RefWebTransformer::process() - Code Metrics - Inspection of "Style after inspection" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Branch — dev (1a31b5)

by Dispositif

created 2020-04-12 15:20 UTC

RefWebTransformer::process() B

↳ Parent: RefWebTransformer

Complexity

Conditions	6
Paths	7

Size

Total Lines	41
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	6
eloc	23
c	1
b	0
f	0
nc	7
nop	1
dl	0
loc	41
rs	8.9297

<?php
/**
 * This file is part of dispositif/wikibot application
 * 2019 © Philippe M. <[email protected]>
 * For the full copyright and MIT license information, please view the LICENSE file.
 */

declare(strict_types=1);

namespace App\Application;

use App\Domain\Models\Wiki\AbstractWikiTemplate;
use App\Domain\Models\Wiki\ArticleTemplate;
use App\Domain\Models\Wiki\LienWebTemplate;
use App\Domain\Models\Wiki\OuvrageTemplate;
use App\Domain\Publisher\WebMapper;
use App\Domain\Utils\WikiTextUtil;
use App\Domain\WikiTemplateFactory;
use App\Infrastructure\Logger;
use Psr\Log\LoggerInterface;
use Symfony\Component\Yaml\Yaml;
use Throwable;

/**
 * Class RefWebTransformer
 *
 * @package App\Application
 */
class RefWebTransformer implements TransformerInterface
{

    public $skipUnauthorised = true;
    /**
     * @var Logger
     */
    protected $log;
    private $config;
    /**
     * @var string|string[]
     */
    private $domain;
    /**
     * @var string
     */
    private $url;
    /**
     * @var WebMapper
     */
    private $mapper;
    /**
     * @var array
     */
    private $data = [];
    /**
     * @var array
     */
    public $summaryLog = [];

    /**
     * RefWebTransformer constructor.
     *
     * @param LoggerInterface $log
     */
    public function __construct(LoggerInterface $log)
    {
        $this->log = $log;
class Alien {}

class Dalek extends Alien {}

class Plot
{
    /** @var  Dalek */
    public $villain;
}

$alien = new Alien();
$plot = new Plot();
if ($alien instanceof Dalek) {
    $plot->villain = $alien;
}

        $this->config = Yaml::parseFile(__DIR__.'/resources/config_presse.yaml');
        $this->data['newspaper'] = json_decode(file_get_contents(__DIR__.'/resources/data_newspapers.json'), true);
        $this->data['scientific domain'] = json_decode(
            file_get_contents(__DIR__.'/resources/data_scientific_domain.json'),
            true
        );
        $this->data['scientific wiki'] = json_decode(
            file_get_contents(__DIR__.'/resources/data_scientific_wiki.json'),
            true
        );

        $this->mapper = new WebMapper(new Logger());
    }

    public function process(string $string)
    {
        if (!$this->isTransformAutorized($string)) {
            return $string;
        }

        $publish = new PublisherAction($this->url);
        try {
            sleep(3);
            $html = $publish->getHTMLSource();
            $htmlData = $publish->extractWebData($html);
            $this->log->debug('htmlData', $htmlData);
        } catch (Throwable $e) {
            // ne pas générer de lien brisé !!
            $this->log->notice('erreur sur extractWebData');

            return $string;
        }

        $mapData = $this->mapper->process($htmlData);

        // check dataValide
        if (empty($mapData) || empty($mapData['url']) || empty($mapData['titre'])) {
            $this->log->info('Mapping incomplet');

            return $string;
        }

        $this->tagAndLog($mapData);
        $this->addSummaryLog($mapData);

        $template = $this->chooseTemplateByData($mapData);

        $mapData = $this->replaceSitenameByConfig($mapData, $template);

        $template->hydrate($mapData, true);

        $serialized = $template->serialize(true);
        $this->log->info($serialized."\n");

        return $serialized;
    }

    protected function isTransformAutorized(string $string): bool
    {
        if (!preg_match('#^http?s://[^ ]+$#i', $string)) {
            return false;
        }
        $this->url = $string;
        $parseURL = parse_url($this->url);
        $this->domain = str_replace('www.', '', $parseURL['host']);

        if (!isset($this->config[$this->domain])) {
            $this->log->info("Domain ".$this->domain." non configuré\n");
            if ($this->skipUnauthorised) {
                return false;
            }
            echo "> Domaine ".Color::LIGHT_RED.$this->domain.Color::NORMAL." non configuré\n";
        }

