodyScriptNodes)) {
            foreach ($this->bodyScriptNodes as $scriptNode) {
                $bodyElement->appendChild($this->importNode($scriptNode));
            }
        }

        $bodyScriptContent = trim($this->bodyScriptContent->__toString());

        if ( ! empty($bodyScriptContent)) {
            $scriptElement = $this->createElement('script', $bodyScriptContent);
            $scriptElement->setAttribute('type', 'text/javascript');
            $bodyElement->appendChild($scriptElement);
        }

        $output = parent::saveHTML($node);

        if ($this->formatOutput === true) {
            $beautifier = new Dom\Beautifier();
            $output = $beautifier->format($output);
        }

        return (string)$output;
    }

    // ------------------------------------------------------------------------

    /**
     * Document::find
     *
     * JQuery style document expression finder.
     *
     * @param string $expression String of document expression.
     *
     * @return Dom\Lists\Nodes
     */
    public function find($expression)
    {
        $xpath = new Dom\XPath($this);

        $xpath->registerNamespace("php", "http://php.net/xpath");
        $xpath->registerPhpFunctions();

        return $xpath->query($expression);
    }

    // ------------------------------------------------------------------------

    /**
     * Document::importSourceNode
     *
     * Import HTML source code into document.
     *
     * @param string $source HTML Source Code.
     *
     * @return \DOMNode|\O2System\Html\Dom\Element
     */
    public function importSourceNode($source)
    {
        $DOMDocument = new self();
        $DOMDocument->loadHTML($source);

        $this->metaNodes->import($DOMDocument->metaNodes);
        $this->headScriptNodes->import($DOMDocument->headScriptNodes);
        $this->bodyScriptNodes->import($DOMDocument->bodyScriptNodes);
        $this->linkNodes->import($DOMDocument->linkNodes);
        $this->styleContent->import($DOMDocument->styleContent);
        $this->headScriptContent->import($DOMDocument->headScriptContent);
        $this->bodyScriptContent->import($DOMDocument->bodyScriptContent);

        $bodyElement = $DOMDocument->getElementsByTagName('body')->item(0);

        if ($bodyElement->firstChild instanceof Dom\Element) {
            return $bodyElement->firstChild;
        } elseif ($bodyElement->firstChild instanceof \DOMText) {
            foreach ($bodyElement->childNodes as $childNode) {
                if ($childNode instanceof Dom\Element) {
                    return $childNode->cloneNode(true);
                    break;
switch ($x) {
    case 1:
        return 'foo';
        break; // This break is not necessary and can be left off.
}
                }
            }
        }

        return $bodyElement;
    }

    // ------------------------------------------------------------------------

    /**
     * Document::loadHTML
     *
     * Load HTML from a string.
     *
     * @see   http://php.net/manual/domdocument.loadhtml.php
     *
     * @param string     $source  <p>
     *                            The HTML string.
     *                            </p>
     * @param int|string $options [optional] <p>
     *                            Since PHP 5.4.0 and Libxml 2.6.0, you may also
     *                            use the options parameter to specify additional Libxml parameters.
     *                            </p>
     *
     * @return bool true on success or false on failure. If called statically, returns a
     * DOMDocument and issues E_STRICT
     * warning.
     * @since 5.0
     */
    public function loadHTML($source, $options = 0)
    {
        // Enables libxml errors handling
        $internalErrorsOptionValue = libxml_use_internal_errors();

        if ($internalErrorsOptionValue === false) {
            libxml_use_internal_errors(true);
        }

        $source = $this->parseHTML($source);

        $DOMDocument = new \DOMDocument();
        $DOMDocument->formatOutput = true;
        $DOMDocument->preserveWhiteSpace = false;

        if ($this->encoding === 'UTF-8') {
            if (function_exists('mb_convert_encoding')) {
                $source = mb_convert_encoding($source, 'HTML-ENTITIES', 'UTF-8');
            } else {
                $source = utf8_decode($source);
            }

            $DOMDocument->encoding = 'UTF-8';
        }

        if (empty($source)) {
            return false;
        }

        $DOMDocument->loadHTML($source, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);

        $headElement = $this->getElementsByTagName('head')->item(0);
        $bodyElement = $this->getElementsByTagName('body')->item(0);

