DomParser::list() - Code Metrics - Inspection of "Dev" - somework/minjust - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#4)

by Igor

created 2020-01-20 14:50 UTC

DomParser::list() A

↳ Parent: DomParser

Complexity

Conditions	1
Paths	1

Size

Total Lines	8
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	6
nc	1
nop	1
dl	0
loc	8
ccs	6
cts	6
cp	1
crap	1
rs	10
c	1
b	0
f	0

<?php

declare(strict_types=1);

namespace SomeWork\Minjust\Parser;

use PHPHtmlParser\Dom;
use PHPHtmlParser\Dom\Collection;
use PHPHtmlParser\Dom\HtmlNode;
use PHPHtmlParser\Exceptions\ChildNotFoundException;
use PHPHtmlParser\Exceptions\NotLoadedException;
use SomeWork\Minjust\Entity\DetailLawyer;
use SomeWork\Minjust\Entity\LawFormation;
use SomeWork\Minjust\Entity\Lawyer;
use SomeWork\Minjust\Exception\BlockNotFoundException;
use SomeWork\Minjust\Exception\RuntimeException;
use SomeWork\Minjust\FindResponse;

/**
 * @see \SomeWork\Minjust\Tests\Unit\DomParserTest
 */
class DomParser implements ParserInterface
{
    /**
     * @var string
     */
    protected const PAGINATION_BLOCK_SELECTOR = 'ul.pagination';

    /**
     * @var string
     */
    protected const CURRENT_PAGE_SELECTOR = 'li.active';

    /**
     * @var string
     */
    protected const PAGINATION_STEP_SELECTOR = 'li';

    /**
     * @var string
     */
    protected const LAWYERS_LIST_BLOCK_SELECTOR = 'table.persons > tbody > tr';

    /**
     * @var string
     */
    protected const LAWYER_DETAIL_SELECTOR = '.floating > p.row';

    /**
     * @var string
     */
    protected const LAWYER_DETAIL_NAME_FIELD = 'label';

    public function list(string $body): FindResponse
    {
        $dom = (new Dom())->loadStr($body);

        return (new FindResponse())
            ->setPage($this->getCurrentPage($dom))
            ->setTotalPage($this->getTotalPage($dom))
            ->setLawyers($this->getListLawyers($dom));
    }

    protected function getCurrentPage(Dom $dom): int
    {
        try {
            /**
             * @var HtmlNode|null $block
             */
            $block = $this
                ->getPagination($dom)
                ->find(static::CURRENT_PAGE_SELECTOR, 0);
        } catch (ChildNotFoundException $exception) {
            throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
        }

        if (null === $block) {
            throw new BlockNotFoundException(static::CURRENT_PAGE_SELECTOR);
        }

        try {
            return (int) $block->firstChild()->text();
        } catch (ChildNotFoundException $exception) {
            throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
        }
    }

    protected function getTotalPage(Dom $dom): int
    {
        try {
            /* @noinspection NullPointerExceptionInspection */
            $collection = $this
                ->getPagination($dom)
                ->find(static::PAGINATION_STEP_SELECTOR)
                ->toArray();
        } catch (ChildNotFoundException $exception) {
            throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
        }

        if (0 === count($collection)) {
            return 1;
        }

        /**
         * @var HtmlNode $last
         */
        $last = end($collection);

        try {
            /**
             * @var HtmlNode|null $link
             */
            $link = $last->find('a', 0);
        } catch (ChildNotFoundException $exception) {
            throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
        }

        if ($link) {
            $href = (string) $link->getAttribute('href');
class A
{
    function getObject()
    {
        return null;
    }

}

$a = new A();
if ($a->getObject()) {
            $matches = [];
            preg_match('/page=([\d]+)/', $href, $matches);

            return (int) $matches[1];
        }

        return $this->getCurrentPage($dom);
    }

    protected function getPagination(Dom $dom): HtmlNode
    {
        static $parsedDom = null;
        static $pagination = null;
        if ($dom !== $parsedDom) {
            $parsedDom = $dom;

            try {
                $pagination = $dom->find(static::PAGINATION_BLOCK_SELECTOR, 0);
            } catch (\Exception $exception) {
                throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
            }

            if (null === $pagination) {
                throw new BlockNotFoundException(static::PAGINATION_BLOCK_SELECTOR);
            }
        }

        return $pagination;
    }

    /**
     * @param \PHPHtmlParser\Dom $dom
     *
     * @return Lawyer[]
     * @throws ChildNotFoundException
     * @throws NotLoadedException
     */
    protected function getListLawyers(Dom $dom): array
    {
        $data = [];
        /**
         * @var Dom\HtmlNode[]|Collection $nodes
         */
        $nodes = $dom->find(static::LAWYERS_LIST_BLOCK_SELECTOR);
        foreach ($nodes as $node) {
            /**
             * @var Dom\HtmlNode[]|Collection $tds
             */
            $tds = $node->find('td');
            $data[] = (new Lawyer())
                ->setRegisterNumber($tds[0]->text())
                ->setFullName($tds[1]->text(true))
                ->setUrl($tds[1]->firstChild()->getAttribute('href'))
                ->setTerritorialSubject($tds[2]->text())
                ->setCertificateNumber($tds[3]->text())
                ->setStatus($tds[4]->text());
        }

        return $data;
    }

    public function detail(string $body): DetailLawyer
    {
        $dom = (new Dom())->loadStr($body);

