GlHtml::getLinksByTagAttribute() - Code Metrics - emmanuelroecker/php-simply-html - Measure and Improve Code Quality continuously with Scrutinizer

GlHtml::getLinksByTagAttribute() A
last analyzed 2016-09-05 15:46 UTC

↳ Parent: GlHtml

Complexity

Conditions	4
Paths	3

Size

Total Lines	10
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	10
rs	9.2
cc	4
eloc	6
nc	3
nop	3

<?php
/**
 * Main Class to manipulate dom
 *
 * PHP version 5.4
 *
 * @category  GLICER
 * @package   GlHtml
 * @author    Emmanuel ROECKER
 * @author    Rym BOUCHAGOUR
 * @copyright 2015 GLICER
 * @license   MIT
 * @link      http://dev.glicer.com/
 *
 * Created : 19/02/15
 * File : GlHtml.php
 *
 */


namespace GlHtml;

use Symfony\Component\CssSelector\CssSelector;
use Symfony\Component\CssSelector\CssSelectorConverter;

/**
 * Class GlHtml
 * @package GlHtml
 */
class GlHtml
{
    /**
     * @var \DOMDocument
     */
    private $dom;

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

    /**
     * @param string $html
     */
    public function __construct($html)
    {
        $html      = self::fixNewlines($html);
        $this->dom = new \DOMDocument();

        $libxml_previous_state = libxml_use_internal_errors(true); //disable warnings
        $this->dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
        libxml_clear_errors();
        libxml_use_internal_errors($libxml_previous_state);

        $this->html = $html;
    }

    /**
     * Unify newlines
     *
     * @param string $text
     *
     * @return string the fixed text
     */
    private static function fixNewlines($text)
    {
        $text = str_replace("\r\n", "\n", $text);
        $text = str_replace("\r", "\n", $text);

        return $text;
    }

    /**
     * return one dom element with $selector css filter
     *
     * @param string $selector CSS 3 Selector
     *
     * @return GlHtmlNode[]
     */
    public function get($selector)
    {
        $xpath = new \DOMXPath($this->dom);

        if (class_exists('Symfony\Component\CssSelector\CssSelector')) {
            $expression = CssSelector::toXPath($selector);
        } else {
            $converter  = new CssSelectorConverter();
            $expression = $converter->toXPath($selector);
        }
        $nodes = $xpath->query($expression);

        $glnodes = [];
        foreach ($nodes as $node) {
            $glnodes[] = new GlHtmlNode($node);
        }

        return $glnodes;
    }

    /**
     * set a list of attributes
     *
     * @param string $selector
     * @param array  $attributes
     */
    public function setAttributes($selector, array $attributes)
    {
        $nodes = $this->get($selector);

        foreach ($nodes as $node) {
            $node->setAttributes($attributes);
        }
    }

    /**
     * @param string $selector
     */
    public function delete($selector)
    {
        $nodes = $this->get($selector);
        foreach ($nodes as $node) {
            $node->delete();
        }
    }

    /**
     * @return string
     */
    public function html()
    {
        return $this->dom->saveHTML();
    }

    public function getText()
    {
        $body = $this->get("body")[0];

        return $body->getText();
    }

    /**
     * @param string $tagname
     * @param string $attribute
     * @param array  $links
     */
    private function getLinksByTagAttribute($tagname, $attribute, array &$links)
    {
        $tagslink = $this->get($tagname);
        foreach ($tagslink as $taglink) {
            $href = $taglink->getAttribute($attribute);
            if (isset($href) && (strlen(trim($href)) > 0)) {
                $links[$href] = $href;
            }
        }
    }

    /**
     * @param bool $all if true get url in text and params
     *
     * @return array
     */
    public function getLinks($all = false)
    {
        $links = [];

        $this->getLinksByTagAttribute("link", "href", $links);
        $this->getLinksByTagAttribute("a", "href", $links);
        $this->getLinksByTagAttribute("script", "src", $links);
        $this->getLinksByTagAttribute("iframe", "src", $links);
        $this->getLinksByTagAttribute("img", "src", $links);