        // Import head attributes
        if (null !== ($sourceHeadElement = $DOMDocument->getElementsByTagName('head')->item(0))) {
            if ($sourceHeadElement->attributes->length > 0) {
                foreach ($sourceHeadElement->attributes as $attribute) {
                    $headElement->setAttribute($attribute->name, $attribute->value);
                }
            }
        }

        // Import body attributes and child nodes
        if (null !== ($sourceBodyElement = $DOMDocument->getElementsByTagName('body')->item(0))) {
            // Import body attributes
            if ($sourceBodyElement->attributes->length > 0) {
                foreach ($sourceBodyElement->attributes as $attribute) {
                    $bodyElement->setAttribute($attribute->name, $attribute->value);
                }
            }

            // Import body child nodes
            foreach ($sourceBodyElement->childNodes as $childNode) {
                $childNode = $this->importNode($childNode, true);
                $bodyElement->appendChild($childNode);
            }
        } elseif ($bodyChildNode = $this->importNode($DOMDocument->firstChild, true)) {
            $bodyElement->appendChild($bodyChildNode);
        }
    }

    // ------------------------------------------------------------------------

    /**
     * Document::parseHTML
     *
     * Parse HTML Source Code.
     *
     * @param string $source HTML Source Code.
     *
     * @return mixed
     */
    private function parseHTML($source)
    {
        $DOMDocument = new \DOMDocument();

        // Has inline script element
        if (preg_match_all('/<script((?:(?!src=).)*?)>(.*?)<\/script>/smix', $source, $matches)) {
            if (isset($matches[ 2 ])) {
                foreach ($matches[ 2 ] as $match) {
                    $script = trim($match);
                    $this->bodyScriptContent[ md5($script) ] = $script . PHP_EOL;
                }
            }
        }

        // Remove all inline script first
        $source = preg_replace('/<script((?:(?!src=).)*?)>(.*?)<\/script>/smix', '', $source);

        $DOMDocument->loadHTML($source);

        $DOMXPath = new \DOMXPath($DOMDocument);
        $metas = $DOMXPath->query('//meta');
        foreach ($metas as $meta) {
            $attributes = [];
            foreach ($meta->attributes as $name => $attribute) {
                $attributes[ $name ] = $attribute->nodeValue;
            }

            $this->metaNodes->createElement($attributes);
        }

        $source = preg_replace('#<meta(.*?)>#is', '', $source); // clean up all inline meta tags

        $links = $DOMXPath->query('//link'); // find all inline link tags
        foreach ($links as $link) {
            $attributes = [];
            foreach ($link->attributes as $name => $attribute) {
                $attributes[ $name ] = $attribute->nodeValue;
            }

            $this->linkNodes->createElement($attributes);
        }

        $source = preg_replace('#<link(.*?)>#is', '', $source); // clean up all inline meta tags

        $scripts = $DOMXPath->query('//head/script'); // find all inline script tags
        foreach ($scripts as $script) {
            $attributes = [];
            foreach ($script->attributes as $name => $attribute) {
                $attributes[ $name ] = $attribute->nodeValue;
            }
        }

        $scripts = $DOMXPath->query('//body/script'); // find all inline script tags
        foreach ($scripts as $script) {
            $attributes = [];
            foreach ($script->attributes as $name => $attribute) {
                $attributes[ $name ] = $attribute->nodeValue;
            }

            if ($script->textContent == '') {
                $this->bodyScriptNodes->createElement($attributes);
            }
        }

        $source = preg_replace('#<script(.*?)>(.*?)</script>#is', '', $source);

        // Has inline style Element
        if (preg_match_all('/((<[\\s\\/]*style\\b[^>]*>)([^>]*)(<\\/style>))/i', $source, $matches)) {
            if (isset($matches[ 3 ])) {
                foreach ($matches[ 3 ] as $match) {
                    $style = trim($match);
                    $this->styleContent[ md5($style) ] = $style . PHP_EOL;
                }
            }

            $source = preg_replace('#<style(.*?)>(.*?)</style>#is', '', $source);
        }

        $codes = $DOMXPath->query('//code');
        foreach ($codes as $code) {
            if ($code->textContent == '') {
                $code = str_replace(['{{php', '/php}}'], ['<?php', '?>'], $code->textContent);
                $code = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\r\n", $code);
                $source = str_replace($code, htmlentities($code), $source);
            }
        }

