ExternRefTransformer::isURLAuthorized() - Code Metrics - Inspection of "Config composer" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 1bcf8b...aa70d7 )

by Dispositif

created 2020-08-12 10:54 UTC

ExternRefTransformer::isURLAuthorized() B

↳ Parent: ExternRefTransformer

Complexity

Conditions	9
Paths	12

Size

Total Lines	40
Code Lines	22

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	90

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	9
eloc	22
c	1
b	0
f	0
nc	12
nop	1
dl	0
loc	40
rs	8.0555
ccs	0
cts	32
cp	0
crap	90

<?php
/**
 * This file is part of dispositif/wikibot application (@github)
 * 2019/2020 © 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\Application\Http\ExternHttpClient;
use App\Domain\ExternDomains;
use App\Domain\ExternPage;
use App\Domain\ExternPageFactory;
use App\Domain\Models\Summary;
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\OptimizerFactory;
use App\Domain\Publisher\ExternMapper;
use App\Domain\Utils\WikiTextUtil;
use App\Domain\WikiTemplateFactory;
use App\Infrastructure\Logger;
use Exception;
use Normalizer;
use Psr\Log\LoggerInterface;
use Symfony\Component\Yaml\Yaml;
use Throwable;

/**
 * todo move Domain
 * Class ExternRefTransformer
 *
 * @package App\Application
 */
class ExternRefTransformer implements TransformerInterface
{
    const HTTP_REQUEST_LOOP_DELAY = 10;
    const LOG_REQUEST_ERROR       = __DIR__.'/resources/external_request_error.log';
    const SKIP_DOMAIN_FILENAME    = __DIR__.'/resources/config_skip_domain.txt';

    public $skipUnauthorised = true;
    /**
     * @var array
     */
    public $summaryLog = [];
    /**
     * @var LoggerInterface
     */
    protected $log;
    private $config;
    /**
     * @var string|string[]
     */
    private $domain;
    /**
     * @var string
     */
    private $url;
    /**
     * @var ExternMapper
     */
    private $mapper;
    /**
     * @var array
     */
    private $data = [];
    /**
     * @var array
     */
    private $skip_domain;
    /**
     * @var ExternPage
     */
    private $externalPage;
    /**
     * @var Summary
     */
    private $summary;

    public function __construct(LoggerInterface $log)
    {
        $this->log = $log;

        $this->importConfigAndData();

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

    /**
     * @param string       $url
     * @param Summary|null $summary
     *
     * @return string
     * @throws Exception
     */
    public function process(string $url, Summary $summary): string
    {
        $this->summary = $summary;
        if (!$this->isURLAuthorized($url)) {
            return $url;
        }
        try {
            sleep(self::HTTP_REQUEST_LOOP_DELAY);
            $this->externalPage = ExternPageFactory::fromURL($url, $this->log);
            $pageData = $this->externalPage->getData();
            $this->log->debug('metaData', $this->externalPage->getData());
        } catch (Exception $e) {
            // "410 gone" => {lien brisé}
            if (preg_match('#410 Gone#i', $e->getMessage())) {
                $this->log->notice('410 page disparue : '.$url);

                return sprintf(
                    '{{Lien brisé |url= %s |titre=%s |brisé le=%s}}',
                    $url,
                    $this->url2TextStyleTitle($url),
                    date('d-m-Y')
                );
            } // 403
            elseif (preg_match('#403 Forbidden#i', $e->getMessage())) {
                $this->log->warning('403 Forbidden : '.$url);
                file_put_contents(self::LOG_REQUEST_ERROR, '403 Forbidden : '.$this->domain."\n", FILE_APPEND);
            } elseif (preg_match('#404 Not Found#i', $e->getMessage())) {
                $this->log->notice('404 Not Found : '.$url);

                return $url;
            } else {
                //  autre : ne pas générer de {lien brisé}, car peut-être 404 temporaire
                $this->log->warning('erreur sur extractWebData '.$e->getMessage());

