Element::parse() - Code Metrics - Inspection of "Added a real life parsing test. Fix a bunch of pa..." - kevintweber/HtmlTokenizer - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( a6e461...7a71e4 )

by Kevin

created 2016-04-16 16:46 UTC

Element::parse() C

↳ Parent: Element

Complexity

Conditions	8
Paths	10

Size

Total Lines	43
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	23
CRAP Score	8

Importance

Changes	7
Bugs	3	Features	5

Metric	Value
c	7
b	3
f	5
dl	0
loc	43
ccs	23
cts	23
cp	1
rs	5.3846
cc	8
eloc	23
nc	10
nop	1
crap	8

<?php

namespace Kevintweber\HtmlTokenizer\Tokens;

use Kevintweber\HtmlTokenizer\Exceptions\ParseException;

class Element extends AbstractToken
{
    /** @var array[Token] */
    private $attributes;

    /** @var array[Token] */
    private $children;

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

    public function __construct(Token $parent = null, $throwOnError = false)
    {
        parent::__construct(Token::ELEMENT, $parent, $throwOnError);

        $this->attributes = array();
        $this->children = array();
        $this->name = null;
    }

    /**
     * Does the parent have an implied closing tag?
     *
     * @param string $html
     *
     * @return boolean
     */
    public function isClosingElementImplied($html)
    {
        $parent = $this->getParent();
        if ($parent === null || !($parent instanceof self)) {
            return false;
        }

        $name = $this->parseElementName($html);
        $parentName = $parent->getName();

        // HEAD: no closing tag.
        if ($name === 'body' && $parentName === 'head') {
            return true;
        }

        // P
        $elementsNotChildrenOfP = array(
            'address',
            'article',
            'aside',
            'blockquote',
            'details',
            'div',
            'dl',
            'fieldset',
            'figcaption',
            'figure',
            'footer',
            'form',
            'h1',
            'h2',
            'h3',
            'h4',
            'h5',
            'h6',
            'header',
            'hgroup',
            'hr',
            'main',
            'menu',
            'nav',
            'ol',
            'p',
            'pre',
            'section',
            'table',
            'ul'
        );
        if ($parentName === 'p' && array_search($name, $elementsNotChildrenOfP) !== false) {
            return true;
        }

        // LI
        if ($parentName == 'li' && $name == 'li') {
            return true;
        }

        // DT and DD
        if (($parentName == 'dt' || $parentName == 'dd') && ($name == 'dt' || $name == 'dd')) {
            return true;
        }

        // RP and RT
        if (($parentName == 'rp' || $parentName == 'rt') && ($name == 'rp' || $name == 'rt')) {
            return true;
        }

        return false;
    }

    /**
     * Will parse this element.
     *
     * @param string $html
     *
     * @return string Remaining HTML.
     */
    public function parse($html)
    {
        $this->name = $this->parseElementName($html);

        // Parse attributes.
        $remainingHtml = substr($html, strlen($this->name) + 1);
        while (strpos($remainingHtml, '>') !== false && preg_match("/^\s*[\/]?>/", $remainingHtml) === 0) {
            $remainingHtml = $this->parseAttribute(trim($remainingHtml));
        }

        // Find position of end of tag.
        $posOfClosingBracket = strpos($remainingHtml, '>');
        if ($posOfClosingBracket === false) {
            if ($this->getThrowOnError()) {
                throw new ParseException('Invalid element: missing closing bracket.');
            }

            return '';
        }

        // Is self-closing?
        $posOfSelfClosingBracket = strpos($remainingHtml, '/>');
        $remainingHtml = trim(substr($remainingHtml, $posOfClosingBracket + 1));
        if ($posOfSelfClosingBracket !== false && $posOfSelfClosingBracket == $posOfClosingBracket - 1) {
            // Self-closing element.
            return $remainingHtml;
        }

        // Lets close those closed-only elements that are left open.
        $closedOnlyElements = array(
            'base',
            'link',
            'meta',
            'hr',
            'br'
        );
        if (array_search($this->name, $closedOnlyElements) !== false) {
            return $remainingHtml;
        }