        $source = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\r\n", $source);

        return $source;
    }

    // ------------------------------------------------------------------------

    /**
     * Document::load
     *
     * Load HTML from a file.
     *
     * @link  http://php.net/manual/domdocument.load.php
     *
     * @param string     $filePath <p>
     *                             The path to the HTML document.
     *                             </p>
     * @param int|string $options  [optional] <p>
     *                             Bitwise OR
     *                             of the libxml option constants.
     *                             </p>
     *
     * @return mixed true on success or false on failure. If called statically, returns a
     * DOMDocument and issues E_STRICT
     * warning.
     * @since 5.0
     */
    public function load($filePath, $options = null)
    {
        if (file_exists($filePath)) {
            return $this->loadHTMLFile($filePath, $options);
        } elseif (is_string($filePath)) {

            return $this->loadHTML($filePath, $options);
        } elseif ( ! empty($filePath)) {
            return parent::load($filePath, $options);
        }
    }

    /**
     * Document::loadHTMLFile
     *
     * Load HTML from a file.
     *
     * @see   http://php.net/manual/domdocument.loadhtmlfile.php
     *
     * @param string     $filePath <p>
     *                             The path to the HTML file.
     *                             </p>
     * @param int|string $options  [optional] <p>
     *
     * Since PHP 5.4.0 and Libxml 2.6.0, you may also
     * use the options parameter to specify additional Libxml parameters.
     * </p>
     *
     * @return bool true on success or false on failure. If called statically, returns a
     * DOMDocument and issues E_STRICT
     * warning.
     * @since 5.0
     */
    public function loadHTMLFile($filePath, $options = 0)
    {
        return $this->loadHTML(file_get_contents($filePath), $options);
    }

    // ------------------------------------------------------------------------

    /**
     * Document::__toString
     *
     * Convert document into HTML source code string.
     *
     * @return string
     */
    public function __toString()
    {
        return $this->saveHTML();
    }
}