        //get all string started with http
        $regexUrl = '/[">\s]+((http|https|ftp|ftps)\:\/\/(.*?))["<\s]+/';
        $urls     = null;
        if (preg_match_all($regexUrl, $this->html, $urls) > 0) {
            $matches = $urls[1];
            foreach ($matches as $url) {
                if (filter_var($url, FILTER_VALIDATE_URL)) {
                    $links[$url] = $url;
                }
            }
        }

        if ($all) {
            //get all params which can be a url
            $regexParam = '/["](.*?)["]/';
            $params     = [];
            if (preg_match_all($regexParam, $this->html, $params) > 0) {
                $urls = $params[1];
                foreach ($urls as $url) {
                    $url = trim($url);
                    if ((strpbrk($url, "/.") !== false) && (strpbrk($url, " ") === false)) {
                        $links[$url] = $url;
                    }
                }
            }
        }

        foreach ($links as $link) {
            $url = parse_url($link);
            if (!((isset($url['host']) && isset($url['scheme'])) || (isset($url['path'])))) {
                unset($links[$link]);
            }
        }

        return $links;
    }

    public function getSentences()
    {
        $sentences = [];

        $body = $this->get("body");
        if (count($body) > 0) {
            $sentences = $body[0]->getSentences();
        }

        $description = $this->get('meta[name="description"]');
        if (count($description) > 0) {

            $description = trim($description[0]->getAttribute("content"));
            if (strlen($description) > 0) {
                array_unshift($sentences, $description);
            }
        }

        $title = $this->get('title');
        if (count($title) > 0) {

            $title = trim($title[0]->getText());
            if (strlen($title) > 0) {
                array_unshift($sentences, $title);
            }
        }

        return $sentences;
    }

    /**
     * @return GlHtmlSummary[]
     */
    public function getSummary()
    {
        $body = $this->get("body")[0];

        $summary  = [];
        $callback = function (GlHtmlNode $childNode) use (&$summary) {
            $nodeName = $childNode->getName();

            if (preg_match('/^h(\d+)$/', $nodeName, $matches)) {
                $summary[] = new GlHtmlSummary($childNode, $matches[1]);
            }
        };

        $body->callChild($callback);

        return $summary;
    }

    private function convertHToTree(
        &$start,
        $summary,
        array &$summaryTree
    ) {
        $end    = count($summary);
        $number = 1;
        while ($start < $end) {
            $text = $summary[$start]->getNode()->getText();
            $id   = $summary[$start]->getNode()->getAttribute('id');

            $summaryTree[$text] = ['id' => $id, 'children' => []];
            if (($start + 1) < $end) {
                $currentLevel = $summary[$start]->getLevel();
                $nextLevel    = $summary[$start + 1]->getLevel();
                if ($nextLevel > $currentLevel) {
                    $start++;
                    $diff = $this->convertHToTree(
                         $start,
                             $summary,
                             $summaryTree[$text]['children']
                    );
                    if ($diff > 0) {
                        return $diff - 1;
                    }
                } else {
                    if ($nextLevel < $currentLevel) {
                        return ($currentLevel - $nextLevel - 1);
                    }
                }
            }
            $number++;
            $start++;
        }
        
        return 0;
    }

    public function getSummaryTree()
    {
        $summary = $this->getSummary();
        $start       = 0;
        $summaryTree = [];
        $this->convertHToTree($start, $summary, $summaryTree);
        reset($summaryTree); //reset array pointer
        
