WebMapper::clean() - 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

WebMapper::clean() A

↳ Parent: WebMapper

Complexity

Conditions	2
Paths	2

Size

Total Lines	8
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	2
eloc	4
c	1
b	0
f	0
nc	2
nop	1
dl	0
loc	8
rs	10

<?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\Domain\Publisher;

use App\Domain\Enums\Language;
use App\Domain\Utils\ArrayProcessTrait;
use App\Infrastructure\Logger;
use DateTime;
use Exception;
use Psr\Log\LoggerInterface;

/**
 * Generic mapper for press/revue article on web.
 * Using JSON-LD and meta tags to obtain {article} data.
 * Class WebMapper
 *
 * @package App\Domain\Publisher
 */
class WebMapper implements MapperInterface
{
    use ArrayProcessTrait, WebOGMapperTrait, WebLDMapperTrait;

    /**
     * @var Logger
     */
    private $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;
}
    }

    public function process($data): array
    {
        $dat = $this->processMapping($data);

        return $this->postProcess($dat);
    }

    protected function processMapping($data): array
    {
        if (!empty($data['JSON-LD'])) {
            if ($this->checkJSONLD($data['JSON-LD'])) {
                return $this->mapArticleDataFromJSONLD($data['JSON-LD']);
            }
            // gestion des multiples objets comme Figaro
            foreach ($data['JSON-LD'] as $dat) {
                if (is_array($dat) && $this->checkJSONLD($dat)) {
                    return $this->mapArticleDataFromJSONLD($dat);
                }
            }
        }
        if (!empty($data['meta'])) {
            // Dublin Core mapping included ;-)
            return $this->mapLienwebFromOpenGraph($data['meta']);
        }

        return [];
    }

    protected function checkJSONLD(array $jsonLD): bool
    {
        return isset($jsonLD['headline']) && isset($jsonLD['@type']);
    }

    /**
     * todo Refac/move domain special mapping
     *
     * @param array $dat
     *
     * @return array
     */
    protected function postProcess(array $dat): array
    {
        $dat = $this->deleteEmptyValueArray($dat);
        if (isset($dat['langue']) && 'fr' === $dat['langue']) {
            unset($dat['langue']);
        }

        if (isset($dat['site']) && $dat['site'] === 'Gallica') {
            unset($dat['format']); // "vidéo"...
        }

        return $dat;
    }

    protected function isAnArticle(?string $str): bool
    {
        if (in_array($str, ['article', 'journalArticle'])) {
            return true;
        }

        return false;
    }

    protected function convertURLaccess($data): ?string
    {
        // NYT, Figaro
        if (isset($data['isAccessibleForFree'])) {
            return $data['isAccessibleForFree'] ? 'ouvert' : 'limité';
        }
        if (isset($data['DC.rights'])) {
            return (in_array($data['DC.rights'], ['free', 'public domain'])) ? 'ouvert' : 'limité';
        }
        if (isset($data['og:article:content_tier'])) {
            return ($data['og:article:content_tier'] === 'free') ? 'ouvert' : 'limité';
        }

        return null;
    }

    /**
     * Réduit le nombre d'auteurs si > 3.
     * En $modeEtAll=true vérification pour "et al.=oui".
     * TODO : wikifyPressAgency()
     *
     * @param string|null $authors
     * @param bool        $modeEtAl
     *
     * @return string|null
     */
    protected function authorsEtAl(?string $authors, $modeEtAl = false): ?string
    {
        if (empty($authors)) {
            return null;
        }
        // conserve juste les 3 premiers auteurs TODO : refactor
        // Bob, Martin ; Yul, Bar ; ... ; ...
        if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
            return ($modeEtAl) ? 'oui' : $matches[1];
        }
        // Bob Martin, Yul Bar, ..., ...,...
        if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
            return ($modeEtAl) ? 'oui' : $matches[1];
        }

        return ($modeEtAl) ? null : $authors;
    }

    protected function convertDCpage(array $meta): ?string
    {
        if (isset($meta['citation_firstpage'])) {
            $page = $meta['citation_firstpage'];
            if (isset($meta['citation_lastpage'])) {
                $page .= '–'.$meta['citation_lastpage'];
            }