1			<?php
2			/**
3			* This file is part of the O2System Framework package.
4			*
5			* For the full copyright and license information, please view the LICENSE
6			* file that was distributed with this source code.
7			*
8			* @author Steeve Andrian Salim
9			* @copyright Copyright (c) Steeve Andrian Salim
10			*/
11
12			// ------------------------------------------------------------------------
13
14			namespace O2System\Html;
15
16			// ------------------------------------------------------------------------
17
18			/**
19			* Class Document
20			*
21			* @package O2System\HTML
22			*/
23			class Document extends \DOMDocument
24			{
25			/**
26			* Document Meta Nodes
27			*
28			* @var \O2System\Html\Dom\Lists\Meta
29			*/
30			public $metaNodes;
31
32			/**
33			* Document Link Nodes
34			*
35			* @var \O2System\Html\Dom\Lists\Asset
36			*/
37			public $linkNodes;
38
39			/**
40			* Document Style Content
41			*
42			* @var \O2System\Html\Dom\Style
43			*/
44			public $styleContent;
45
46			/**
47			* Document Script Nodes
48			*
49			* @var \O2System\Html\Dom\Lists\Asset
50			*/
51			public $headScriptNodes;
52
53			/**
54			* Document Script Content
55			*
56			* @var \O2System\Html\Dom\Script
57			*/
58			public $headScriptContent;
59
60			/**
61			* Document Script Nodes
62			*
63			* @var \O2System\Html\Dom\Lists\Asset
64			*/
65			public $bodyScriptNodes;
66
67			/**
68			* Document Script Content
69			*
70			* @var \O2System\Html\Dom\Script
71			*/
72			public $bodyScriptContent;
73
74			// ------------------------------------------------------------------------
75
76			/**
77			* Document::__construct
78			*
79			* @param string $version Document version.
80			* @param string $encoding Document encoding.
81			*
82			* @return Document
83			*/
84			public function __construct($version = '1.0', $encoding = 'UTF-8')
85			{
86			language()
87			->addFilePath(__DIR__ . DIRECTORY_SEPARATOR)
88			->loadFile('html');
89
90			parent::__construct($version, $encoding);
91
92			$this->registerNodeClass('DOMElement', '\O2System\Html\Dom\Element');
93			$this->registerNodeClass('DOMAttr', '\O2System\Html\Dom\Attr');
94
95			$this->formatOutput = true;
96
97			$this->metaNodes = new Dom\Lists\Meta($this);
98
99			$this->linkNodes = new Dom\Lists\Asset($this);
100			$this->linkNodes->element = 'link';
101
102			$this->styleContent = new Dom\Style();
103
104			$this->headScriptNodes = new Dom\Lists\Asset($this);
105			$this->headScriptNodes->element = 'script';
106			$this->headScriptContent = new Dom\Script();
107
108			$this->bodyScriptNodes = new Dom\Lists\Asset($this);
109			$this->bodyScriptNodes->element = 'script';
110			$this->bodyScriptContent = new Dom\Script();
111
112			$this->loadHTMLTemplate();
113			}
114
115			// ------------------------------------------------------------------------
116
117			/**
118			* Document::loadHTMLTemplate
119			*
120			* Load HTML template from a file.
121			*
122			* @return void
123			*/
124			protected function loadHTMLTemplate()
125			{
126			$htmlTemplate = <<<HTML
127			<!DOCTYPE html>
128			<html lang="en">
129			<head>
130			<meta charset="UTF-8">
131			<title>O2System HTML</title>
132			</head>
133			<body>
134			</body>
135			</html>
136			HTML;
137
138			parent::loadHTML($htmlTemplate);
139			}
140
141			// ------------------------------------------------------------------------
142
143			/**
144			* Document::__get
145			*
146			* @param string $tagName The document tag element.
147			*
148			* @return mixed The value at the specified index or false.
149			*/
150			public function &__get($tagName)
151			{
152			$getDocument[ $tagName ] = null;
			0 ignored issues – show Comprehensibility Best Practice introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report `$getDocument` was never initialized. Although not strictly required by PHP, it is generally a good practice to add `$getDocument = array();` before regardless. Loading history...
153
154			if (in_array($tagName, ['html', 'head', 'body', 'title'])) {
155			$getDocument[ $tagName ] = $this->getElementsByTagName($tagName)->item(0);
156			}
157
158			return $getDocument[ $tagName ];
159			}
160
161			// ------------------------------------------------------------------------
162
163			/**
164			* Document::saveHTMLFile
165			*
166			* Dumps the internal document into a file using HTML formatting
167			*
168			* @see http://php.net/manual/domdocument.savehtmlfile.php
169			*
170			* @param string $filePath <p>
171			* The path to the saved HTML document.
172			* </p>
173			*
174			* @return int the number of bytes written or false if an error occurred.
175			* @since 5.0
176			*/
177			public function saveHTMLFile($filePath)
178			{
179			if ( ! is_string($filePath)) {
			0 ignored issues – show introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report The condition `is_string($filePath)` is always `true`. Loading history...