                //file_put_contents(self::LOG_REQUEST_ERROR, $this->domain."\n", FILE_APPEND);

                return $url;
            }
        }

        if (empty($pageData)
            || (empty($pageData['JSON-LD']) && empty($pageData['meta']))
        ) {
            $this->log->notice('SKIP no metadata : '.$url);

            return $url;
        }

        if (isset($pageData['robots']) && strpos($pageData['robots'], 'noindex') !== false) {
            $this->log->notice('SKIP robots: noindex : '.$url);

            return $url;
        }

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

        // check dataValide
        // Pas de skip domaine car s'agit peut-être d'un 404 ou erreur juste sur cette URL
        if (empty($mapData) || empty($mapData['url']) || empty($mapData['titre'])) {
            $this->log->info('Mapping incomplet : '.$url);

            return $url;
        }

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

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

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


        if ($template instanceof ArticleTemplate) {
            unset($mapData['site']);
        }
        unset($mapData['DATA-TYPE']); // ugly
        unset($mapData['DATA-ARTICLE']); // ugly
        unset($mapData['url-access']);

        $template->hydrate($mapData);

        $optimizer = OptimizerFactory::fromTemplate($template);
        $optimizer->doTasks();
        $templateOptimized = $optimizer->getOptiTemplate();

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

        return Normalizer::normalize($serialized);
    }

    /**
     * @param string $url
     *
     * @return bool
     * @throws Exception
     */
    protected function isURLAuthorized(string $url): bool
    {
        if (!ExternHttpClient::isWebURL($url)) {
            $this->log->debug('Skip : not an URL : '.$url);

            return false;
        }

        if ($this->hasForbiddenFilenameExtension($url)) {
            return false;
        }

        $this->url = $url;
        $this->domain = ExternDomains::extractSubDomain($this->url);

        if (in_array($this->domain, $this->skip_domain)) {
            $this->log->notice("Skip domain ".$this->domain);

            return false;
        }

        if (!isset($this->config[$this->domain])) {
            $this->log->debug("Domain ".$this->domain." non configuré");
            if ($this->skipUnauthorised) {
                return false;
            }
        } else {
            $this->log->debug("Domain ".$this->domain." configuré");
        }

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

            return false;
        }

        return true;
    }

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

        if (isset($mapData['DATA-ARTICLE']) && $mapData['DATA-ARTICLE']) {
            $this->log->notice("Article OK");
        }
        if (isset($this->data['newspaper'][$this->domain])) {
            $this->log->notice('PRESSE');
            $this->summary->memo['presse'] = true;
        }
        if ($this->isScientificDomain()) {
            $this->log->notice('SCIENCE');
            $this->summary->memo['science'] = true;
        }
        if (!isset($this->summary->memo['sites'])
            || !in_array($this->externalPage->getPrettyDomainName(), $this->summary->memo['sites'])
        ) {
            $this->summary->memo['sites'][] = $this->externalPage->getPrettyDomainName();
        }
    }

    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
     *
     * @param array $mapData
     *
     * @return AbstractWikiTemplate
     * @throws Exception
     */
    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 (!empty($mapData['doi'])) {
            $templateName = 'article';
        }

        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') {
            $templateName = 'lien web';
        }
        // date obligatoire pour {article}
        if (!isset($mapData['date'])) {
            $templateName = 'lien web';
        }

        $template = WikiTemplateFactory::create($templateName);
        $template->userSeparator = " |";
        $this->summary->memo['count '.$templateName] = 1 + ($this->summary->memo['count '.$templateName] ?? 0);