            return (string)$page;
        }

        return null;
    }

    protected function clean(?string $str = null): ?string
    {
        if ($str === null) {
            return null;
        }
        $str = str_replace(['&apos;', "\n", "&#10;", "|"], ["'", '', ' ', '/'], $str);

        return html_entity_decode($str);
    }

    protected function convertOGtype2format(?string $ogType): ?string
    {
        if (empty($ogType)) {
            return null;
        }
        // og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
        if (strpos($ogType, 'video') !== false) {
            return 'vidéo';
        }
        if (strpos($ogType, 'book') !== false) {
            return 'livre';
        }

        return null;
    }

    /**
     * https://developers.facebook.com/docs/internationalization#locales
     */
    protected function convertLangue(?string $lang = null): ?string
    {
        if (empty($lang)) {
            return null;
        }
        // en_GB
        if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
            return $matches[1];
        }

        return Language::all2wiki($lang);
    }

    protected function convertAuteur($data, $indice)
    {
        // author=Bob
        if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
            return html_entity_decode($data['author']);
        }

        // author ['name'=>'Bob','@type'=>'Person']
        if (0 === $indice
            && isset($data['author'])
            && isset($data['author']['name'])
            && (!isset($data['author']['@type'])
                || 'Person' === $data['author']['@type'])
        ) {
            if (is_string($data['author']['name'])) {
                return html_entity_decode($data['author']['name']);
            }

            return html_entity_decode($data['author']['name'][0]);
        }

        // author [ 0 => ['name'=>'Bob'], 1=> ...]
        if (isset($data['author']) && isset($data['author'][$indice])
            && (!isset($data['author'][$indice]['@type'])
                || 'Person' === $data['author'][$indice]['@type'])
        ) {
            if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
                return html_entity_decode($data['author'][$indice]['name']);
            }

            // "author" => [ "@type" => "Person", "name" => [] ]
            return html_entity_decode($data['author'][$indice]['name'][0]);
        }

        return null;
    }

    protected function convertInstitutionnel($data)
    {
        if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
            && 'Person' !== $data['author'][0]['@type']
        ) {
            return html_entity_decode($data['author'][0]['name']);
        }

        return null;
    }

    /**
     * @param string $str
     *
     * @return string
     */
    protected function convertDate(?string $str): ?string
    {
        if (empty($str)) {
            return null;
        }

        // "2012"
        if (preg_match('#^[12][0-9]{3}$#', $str)) {
            return $str;
        }

        try {
            $date = new DateTime($str);
        } catch (Exception $e) {
            dump('EXCEPTION DATE');

            return $str;
        }

        return $date->format('d-m-Y');
    }

    /**
     * Wikification des noms/acronymes d'agences de presse.
     *
     * @param string $str
     *
     * @return string
     */
    protected function wikifyPressAgency(?string $str): ?string
    {
        if (empty($str)) {
            return null;
        }
        // skip potential wikilinks
        if (strpos($str, '[') !== false) {
            return $str;
        }
        $str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse|AFP]]', $str);
        $str = str_replace('Reuters', '[[Reuters]]', $str);
        $str = str_replace('Associated Press', '[[Associated Press]]', $str);
        $str = preg_replace('#\b(PA)\b#', '[[Press Association|PA]]', $str);
        $str = preg_replace('#\b(AP)\b#', '[[Associated Press|AP]]', $str);
        $str = str_replace('Xinhua', '[[Xinhua]]', $str);
        $str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse|ATS]]', $str);
        $str = preg_replace('#\b(PC|CP)\b#', '[[La Presse canadienne|PC]]', $str);

