ExternConverterTrait::convertDCpage() - Code Metrics - Inspection of "Feature : mapping "accès url" for externalLinks" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Push — master ( 319840...0bb54c )

by Dispositif

created 2020-08-26 15:55 UTC

ExternConverterTrait::convertDCpage() A

↳ Parent: ExternConverterTrait

Complexity

Conditions	3
Paths	3

Size

Total Lines	12
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	4.125

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	3
eloc	6
nc	3
nop	1
dl	0
loc	12
ccs	3
cts	6
cp	0.5
crap	4.125
rs	10
c	1
b	0
f	0

<?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\Domain\Publisher;

use App\Domain\Enums\Language;
use DateTime;
use Exception;

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

        return false;
    }

    /**
     * mapping "accès url" : libre, inscription, limité, payant/abonnement.
     * https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Le_Bistro/25_ao%C3%BBt_2020#Lien_externe_:_paramètre_pour_accessibilité_restreinte_(abonnement,_article_payant)
     *
     * @param $data
     *
     * @return string|null
     */
    protected function convertURLaccess($data): ?string
    {
        // https://developers.facebook.com/docs/instant-articles/subscriptions/content-tiering/?locale=fr_FR
        if (isset($data['og:article:content_tier'])) {
            switch (strtolower($data['og:article:content_tier'])) {
                case 'free':
                    return 'libre';
                case 'locked':
                    return 'payant';
                case 'metered':
                    return 'limité';
            }
        }

        // NYT, Figaro
        // Todo : Si pas libre => limité ou payant ?
        if (isset($data['isAccessibleForFree'])) {
            return ($this->sameAsTrue($data['isAccessibleForFree'])) ? 'libre' : 'limité';
        }

        if (isset($data['DC.rights'])) {
            if (in_array(strtolower($data['DC.rights']), ['free', 'public domain'])) {
                return 'libre';
            }
        }

        // TODO : https://terms.tdwg.org/wiki/dcterms:accessRights
        // "	Information about who access the resource or an indication of its security status."

        return null;
    }

    protected function sameAsTrue($str = null): bool
    {
        if ($str === null) {
            return false;
        }
        if (is_bool($str)) {
            return $str;
        }
        $str = strtolower($str);
        if (in_array($str, ['true', '1', 'yes', 'oui', 'ok'])) {
            return true;
        }

        return false;
    }

    /**
     * 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 2 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;
    }

    public function cleanAuthor(?string $str = null): ?string
    {
        if ($str === null) {
            return null;
        }
        $str = $this->clean($str);
        // "https://www.facebook.com/search/top/?q=..."
        if (preg_match('#^https?://.+#i', $str)) {
            return null;
        }
        // "Par Bob"
        if (preg_match('#^Par (.+)$#i', $str, $matches)) {
            return $matches[1];
        }

        return $str;
    }

    /**
     * Note : à appliquer AVANT wikification (sinon bug sur | )
     *
     * @param string|null $str
     *
     * @return string|null
     */
    public function clean(?string $str = null): ?string
    {
        if ($str === null) {
            return null;
        }
        $str = $this->stripEmailAdress($str);

        $str = str_replace(
            ['|', "\n", "\t", '&#x27;', '&#39;', '&#039;', '&apos;', "\n", "&#10;", "&eacute;", '©'],
            [
                '/',
                ' ',
                ' ',
                "’",
                "'",
                "'",
                "'",
                '',
                ' ',
                "é",
                '',
            ],
            $str
        );

        return html_entity_decode($str);
    }

    public function stripEmailAdress(?string $str = null): ?string
    {
        if ($str === null) {
            return null;
        }

        return preg_replace('# ?[^ ]+@[^ ]+\.[A-Z]+#i', '', $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
     * @param string|null $lang
     *
     * @return string|null
     */
    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" => [] ]
            if (isset($data['author'][$indice]['name'][0])) {
                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
     * @throws Exception
     */
    protected function convertDate(?string $str): ?string
    {
        if (empty($str)) {
            return null;
        }
        $str = str_replace(' 00:00:00', '', $str);
        $str = str_replace('/', '-', $str);

        // "2012"
        if (preg_match('#^[12][0-9]{3}$#', $str)) {
            return $str;
        }
        // "1775-1783" (Gallica)
        if (preg_match('#^[12][0-9]{3}-[12][0-9]{3}$#', $str)) {
            return $str;
        }

        try {
            $date = new DateTime($str);
        } catch (Exception $e) {
            // 23/11/2015 00:00:00
            dump('EXCEPTION DATE');