        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'];
        }

        // from logic
        if (empty($mapData['site']) && $template instanceof LienWebTemplate) {
            try {
                $mapData['site'] = $this->externalPage->getPrettyDomainName();
            } catch (Throwable $e) {
                unset($e);
            }
        }

        return $mapData;
    }

    private function replaceURLbyOriginal(array $mapData): array
    {
        $mapData['url'] = $this->url;

        return $mapData;
    }

    /**
     * todo move ?
     * URL => "parismatch.com/People/bla…"
     *
     * @param string $url
     *
     * @return string
     */
    public function url2TextStyleTitle(string $url): string
    {
        $text = str_replace(['https://', 'http://', 'www.'], '', $url);
        if (strlen($text) > 30) {
            $text = substr($text, 0, 30).'…';
        }

        return $text;
    }

    /**
     * Skip PDF GIF etc
     * https://fr.wikipedia.org/wiki/Liste_d%27extensions_de_fichiers
     *
     * @param string $url
     *
     * @return bool
     */
    private function hasForbiddenFilenameExtension(string $url): bool
    {
        if (preg_match(
            '#\.(pdf|jpg|jpeg|gif|png|xls|xlsx|xlr|xml|xlt|xlsx|txt|csv|js|docx|exe|gz|zip|ini|movie|mp3|mp4|ogg|raw|rss|tar|tgz|wma)$#i',
            $url
        )
        ) {
            return true;
        }

        return false;
    }

    protected function importConfigAndData(): void
    {
        // todo REFAC DataObject[]
        $this->config = Yaml::parseFile(__DIR__.'/resources/config_presse.yaml');
        $skipFromFile = file(
            self::SKIP_DOMAIN_FILENAME,
            FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES
        );
        $this->skip_domain = ($skipFromFile) ? $skipFromFile : [];