        // Open element.
        return $this->parseContents($remainingHtml);
    }

    /**
     * Will parse attributes.
     *
     * @param string $html
     *
     * @return string Remaining HTML.
     */
    private function parseAttribute($html)
    {
        $remainingHtml = trim($html);

        // Will match the first entire name/value attribute pair.
        preg_match(
            "/((([a-z0-9\-_]+:)?[a-z0-9\-_]+)(\s*=\s*)?)/i",
            $remainingHtml,
            $attributeMatches
        );

        $name = $attributeMatches[2];
        $remainingHtml = substr(strstr($remainingHtml, $name), strlen($name));
        if (preg_match("/^\s*=\s*/", $remainingHtml) === 0) {
            // Valueless attribute.
            $this->attributes[trim($name)] = true;
        } else {
            $remainingHtml = ltrim($remainingHtml, ' =');
            if ($remainingHtml[0] === "'" || $remainingHtml[0] === '"') {
                // Quote enclosed attribute value.
                $valueMatchSuccessful = preg_match(
                    "/" . $remainingHtml[0] . "(.*?(?<!\\\))" . $remainingHtml[0] . "/s",
                    $remainingHtml,
                    $valueMatches
                );
                if ($valueMatchSuccessful !== 1) {
                    if ($this->getThrowOnError()) {
                        throw new ParseException('Invalid value encapsulation.');
                    }

                    return '';
                }

                $value = $valueMatches[1];
            } else {
                // No quotes enclosing the attribute value.
                preg_match("/(\s*([^>\s]*(?<!\/)))/", $remainingHtml, $valueMatches);
                $value = $valueMatches[2];
            }

            $this->attributes[trim($name)] = $value;

            // Determine remaining html.
            $posOfAttributeValue = strpos($html, $value);
            $remainingHtml = trim(
                substr($html, $posOfAttributeValue + strlen($value))
            );
            $remainingHtml = ltrim($remainingHtml, '\'"/ ');
        }

        return $remainingHtml;
    }

    /**
     * Will parse the contents of this element.
     *
     * @param string $html
     *
     * @return string Remaining HTML.
     */
    private function parseContents($html)
    {
        $remainingHtml = trim($html);
        if ($remainingHtml == '') {
            return '';
        }

        // Nothing to parse inside a script tag.
        if ($this->name == 'script') {
            return $this->parseScriptContents($remainingHtml);
        }

        // Parse contents one token at a time.
        while (preg_match("/^<\/\s*" . $this->name . "\s*>/is", $remainingHtml) === 0) {
            $token = TokenFactory::buildFromHtml(
                $remainingHtml,
                $this,
                $this->getThrowOnError()
            );

            if ($token === false || $token->isClosingElementImplied($remainingHtml)) {
                return $remainingHtml;
            }

            $remainingHtml = trim($token->parse($remainingHtml));
            $this->children[] = $token;
        }

        // Remove remaining closing tag.
        $posOfClosingBracket = strpos($remainingHtml, '>');

        return substr($remainingHtml, $posOfClosingBracket + 1);
    }

    /**
     * Will get the element name from the html string.
     *
     * @param $html string
     *
     * @return string The element name.
     */
    private function parseElementName($html)
    {
        $elementMatchSuccessful = preg_match(
            "/^(<(([a-z0-9\-]+:)?[a-z0-9\-]+))/i",
            $html,
            $elementMatches
        );
        if ($elementMatchSuccessful !== 1) {
            if ($this->getThrowOnError()) {
                throw new ParseException('Invalid element name.');
            }

            return '';
        }

        return strtolower($elementMatches[2]);
    }

    /**
     * Will parse the script contents correctly.
     *
     * @param $html string
     *
     * @return string The remaining HTML.
     */
    private function parseScriptContents($html)
    {
        $remainingHtml = trim($html);

        $matchingResult = preg_match("/(<\/\s*script\s*>)/i", $html, $endOfScriptMatches);
        if ($matchingResult === 0) {
            $value = $remainingHtml;
            $remainingHtml = '';
        } else {
            $closingTag = $endOfScriptMatches[1];
            $value = trim(
                substr($remainingHtml, 0, strpos($remainingHtml, $closingTag))
            );
            $remainingHtml = substr(
                strstr($remainingHtml, $closingTag),
                strlen($closingTag)
            );
        }