            return '<!-- '.$str.' -->';
        }

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

    /**
     * Wikification des noms/acronymes d'agences de presse.
     * Note : utiliser APRES clean() et cleanAuthor() sinon bug "|"
     *
     * @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 (@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
11		namespace App\Domain\Publisher;
12
13		use App\Domain\Enums\Language;
14		use DateTime;
15		use Exception;
16
17		trait ExternConverterTrait
18		{
19	5	protected function isAnArticle(?string $str): bool
20		{
21	5	if (in_array($str, ['article', 'journalArticle'])) {
22	5	return true;
23		}
24
25		return false;
26		}
27
28	5	/**
29		* mapping "accès url" : libre, inscription, limité, payant/abonnement.
30		* https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Le_Bistro/25_ao%C3%BBt_2020#Lien_externe_:_paramètre_pour_accessibilité_restreinte_(abonnement,_article_payant)
31	5	*
32	2	* @param $data
33		*
34	5	* @return string\|null
35	1	*/
36		protected function convertURLaccess($data): ?string
37	4	{
38	1	// https://developers.facebook.com/docs/instant-articles/subscriptions/content-tiering/?locale=fr_FR
39		if (isset($data['og:article:content_tier'])) {
40		switch (strtolower($data['og:article:content_tier'])) {
41	4	case 'free':
42		return 'libre';
43		case 'locked':
44		return 'payant';
45		case 'metered':
46		return 'limité';
47		}
48		}
49
50		// NYT, Figaro
51		// Todo : Si pas libre => limité ou payant ?
52		if (isset($data['isAccessibleForFree'])) {
53		return ($this->sameAsTrue($data['isAccessibleForFree'])) ? 'libre' : 'limité';
54	5	}
55
56	5	if (isset($data['DC.rights'])) {
57	3	if (in_array(strtolower($data['DC.rights']), ['free', 'public domain'])) {
58		return 'libre';
59		}
60		}
61	2
62	1	// TODO : https://terms.tdwg.org/wiki/dcterms:accessRights
63		// " Information about who access the resource or an indication of its security status."
64
65	1	return null;
66	1	}
67
68		protected function sameAsTrue($str = null): bool
69		{
70		if ($str === null) {
71		return false;
72	5	}
73		if (is_bool($str)) {
74	5	return $str;
75	2	}
76	2	$str = strtolower($str);
77	1	if (in_array($str, ['true', '1', 'yes', 'oui', 'ok'])) {
78		return true;
79		}
80	2
81		return false;
82		}
83	3
84		/**
85		* Réduit le nombre d'auteurs si > 3.
86		* En $modeEtAll=true vérification pour "et al.=oui".
87	5	* TODO : wikifyPressAgency()
88		*
89	5	* @param string\|null $authors
90	5	* @param bool $modeEtAl
91		*
92	5	* @return string\|null
93	5	*/
94		protected function authorsEtAl(?string $authors, $modeEtAl = false): ?string
95	5	{
96		if (empty($authors)) {
97		return null;
98		}
99		// conserve juste les 2 premiers auteurs TODO : refactor
100		// Bob, Martin ; Yul, Bar ; ... ; ...
101		if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
102		return ($modeEtAl) ? 'oui' : $matches[1];
103	5	}
104		// Bob Martin, Yul Bar, ..., ...,...
105		if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
106	5	return ($modeEtAl) ? 'oui' : $matches[1];
107		}
108
109	5	return ($modeEtAl) ? null : $authors;
110		}
111	5
112		protected function convertDCpage(array $meta): ?string
113		{
114		if (isset($meta['citation_firstpage'])) {
115	5	$page = $meta['citation_firstpage'];
116		if (isset($meta['citation_lastpage'])) {
117		$page .= '–'.$meta['citation_lastpage'];
118	5	}
119
120		return (string)$page;
121		}
122	5
123		return null;
124		}
125
126		public function cleanAuthor(?string $str = null): ?string
127		{
128		if ($str === null) {
129		return null;
130		}
131	5	$str = $this->clean($str);
132		// "https://www.facebook.com/search/top/?q=..."
133	5	if (preg_match('#^https?://.+#i', $str)) {
134	2	return null;
135		}
136		// "Par Bob"
137	3	if (preg_match('#^Par (.+)$#i', $str, $matches)) {
138	2	return $matches[1];
139		}
140
141	1	return $str;
142		}
143
144	3	/**
145		* Note : à appliquer AVANT wikification (sinon bug sur \| )
146		*
147	3	* @param string\|null $str
148		*
149		* @return string\|null
150		*/
151		public function clean(?string $str = null): ?string
152	3	{
153	3	if ($str === null) {
154	3	return null;
155		}
156	3	$str = $this->stripEmailAdress($str);
157
158		$str = str_replace(
159		['\|', "\n", "\t", ''', ''', ''', ''', "\n", " ", "é", '©'],
160		[
161		'/',
162		' ',
163		' ',
164		"’",
165		"'",
166	3	"'",
167	2	"'",
168	3	'',
169		' ',
170	2	"é",
171	2	'',
172		],
173		$str
174		);
175
176		return html_entity_decode($str);
177		}
178	3