        $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
        );
    }

}


1			<?php
2			/**
3			* This file is part of dispositif/wikibot application (@github)
4			* 2019/2020 © 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\Application\Http\ExternHttpClient;
13			use App\Domain\ExternDomains;
14			use App\Domain\ExternPage;
15			use App\Domain\ExternPageFactory;
16			use App\Domain\Models\Summary;
17			use App\Domain\Models\Wiki\AbstractWikiTemplate;
18			use App\Domain\Models\Wiki\ArticleTemplate;
19			use App\Domain\Models\Wiki\LienWebTemplate;
20			use App\Domain\Models\Wiki\OuvrageTemplate;
21			use App\Domain\OptimizerFactory;
22			use App\Domain\Publisher\ExternMapper;
23			use App\Domain\Utils\WikiTextUtil;
24			use App\Domain\WikiTemplateFactory;
25			use App\Infrastructure\Logger;
26			use Exception;
27			use Normalizer;
28			use Psr\Log\LoggerInterface;
29			use Symfony\Component\Yaml\Yaml;
30			use Throwable;
31
32			/**
33			* todo move Domain
34			* Class ExternRefTransformer
35			*
36			* @package App\Application
37			*/
38			class ExternRefTransformer implements TransformerInterface
39			{
40			const HTTP_REQUEST_LOOP_DELAY = 10;
41			const LOG_REQUEST_ERROR = __DIR__.'/resources/external_request_error.log';
42			const SKIP_DOMAIN_FILENAME = __DIR__.'/resources/config_skip_domain.txt';
43
44			public $skipUnauthorised = true;
45			/**
46			* @var array
47			*/
48			public $summaryLog = [];
49			/**
50			* @var LoggerInterface
51			*/
52			protected $log;
53			private $config;
54			/**
55			* @var string\|string[]
56			*/
57			private $domain;
58			/**
59			* @var string
60			*/
61			private $url;
62			/**
63			* @var ExternMapper
64			*/
65			private $mapper;
66			/**
67			* @var array
68			*/
69			private $data = [];
70			/**
71			* @var array
72			*/
73			private $skip_domain;
74			/**
75			* @var ExternPage
76			*/
77			private $externalPage;
78			/**
79			* @var Summary
80			*/
81			private $summary;
82
83			public function __construct(LoggerInterface $log)
84			{
85			$this->log = $log;
86
87			$this->importConfigAndData();
88
89			$this->mapper = new ExternMapper(new Logger());
90			}
91
92			/**
93			* @param string $url
94			* @param Summary\|null $summary
95			*
96			* @return string
97			* @throws Exception
98			*/
99			public function process(string $url, Summary $summary): string
100			{
101			$this->summary = $summary;
102			if (!$this->isURLAuthorized($url)) {
103			return $url;
104			}
105			try {
106			sleep(self::HTTP_REQUEST_LOOP_DELAY);
107			$this->externalPage = ExternPageFactory::fromURL($url, $this->log);
108			$pageData = $this->externalPage->getData();
109			$this->log->debug('metaData', $this->externalPage->getData());
110			} catch (Exception $e) {
111			// "410 gone" => {lien brisé}
112			if (preg_match('#410 Gone#i', $e->getMessage())) {
113			$this->log->notice('410 page disparue : '.$url);
114
115			return sprintf(
116			'{{Lien brisé \|url= %s \|titre=%s \|brisé le=%s}}',
117			$url,
118			$this->url2TextStyleTitle($url),
119			date('d-m-Y')
120			);
121			} // 403
122			elseif (preg_match('#403 Forbidden#i', $e->getMessage())) {
123			$this->log->warning('403 Forbidden : '.$url);
124			file_put_contents(self::LOG_REQUEST_ERROR, '403 Forbidden : '.$this->domain."\n", FILE_APPEND);
125			} elseif (preg_match('#404 Not Found#i', $e->getMessage())) {
126			$this->log->notice('404 Not Found : '.$url);
127
128			return $url;
129			} else {
130			// autre : ne pas générer de {lien brisé}, car peut-être 404 temporaire
131			$this->log->warning('erreur sur extractWebData '.$e->getMessage());
132
133			//file_put_contents(self::LOG_REQUEST_ERROR, $this->domain."\n", FILE_APPEND);
134
135			return $url;
136			}
137			}
138
139			if (empty($pageData)
140			\|\| (empty($pageData['JSON-LD']) && empty($pageData['meta']))
141			) {
142			$this->log->notice('SKIP no metadata : '.$url);
143
144			return $url;
145			}
146
147			if (isset($pageData['robots']) && strpos($pageData['robots'], 'noindex') !== false) {
148			$this->log->notice('SKIP robots: noindex : '.$url);
149
150			return $url;
151			}
152
153			$mapData = $this->mapper->process($pageData);
154
155			// check dataValide
156			// Pas de skip domaine car s'agit peut-être d'un 404 ou erreur juste sur cette URL
157			if (empty($mapData) \|\| empty($mapData['url']) \|\| empty($mapData['titre'])) {
