Element::parse() - Code Metrics - Inspection of "Small refactoring. Better tests." - kevintweber/HtmlTokenizer - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( b928ab...26824d )

by Kevin

created 2016-04-16 03:23 UTC

Element::parse() C

↳ Parent: Element

Complexity

Conditions	7
Paths	8

Size

Total Lines	31
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	16
CRAP Score	7

Importance

Changes	4
Bugs	2	Features	2

Metric	Value
c	4
b	2
f	2
dl	0
loc	31
ccs	16
cts	16
cp	1
rs	6.7272
cc	7
eloc	15
nc	8
nop	1
crap	7

<?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;
    }

    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;
        }

        // Closed-only elements.
        // Closing tags not required.  We will close them now.
        if ($this->isClosedOnlyElement($parentName)) {
            return true;
        }

        // P
        if ($parentName === 'p') {
            switch ($name) {
            case 'address':
            case 'article':
            case 'aside':
            case 'blockquote':
            case 'details':
            case 'div':
            case 'dl':
            case 'fieldset':
            case 'figcaption':
            case 'figure':
            case 'footer':
            case 'form':
            case 'h1':
            case 'h2':
            case 'h3':
            case 'h4':
            case 'h5':
            case 'h6':
            case 'header':
            case 'hgroup':
            case 'hr':
            case 'main':
            case 'menu':
            case 'nav':
            case 'ol':
            case 'p':
            case 'pre':
            case 'section':
            case 'table':
            case 'ul':
                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;
    }

    private function isClosedOnlyElement($name)
    {
        return $name == 'base' ||
            $name == 'link' ||
            $name == 'meta' ||
            $name == 'hr' ||
            $name == 'br';
    }

    public static function isMatch($html)
    {
        return preg_match("/^<[a-zA-Z]/", $html) === 1;
    }

    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 = $this->parseAttribute($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;
        }

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

    private function parseAttribute($html)
    {
        // Will match the first entire name/value attribute pair.
        $attrMatchSuccessful = preg_match(
            "/(\s*([^>\s]*))/",
            $html,
            $attributeMatches
        );
        if ($attrMatchSuccessful !== 1) {
            if ($this->getThrowOnError()) {
                throw new ParseException('Invalid attribute.');
            }

            return '';
        }

        $posOfEqualsSign = strpos($attributeMatches[2], '=');
        if ($posOfEqualsSign === false) {
            // Valueless attribute.
            $this->attributes[trim($attributeMatches[2])] = true;
        } else {
            list($name, $value) = explode('=', $attributeMatches[2]);
            if ($value[0] === "'" || $value[0] === '"') {
                $valueMatchSuccessful = preg_match(
                    "/" . $value[0] . "(.*?(?<!\\\))" . $value[0] . "/s",
                    $value,
                    $valueMatches
                );
                if ($valueMatchSuccessful !== 1) {
                    if ($this->getThrowOnError()) {
                        throw new ParseException('Invalid value encapsulation.');
                    }

                    return '';
                }

                $value = $valueMatches[1];
            }

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

        // Return the html minus the current attribute.
        $posOfAttribute = strpos($html, $attributeMatches[2]);

        return substr($html, $posOfAttribute + strlen($attributeMatches[2]));
    }

    private function parseContents($html)
    {
        $remainingHtml = trim($html);
        if ($remainingHtml == '') {
            return '';
        }

