ExternConverterTrait::convertDate() - Code Metrics - Inspection of "Style" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 7c3be8...26e53d )

by Dispositif

created 2020-08-31 15:37 UTC

ExternConverterTrait::convertDate() B

↳ Parent: ExternConverterTrait

Complexity

Conditions	8
Paths	6

Size

Total Lines	29
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	72

Importance

Changes	4
Bugs	0	Features	0

Metric	Value
cc	8
eloc	15
c	4
b	0
f	0
nc	6
nop	1
dl	0
loc	29
ccs	0
cts	0
cp	0
crap	72
rs	8.4444

<?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' : 'payant';
        }

        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."
        // Values are a mystery...
        if (isset($data['DC.accessRights'])) {
            if (in_array(strtolower($data['DC.accessRights']), ['free', 'public domain', 'public', 'available'])) {
                return 'libre';
            }
        }

        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;
    }

    /**
     * todo move to generalize as utility
     *
     * @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 $e) {
            // 23/11/2015 00:00:00
            if (isset($this) && isset($this->log) && method_exists($this->log, 'notice')) {
                $this->log->notice('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($data['isAccessibleForFree'])) ? 'libre' : 'payant';
54	5	}
55
56	5	if (isset($data['DC.rights'])) {
57	3	if (in_array(strtolower($data['DC.rights']), ['free', 'public domain'])) {
58		return 'libre';
59		}
60		}
61	2
62	1	// TODO : https://terms.tdwg.org/wiki/dcterms:accessRights
63		// " Information about who access the resource or an indication of its security status."
64		// Values are a mystery...
65	1	if (isset($data['DC.accessRights'])) {
66	1	if (in_array(strtolower($data['DC.accessRights']), ['free', 'public domain', 'public', 'available'])) {
67		return 'libre';
68		}
69		}
70
71		return null;
72	5	}
73
74	5	protected function sameAsTrue($str = null): bool
75	2	{
76	2	if ($str === null) {
77	1	return false;
78		}
79		if (is_bool($str)) {
80	2	return $str;
81		}
82		$str = strtolower($str);
83	3	if (in_array($str, ['true', '1', 'yes', 'oui', 'ok'])) {
84		return true;
85		}
86
87	5	return false;
88		}
89	5
90	5	/**
91		* Réduit le nombre d'auteurs si > 3.
92	5	* En $modeEtAll=true vérification pour "et al.=oui".
93	5	* TODO : wikifyPressAgency()
94		*
95	5	* @param string\|null $authors
96		* @param bool $modeEtAl
97		*
98		* @return string\|null
99		*/
100		protected function authorsEtAl(?string $authors, $modeEtAl = false): ?string
101		{
102		if (empty($authors)) {
103	5	return null;
104		}
105		// conserve juste les 2 premiers auteurs TODO : refactor
106	5	// Bob, Martin ; Yul, Bar ; ... ; ...
107		if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
108		return ($modeEtAl) ? 'oui' : $matches[1];
109	5	}
110		// Bob Martin, Yul Bar, ..., ...,...
111	5	if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
112		return ($modeEtAl) ? 'oui' : $matches[1];
113		}
114
115	5	return ($modeEtAl) ? null : $authors;
116		}
117
118	5	protected function convertDCpage(array $meta): ?string
119		{
120		if (isset($meta['citation_firstpage'])) {
121		$page = $meta['citation_firstpage'];
122	5	if (isset($meta['citation_lastpage'])) {
123		$page .= '–'.$meta['citation_lastpage'];
124		}
125
126		return (string)$page;
127		}
128
129		return null;
130		}
131	5
132		public function cleanAuthor(?string $str = null): ?string
133	5	{
134	2	if ($str === null) {
135		return null;
136		}
137	3	$str = $this->clean($str);
138	2	// "https://www.facebook.com/search/top/?q=..."
139		if (preg_match('#^https?://.+#i', $str)) {
140		return null;
141	1	}
142		// "Par Bob"
143		if (preg_match('#^Par (.+)$#i', $str, $matches)) {
144	3	return $matches[1];
145		}
146
147	3	return $str;
148		}
149
150		/**
151		* Note : à appliquer AVANT wikification (sinon bug sur \| )
152	3	*
153	3	* @param string\|null $str
154	3	*
155		* @return string\|null
156	3	*/
157		public function clean(?string $str = null): ?string
158		{
159		if ($str === null) {
160		return null;
161		}
162		$str = $this->stripEmailAdress($str);
163
164		$str = str_replace(
165		['\|', "\n", "\t", ''', ''', ''', ''', "\n", " ", "é", '©'],
166	3	[
167	2	'/',
168	3	' ',
169		' ',
170	2	"’",
171	2	"'",
172		"'",
173		"'",
174		'',
175		' ',
176		"é",
177		'',
178	3	],
179		$str
180		);
181	3
182		return html_entity_decode($str);
183	3	}
184	3
185		public function stripEmailAdress(?string $str = null): ?string
186		{
187		if ($str === null) {