180			throw new \InvalidArgumentException('The filename argument must be of type string');
181			}
182
183			if ( ! is_writable($filePath)) {
184			return false;
			0 ignored issues – show Bug Best Practice introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report The expression `return false` returns the type `false` which is incompatible with the documented return type `integer`. Loading history...
185			}
186
187			$result = $this->saveHTML();
188			file_put_contents($filePath, $result);
189			$bytesWritten = filesize($filePath);
190
191			if ($bytesWritten === strlen($result)) {
192			return $bytesWritten;
193			}
194
195			return false;
			0 ignored issues – show Bug Best Practice introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report The expression `return false` returns the type `false` which is incompatible with the documented return type `integer`. Loading history...
196			}
197
198			// ------------------------------------------------------------------------
199
200			/**
201			* Document::saveHTML
202			*
203			* Dumps the internal document into a string using HTML formatting.
204			*
205			* @see http://php.net/manual/domdocument.savehtml.php
206			*
207			* @param \DOMNode $node [optional] parameter to output a subset of the document.
208			*
209			* @return string the HTML, or false if an error occurred.
210			* @since 5.0
211			*/
212			public function saveHTML(\DOMNode $node = null)
213			{
214			$headElement = $this->getElementsByTagName('head')->item(0);
215
216			$styleContent = trim($this->styleContent->__toString());
217
218			if ( ! empty($styleContent)) {
219			$styleElement = $this->createElement('style', $styleContent);
220			$styleElement->setAttribute('type', 'text/css');
221			$headElement->appendChild($styleElement);
222			}
223
224			$titleElement = $this->getElementsByTagName('title')->item(0);
225
226			// Insert Meta
227			if ($this->metaNodes->count()) {
228			$metaNodes = array_reverse($this->metaNodes->getArrayCopy());
229
230			foreach ($metaNodes as $metaNode) {
231			$headElement->insertBefore($this->importNode($metaNode), $titleElement);
232			}
233			}
234
235			// Insert Link
236			if ($this->linkNodes->count()) {
237			foreach ($this->linkNodes as $linkNode) {
238			$headElement->appendChild($this->importNode($linkNode));
239			}
240			}
241
242			// Insert Head Script
243			if (count($this->headScriptNodes)) {
244			foreach ($this->headScriptNodes as $scriptNode) {
245			$headElement->appendChild($this->importNode($scriptNode));
246			}
247			}
248
249			$headScriptContent = trim($this->headScriptContent->__toString());
250
251			if ( ! empty($headScriptContent)) {
252			$scriptElement = $this->createElement('script', $headScriptContent);
253			$scriptElement->setAttribute('type', 'text/javascript');
254			$headElement->appendChild($scriptElement);
255			}
256
257			$bodyElement = $this->getElementsByTagName('body')->item(0);
258
259			// Insert Body Script
260			if (count($this->bodyScriptNodes)) {
261			foreach ($this->bodyScriptNodes as $scriptNode) {
262			$bodyElement->appendChild($this->importNode($scriptNode));
263			}
264			}
265
266			$bodyScriptContent = trim($this->bodyScriptContent->__toString());
267
268			if ( ! empty($bodyScriptContent)) {
269			$scriptElement = $this->createElement('script', $bodyScriptContent);
270			$scriptElement->setAttribute('type', 'text/javascript');
271			$bodyElement->appendChild($scriptElement);
272			}
273
274			$output = parent::saveHTML($node);
275
276			if ($this->formatOutput === true) {
277			$beautifier = new Dom\Beautifier();
278			$output = $beautifier->format($output);
279			}
280
281			return (string)$output;
282			}
283
284			// ------------------------------------------------------------------------
285
286			/**
287			* Document::find
288			*
289			* JQuery style document expression finder.
290			*
291			* @param string $expression String of document expression.
292			*
293			* @return Dom\Lists\Nodes
294			*/
295			public function find($expression)
296			{
297			$xpath = new Dom\XPath($this);
298
299			$xpath->registerNamespace("php", "http://php.net/xpath");
300			$xpath->registerPhpFunctions();
301
302			return $xpath->query($expression);
303			}
304
305			// ------------------------------------------------------------------------
306
307			/**
308			* Document::importSourceNode
309			*
310			* Import HTML source code into document.
311			*
312			* @param string $source HTML Source Code.
313			*
314			* @return \DOMNode\|\O2System\Html\Dom\Element
315			*/
316			public function importSourceNode($source)
317			{
318			$DOMDocument = new self();
319			$DOMDocument->loadHTML($source);
320
321			$this->metaNodes->import($DOMDocument->metaNodes);
322			$this->headScriptNodes->import($DOMDocument->headScriptNodes);
323			$this->bodyScriptNodes->import($DOMDocument->bodyScriptNodes);
324			$this->linkNodes->import($DOMDocument->linkNodes);