        /**
         * @var Dom\HtmlNode[] $nodes
         */
        $nodes = $dom->find(static::LAWYER_DETAIL_SELECTOR)->toArray();

        $nodes = array_filter($nodes, static function (HtmlNode $htmlNode) {
            return strpos($htmlNode->getAttribute('class'), static::LAWYER_DETAIL_NAME_FIELD) === false;
class A
{
    function getObject()
    {
        return null;
    }

}

$a = new A();
if ($a->getObject()) {
        });

        $nodes = array_values($nodes);

        $lawyer = (new DetailLawyer())
            ->setChamberOfLaw(trim($nodes[2]->text()));

        if (($organizationForm = trim($nodes[3]->text())) !== '') {
            $lawyer->setLawFormation(
                (new LawFormation())
                    ->setOrganizationalForm($organizationForm)
                    ->setName(trim($nodes[4]->text()))
                    ->setAddress(trim($nodes[5]->text()))
                    ->setPhone(trim($nodes[6]->text()))
                    ->setEmail(trim($nodes[7]->text()))
            );
        }

        return $lawyer;
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		namespace SomeWork\Minjust\Parser;
6
7		use PHPHtmlParser\Dom;
8		use PHPHtmlParser\Dom\Collection;
9		use PHPHtmlParser\Dom\HtmlNode;
10		use PHPHtmlParser\Exceptions\ChildNotFoundException;
11		use PHPHtmlParser\Exceptions\NotLoadedException;
12		use SomeWork\Minjust\Entity\DetailLawyer;
13		use SomeWork\Minjust\Entity\LawFormation;
14		use SomeWork\Minjust\Entity\Lawyer;
15		use SomeWork\Minjust\Exception\BlockNotFoundException;
16		use SomeWork\Minjust\Exception\RuntimeException;
17		use SomeWork\Minjust\FindResponse;
18
19		/**
20		* @see \SomeWork\Minjust\Tests\Unit\DomParserTest
21		*/
22		class DomParser implements ParserInterface
23		{
24		/**
25		* @var string
26		*/
27		protected const PAGINATION_BLOCK_SELECTOR = 'ul.pagination';
28
29		/**
30		* @var string
31		*/
32		protected const CURRENT_PAGE_SELECTOR = 'li.active';
33
34		/**
35		* @var string
36		*/
37		protected const PAGINATION_STEP_SELECTOR = 'li';
38
39		/**
40		* @var string
41		*/
42		protected const LAWYERS_LIST_BLOCK_SELECTOR = 'table.persons > tbody > tr';
43
44		/**
45		* @var string
46		*/
47		protected const LAWYER_DETAIL_SELECTOR = '.floating > p.row';
48
49		/**
50		* @var string
51		*/
52		protected const LAWYER_DETAIL_NAME_FIELD = 'label';
53
54	5	public function list(string $body): FindResponse
55		{
56	5	$dom = (new Dom())->loadStr($body);
57
58	5	return (new FindResponse())
59	5	->setPage($this->getCurrentPage($dom))
60	5	->setTotalPage($this->getTotalPage($dom))
61	5	->setLawyers($this->getListLawyers($dom));
62		}
63
64	5	protected function getCurrentPage(Dom $dom): int
65		{
66		try {
67		/**
68		* @var HtmlNode\|null $block
69		*/
70		$block = $this
71	5	->getPagination($dom)
72	5	->find(static::CURRENT_PAGE_SELECTOR, 0);
73		} catch (ChildNotFoundException $exception) {
74		throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
75		}
76
77	5	if (null === $block) {
78		throw new BlockNotFoundException(static::CURRENT_PAGE_SELECTOR);
79		}
80
81		try {
82	5	return (int) $block->firstChild()->text();
83		} catch (ChildNotFoundException $exception) {
84		throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
85		}
86		}
87
88	4	protected function getTotalPage(Dom $dom): int
89		{
90		try {
91		/* @noinspection NullPointerExceptionInspection */
92		$collection = $this
93	4	->getPagination($dom)
94	4	->find(static::PAGINATION_STEP_SELECTOR)
95	4	->toArray();
96		} catch (ChildNotFoundException $exception) {
97		throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
98		}
99
100	4	if (0 === count($collection)) {
101		return 1;
102		}
103
104		/**
105		* @var HtmlNode $last
106		*/
107	4	$last = end($collection);
108
109		try {
110		/**
111		* @var HtmlNode\|null $link
112		*/
113	4	$link = $last->find('a', 0);
114		} catch (ChildNotFoundException $exception) {
115		throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
116		}
117
118	4	if ($link) {