158			$this->log->info('Mapping incomplet : '.$url);
159
160			return $url;
161			}
162
163			$this->tagAndLog($mapData);
164			$this->addSummaryLog($mapData);
165
166			$template = $this->chooseTemplateByData($mapData);
167
168			$mapData = $this->replaceSitenameByConfig($mapData, $template);
169			$mapData = $this->replaceURLbyOriginal($mapData);
170
171
172			if ($template instanceof ArticleTemplate) {
173			unset($mapData['site']);
174			}
175			unset($mapData['DATA-TYPE']); // ugly
176			unset($mapData['DATA-ARTICLE']); // ugly
177			unset($mapData['url-access']);
178
179			$template->hydrate($mapData);
180
181			$optimizer = OptimizerFactory::fromTemplate($template);
182			$optimizer->doTasks();
183			$templateOptimized = $optimizer->getOptiTemplate();
184
185			$serialized = $templateOptimized->serialize(true);
186			$this->log->info($serialized."\n");
187
188			return Normalizer::normalize($serialized);
189			}
190
191			/**
192			* @param string $url
193			*
194			* @return bool
195			* @throws Exception
196			*/
197			protected function isURLAuthorized(string $url): bool
198			{
199			if (!ExternHttpClient::isWebURL($url)) {
200			$this->log->debug('Skip : not an URL : '.$url);
201
202			return false;
203			}
204
205			if ($this->hasForbiddenFilenameExtension($url)) {
206			return false;
207			}
208
209			$this->url = $url;
210			$this->domain = ExternDomains::extractSubDomain($this->url);
211
212			if (in_array($this->domain, $this->skip_domain)) {
213			$this->log->notice("Skip domain ".$this->domain);
214
215			return false;
216			}
217
218			if (!isset($this->config[$this->domain])) {
219			$this->log->debug("Domain ".$this->domain." non configuré");
220			if ($this->skipUnauthorised) {
221			return false;
222			}
223			} else {
224			$this->log->debug("Domain ".$this->domain." configuré");
225			}
226
227			$this->config[$this->domain] = $this->config[$this->domain] ?? [];
228			$this->config[$this->domain] = is_array($this->config[$this->domain]) ? $this->config[$this->domain] : [];
229
230			if ($this->config[$this->domain] === 'deactivated' \|\| isset($this->config[$this->domain]['deactivated'])) {
231			$this->log->info("Domain ".$this->domain." desactivé\n");
232
233			return false;
234			}
235
236			return true;
237			}
238
239			private function tagAndLog(array $mapData)
240			{
241			$this->log->debug('mapData', $mapData);
242			$this->summary->citationNumber++;
243
244			if (isset($mapData['DATA-ARTICLE']) && $mapData['DATA-ARTICLE']) {
245			$this->log->notice("Article OK");
246			}
247			if (isset($this->data['newspaper'][$this->domain])) {
248			$this->log->notice('PRESSE');
249			$this->summary->memo['presse'] = true;
250			}
251			if ($this->isScientificDomain()) {
252			$this->log->notice('SCIENCE');
253			$this->summary->memo['science'] = true;
254			}
255			if (!isset($this->summary->memo['sites'])
256			\|\| !in_array($this->externalPage->getPrettyDomainName(), $this->summary->memo['sites'])
257			) {
258			$this->summary->memo['sites'][] = $this->externalPage->getPrettyDomainName();
259			}
260			}
261
262			private function isScientificDomain(): bool
263			{
264			if (isset($this->data['scientific domain'][$this->domain])) {
265			return true;
266			}
267			if (strpos('.revues.org', $this->domain) > 0) {
268			return true;
269			}
270
271			return false;
272			}
273
274			private function addSummaryLog(array $mapData)
275			{
276			$this->summaryLog[] = $mapData['site'] ?? $mapData['périodique'] ?? '?';
277			}
278
279			/**
280			* todo refac lisible
281			*
282			* @param array $mapData
283			*
284			* @return AbstractWikiTemplate
285			* @throws Exception
286			*/
287			private function chooseTemplateByData(array $mapData): AbstractWikiTemplate
288			{
289			// Logique : choix template
290			$this->config[$this->domain]['template'] = $this->config[$this->domain]['template'] ?? [];
291			$mapData['DATA-ARTICLE'] = $mapData['DATA-ARTICLE'] ?? false;
292
293			if (!empty($mapData['doi'])) {
294			$templateName = 'article';
295			}
296
297			if ($this->config[$this->domain]['template'] === 'article'
298			\|\| ($this->config[$this->domain]['template'] === 'auto' && $mapData['DATA-ARTICLE'])
299			\|\| ($mapData['DATA-ARTICLE'] && !empty($this->data['newspaper'][$this->domain]))
300			\|\| $this->isScientificDomain()
301			) {
302			$templateName = 'article';
303			}