        // Handle no contents.
        if ($value == '') {
            return $remainingHtml;
        }

        $text = new Text($this, $this->getThrowOnError(), $value);
        $this->children[] = $text;

        return $remainingHtml;
    }

    /**
     * Getter for 'attributes'.
     *
     * @return array
     */
    public function getAttributes()
    {
        return $this->attributes;
    }

    /**
     * @return boolean
     */
    public function hasAttributes()
    {
        return !empty($this->attributes);
    }

    /**
     * Getter for 'children'.
     *
     * @return array
     */
    public function getChildren()
    {
        return $this->children;
    }

    /**
     * @return boolean
     */
    public function hasChildren()
    {
        return !empty($this->children);
    }

    /**
     * Getter for 'name'.
     *
     * @return string
     */
    public function getName()
    {
        return $this->name;
    }

    public function toArray()
    {
        $result = array(
            'type' => 'element',
            'name' => $this->name
        );

        if (!empty($this->attributes)) {
            $result['attributes'] = array();
            foreach ($this->attributes as $name => $value) {
                $result['attributes'][$name] = $value;
            }
        }

        if (!empty($this->children)) {
            $result['children'] = array();
            foreach ($this->children as $child) {
                $result['children'][] = $child->toArray();
            }
        }

        return $result;
    }
}


1		<?php
2
3		namespace Kevintweber\HtmlTokenizer\Tokens;
4
5		use Kevintweber\HtmlTokenizer\Exceptions\ParseException;
6
7		class Element extends AbstractToken
8		{
9		/** @var array[Token] */
10		private $attributes;
11
12		/** @var array[Token] */
13		private $children;
14
15		/** @var string */
16		private $name;
17
18	51	public function __construct(Token $parent = null, $throwOnError = false)
19		{
20	51	parent::__construct(Token::ELEMENT, $parent, $throwOnError);
21
22	51	$this->attributes = array();
23	51	$this->children = array();
24	51	$this->name = null;
25	51	}
26
27		/**
28		* Does the parent have an implied closing tag?
29		*
30		* @param string $html
31		*
32		* @return boolean
33		*/
34	20	public function isClosingElementImplied($html)
35		{
36	20	$parent = $this->getParent();
37	20	if ($parent === null \|\| !($parent instanceof self)) {
38	1	return false;
39		}
40
41	19	$name = $this->parseElementName($html);
42	19	$parentName = $parent->getName();
43
44		// HEAD: no closing tag.
45	19	if ($name === 'body' && $parentName === 'head') {
46	2	return true;
47		}
48
49		// P
50		$elementsNotChildrenOfP = array(
51	18	'address',
52	18	'article',
53	18	'aside',
54	18	'blockquote',
55	18	'details',
56	18	'div',
57	18	'dl',
58	18	'fieldset',
59	18	'figcaption',
60	18	'figure',
61	18	'footer',
62	18	'form',
63	18	'h1',
64	18	'h2',
65	18	'h3',
66	18	'h4',
67	18	'h5',
68	18	'h6',
69	18	'header',
70	18	'hgroup',
71	18	'hr',
72	18	'main',
73	18	'menu',
74	18	'nav',
75	18	'ol',
76	18	'p',
77	18	'pre',
78	18	'section',
79	18	'table',
80		'ul'
81	18	);
82	18	if ($parentName === 'p' && array_search($name, $elementsNotChildrenOfP) !== false) {
83	3	return true;
84		}
85
86		// LI
87	15	if ($parentName == 'li' && $name == 'li') {
88	1	return true;
89		}
90
91		// DT and DD
92	14	if (($parentName == 'dt' \|\| $parentName == 'dd') && ($name == 'dt' \|\| $name == 'dd')) {
93	4	return true;
94		}
95
96		// RP and RT
97	10	if (($parentName == 'rp' \|\| $parentName == 'rt') && ($name == 'rp' \|\| $name == 'rt')) {
98	4	return true;
99		}
100
101	6	return false;
102		}
103
104		/**