        $this->config[$this->domain] = $this->config[$this->domain] ?? [];
        $this->config[$this->domain] = is_array($this->config[$this->domain]) ? $this->config[$this->domain] : [];

        if ($this->config[$this->domain] === 'desactived' || isset($this->config[$this->domain]['desactived'])) {
            $this->log->info("Domain ".$this->domain." desactivé\n");

            return false;
        }

        return true;
    }

    private function tagAndLog(array $mapData)
    {
        $this->log->debug('mapData', $mapData);

        if (isset($mapData['DATA-ARTICLE']) && $mapData['DATA-ARTICLE']) {
            $this->log->notice("Article OK");
        }
        if (isset($this->data['newspaper'][$this->domain])) {
            $this->log->notice('PRESSE');
        }
        if ($this->isScientificDomain()) {
            $this->log->notice('SCIENCE');
        }
    }

    private function isScientificDomain(): bool
    {
        if (isset($this->data['scientific domain'][$this->domain])) {
            return true;
        }
        if (strpos('.revues.org', $this->domain) > 0) {
            return true;
        }

        return false;
    }

    private function addSummaryLog(array $mapData)
    {
        $this->summaryLog[] = $mapData['site'] ?? $mapData['périodique'] ?? '?';
    }

    /**
     * todo refac lisible
     */
    private function chooseTemplateByData(array $mapData): AbstractWikiTemplate
    {
        // Logique : choix template
        $this->config[$this->domain]['template'] = $this->config[$this->domain]['template'] ?? [];
        $mapData['DATA-ARTICLE'] = $mapData['DATA-ARTICLE'] ?? false;

        if ($this->config[$this->domain]['template'] === 'article'
            || ($this->config[$this->domain]['template'] === 'auto' && $mapData['DATA-ARTICLE'])
            || ($mapData['DATA-ARTICLE'] && !empty($this->data['newspaper'][$this->domain]))
            || $this->isScientificDomain()
        ) {
            $templateName = 'article';
        }
        if (!isset($templateName) || $this->config[$this->domain]['template'] === 'lien web') {
            $templateName = 'lien web';
        }
        $template = WikiTemplateFactory::create($templateName);
        $template->userSeparator = " |";

        return $template;
    }

    /**
     * Logique : remplacement titre périodique ou nom du site
     *
     * @param array $mapData
     * @param       $template
     *
     * @return array
     */
    private function replaceSitenameByConfig(array $mapData, $template): array
    {
        // from wikidata URL of newspapers
        if (!empty($this->data['newspaper'][$this->domain])) {
            $frwiki = $this->data['newspaper'][$this->domain]['frwiki'];
            $label = $this->data['newspaper'][$this->domain]['fr'];
            if (isset($mapData['site']) || $template instanceof LienWebTemplate) {
                $mapData['site'] = WikiTextUtil::wikilink($label, $frwiki);
            }
            if (isset($mapData['périodique']) || $template instanceof ArticleTemplate) {
                $mapData['périodique'] = WikiTextUtil::wikilink($label, $frwiki);
            }
        }

        // from wikidata of scientific journals
        if (isset($mapData['périodique']) && isset($this->data['scientific wiki'][$mapData['périodique']])) {
            $mapData['périodique'] = WikiTextUtil::wikilink(
                $mapData['périodique'],
                $this->data['scientific wiki'][$mapData['périodique']]
            );
        }

        // from YAML config
        if (!empty($this->config[$this->domain]['site']) && $template instanceof LienWebTemplate) {
            $mapData['site'] = $this->config[$this->domain]['site'];
        }
        if (!empty($this->config[$this->domain]['périodique'])
            && (!empty($mapData['périodique'])
                || $template instanceof OuvrageTemplate)
        ) {
            $mapData['périodique'] = $this->config[$this->domain]['périodique'];
        }