304			if (!isset($templateName) \|\| $this->config[$this->domain]['template'] === 'lien web') {
305			$templateName = 'lien web';
306			}
307			// date obligatoire pour {article}
308			if (!isset($mapData['date'])) {
309			$templateName = 'lien web';
310			}
311
312			$template = WikiTemplateFactory::create($templateName);
313			$template->userSeparator = " \|";
314			$this->summary->memo['count '.$templateName] = 1 + ($this->summary->memo['count '.$templateName] ?? 0);
315
316			return $template;
317			}
318
319			/**
320			* Logique : remplacement titre périodique ou nom du site
321			*
322			* @param array $mapData
323			* @param $template
324			*
325			* @return array
326			*/
327			private function replaceSitenameByConfig(array $mapData, $template): array
328			{
329			// from wikidata URL of newspapers
330			if (!empty($this->data['newspaper'][$this->domain])) {
331			$frwiki = $this->data['newspaper'][$this->domain]['frwiki'];
332			$label = $this->data['newspaper'][$this->domain]['fr'];
333			if (isset($mapData['site']) \|\| $template instanceof LienWebTemplate) {
334			$mapData['site'] = WikiTextUtil::wikilink($label, $frwiki);
335			}
336			if (isset($mapData['périodique']) \|\| $template instanceof ArticleTemplate) {
337			$mapData['périodique'] = WikiTextUtil::wikilink($label, $frwiki);
338			}
339			}
340
341			// from wikidata of scientific journals
342			if (isset($mapData['périodique']) && isset($this->data['scientific wiki'][$mapData['périodique']])) {
343			$mapData['périodique'] = WikiTextUtil::wikilink(
344			$mapData['périodique'],
345			$this->data['scientific wiki'][$mapData['périodique']]
346			);
347			}
348
349			// from YAML config
350			if (!empty($this->config[$this->domain]['site']) && $template instanceof LienWebTemplate) {
351			$mapData['site'] = $this->config[$this->domain]['site'];
352			}
353			if (!empty($this->config[$this->domain]['périodique'])
354			&& (!empty($mapData['périodique'])
355			\|\| $template instanceof OuvrageTemplate)
356			) {
357			$mapData['périodique'] = $this->config[$this->domain]['périodique'];
358			}
359
360			// from logic
361			if (empty($mapData['site']) && $template instanceof LienWebTemplate) {
362			try {
363			$mapData['site'] = $this->externalPage->getPrettyDomainName();
364			} catch (Throwable $e) {
365			unset($e);
366			}
367			}
368
369			return $mapData;
370			}
371
372			private function replaceURLbyOriginal(array $mapData): array
373			{
374			$mapData['url'] = $this->url;
375
376			return $mapData;
377			}
378
379			/**
380			* todo move ?
381			* URL => "parismatch.com/People/bla…"
382			*
383			* @param string $url
384			*
385			* @return string
386			*/
387			public function url2TextStyleTitle(string $url): string
388			{
389			$text = str_replace(['https://', 'http://', 'www.'], '', $url);
390			if (strlen($text) > 30) {
391			$text = substr($text, 0, 30).'…';
392			}
393
394			return $text;
395			}
396
397			/**
398			* Skip PDF GIF etc
399			* https://fr.wikipedia.org/wiki/Liste_d%27extensions_de_fichiers
400			*
401			* @param string $url
402			*
403			* @return bool
404			*/
405			private function hasForbiddenFilenameExtension(string $url): bool
406			{
407			if (preg_match(
408			'#\.(pdf\|jpg\|jpeg\|gif\|png\|xls\|xlsx\|xlr\|xml\|xlt\|xlsx\|txt\|csv\|js\|docx\|exe\|gz\|zip\|ini\|movie\|mp3\|mp4\|ogg\|raw\|rss\|tar\|tgz\|wma)$#i',
409			$url
410			)
411			) {
412			return true;
413			}
414
415			return false;
416			}
417
418			protected function importConfigAndData(): void
419			{
420			// todo REFAC DataObject[]
421			$this->config = Yaml::parseFile(__DIR__.'/resources/config_presse.yaml');
422			$skipFromFile = file(
423			self::SKIP_DOMAIN_FILENAME,
424			FILE_IGNORE_NEW_LINES \| FILE_SKIP_EMPTY_LINES
425			);
426			$this->skip_domain = ($skipFromFile) ? $skipFromFile : [];
427
428			$this->data['newspaper'] = json_decode(file_get_contents(__DIR__.'/resources/data_newspapers.json'), true);
429			$this->data['scientific domain'] = json_decode(
430			file_get_contents(__DIR__.'/resources/data_scientific_domain.json'),
431			true
432			);
433			$this->data['scientific wiki'] = json_decode(
434			file_get_contents(__DIR__.'/resources/data_scientific_wiki.json'),
435			true
436			);
437			}
438
439			}
440

Dispositif / Wikibot

Push — master ( 1bcf8b...aa70d7 )

ExternRefTransformer::isURLAuthorized() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like