        // 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]);
    }

    public function getAttributes()
    {
        return $this->attributes;
    }

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

    public function getChildren()
    {
        return $this->children;
    }

    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	46	public function __construct(Token $parent = null, $throwOnError = false)
19		{
20	46	parent::__construct(Token::ELEMENT, $parent, $throwOnError);
21
22	46	$this->attributes = array();
23	46	$this->children = array();
24	46	$this->name = null;
25	46	}
26
27	22	public function isClosingElementImplied($html)
28		{
29	22	$parent = $this->getParent();
30	22	if ($parent === null \|\| !($parent instanceof self)) {
31	1	return false;
32		}
33
34	21	$name = $this->parseElementName($html);
35	21	$parentName = $parent->getName();
36
37		// HEAD: no closing tag.
38	21	if ($name === 'body' && $parentName === 'head') {
39	2	return true;
40		}
41
42		// Closed-only elements.
43		// Closing tags not required. We will close them now.
44	20	if ($this->isClosedOnlyElement($parentName)) {
45	6	return true;
46		}
47
48		// P
49	15	if ($parentName === 'p') {
50		switch ($name) {
51	4	case 'address':
52	4	case 'article':
53	4	case 'aside':
54	4	case 'blockquote':
55	4	case 'details':
56	4	case 'div':
57	4	case 'dl':
58	4	case 'fieldset':
59	4	case 'figcaption':
60	4	case 'figure':
61	4	case 'footer':
62	4	case 'form':
63	4	case 'h1':
64	4	case 'h2':
65	4	case 'h3':
66	4	case 'h4':
67	4	case 'h5':
68	4	case 'h6':
69	4	case 'header':
70	4	case 'hgroup':
71	4	case 'hr':
72	4	case 'main':
73	4	case 'menu':
74	4	case 'nav':
75	4	case 'ol':
76	4	case 'p':
77	4	case 'pre':
78	4	case 'section':
79	4	case 'table':
80	4	case 'ul':
81	3	return true;
82		}
83	1	}
84
85		// LI
86	12	if ($parentName == 'li' && $name == 'li') {
87	1	return true;
88		}
89
90		// DT and DD
91	11	if (($parentName == 'dt' \|\| $parentName == 'dd') && ($name == 'dt' \|\| $name == 'dd')) {
92	4	return true;
93		}
94
95		// RP and RT
96	7	if (($parentName == 'rp' \|\| $parentName == 'rt') && ($name == 'rp' \|\| $name == 'rt')) {
97	4	return true;
98		}
99
100	3	return false;
101		}
102
103	20	private function isClosedOnlyElement($name)
104		{
105	20	return $name == 'base' \|\|
106	19	$name == 'link' \|\|
107	18	$name == 'meta' \|\|
108	17	$name == 'hr' \|\|
109	20	$name == 'br';
110		}
111
112	14	public static function isMatch($html)
113		{
114	14	return preg_match("/^<[a-zA-Z]/", $html) === 1;
115		}
116
117	25	public function parse($html)
118		{
119	25	$this->name = $this->parseElementName($html);
120
121		// Parse attributes.
122	25	$remainingHtml = substr($html, strlen($this->name) + 1);
123	25	while (strpos($remainingHtml, '>') !== false && preg_match("/^\s*[\/]?>/", $remainingHtml) === 0) {
124	12	$remainingHtml = $this->parseAttribute($remainingHtml);
125	12	}
126
127		// Find position of end of tag.
128	25	$posOfClosingBracket = strpos($remainingHtml, '>');
129	25	if ($posOfClosingBracket === false) {
130	2	if ($this->getThrowOnError()) {
131	1	throw new ParseException('Invalid element: missing closing bracket.');
132		}
133
134	1	return '';
135		}
136
137		// Is self-closing?
138	23	$posOfSelfClosingBracket = strpos($remainingHtml, '/>');
139	23	$remainingHtml = trim(substr($remainingHtml, $posOfClosingBracket + 1));
140	23	if ($posOfSelfClosingBracket !== false && $posOfSelfClosingBracket == $posOfClosingBracket - 1) {
141		// Self-closing element.
142	13	return $remainingHtml;
143		}
144
145		// Open element.
146	11	return $this->parseContents($remainingHtml);
147		}
148
149	12	private function parseAttribute($html)
150		{
151		// Will match the first entire name/value attribute pair.
152	12	$attrMatchSuccessful = preg_match(