        return $mapData;
    }

}


1			<?php
2			/**
3			* This file is part of dispositif/wikibot application
4			* 2019 © Philippe M. <[email protected]>
5			* For the full copyright and MIT license information, please view the LICENSE file.
6			*/
7
8			declare(strict_types=1);
9
10			namespace App\Application;
11
12			use App\Domain\Models\Wiki\AbstractWikiTemplate;
13			use App\Domain\Models\Wiki\ArticleTemplate;
14			use App\Domain\Models\Wiki\LienWebTemplate;
15			use App\Domain\Models\Wiki\OuvrageTemplate;
16			use App\Domain\Publisher\WebMapper;
17			use App\Domain\Utils\WikiTextUtil;
18			use App\Domain\WikiTemplateFactory;
19			use App\Infrastructure\Logger;
20			use Psr\Log\LoggerInterface;
21			use Symfony\Component\Yaml\Yaml;
22			use Throwable;
23
24			/**
25			* Class RefWebTransformer
26			*
27			* @package App\Application
28			*/
29			class RefWebTransformer implements TransformerInterface
30			{
31
32			public $skipUnauthorised = true;
33			/**
34			* @var Logger
35			*/
36			protected $log;
37			private $config;
38			/**
39			* @var string\|string[]
40			*/
41			private $domain;
42			/**
43			* @var string
44			*/
45			private $url;
46			/**
47			* @var WebMapper
48			*/
49			private $mapper;
50			/**
51			* @var array
52			*/
53			private $data = [];
54			/**
55			* @var array
56			*/
57			public $summaryLog = [];
58
59			/**
60			* RefWebTransformer constructor.
61			*
62			* @param LoggerInterface $log
63			*/
64			public function __construct(LoggerInterface $log)
65			{
66			$this->log = $log;
			1 ignored issue – show Documentation Bug introduced 2020-04-12 15:04 UTC by Report Bug Copy Issue Report `$log` is of type `Psr\Log\LoggerInterface`, but the property `$log` was declared to be of type `App\Infrastructure\Logger`. Are you sure that you always receive this specific sub-class here, or does it make sense to add an `instanceof` check? Our type inference engine has found a suspicous assignment of a value to a property. This check raises an issue when a value that can be of a given class or a super-class is assigned to a property that is type hinted more strictly. Either this assignment is in error or an instanceof check should be added for that assignment. class Alien {} class Dalek extends Alien {} class Plot { /** @var Dalek */ public $villain; } $alien = new Alien(); $plot = new Plot(); if ($alien instanceof Dalek) { $plot->villain = $alien; } Loading history...
67
68			$this->config = Yaml::parseFile(__DIR__.'/resources/config_presse.yaml');
69			$this->data['newspaper'] = json_decode(file_get_contents(__DIR__.'/resources/data_newspapers.json'), true);
70			$this->data['scientific domain'] = json_decode(
71			file_get_contents(__DIR__.'/resources/data_scientific_domain.json'),
72			true
73			);
74			$this->data['scientific wiki'] = json_decode(
75			file_get_contents(__DIR__.'/resources/data_scientific_wiki.json'),
76			true
77			);
78
79			$this->mapper = new WebMapper(new Logger());
80			}
81
82			public function process(string $string)
83			{
84			if (!$this->isTransformAutorized($string)) {
85			return $string;
86			}
87
88			$publish = new PublisherAction($this->url);
89			try {
90			sleep(3);
91			$html = $publish->getHTMLSource();
92			$htmlData = $publish->extractWebData($html);
93			$this->log->debug('htmlData', $htmlData);
94			} catch (Throwable $e) {
95			// ne pas générer de lien brisé !!
96			$this->log->notice('erreur sur extractWebData');
97
98			return $string;
99			}
100
101			$mapData = $this->mapper->process($htmlData);
102
103			// check dataValide
104			if (empty($mapData) \|\| empty($mapData['url']) \|\| empty($mapData['titre'])) {
105			$this->log->info('Mapping incomplet');
106
107			return $string;
108			}
109
110			$this->tagAndLog($mapData);
111			$this->addSummaryLog($mapData);
112
113			$template = $this->chooseTemplateByData($mapData);
114
115			$mapData = $this->replaceSitenameByConfig($mapData, $template);
116
117			$template->hydrate($mapData, true);
118
119			$serialized = $template->serialize(true);
120			$this->log->info($serialized."\n");
121
122			return $serialized;
123			}
124
125			protected function isTransformAutorized(string $string): bool
126			{
127			if (!preg_match('#^http?s://[^ ]+$#i', $string)) {
128			return false;
129			}
130			$this->url = $string;
131			$parseURL = parse_url($this->url);