325			$this->styleContent->import($DOMDocument->styleContent);
326			$this->headScriptContent->import($DOMDocument->headScriptContent);
327			$this->bodyScriptContent->import($DOMDocument->bodyScriptContent);
328
329			$bodyElement = $DOMDocument->getElementsByTagName('body')->item(0);
330
331			if ($bodyElement->firstChild instanceof Dom\Element) {
332			return $bodyElement->firstChild;
333			} elseif ($bodyElement->firstChild instanceof \DOMText) {
334			foreach ($bodyElement->childNodes as $childNode) {
335			if ($childNode instanceof Dom\Element) {
336			return $childNode->cloneNode(true);
337			break;
			0 ignored issues – show Unused Code introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report `break` is not strictly necessary here and could be removed. The `break` statement is not necessary if it is preceded for example by a `return` statement: switch ($x) { case 1: return 'foo'; break; // This break is not necessary and can be left off. } If you would like to keep this construct to be consistent with other `case` statements, you can safely mark this issue as a false-positive. Loading history...
338			}
339			}
340			}
341
342			return $bodyElement;
343			}
344
345			// ------------------------------------------------------------------------
346
347			/**
348			* Document::loadHTML
349			*
350			* Load HTML from a string.
351			*
352			* @see http://php.net/manual/domdocument.loadhtml.php
353			*
354			* @param string $source <p>
355			* The HTML string.
356			* </p>
357			* @param int\|string $options [optional] <p>
358			* Since PHP 5.4.0 and Libxml 2.6.0, you may also
359			* use the options parameter to specify additional Libxml parameters.
360			* </p>
361			*
362			* @return bool true on success or false on failure. If called statically, returns a
363			* DOMDocument and issues E_STRICT
364			* warning.
365			* @since 5.0
366			*/
367			public function loadHTML($source, $options = 0)
368			{
369			// Enables libxml errors handling
370			$internalErrorsOptionValue = libxml_use_internal_errors();
371
372			if ($internalErrorsOptionValue === false) {
373			libxml_use_internal_errors(true);
374			}
375
376			$source = $this->parseHTML($source);
377
378			$DOMDocument = new \DOMDocument();
379			$DOMDocument->formatOutput = true;
380			$DOMDocument->preserveWhiteSpace = false;
381
382			if ($this->encoding === 'UTF-8') {
383			if (function_exists('mb_convert_encoding')) {
384			$source = mb_convert_encoding($source, 'HTML-ENTITIES', 'UTF-8');
385			} else {
386			$source = utf8_decode($source);
387			}
388
389			$DOMDocument->encoding = 'UTF-8';
390			}
391
392			if (empty($source)) {
393			return false;
394			}
395
396			$DOMDocument->loadHTML($source, LIBXML_HTML_NOIMPLIED \| LIBXML_HTML_NODEFDTD);
397
398			$headElement = $this->getElementsByTagName('head')->item(0);
399			$bodyElement = $this->getElementsByTagName('body')->item(0);
400
401			// Import head attributes
402			if (null !== ($sourceHeadElement = $DOMDocument->getElementsByTagName('head')->item(0))) {
403			if ($sourceHeadElement->attributes->length > 0) {
404			foreach ($sourceHeadElement->attributes as $attribute) {
405			$headElement->setAttribute($attribute->name, $attribute->value);
406			}
407			}
408			}
409
410			// Import body attributes and child nodes
411			if (null !== ($sourceBodyElement = $DOMDocument->getElementsByTagName('body')->item(0))) {
412			// Import body attributes
413			if ($sourceBodyElement->attributes->length > 0) {
414			foreach ($sourceBodyElement->attributes as $attribute) {
415			$bodyElement->setAttribute($attribute->name, $attribute->value);
416			}
417			}
418
419			// Import body child nodes
420			foreach ($sourceBodyElement->childNodes as $childNode) {
421			$childNode = $this->importNode($childNode, true);
422			$bodyElement->appendChild($childNode);
423			}
424			} elseif ($bodyChildNode = $this->importNode($DOMDocument->firstChild, true)) {
425			$bodyElement->appendChild($bodyChildNode);
426			}
427			}
428
429			// ------------------------------------------------------------------------
430
431			/**
432			* Document::parseHTML
433			*
434			* Parse HTML Source Code.
435			*
436			* @param string $source HTML Source Code.
437			*
438			* @return mixed
439			*/
440			private function parseHTML($source)
441			{
442			$DOMDocument = new \DOMDocument();
443
444			// Has inline script element
445			if (preg_match_all('/<script((?:(?!src=).)?)>(.?)<\/script>/smix', $source, $matches)) {
446			if (isset($matches[ 2 ])) {
447			foreach ($matches[ 2 ] as $match) {
448			$script = trim($match);
449			$this->bodyScriptContent[ md5($script) ] = $script . PHP_EOL;
450			}
451			}
452			}
453
454			// Remove all inline script first
455			$source = preg_replace('/<script((?:(?!src=).)?)>(.?)<\/script>/smix', '', $source);
456
457			$DOMDocument->loadHTML($source);
458
459			$DOMXPath = new \DOMXPath($DOMDocument);
