HtmlSplitter::cleanContent() - Code Metrics - Inspection of "Apply fixes from StyleCI" - algolia/scout-extended - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — feat/html-splitter (#180)

by Nuno

created 2019-04-12 08:22 UTC

HtmlSplitter::cleanContent() A

↳ Parent: HtmlSplitter

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

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

<?php

declare(strict_types=1);

/**
 * This file is part of Scout Extended.
 *
 * (c) Algolia Team <[email protected]>
 *
 *  For the full copyright and license information, please view the LICENSE
 *  file that was distributed with this source code.
 */

namespace Algolia\ScoutExtended\Splitters;

use DOMXPath;
use DOMDocument;
use Algolia\ScoutExtended\Contracts\SplitterContract;

/**
 * Class HtmlSplitter.
 */
class HtmlSplitter implements SplitterContract
{
    /**
     * The list of html tags.
     *
     * @var string[]
     */
    protected $nodes = [
        'h1',
        'h2',
        'h3',
        'h4',
        'h5',
        'h6',
        'p',
    ];

    /**
     * String for key check purpose.
     *
     * @const string IMPORTANCE
     */
    private const IMPORTANCE = 'importance';

    /**
     * String for exception purpose.
     *
     * @const string PARAGRAPH
     */
    private const PARAGRAPH = 'p';

    /**
     * Creates a new instance of the class.
     *
     * @param array $nodes
     *
     * @return void
     */
    public function __construct(array $nodes = null)
    {
        if ($nodes !== null) {
            $this->nodes = $nodes;
        }
    }

    /**
     * Find weight of current nodes.
     *
     * @param array<string, string> $object
     *
     * @return int
     */
    public function findWeight(array $object): int
    {
        return (int) array_search((key($object)), $this->nodes, true);
    }

    /**
     * Add object to queue.
     *
     * @param array<array<string, string>> $object
     * @param array $queue
     *
     * @return array
     */
    public function addObjectToQueue(array $object, array $queue): array
    {
        if (count($queue) === 0) {
            $queue[] = $object;

            return $queue;
        }

        if ($this->findWeight($object) > $this->findWeight(end($queue))) {
            $queue[] = $object;

            return $queue;
        }

        array_pop($queue);

        return $this->addObjectToQueue($object, $queue);
    }

    /**
     * Importance formula.
     * Give integer from tags ranking.
     *
     * @param \DOMElement $node
     * @param array<array<array<string, string>, <array<string, int>>> $queue

     *
     * @return int
     */
    public function importanceWeight(\DOMElement $node, array $queue): int
    {
        if ($node->nodeName === self::PARAGRAPH) {
            if (empty(end($queue))) {
                return 0;
            }
            if (key(end($queue)) === self::PARAGRAPH) {
                $key = key(prev($queue));
            } else {
                $key = key(end($queue));
            }

            return (int) (count($this->nodes) - 1) + (int) array_search($key, $this->nodes, true);
        }

        return (int) array_search($node->nodeName, $this->nodes, true);
    }

    /**
     * Clean Records to have a correct format.
     *
     *
     * @param array<array<array<string, string>, <array<string, int>>> $objects

     *
     * @return array
     */
    public function cleanRecords(array $objects): array
    {
        $records = [];
        foreach ($objects as $object) {
            foreach ($object as $data) {
                foreach ($data as $key => $value) {
                    $record[$key] = $value;
                    if ($key === self::IMPORTANCE) {
                        $records[] = $record;
                        $record = [];
                    }
                }
            }
        }

        return $records;
    }

    /**
     * Clean Content from Html tag.
     * Remove space at the begin and end, useless space, return.
     *
     * @param string $content
     *
     * @return string
     */
    public function cleanContent(string $content): string
    {
        return trim(preg_replace('/\s+/', ' ', str_replace('\n', '', $content)));
    }

    /**
     * Acts a static factory.
     *
     * @param string|array<string> $tags
     *
     * @return static
     */
    public static function by($tags)
    {
        return new static((array) $tags);
    }

    /**
     * Splits the given value.
     *
     * @param object $searchable
     * @param string $value
     *
     * @return array
     */
    public function split($searchable, $value): array
    {
        $dom = new DOMDocument();
        try {
            $dom->loadHTML($value);
        } catch (\ErrorException $exception) {

        }

        $xpath = new DOMXpath($dom);
        $queue = [];
        $objects = [];
        $xpathQuery = '//'.implode(' | //', $this->nodes);
        $nodes = $xpath->query($xpathQuery);

        foreach ($nodes as $node) {
            $content = $this->cleanContent($node->textContent);
            $object = [$node->nodeName => $content];
            $importance = $this->importanceWeight($node, $queue);
            $queue = $this->addObjectToQueue($object, $queue);
            $cloneQueue = $queue;
            $cloneQueue[] = [self::IMPORTANCE => $importance];
            $objects[] = $cloneQueue;
        }