132			$this->domain = str_replace('www.', '', $parseURL['host']);
133
134			if (!isset($this->config[$this->domain])) {
135			$this->log->info("Domain ".$this->domain." non configuré\n");
136			if ($this->skipUnauthorised) {
137			return false;
138			}
139			echo "> Domaine ".Color::LIGHT_RED.$this->domain.Color::NORMAL." non configuré\n";
140			}
141
142			$this->config[$this->domain] = $this->config[$this->domain] ?? [];
143			$this->config[$this->domain] = is_array($this->config[$this->domain]) ? $this->config[$this->domain] : [];
144
145			if ($this->config[$this->domain] === 'desactived' \|\| isset($this->config[$this->domain]['desactived'])) {
146			$this->log->info("Domain ".$this->domain." desactivé\n");
147
148			return false;
149			}
150
151			return true;
152			}
153
154			private function tagAndLog(array $mapData)
155			{
156			$this->log->debug('mapData', $mapData);
157
158			if (isset($mapData['DATA-ARTICLE']) && $mapData['DATA-ARTICLE']) {
159			$this->log->notice("Article OK");
160			}
161			if (isset($this->data['newspaper'][$this->domain])) {
162			$this->log->notice('PRESSE');
163			}
164			if ($this->isScientificDomain()) {
165			$this->log->notice('SCIENCE');
166			}
167			}
168
169			private function isScientificDomain(): bool
170			{
171			if (isset($this->data['scientific domain'][$this->domain])) {
172			return true;
173			}
174			if (strpos('.revues.org', $this->domain) > 0) {
175			return true;
176			}
177
178			return false;
179			}
180
181			private function addSummaryLog(array $mapData)
182			{
183			$this->summaryLog[] = $mapData['site'] ?? $mapData['périodique'] ?? '?';
184			}
185
186			/**
187			* todo refac lisible
188			*/
189			private function chooseTemplateByData(array $mapData): AbstractWikiTemplate
190			{
191			// Logique : choix template
192			$this->config[$this->domain]['template'] = $this->config[$this->domain]['template'] ?? [];
193			$mapData['DATA-ARTICLE'] = $mapData['DATA-ARTICLE'] ?? false;
194
195			if ($this->config[$this->domain]['template'] === 'article'
196			\|\| ($this->config[$this->domain]['template'] === 'auto' && $mapData['DATA-ARTICLE'])
197			\|\| ($mapData['DATA-ARTICLE'] && !empty($this->data['newspaper'][$this->domain]))
198			\|\| $this->isScientificDomain()
199			) {
200			$templateName = 'article';
201			}
202			if (!isset($templateName) \|\| $this->config[$this->domain]['template'] === 'lien web') {
203			$templateName = 'lien web';
204			}
205			$template = WikiTemplateFactory::create($templateName);
206			$template->userSeparator = " \|";
207
208			return $template;
209			}
210
211			/**
212			* Logique : remplacement titre périodique ou nom du site
213			*
214			* @param array $mapData
215			* @param $template
216			*
217			* @return array
218			*/
219			private function replaceSitenameByConfig(array $mapData, $template): array
220			{
221			// from wikidata URL of newspapers
222			if (!empty($this->data['newspaper'][$this->domain])) {
223			$frwiki = $this->data['newspaper'][$this->domain]['frwiki'];
224			$label = $this->data['newspaper'][$this->domain]['fr'];
225			if (isset($mapData['site']) \|\| $template instanceof LienWebTemplate) {
226			$mapData['site'] = WikiTextUtil::wikilink($label, $frwiki);
227			}
228			if (isset($mapData['périodique']) \|\| $template instanceof ArticleTemplate) {
229			$mapData['périodique'] = WikiTextUtil::wikilink($label, $frwiki);
230			}
231			}
232
233			// from wikidata of scientific journals
234			if (isset($mapData['périodique']) && isset($this->data['scientific wiki'][$mapData['périodique']])) {
235			$mapData['périodique'] = WikiTextUtil::wikilink(
236			$mapData['périodique'],
237			$this->data['scientific wiki'][$mapData['périodique']]
238			);
239			}
240
241			// from YAML config
242			if (!empty($this->config[$this->domain]['site']) && $template instanceof LienWebTemplate) {
243			$mapData['site'] = $this->config[$this->domain]['site'];
244			}
245			if (!empty($this->config[$this->domain]['périodique'])
246			&& (!empty($mapData['périodique'])
247			\|\| $template instanceof OuvrageTemplate)
248			) {
249			$mapData['périodique'] = $this->config[$this->domain]['périodique'];
250			}
251
252			return $mapData;
253			}
254
255			}
256

Dispositif / Wikibot

Branch — dev (1a31b5)

RefWebTransformer::process() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like