        return $summaryTree;
    }
}


1		<?php
2		/**
3		* Main Class to manipulate dom
4		*
5		* PHP version 5.4
6		*
7		* @category GLICER
8		* @package GlHtml
9		* @author Emmanuel ROECKER
10		* @author Rym BOUCHAGOUR
11		* @copyright 2015 GLICER
12		* @license MIT
13		* @link http://dev.glicer.com/
14		*
15		* Created : 19/02/15
16		* File : GlHtml.php
17		*
18		*/
19
20
21		namespace GlHtml;
22
23		use Symfony\Component\CssSelector\CssSelector;
24		use Symfony\Component\CssSelector\CssSelectorConverter;
25
26		/**
27		* Class GlHtml
28		* @package GlHtml
29		*/
30		class GlHtml
31		{
32		/**
33		* @var \DOMDocument
34		*/
35		private $dom;
36
37		/**
38		* @var string
39		*/
40		private $html;
41
42		/**
43		* @param string $html
44		*/
45		public function __construct($html)
46		{
47		$html = self::fixNewlines($html);
48		$this->dom = new \DOMDocument();
49
50		$libxml_previous_state = libxml_use_internal_errors(true); //disable warnings
51		$this->dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
52		libxml_clear_errors();
53		libxml_use_internal_errors($libxml_previous_state);
54
55		$this->html = $html;
56		}
57
58		/**
59		* Unify newlines
60		*
61		* @param string $text
62		*
63		* @return string the fixed text
64		*/
65		private static function fixNewlines($text)
66		{
67		$text = str_replace("\r\n", "\n", $text);
68		$text = str_replace("\r", "\n", $text);
69
70		return $text;
71		}
72
73		/**
74		* return one dom element with $selector css filter
75		*
76		* @param string $selector CSS 3 Selector
77		*
78		* @return GlHtmlNode[]
79		*/
80		public function get($selector)
81		{
82		$xpath = new \DOMXPath($this->dom);
83
84		if (class_exists('Symfony\Component\CssSelector\CssSelector')) {
85		$expression = CssSelector::toXPath($selector);
86		} else {
87		$converter = new CssSelectorConverter();
88		$expression = $converter->toXPath($selector);
89		}
90		$nodes = $xpath->query($expression);
91
92		$glnodes = [];
93		foreach ($nodes as $node) {
94		$glnodes[] = new GlHtmlNode($node);
95		}
96
97		return $glnodes;
98		}
99
100		/**
101		* set a list of attributes
102		*
103		* @param string $selector
104		* @param array $attributes
105		*/
106		public function setAttributes($selector, array $attributes)
107		{
108		$nodes = $this->get($selector);
109
110		foreach ($nodes as $node) {
111		$node->setAttributes($attributes);
112		}
113		}
114
115		/**
116		* @param string $selector
117		*/
118		public function delete($selector)
119		{
120		$nodes = $this->get($selector);
121		foreach ($nodes as $node) {
122		$node->delete();
123		}
124		}
125
126		/**
127		* @return string
128		*/
129		public function html()
130		{
131		return $this->dom->saveHTML();
132		}
133
134		public function getText()
135		{
136		$body = $this->get("body")[0];
137
138		return $body->getText();
139		}
140
141		/**
142		* @param string $tagname
143		* @param string $attribute
144		* @param array $links
145		*/
146		private function getLinksByTagAttribute($tagname, $attribute, array &$links)
147		{
148		$tagslink = $this->get($tagname);
149		foreach ($tagslink as $taglink) {
150		$href = $taglink->getAttribute($attribute);
151		if (isset($href) && (strlen(trim($href)) > 0)) {
152		$links[$href] = $href;
153		}
154		}
155		}
156
157		/**
158		* @param bool $all if true get url in text and params
159		*
160		* @return array
161		*/
162		public function getLinks($all = false)
163		{
164		$links = [];
165
166		$this->getLinksByTagAttribute("link", "href", $links);
167		$this->getLinksByTagAttribute("a", "href", $links);
168		$this->getLinksByTagAttribute("script", "src", $links);
169		$this->getLinksByTagAttribute("iframe", "src", $links);
170		$this->getLinksByTagAttribute("img", "src", $links);
171
172		//get all string started with http
173		$regexUrl = '/[">\s]+((http\|https\|ftp\|ftps)\:\/\/(.*?))["<\s]+/';
174		$urls = null;