        return $this->cleanRecords($objects);
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		/**
6		* This file is part of Scout Extended.
7		*
8		* (c) Algolia Team <[email protected]>
9		*
10		* For the full copyright and license information, please view the LICENSE
11		* file that was distributed with this source code.
12		*/
13
14		namespace Algolia\ScoutExtended\Splitters;
15
16		use DOMXPath;
17		use DOMDocument;
18		use Algolia\ScoutExtended\Contracts\SplitterContract;
19
20		/**
21		* Class HtmlSplitter.
22		*/
23		class HtmlSplitter implements SplitterContract
24		{
25		/**
26		* The list of html tags.
27		*
28		* @var string[]
29		*/
30		protected $nodes = [
31		'h1',
32		'h2',
33		'h3',
34		'h4',
35		'h5',
36		'h6',
37		'p',
38		];
39
40		/**
41		* String for key check purpose.
42		*
43		* @const string IMPORTANCE
44		*/
45		private const IMPORTANCE = 'importance';
46
47		/**
48		* String for exception purpose.
49		*
50		* @const string PARAGRAPH
51		*/
52		private const PARAGRAPH = 'p';
53
54		/**
55		* Creates a new instance of the class.
56		*
57		* @param array $nodes
58		*
59		* @return void
60		*/
61	5	public function __construct(array $nodes = null)
62		{
63	5	if ($nodes !== null) {
64	2	$this->nodes = $nodes;
65		}
66	5	}
67
68		/**
69		* Find weight of current nodes.
70		*
71		* @param array<string, string> $object
72		*
73		* @return int
74		*/
75	5	public function findWeight(array $object): int
76		{
77	5	return (int) array_search((key($object)), $this->nodes, true);
78		}
79
80		/**
81		* Add object to queue.
82		*
83		* @param array<array<string, string>> $object
84		* @param array $queue
85		*
86		* @return array
87		*/
88	5	public function addObjectToQueue(array $object, array $queue): array
89		{
90	5	if (count($queue) === 0) {
91	5	$queue[] = $object;
92
93	5	return $queue;
94		}
95
96	5	if ($this->findWeight($object) > $this->findWeight(end($queue))) {
97	3	$queue[] = $object;
98
99	3	return $queue;
100		}
101
102	5	array_pop($queue);
103
104	5	return $this->addObjectToQueue($object, $queue);
105		}
106
107		/**
108		* Importance formula.
109		* Give integer from tags ranking.
110		*
111		* @param \DOMElement $node
112		* @param array<array<array<string, string>, <array<string, int>>> $queue
		0 ignored issues – show Documentation Bug introduced 2019-04-12 08:30 UTC by Report Bug Copy Issue Report The doc comment `array<array<array<string..., <array<string, int>>>` at position `11` could not be parsed: Unknown type name '<' at position 11 in array<array<array<string, string>, <array<string, int>>>. Loading history...
113		*
114		* @return int
115		*/
116	5	public function importanceWeight(\DOMElement $node, array $queue): int
117		{
118	5	if ($node->nodeName === self::PARAGRAPH) {
119	3	if (empty(end($queue))) {
120		return 0;
121		}
122	3	if (key(end($queue)) === self::PARAGRAPH) {
123	2	$key = key(prev($queue));
124		} else {
125	3	$key = key(end($queue));
126		}
127
128	3	return (int) (count($this->nodes) - 1) + (int) array_search($key, $this->nodes, true);
129		}
130
131	5	return (int) array_search($node->nodeName, $this->nodes, true);
132		}
133
134		/**
135		* Clean Records to have a correct format.
136		*
137		*
138		* @param array<array<array<string, string>, <array<string, int>>> $objects
		0 ignored issues – show Documentation Bug introduced 2019-04-12 08:30 UTC by Report Bug Copy Issue Report The doc comment `array<array<array<string..., <array<string, int>>>` at position `11` could not be parsed: Unknown type name '<' at position 11 in array<array<array<string, string>, <array<string, int>>>. Loading history...
139		*
140		* @return array
141		*/
142	5	public function cleanRecords(array $objects): array
143		{
144	5	$records = [];
145	5	foreach ($objects as $object) {
146	5	foreach ($object as $data) {
147	5	foreach ($data as $key => $value) {
148	5	$record[$key] = $value;
149	5	if ($key === self::IMPORTANCE) {
150	5	$records[] = $record;
151	5	$record = [];
152		}
153		}
154		}
155		}
156
157	5	return $records;
158		}
159
160		/**
161		* Clean Content from Html tag.
162		* Remove space at the begin and end, useless space, return.
163		*
164		* @param string $content
165		*
166		* @return string
167		*/
168	5	public function cleanContent(string $content): string
169		{
170	5	return trim(preg_replace('/\s+/', ' ', str_replace('\n', '', $content)));
171		}
172
173		/**
174		* Acts a static factory.
175		*
176		* @param string\|array<string> $tags
177		*
178		* @return static
179		*/
180	2	public static function by($tags)
181		{
182	2	return new static((array) $tags);
183		}
184
185		/**
186		* Splits the given value.
187		*
188		* @param object $searchable
189		* @param string $value
190		*
191		* @return array
192		*/
193	5	public function split($searchable, $value): array
194		{
195	5	$dom = new DOMDocument();
196		try {
197	5	$dom->loadHTML($value);
198	2	} catch (\ErrorException $exception) {
		0 ignored issues – show Coding Style Comprehensibility introduced 2019-04-12 08:30 UTC by Report Bug Copy Issue Report Consider adding a comment why this CATCH block is empty. Loading history...
199		}
200
201	5	$xpath = new DOMXpath($dom);
202	5	$queue = [];
203	5	$objects = [];
204	5	$xpathQuery = '//'.implode(' \| //', $this->nodes);
205	5	$nodes = $xpath->query($xpathQuery);
206
207	5	foreach ($nodes as $node) {
208	5	$content = $this->cleanContent($node->textContent);
209	5	$object = [$node->nodeName => $content];
210	5	$importance = $this->importanceWeight($node, $queue);
211	5	$queue = $this->addObjectToQueue($object, $queue);
212	5	$cloneQueue = $queue;
213	5	$cloneQueue[] = [self::IMPORTANCE => $importance];
214	5	$objects[] = $cloneQueue;
215		}
216
217	5	return $this->cleanRecords($objects);
218		}
219		}
220

algolia / scout-extended

Pull Request — feat/html-splitter (#180)

HtmlSplitter::cleanContent() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like