179		public function stripEmailAdress(?string $str = null): ?string
180		{
181	3	if ($str === null) {
182		return null;
183	3	}
184	3
185		return preg_replace('# ?[^ ]+@[^ ]+\.[A-Z]+#i', '', $str);
186		}
187
188		protected function convertOGtype2format(?string $ogType): ?string
189	3	{
190		if (empty($ogType)) {
191		return null;
192		}
193		// og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
194		if (strpos($ogType, 'video') !== false) {
195		return 'vidéo';
196		}
197		if (strpos($ogType, 'book') !== false) {
198	5	return 'livre';
199		}
200	5
201		return null;
202		}
203
204		/**
205	5	* https://developers.facebook.com/docs/internationalization#locales
206		* @param string\|null $lang
207		*
208		* @return string\|null
209		*/
210	5	protected function convertLangue(?string $lang = null): ?string
211		{
212		if (empty($lang)) {
213		return null;
214		}
215		// en_GB
216		if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
217	5	return $matches[1];
218		}
219
220		return Language::all2wiki($lang);
221		}
222
223		protected function convertAuteur($data, $indice)
224		{
225		// author=Bob
226		if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
227	5	return html_entity_decode($data['author']);
228		}
229	5
230	3	// author ['name'=>'Bob','@type'=>'Person']
231		if (0 === $indice
232		&& isset($data['author'])
233	4	&& isset($data['author']['name'])
234		&& (!isset($data['author']['@type'])
235		\|\| 'Person' === $data['author']['@type'])
236	4	) {
237	4	if (is_string($data['author']['name'])) {
238	4	return html_entity_decode($data['author']['name']);
239	4	}
240	4
241	4	return html_entity_decode($data['author']['name'][0]);
242	4	}
243	4
244		// author [ 0 => ['name'=>'Bob'], 1=> ...]
245	4	if (isset($data['author']) && isset($data['author'][$indice])
246		&& (!isset($data['author'][$indice]['@type'])
247		\|\| 'Person' === $data['author'][$indice]['@type'])
248		) {
249		if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
250		return html_entity_decode($data['author'][$indice]['name']);
251		}
252
253		// "author" => [ "@type" => "Person", "name" => [] ]
254		if (isset($data['author'][$indice]['name'][0])) {
255		return html_entity_decode($data['author'][$indice]['name'][0]);
256		}
257		}
258
259		return null;
260		}
261
262		protected function convertInstitutionnel($data)
263		{
264		if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
265		&& 'Person' !== $data['author'][0]['@type']
266		) {
267		return html_entity_decode($data['author'][0]['name']);
268		}
269
270		return null;
271		}
272
273		/**
274		* @param string $str
275		*
276		* @return string
277		* @throws Exception
278		*/
279		protected function convertDate(?string $str): ?string
280		{
281		if (empty($str)) {
282		return null;
283		}
284		$str = str_replace(' 00:00:00', '', $str);
285		$str = str_replace('/', '-', $str);
286
287		// "2012"
288		if (preg_match('#^[12][0-9]{3}$#', $str)) {
289		return $str;
290		}
291		// "1775-1783" (Gallica)
292		if (preg_match('#^[12][0-9]{3}-[12][0-9]{3}$#', $str)) {
293		return $str;
294		}
295
296		try {
297		$date = new DateTime($str);
298		} catch (Exception $e) {
299		// 23/11/2015 00:00:00
300		dump('EXCEPTION DATE');
301
302		return '<!-- '.$str.' -->';
303		}
304
305		return $date->format('d-m-Y');
306		}
307
308		/**
309		* Wikification des noms/acronymes d'agences de presse.
310		* Note : utiliser APRES clean() et cleanAuthor() sinon bug "\|"
311		*
312		* @param string $str
313		*
314		* @return string
315		*/
316		protected function wikifyPressAgency(?string $str): ?string
317		{
318		if (empty($str)) {
319		return null;
320		}
321		// skip potential wikilinks
322		if (strpos($str, '[') !== false) {
323		return $str;
324		}
325		$str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse\|AFP]]', $str);
326		$str = str_replace('Reuters', '[[Reuters]]', $str);
327		$str = str_replace('Associated Press', '[[Associated Press]]', $str);
328		$str = preg_replace('#\b(PA)\b#', '[[Press Association\|PA]]', $str);
329		$str = preg_replace('#\b(AP)\b#', '[[Associated Press\|AP]]', $str);
330		$str = str_replace('Xinhua', '[[Xinhua]]', $str);
331		$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse\|ATS]]', $str);
332		$str = preg_replace('#\b(PC\|CP)\b#', '[[La Presse canadienne\|PC]]', $str);
333
334		return $str;
335		}
336
337		}
338

Dispositif / Wikibot

Push — master ( 319840...0bb54c )

ExternConverterTrait::convertDCpage() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like