153	12	"/(\s([^>\s]))/",
154	12	$html,
155		$attributeMatches
156	12	);
157	12	if ($attrMatchSuccessful !== 1) {
158		if ($this->getThrowOnError()) {
159		throw new ParseException('Invalid attribute.');
160		}
161
162		return '';
163		}
164
165	12	$posOfEqualsSign = strpos($attributeMatches[2], '=');
166	12	if ($posOfEqualsSign === false) {
167		// Valueless attribute.
168	2	$this->attributes[trim($attributeMatches[2])] = true;
169	2	} else {
170	11	list($name, $value) = explode('=', $attributeMatches[2]);
171	11	if ($value[0] === "'" \|\| $value[0] === '"') {
172	10	$valueMatchSuccessful = preg_match(
173	10	"/" . $value[0] . "(.*?(?<!\\\))" . $value[0] . "/s",
174	10	$value,
175		$valueMatches
176	10	);
177	10	if ($valueMatchSuccessful !== 1) {
178		if ($this->getThrowOnError()) {
179		throw new ParseException('Invalid value encapsulation.');
180		}
181
182		return '';
183		}
184
185	10	$value = $valueMatches[1];
186	10	}
187
188	11	$this->attributes[trim($name)] = trim($value);
189		}
190
191		// Return the html minus the current attribute.
192	12	$posOfAttribute = strpos($html, $attributeMatches[2]);
193
194	12	return substr($html, $posOfAttribute + strlen($attributeMatches[2]));
195		}
196
197	11	private function parseContents($html)
198		{
199	11	$remainingHtml = trim($html);
200	11	if ($remainingHtml == '') {
201	1	return '';
202		}
203
204		// Parse contents one token at a time.
205	10	while (preg_match("/^<\/\s" . $this->name . "\s>/is", $remainingHtml) === 0) {
206	8	$token = TokenFactory::buildFromHtml(
207	8	$remainingHtml,
208	8	$this,
209	8	$this->getThrowOnError()
210	8	);
211
212	8	if ($token === false \|\| $token->isClosingElementImplied($remainingHtml)) {
213	2	return $remainingHtml;
214		}
215
216	8	$remainingHtml = trim($token->parse($remainingHtml));
217	8	$this->children[] = $token;
218	8	}
219
220		// Remove remaining closing tag.
221	10	$posOfClosingBracket = strpos($remainingHtml, '>');
222
223	10	return substr($remainingHtml, $posOfClosingBracket + 1);
224		}
225
226		/**
227		* Will get the element name from the html string.
228		*
229		* @param $html string
230		*
231		* @return string The element name.
232		*/
233	44	private function parseElementName($html)
234		{
235	44	$elementMatchSuccessful = preg_match(
236	44	"/^(<(([a-z0-9\-]+:)?[a-z0-9\-]+))/i",
237	44	$html,
238		$elementMatches
239	44	);
240	44	if ($elementMatchSuccessful !== 1) {
241		if ($this->getThrowOnError()) {
242		throw new ParseException('Invalid element name.');
243		}
244
245		return '';
246		}
247
248	44	return strtolower($elementMatches[2]);
249		}
250
251	1	public function getAttributes()
252		{
253	1	return $this->attributes;
254		}
255
256	1	public function hasAttributes()
257		{
258	1	return !empty($this->attributes);
259		}
260
261	1	public function getChildren()
262		{
263	1	return $this->children;
264		}
265
266	1	public function hasChildren()
267		{
268	1	return !empty($this->children);
269		}
270
271		/**
272		* Getter for 'name'.
273		*
274		* @return string
275		*/
276	30	public function getName()
277		{
278	30	return $this->name;
279		}
280
281	13	public function toArray()
282		{
283		$result = array(
284	13	'type' => 'element',
285	13	'name' => $this->name
286	13	);
287
288	13	if (!empty($this->attributes)) {
289	7	$result['attributes'] = array();
290	7	foreach ($this->attributes as $name => $value) {
291	7	$result['attributes'][$name] = $value;
292	7	}
293	7	}
294
295	13	if (!empty($this->children)) {
296	4	$result['children'] = array();
297	4	foreach ($this->children as $child) {
298	4	$result['children'][] = $child->toArray();
299	4	}
300	4	}
301
302	13	return $result;
303		}
304		}
305

kevintweber / HtmlTokenizer

Push — master ( b928ab...26824d )

Element::parse() C

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like