105		* Will parse this element.
106		*
107		* @param string $html
108		*
109		* @return string Remaining HTML.
110		*/
111	35	public function parse($html)
112		{
113	35	$this->name = $this->parseElementName($html);
114
115		// Parse attributes.
116	35	$remainingHtml = substr($html, strlen($this->name) + 1);
117	35	while (strpos($remainingHtml, '>') !== false && preg_match("/^\s*[\/]?>/", $remainingHtml) === 0) {
118	18	$remainingHtml = $this->parseAttribute(trim($remainingHtml));
119	18	}
120
121		// Find position of end of tag.
122	35	$posOfClosingBracket = strpos($remainingHtml, '>');
123	35	if ($posOfClosingBracket === false) {
124	4	if ($this->getThrowOnError()) {
125	1	throw new ParseException('Invalid element: missing closing bracket.');
126		}
127
128	3	return '';
129		}
130
131		// Is self-closing?
132	31	$posOfSelfClosingBracket = strpos($remainingHtml, '/>');
133	31	$remainingHtml = trim(substr($remainingHtml, $posOfClosingBracket + 1));
134	31	if ($posOfSelfClosingBracket !== false && $posOfSelfClosingBracket == $posOfClosingBracket - 1) {
135		// Self-closing element.
136	6	return $remainingHtml;
137		}
138
139		// Lets close those closed-only elements that are left open.
140		$closedOnlyElements = array(
141	26	'base',
142	26	'link',
143	26	'meta',
144	26	'hr',
145		'br'
146	26	);
147	26	if (array_search($this->name, $closedOnlyElements) !== false) {
148	4	return $remainingHtml;
149		}
150
151		// Open element.
152	26	return $this->parseContents($remainingHtml);
153		}
154
155		/**
156		* Will parse attributes.
157		*
158		* @param string $html
159		*
160		* @return string Remaining HTML.
161		*/
162	18	private function parseAttribute($html)
163		{
164	18	$remainingHtml = trim($html);
165
166		// Will match the first entire name/value attribute pair.
167	18	preg_match(
168	18	"/((([a-z0-9\-_]+:)?[a-z0-9\-_]+)(\s=\s)?)/i",
169	18	$remainingHtml,
170		$attributeMatches
171	18	);
172
173	18	$name = $attributeMatches[2];
174	18	$remainingHtml = substr(strstr($remainingHtml, $name), strlen($name));
175	18	if (preg_match("/^\s=\s/", $remainingHtml) === 0) {
176		// Valueless attribute.
177	3	$this->attributes[trim($name)] = true;
178	3	} else {
179	17	$remainingHtml = ltrim($remainingHtml, ' =');
180	17	if ($remainingHtml[0] === "'" \|\| $remainingHtml[0] === '"') {
181		// Quote enclosed attribute value.
182	14	$valueMatchSuccessful = preg_match(
183	14	"/" . $remainingHtml[0] . "(.*?(?<!\\\))" . $remainingHtml[0] . "/s",
184	14	$remainingHtml,
185		$valueMatches
186	14	);
187	14	if ($valueMatchSuccessful !== 1) {
188	1	if ($this->getThrowOnError()) {
189	1	throw new ParseException('Invalid value encapsulation.');
190		}
191
192	1	return '';
193		}
194
195	13	$value = $valueMatches[1];
196	13	} else {
197		// No quotes enclosing the attribute value.
198	6	preg_match("/(\s([^>\s](?<!\/)))/", $remainingHtml, $valueMatches);
199	6	$value = $valueMatches[2];
200		}
201
202	16	$this->attributes[trim($name)] = $value;
203
204		// Determine remaining html.
205	16	$posOfAttributeValue = strpos($html, $value);
206	16	$remainingHtml = trim(
207	16	substr($html, $posOfAttributeValue + strlen($value))
208	16	);
209	16	$remainingHtml = ltrim($remainingHtml, '\'"/ ');
210		}
211
212	17	return $remainingHtml;
213		}
214
215		/**
216		* Will parse the contents of this element.
217		*
218		* @param string $html
219		*
220		* @return string Remaining HTML.
221		*/
222	26	private function parseContents($html)
223		{
224	26	$remainingHtml = trim($html);
225	26	if ($remainingHtml == '') {
226	10	return '';
227		}
228
229		// Nothing to parse inside a script tag.
230	16	if ($this->name == 'script') {