119	2	$href = (string) $link->getAttribute('href');
		0 ignored issues – show Bug introduced 2020-01-20 14:54 UTC by Report Bug Copy Issue Report Are you sure the usage of `$link->getAttribute('href')` targeting `PHPHtmlParser\Dom\AbstractNode::getAttribute()` seems to always return null. This check looks for function or method calls that always return null and whose return value is used. class A { function getObject() { return null; } } $a = new A(); if ($a->getObject()) { The method `getObject()` can return nothing but null, so it makes no sense to use the return value. The reason is most likely that a function or method is imcomplete or has been reduced for debug purposes. Loading history...
120	2	$matches = [];
121	2	preg_match('/page=([\d]+)/', $href, $matches);
122
123	2	return (int) $matches[1];
124		}
125
126	2	return $this->getCurrentPage($dom);
127		}
128
129		protected function getPagination(Dom $dom): HtmlNode
130		{
131		static $parsedDom = null;
132		static $pagination = null;
133		if ($dom !== $parsedDom) {
134		$parsedDom = $dom;
135
136		try {
137		$pagination = $dom->find(static::PAGINATION_BLOCK_SELECTOR, 0);
138		} catch (\Exception $exception) {
139		throw new RuntimeException($exception->getMessage(), $exception->getCode(), $exception);
140		}
141
142		if (null === $pagination) {
143		throw new BlockNotFoundException(static::PAGINATION_BLOCK_SELECTOR);
144		}
145		}
146
147		return $pagination;
148		}
149
150		/**
151		* @param \PHPHtmlParser\Dom $dom
152		*
153		* @return Lawyer[]
154		* @throws ChildNotFoundException
155		* @throws NotLoadedException
156		*/
157	5	protected function getListLawyers(Dom $dom): array
158		{
159	5	$data = [];
160		/**
161		* @var Dom\HtmlNode[]\|Collection $nodes
162		*/
163	5	$nodes = $dom->find(static::LAWYERS_LIST_BLOCK_SELECTOR);
164	5	foreach ($nodes as $node) {
165		/**
166		* @var Dom\HtmlNode[]\|Collection $tds
167		*/
168	5	$tds = $node->find('td');
169	5	$data[] = (new Lawyer())
170	5	->setRegisterNumber($tds[0]->text())
171	5	->setFullName($tds[1]->text(true))
172	5	->setUrl($tds[1]->firstChild()->getAttribute('href'))
173	5	->setTerritorialSubject($tds[2]->text())
174	5	->setCertificateNumber($tds[3]->text())
175	5	->setStatus($tds[4]->text());
176		}
177
178	5	return $data;
179		}
180
181	2	public function detail(string $body): DetailLawyer
182		{
183	2	$dom = (new Dom())->loadStr($body);
184
185		/**
186		* @var Dom\HtmlNode[] $nodes
187		*/
188	2	$nodes = $dom->find(static::LAWYER_DETAIL_SELECTOR)->toArray();
189
190		$nodes = array_filter($nodes, static function (HtmlNode $htmlNode) {
191	2	return strpos($htmlNode->getAttribute('class'), static::LAWYER_DETAIL_NAME_FIELD) === false;
		0 ignored issues – show Bug introduced 2020-01-20 14:54 UTC by Report Bug Copy Issue Report Are you sure the usage of `$htmlNode->getAttribute('class')` targeting `PHPHtmlParser\Dom\AbstractNode::getAttribute()` seems to always return null. This check looks for function or method calls that always return null and whose return value is used. class A { function getObject() { return null; } } $a = new A(); if ($a->getObject()) { The method `getObject()` can return nothing but null, so it makes no sense to use the return value. The reason is most likely that a function or method is imcomplete or has been reduced for debug purposes. Loading history...
192	2	});
193
194	2	$nodes = array_values($nodes);
195
196	2	$lawyer = (new DetailLawyer())
197	2	->setChamberOfLaw(trim($nodes[2]->text()));
198
199	2	if (($organizationForm = trim($nodes[3]->text())) !== '') {
200	1	$lawyer->setLawFormation(
201	1	(new LawFormation())
202	1	->setOrganizationalForm($organizationForm)
203	1	->setName(trim($nodes[4]->text()))
204	1	->setAddress(trim($nodes[5]->text()))
205	1	->setPhone(trim($nodes[6]->text()))
206	1	->setEmail(trim($nodes[7]->text()))
207		);
208		}
209
210	2	return $lawyer;
211		}
212		}
213

somework / minjust

Pull Request — master (#4)

DomParser::list() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like