ExternConverterTrait::cleanSEOTitle() - Code Metrics - Inspection of "style, cosmetic" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( dff8a4...2556d0 )

by Dispositif

created 2023-04-08 11:33 UTC

ExternConverterTrait::cleanSEOTitle() A

↳ Parent: ExternConverterTrait

Complexity

Conditions	5
Paths	2

Size

Total Lines	14
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	9.664

Importance

Changes	1
Bugs	1	Features	0

Metric	Value
cc	5
eloc	7
c	1
b	1
f	0
nc	2
nop	2
dl	0
loc	14
ccs	3
cts	7
cp	0.4286
crap	9.664
rs	9.6111

<?php
/*
 * This file is part of dispositif/wikibot application (@github)
 * 2019-2023 © Philippe M./Irønie  <[email protected]>
 * For the full copyright and MIT license information, 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 $str): bool
    {
        return in_array($str, ['article', 'journalArticle']);
    }

    /**
     * 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($data['DC.rights']) && in_array(strtolower($data['DC.rights']), ['free', 'public domain', 'domaine public'])) {
            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($data['DC.accessRights']) && in_array(
                strtolower($data['DC.accessRights']),
                [
                    'free',
                    'public domain',
                    'public',
                    'domaine 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);
        return in_array($str, ['true', '1', 'yes', 'oui', 'ok']);
    }

    /**
     * Réduit le nombre d'auteurs si > 3.
     * En $modeEtAll=true vérification pour "et al.=oui".
     * TODO : wikifyPressAgency()
     */
    protected function authorsEtAl(?string $authors, bool $modeEtAl = false): ?string
    {
        if (empty($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",
                "\r",
                '&#x27;',
                '&#39;',
                '&#039;',
                '&apos;',
                "\n",
                "&#10;",
                "&eacute;",
                '©',
                '{{',
                '}}',
                '[[',
                ']]',
            ],
            [
                '/',
                ' ',
                ' ',
                '',
                "’",
                "'",
                "'",
                "'",
                '',
                ' ',
                "é",
                '',
                '',
                '',
                '',
                '',
            ],
            $str
        );

        $str = html_entity_decode($str);
        $str = strip_tags($str);

        return trim($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, $indice): ?string
    {
        // author=Bob
        if (isset($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($data): ?string
    {
        if (isset($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
     * @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]\d{3}$#', $str)) {
            return $str;
        }
        // "1775-1783" (Gallica)
        if (preg_match('#^[12]\d{3}-[12]\d{3}$#', $str)) {
            return $str;
        }

        try {
            $date = new DateTime($str);
        } catch (Exception $e) {
            // 23/11/2015 00:00:00
            if (isset($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 "|"
     */
    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);

        return preg_replace('#\b(PC|CP)\b#', '[[La Presse canadienne|PC]]', $str);
    }
}