        return $str;
    }
}


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\Domain\Publisher;
11
12			use App\Domain\Enums\Language;
13			use App\Domain\Utils\ArrayProcessTrait;
14			use App\Infrastructure\Logger;
15			use DateTime;
16			use Exception;
17			use Psr\Log\LoggerInterface;
18
19			/**
20			* Generic mapper for press/revue article on web.
21			* Using JSON-LD and meta tags to obtain {article} data.
22			* Class WebMapper
23			*
24			* @package App\Domain\Publisher
25			*/
26			class WebMapper implements MapperInterface
27			{
28			use ArrayProcessTrait, WebOGMapperTrait, WebLDMapperTrait;
29
30			/**
31			* @var Logger
32			*/
33			private $log;
34
35			public function __construct(LoggerInterface $log)
36			{
37			$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 = $alien; } Loading history...
38			}
39
40			public function process($data): array
41			{
42			$dat = $this->processMapping($data);
43
44			return $this->postProcess($dat);
45			}
46
47			protected function processMapping($data): array
48			{
49			if (!empty($data['JSON-LD'])) {
50			if ($this->checkJSONLD($data['JSON-LD'])) {
51			return $this->mapArticleDataFromJSONLD($data['JSON-LD']);
52			}
53			// gestion des multiples objets comme Figaro
54			foreach ($data['JSON-LD'] as $dat) {
55			if (is_array($dat) && $this->checkJSONLD($dat)) {
56			return $this->mapArticleDataFromJSONLD($dat);
57			}
58			}
59			}
60			if (!empty($data['meta'])) {
61			// Dublin Core mapping included ;-)
62			return $this->mapLienwebFromOpenGraph($data['meta']);
63			}
64
65			return [];
66			}
67
68			protected function checkJSONLD(array $jsonLD): bool
69			{
70			return isset($jsonLD['headline']) && isset($jsonLD['@type']);
71			}
72
73			/**
74			* todo Refac/move domain special mapping
75			*
76			* @param array $dat
77			*
78			* @return array
79			*/
80			protected function postProcess(array $dat): array
81			{
82			$dat = $this->deleteEmptyValueArray($dat);
83			if (isset($dat['langue']) && 'fr' === $dat['langue']) {
84			unset($dat['langue']);
85			}
86
87			if (isset($dat['site']) && $dat['site'] === 'Gallica') {
88			unset($dat['format']); // "vidéo"...
89			}
90
91			return $dat;
92			}
93
94			protected function isAnArticle(?string $str): bool
95			{
96			if (in_array($str, ['article', 'journalArticle'])) {
97			return true;
98			}
99
100			return false;
101			}
102
103			protected function convertURLaccess($data): ?string
104			{
105			// NYT, Figaro
106			if (isset($data['isAccessibleForFree'])) {
107			return $data['isAccessibleForFree'] ? 'ouvert' : 'limité';
108			}
109			if (isset($data['DC.rights'])) {
110			return (in_array($data['DC.rights'], ['free', 'public domain'])) ? 'ouvert' : 'limité';
111			}
112			if (isset($data['og:article:content_tier'])) {
113			return ($data['og:article:content_tier'] === 'free') ? 'ouvert' : 'limité';
114			}
115
116			return null;
117			}
118
119			/**
120			* Réduit le nombre d'auteurs si > 3.
121			* En $modeEtAll=true vérification pour "et al.=oui".
122			* TODO : wikifyPressAgency()
123			*
124			* @param string\|null $authors
125			* @param bool $modeEtAl
126			*
127			* @return string\|null
128			*/
129			protected function authorsEtAl(?string $authors, $modeEtAl = false): ?string
130			{
131			if (empty($authors)) {
132			return null;
133			}
134			// conserve juste les 3 premiers auteurs TODO : refactor
135			// Bob, Martin ; Yul, Bar ; ... ; ...
136			if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
137			return ($modeEtAl) ? 'oui' : $matches[1];
138			}
139			// Bob Martin, Yul Bar, ..., ...,...
140			if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
141			return ($modeEtAl) ? 'oui' : $matches[1];
142			}
143
144			return ($modeEtAl) ? null : $authors;
145			}
146
147			protected function convertDCpage(array $meta): ?string
148			{
149			if (isset($meta['citation_firstpage'])) {
150			$page = $meta['citation_firstpage'];
151			if (isset($meta['citation_lastpage'])) {
152			$page .= '–'.$meta['citation_lastpage'];
153			}
154
155			return (string)$page;
156			}
157
158			return null;
159			}
160
