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

Passed

Push — master ( 95294c...e75f8e )

by Dispositif

created 2020-07-26 11:00 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
ccs	0
cts	30
cp	0
crap	90
rs	8.0555

<?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\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 Codedungeon\PHPCliColors\Color;
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';
    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;

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

        // todo REFAC DataObject[]
        $this->config = Yaml::parseFile(__DIR__.'/resources/config_presse.yaml');
        $skipFromFile = file(
            __DIR__.'/resources/config_skip_domain.txt',
            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
        );

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

    /**
     * @param string $url
     *
     * @return string
     * @throws Exception
     */
    public function process(string $url): string
    {
        if (!$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())) {
                $this->log->notice('410 page définitivement disparue : '.$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 sur extractWebData');

                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']))
        ) {
            // site avec HTML pourri
            return $url;
        }

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

            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');

            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

        $template->hydrate($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])) {
            $this->log->info("Domain ".$this->domain." non configuré\n");
            if ($this->skipUnauthorised) {
                return false;
            }
        } else {
            echo "> Domaine ".Color::LIGHT_GREEN.$this->domain.Color::NORMAL." 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] === '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);

        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
     *
     * @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 = " |";

        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;
    }

}


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

Dispositif / Wikibot

Push — master ( 95294c...e75f8e )

ExternRefTransformer::isURLAuthorized() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like