175		if (preg_match_all($regexUrl, $this->html, $urls) > 0) {
176		$matches = $urls[1];
177		foreach ($matches as $url) {
178		if (filter_var($url, FILTER_VALIDATE_URL)) {
179		$links[$url] = $url;
180		}
181		}
182		}
183
184		if ($all) {
185		//get all params which can be a url
186		$regexParam = '/["](.*?)["]/';
187		$params = [];
188		if (preg_match_all($regexParam, $this->html, $params) > 0) {
189		$urls = $params[1];
190		foreach ($urls as $url) {
191		$url = trim($url);
192		if ((strpbrk($url, "/.") !== false) && (strpbrk($url, " ") === false)) {
193		$links[$url] = $url;
194		}
195		}
196		}
197		}
198
199		foreach ($links as $link) {
200		$url = parse_url($link);
201		if (!((isset($url['host']) && isset($url['scheme'])) \|\| (isset($url['path'])))) {
202		unset($links[$link]);
203		}
204		}
205
206		return $links;
207		}
208
209		public function getSentences()
210		{
211		$sentences = [];
212
213		$body = $this->get("body");
214		if (count($body) > 0) {
215		$sentences = $body[0]->getSentences();
216		}
217
218		$description = $this->get('meta[name="description"]');
219	View Code Duplication	if (count($description) > 0) {
		0 ignored issues – show Duplication introduced 2016-01-03 11:22 UTC by Report Bug Copy Issue Report This code seems to be duplicated across 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...
220		$description = trim($description[0]->getAttribute("content"));
221		if (strlen($description) > 0) {
222		array_unshift($sentences, $description);
223		}
224		}
225
226		$title = $this->get('title');
227	View Code Duplication	if (count($title) > 0) {
		0 ignored issues – show Duplication introduced 2016-01-03 11:22 UTC by Report Bug Copy Issue Report This code seems to be duplicated across 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...
228		$title = trim($title[0]->getText());
229		if (strlen($title) > 0) {
230		array_unshift($sentences, $title);
231		}
232		}
233
234		return $sentences;
235		}
236
237		/**
238		* @return GlHtmlSummary[]
239		*/
240		public function getSummary()
241		{
242		$body = $this->get("body")[0];
243
244		$summary = [];
245		$callback = function (GlHtmlNode $childNode) use (&$summary) {
246		$nodeName = $childNode->getName();
247
248		if (preg_match('/^h(\d+)$/', $nodeName, $matches)) {
249		$summary[] = new GlHtmlSummary($childNode, $matches[1]);
250		}
251		};
252
253		$body->callChild($callback);
254
255		return $summary;
256		}
257
258		private function convertHToTree(
259		&$start,
260		$summary,
261		array &$summaryTree
262		) {
263		$end = count($summary);
264		$number = 1;
265		while ($start < $end) {
266		$text = $summary[$start]->getNode()->getText();
267		$id = $summary[$start]->getNode()->getAttribute('id');
268
269		$summaryTree[$text] = ['id' => $id, 'children' => []];
270		if (($start + 1) < $end) {
271		$currentLevel = $summary[$start]->getLevel();
272		$nextLevel = $summary[$start + 1]->getLevel();
273		if ($nextLevel > $currentLevel) {
274		$start++;
275		$diff = $this->convertHToTree(
276		$start,
277		$summary,
278		$summaryTree[$text]['children']
279		);
280		if ($diff > 0) {
281		return $diff - 1;
282		}
283		} else {
284		if ($nextLevel < $currentLevel) {
285		return ($currentLevel - $nextLevel - 1);
286		}
287		}
288		}
289		$number++;
290		$start++;
291		}
292
293		return 0;
294		}
295
296		public function getSummaryTree()
297		{
298		$summary = $this->getSummary();
299		$start = 0;
300		$summaryTree = [];
301		$this->convertHToTree($start, $summary, $summaryTree);
302		reset($summaryTree); //reset array pointer
303
304		return $summaryTree;
305		}
306		}
307

emmanuelroecker / php-simply-html

GitHub Access Token became invalid

GlHtml::getLinksByTagAttribute() A last analyzed 2016-09-05 15:46 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

GlHtml::getLinksByTagAttribute() A
last analyzed 2016-09-05 15:46 UTC