161			protected function clean(?string $str = null): ?string
162			{
163			if ($str === null) {
164			return null;
165			}
166			$str = str_replace([''', "\n", " ", "\|"], ["'", '', ' ', '/'], $str);
167
168			return html_entity_decode($str);
169			}
170
171			protected function convertOGtype2format(?string $ogType): ?string
172			{
173			if (empty($ogType)) {
174			return null;
175			}
176			// og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
177			if (strpos($ogType, 'video') !== false) {
178			return 'vidéo';
179			}
180			if (strpos($ogType, 'book') !== false) {
181			return 'livre';
182			}
183
184			return null;
185			}
186
187			/**
188			* https://developers.facebook.com/docs/internationalization#locales
189			*/
190			protected function convertLangue(?string $lang = null): ?string
191			{
192			if (empty($lang)) {
193			return null;
194			}
195			// en_GB
196			if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
197			return $matches[1];
198			}
199
200			return Language::all2wiki($lang);
201			}
202
203			protected function convertAuteur($data, $indice)
204			{
205			// author=Bob
206			if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
207			return html_entity_decode($data['author']);
208			}
209
210			// author ['name'=>'Bob','@type'=>'Person']
211			if (0 === $indice
212			&& isset($data['author'])
213			&& isset($data['author']['name'])
214			&& (!isset($data['author']['@type'])
215			\|\| 'Person' === $data['author']['@type'])
216			) {
217			if (is_string($data['author']['name'])) {
218			return html_entity_decode($data['author']['name']);
219			}
220
221			return html_entity_decode($data['author']['name'][0]);
222			}
223
224			// author [ 0 => ['name'=>'Bob'], 1=> ...]
225			if (isset($data['author']) && isset($data['author'][$indice])
226			&& (!isset($data['author'][$indice]['@type'])
227			\|\| 'Person' === $data['author'][$indice]['@type'])
228			) {
229			if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
230			return html_entity_decode($data['author'][$indice]['name']);
231			}
232
233			// "author" => [ "@type" => "Person", "name" => [] ]
234			return html_entity_decode($data['author'][$indice]['name'][0]);
235			}
236
237			return null;
238			}
239
240			protected function convertInstitutionnel($data)
241			{
242			if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
243			&& 'Person' !== $data['author'][0]['@type']
244			) {
245			return html_entity_decode($data['author'][0]['name']);
246			}
247
248			return null;
249			}
250
251			/**
252			* @param string $str
253			*
254			* @return string
255			*/
256			protected function convertDate(?string $str): ?string
257			{
258			if (empty($str)) {
259			return null;
260			}
261
262			// "2012"
263			if (preg_match('#^[12][0-9]{3}$#', $str)) {
264			return $str;
265			}
266
267			try {
268			$date = new DateTime($str);
269			} catch (Exception $e) {
270			dump('EXCEPTION DATE');
271
272			return $str;
273			}
274
275			return $date->format('d-m-Y');
276			}
277
278			/**
279			* Wikification des noms/acronymes d'agences de presse.
280			*
281			* @param string $str
282			*
283			* @return string
284			*/
285			protected function wikifyPressAgency(?string $str): ?string
286			{
287			if (empty($str)) {
288			return null;
289			}
290			// skip potential wikilinks
291			if (strpos($str, '[') !== false) {
292			return $str;
293			}
294			$str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse\|AFP]]', $str);
295			$str = str_replace('Reuters', '[[Reuters]]', $str);
296			$str = str_replace('Associated Press', '[[Associated Press]]', $str);
297			$str = preg_replace('#\b(PA)\b#', '[[Press Association\|PA]]', $str);
298			$str = preg_replace('#\b(AP)\b#', '[[Associated Press\|AP]]', $str);
299			$str = str_replace('Xinhua', '[[Xinhua]]', $str);
300			$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse\|ATS]]', $str);
301			$str = preg_replace('#\b(PC\|CP)\b#', '[[La Presse canadienne\|PC]]', $str);
302
303			return $str;
304			}
305			}
306

Dispositif / Wikibot

Branch — dev (1a31b5)

WebMapper::clean() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like