231	3	return $this->parseScriptContents($remainingHtml);
232		}
233
234		// Parse contents one token at a time.
235	14	while (preg_match("/^<\/\s" . $this->name . "\s>/is", $remainingHtml) === 0) {
236	12	$token = TokenFactory::buildFromHtml(
237	12	$remainingHtml,
238	12	$this,
239	12	$this->getThrowOnError()
240	12	);
241
242	12	if ($token === false \|\| $token->isClosingElementImplied($remainingHtml)) {
243	3	return $remainingHtml;
244		}
245
246	11	$remainingHtml = trim($token->parse($remainingHtml));
247	11	$this->children[] = $token;
248	11	}
249
250		// Remove remaining closing tag.
251	13	$posOfClosingBracket = strpos($remainingHtml, '>');
252
253	13	return substr($remainingHtml, $posOfClosingBracket + 1);
254		}
255
256		/**
257		* Will get the element name from the html string.
258		*
259		* @param $html string
260		*
261		* @return string The element name.
262		*/
263	49	private function parseElementName($html)
264		{
265	49	$elementMatchSuccessful = preg_match(
266	49	"/^(<(([a-z0-9\-]+:)?[a-z0-9\-]+))/i",
267	49	$html,
268		$elementMatches
269	49	);
270	49	if ($elementMatchSuccessful !== 1) {
271	1	if ($this->getThrowOnError()) {
272	1	throw new ParseException('Invalid element name.');
273		}
274
275	1	return '';
276		}
277
278	48	return strtolower($elementMatches[2]);
279		}
280
281		/**
282		* Will parse the script contents correctly.
283		*
284		* @param $html string
285		*
286		* @return string The remaining HTML.
287		*/
288	3	private function parseScriptContents($html)
289		{
290	3	$remainingHtml = trim($html);
291
292	3	$matchingResult = preg_match("/(<\/\sscript\s>)/i", $html, $endOfScriptMatches);
293	3	if ($matchingResult === 0) {
294	1	$value = $remainingHtml;
295	1	$remainingHtml = '';
296	1	} else {
297	2	$closingTag = $endOfScriptMatches[1];
298	2	$value = trim(
299	2	substr($remainingHtml, 0, strpos($remainingHtml, $closingTag))
300	2	);
301	2	$remainingHtml = substr(
302	2	strstr($remainingHtml, $closingTag),
303	2	strlen($closingTag)
304	2	);
305		}
306
307		// Handle no contents.
308	3	if ($value == '') {
309	1	return $remainingHtml;
310		}
311
312	3	$text = new Text($this, $this->getThrowOnError(), $value);
313	3	$this->children[] = $text;
314
315	3	return $remainingHtml;
316		}
317
318		/**
319		* Getter for 'attributes'.
320		*
321		* @return array
322		*/
323	1	public function getAttributes()
324		{
325	1	return $this->attributes;
326		}
327
328		/**
329		* @return boolean
330		*/
331	1	public function hasAttributes()
332		{
333	1	return !empty($this->attributes);
334		}
335
336		/**
337		* Getter for 'children'.
338		*
339		* @return array
340		*/
341	1	public function getChildren()
342		{
343	1	return $this->children;
344		}
345
346		/**
347		* @return boolean
348		*/
349	1	public function hasChildren()
350		{
351	1	return !empty($this->children);
352		}
353
354		/**
355		* Getter for 'name'.
356		*
357		* @return string
358		*/
359	29	public function getName()
360		{
361	29	return $this->name;
362		}
363
364	20	public function toArray()
365		{
366		$result = array(
367	20	'type' => 'element',
368	20	'name' => $this->name
369	20	);
370
371	20	if (!empty($this->attributes)) {
372	12	$result['attributes'] = array();
373	12	foreach ($this->attributes as $name => $value) {
374	12	$result['attributes'][$name] = $value;
375	12	}
376	12	}
377
378	20	if (!empty($this->children)) {
379	9	$result['children'] = array();
380	9	foreach ($this->children as $child) {
381	9	$result['children'][] = $child->toArray();
382	9	}
383	9	}
384
385	20	return $result;
386		}
387		}
388

kevintweber / HtmlTokenizer

Push — master ( a6e461...7a71e4 )

Element::parse() C

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like