460			$metas = $DOMXPath->query('//meta');
461			foreach ($metas as $meta) {
462			$attributes = [];
463			foreach ($meta->attributes as $name => $attribute) {
464			$attributes[ $name ] = $attribute->nodeValue;
465			}
466
467			$this->metaNodes->createElement($attributes);
468			}
469
470			$source = preg_replace('#<meta(.*?)>#is', '', $source); // clean up all inline meta tags
471
472			$links = $DOMXPath->query('//link'); // find all inline link tags
473			foreach ($links as $link) {
474			$attributes = [];
475			foreach ($link->attributes as $name => $attribute) {
476			$attributes[ $name ] = $attribute->nodeValue;
477			}
478
479			$this->linkNodes->createElement($attributes);
480			}
481
482			$source = preg_replace('#<link(.*?)>#is', '', $source); // clean up all inline meta tags
483
484			$scripts = $DOMXPath->query('//head/script'); // find all inline script tags
485			foreach ($scripts as $script) {
486			$attributes = [];
487			foreach ($script->attributes as $name => $attribute) {
488			$attributes[ $name ] = $attribute->nodeValue;
489			}
490			}
491
492			$scripts = $DOMXPath->query('//body/script'); // find all inline script tags
493			foreach ($scripts as $script) {
494			$attributes = [];
495			foreach ($script->attributes as $name => $attribute) {
496			$attributes[ $name ] = $attribute->nodeValue;
497			}
498
499			if ($script->textContent == '') {
500			$this->bodyScriptNodes->createElement($attributes);
501			}
502			}
503
504			$source = preg_replace('#<script(.?)>(.?)</script>#is', '', $source);
505
506			// Has inline style Element
507			if (preg_match_all('/((<[\\s\\/]style\\b[^>]>)([^>]*)(<\\/style>))/i', $source, $matches)) {
508			if (isset($matches[ 3 ])) {
509			foreach ($matches[ 3 ] as $match) {
510			$style = trim($match);
511			$this->styleContent[ md5($style) ] = $style . PHP_EOL;
512			}
513			}
514
515			$source = preg_replace('#<style(.?)>(.?)</style>#is', '', $source);
516			}
517
518			$codes = $DOMXPath->query('//code');
519			foreach ($codes as $code) {
520			if ($code->textContent == '') {
521			$code = str_replace(['{{php', '/php}}'], ['<?php', '?>'], $code->textContent);
522			$code = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\r\n", $code);
523			$source = str_replace($code, htmlentities($code), $source);
524			}
525			}
526
527			$source = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\r\n", $source);
528
529			return $source;
530			}
531
532			// ------------------------------------------------------------------------
533
534			/**
535			* Document::load
536			*
537			* Load HTML from a file.
538			*
539			* @link http://php.net/manual/domdocument.load.php
540			*
541			* @param string $filePath <p>
542			* The path to the HTML document.
543			* </p>
544			* @param int\|string $options [optional] <p>
545			* Bitwise OR
546			* of the libxml option constants.
547			* </p>
548			*
549			* @return mixed true on success or false on failure. If called statically, returns a
550			* DOMDocument and issues E_STRICT
551			* warning.
552			* @since 5.0
553			*/
554			public function load($filePath, $options = null)
555			{
556			if (file_exists($filePath)) {
557			return $this->loadHTMLFile($filePath, $options);
558			} elseif (is_string($filePath)) {
			0 ignored issues – show introduced 2019-01-16 12:18 UTC by Report Bug Copy Issue Report The condition `is_string($filePath)` is always `true`. Loading history...
559			return $this->loadHTML($filePath, $options);
560			} elseif ( ! empty($filePath)) {
561			return parent::load($filePath, $options);
562			}
563			}
564
565			/**
566			* Document::loadHTMLFile
567			*
568			* Load HTML from a file.
569			*
570			* @see http://php.net/manual/domdocument.loadhtmlfile.php
571			*
572			* @param string $filePath <p>
573			* The path to the HTML file.
574			* </p>
575			* @param int\|string $options [optional] <p>
576			*
577			* Since PHP 5.4.0 and Libxml 2.6.0, you may also
578			* use the options parameter to specify additional Libxml parameters.
579			* </p>
580			*
581			* @return bool true on success or false on failure. If called statically, returns a
582			* DOMDocument and issues E_STRICT
583			* warning.
584			* @since 5.0
585			*/
586			public function loadHTMLFile($filePath, $options = 0)
587			{
588			return $this->loadHTML(file_get_contents($filePath), $options);
589			}
590
591			// ------------------------------------------------------------------------
592
593			/**
594			* Document::__toString
595			*
596			* Convert document into HTML source code string.
597			*
598			* @return string
599			*/
600			public function __toString()
601			{
602			return $this->saveHTML();
603			}
604			}

o2system / html

GitHub Access Token became invalid

Document::loadHTMLFile() A last analyzed 2019-09-25 13:07 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

Document::loadHTMLFile() A
last analyzed 2019-09-25 13:07 UTC