188		return null;
189	3	}
190
191		return preg_replace('# ?[^ ]+@[^ ]+\.[A-Z]+#i', '', $str);
192		}
193
194		protected function convertOGtype2format(?string $ogType): ?string
195		{
196		if (empty($ogType)) {
197		return null;
198	5	}
199		// og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
200	5	if (strpos($ogType, 'video') !== false) {
201		return 'vidéo';
202		}
203		if (strpos($ogType, 'book') !== false) {
204		return 'livre';
205	5	}
206
207		return null;
208		}
209
210	5	/**
211		* https://developers.facebook.com/docs/internationalization#locales
212		* @param string\|null $lang
213		*
214		* @return string\|null
215		*/
216		protected function convertLangue(?string $lang = null): ?string
217	5	{
218		if (empty($lang)) {
219		return null;
220		}
221		// en_GB
222		if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
223		return $matches[1];
224		}
225
226		return Language::all2wiki($lang);
227	5	}
228
229	5	protected function convertAuteur($data, $indice)
230	3	{
231		// author=Bob
232		if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
233	4	return html_entity_decode($data['author']);
234		}
235
236	4	// author ['name'=>'Bob','@type'=>'Person']
237	4	if (0 === $indice
238	4	&& isset($data['author'])
239	4	&& isset($data['author']['name'])
240	4	&& (!isset($data['author']['@type'])
241	4	\|\| 'Person' === $data['author']['@type'])
242	4	) {
243	4	if (is_string($data['author']['name'])) {
244		return html_entity_decode($data['author']['name']);
245	4	}
246
247		return html_entity_decode($data['author']['name'][0]);
248		}
249
250		// author [ 0 => ['name'=>'Bob'], 1=> ...]
251		if (isset($data['author']) && isset($data['author'][$indice])
252		&& (!isset($data['author'][$indice]['@type'])
253		\|\| 'Person' === $data['author'][$indice]['@type'])
254		) {
255		if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
256		return html_entity_decode($data['author'][$indice]['name']);
257		}
258
259		// "author" => [ "@type" => "Person", "name" => [] ]
260		if (isset($data['author'][$indice]['name'][0])) {
261		return html_entity_decode($data['author'][$indice]['name'][0]);
262		}
263		}
264
265		return null;
266		}
267
268		protected function convertInstitutionnel($data)
269		{
270		if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
271		&& 'Person' !== $data['author'][0]['@type']
272		) {
273		return html_entity_decode($data['author'][0]['name']);
274		}
275
276		return null;
277		}
278
279		/**
280		* todo move to generalize as utility
281		*
282		* @param string $str
283		*
284		* @return string
285		* @throws Exception
286		*/
287		protected function convertDate(?string $str): ?string
288		{
289		if (empty($str)) {
290		return null;
291		}
292		$str = str_replace(' 00:00:00', '', $str);
293		$str = str_replace('/', '-', $str);
294
295		// "2012"
296		if (preg_match('#^[12][0-9]{3}$#', $str)) {
297		return $str;
298		}
299		// "1775-1783" (Gallica)
300		if (preg_match('#^[12][0-9]{3}-[12][0-9]{3}$#', $str)) {
301		return $str;
302		}
303
304		try {
305		$date = new DateTime($str);
306		} catch (Exception $e) {
307		// 23/11/2015 00:00:00
308		if (isset($this) && isset($this->log) && method_exists($this->log, 'notice')) {
309		$this->log->notice('EXCEPTION DATE');
310		}
311
312		return '<!-- '.$str.' -->';
313		}
314
315		return $date->format('d-m-Y');
316		}
317
318		/**
319		* Wikification des noms/acronymes d'agences de presse.
320		* Note : utiliser APRES clean() et cleanAuthor() sinon bug "\|"
321		*
322		* @param string $str
323		*
324		* @return string
325		*/
326		protected function wikifyPressAgency(?string $str): ?string
327		{
328		if (empty($str)) {
329		return null;
330		}
331		// skip potential wikilinks
332		if (strpos($str, '[') !== false) {
333		return $str;
334		}
335		$str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse\|AFP]]', $str);
336		$str = str_replace('Reuters', '[[Reuters]]', $str);
337		$str = str_replace('Associated Press', '[[Associated Press]]', $str);
338		$str = preg_replace('#\b(PA)\b#', '[[Press Association\|PA]]', $str);
339		$str = preg_replace('#\b(AP)\b#', '[[Associated Press\|AP]]', $str);
340		$str = str_replace('Xinhua', '[[Xinhua]]', $str);
341		$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse\|ATS]]', $str);
342		$str = preg_replace('#\b(PC\|CP)\b#', '[[La Presse canadienne\|PC]]', $str);
343
344		return $str;
345		}
346
347		}
348

Dispositif / Wikibot

Push — master ( 7c3be8...26e53d )

ExternConverterTrait::convertDate() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like