1		<?php
2		/*
3		* This file is part of dispositif/wikibot application (@github)
4		* 2019-2023 © Philippe M./Irønie <[email protected]>
5		* For the full copyright and MIT license information, 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 DateTime;
14		use Exception;
15
16		trait ExternConverterTrait
17		{
18		protected function isAnArticle(?string $str): bool
19	5	{
20		return in_array($str, ['article', 'journalArticle']);
21	5	}
22	5
23		/**
24		* mapping "accès url" : libre, inscription, limité, payant/abonnement.
25		* https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Le_Bistro/25_ao%C3%BBt_2020#Lien_externe_:_paramètre_pour_accessibilité_restreinte_(abonnement,_article_payant)
26		*
27		* @param $data
28	5	*
29		* @return string\|null
30		*/
31	5	protected function convertURLaccess($data): ?string
32	2	{
33		// https://developers.facebook.com/docs/instant-articles/subscriptions/content-tiering/?locale=fr_FR
34	5	if (isset($data['og:article:content_tier'])) {
35	1	switch (strtolower($data['og:article:content_tier'])) {
36		case 'free':
37	4	return 'libre';
38	1	case 'locked':
39		return 'payant';
40		case 'metered':
41	4	return 'limité';
42		}
43		}
44
45		// NYT, Figaro
46		// Todo : Si pas libre => limité ou payant ?
47		if (isset($data['isAccessibleForFree'])) {
48		return ($this->sameAsTrue($data['isAccessibleForFree'])) ? 'libre' : 'payant';
49		}
50
51		if (isset($data['DC.rights']) && in_array(strtolower($data['DC.rights']), ['free', 'public domain', 'domaine public'])) {
52		return 'libre';
53		}
54	5
55		// TODO : https://terms.tdwg.org/wiki/dcterms:accessRights
56	5	// " Information about who access the resource or an indication of its security status."
57	3	// Values are a mystery...
58		if (isset($data['DC.accessRights']) && in_array(
59		strtolower($data['DC.accessRights']),
60		[
61	2	'free',
62	1	'public domain',
63		'public',
64		'domaine public',
65	1	'available',
66	1	]
67		)) {
68		return 'libre';
69		}
70
71		return null;
72	5	}
73
74	5	protected function sameAsTrue($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	return in_array($str, ['true', '1', 'yes', 'oui', 'ok']);
84		}
85
86		/**
87	5	* Réduit le nombre d'auteurs si > 3.
88		* En $modeEtAll=true vérification pour "et al.=oui".
89	5	* TODO : wikifyPressAgency()
90	5	*/
91		protected function authorsEtAl(?string $authors, bool $modeEtAl = false): ?string
92	5	{
93	5	if (empty($authors)) {
94		return null;
95	5	}
96		// conserve juste les 2 premiers auteurs TODO : refactor
97		// Bob, Martin ; Yul, Bar ; ... ; ...
98		if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
99		return ($modeEtAl) ? 'oui' : $matches[1];
100		}
101		// Bob Martin, Yul Bar, ..., ...,...
102		if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
103	5	return ($modeEtAl) ? 'oui' : $matches[1];
104		}
105
106	5	return ($modeEtAl) ? null : $authors;
107		}
108
109	5	protected function convertDCpage(array $meta): ?string
110		{
111	5	if (isset($meta['citation_firstpage'])) {
112		$page = $meta['citation_firstpage'];
113		if (isset($meta['citation_lastpage'])) {
114		$page .= '–' . $meta['citation_lastpage'];
115	5	}
116
117		return (string)$page;
118	5	}
119
120		return null;
121		}
122	5
123		public function cleanAuthor(?string $str = null): ?string
124		{
125		if ($str === null) {
126		return null;
127		}
128		$str = $this->clean($str);
129		// "https://www.facebook.com/search/top/?q=..."
130		if (preg_match('#^https?://.+#i', $str)) {
131	5	return null;
132		}
133	5	// "Par Bob"
134	2	if (preg_match('#^Par (.+)$#i', $str, $matches)) {
135		return $matches[1];
136		}
137	3
138	2	return $str;
139		}
140
141	1	/**
142		* Note : à appliquer AVANT wikification (sinon bug sur \| )
143		*
144	3	* @param string\|null $str
145		*
146		* @return string\|null
147	3	*/
148		public function clean(?string $str = null): ?string
149		{
150		if ($str === null) {
151		return null;
152	3	}
153	3	$str = $this->stripEmailAdress($str);
154	3
155		$str = str_replace(
156	3	[
157		'\|',
158		"\n",
159		"\t",
160		"\r",
161		''',
162		''',
163		''',
164		''',
165		"\n",
166	3	" ",
167	2	"é",
168	3	'©',
169		'{{',
170	2	'}}',
171	2	'[[',
172		']]',
173		],
174		[
175		'/',
176		' ',
177		' ',
178	3	'',
179		"’",
180		"'",
181	3	"'",
182		"'",
183	3	'',
184	3	' ',
185		"é",
186		'',
187		'',
188		'',
189	3	'',
190		'',
191		],
192		$str
193		);
194
195		$str = html_entity_decode($str);
196		$str = strip_tags($str);
197
198	5	return trim($str);
199		}
200	5
201		public function stripEmailAdress(?string $str = null): ?string
202		{
203		if ($str === null) {
204		return null;
205	5	}
206
207		return preg_replace('# ?[^ ]+@[^ ]+\.[A-Z]+#i', '', $str);
208		}
209
210	5	protected function convertOGtype2format(?string $ogType): ?string
211		{
212		if (empty($ogType)) {
213		return null;
214		}
215		// og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
216		if (strpos($ogType, 'video') !== false) {
217	5	return 'vidéo';
218		}
219		if (strpos($ogType, 'book') !== false) {
220		return 'livre';
221		}
222
223		return null;
224		}
225
226		/**
227	5	* https://developers.facebook.com/docs/internationalization#locales
228		* @param string\|null $lang
229	5	*
230	3	* @return string\|null
231		*/
232		protected function convertLangue(?string $lang = null): ?string
233	4	{
234		if (empty($lang)) {
235		return null;
236	4	}
237	4	// en_GB
238	4	if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
239	4	return $matches[1];
240	4	}
241	4
242	4	return Language::all2wiki($lang);
243	4	}
244
245	4	protected function convertAuteur($data, $indice): ?string
246		{
247		// author=Bob
248		if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
249		return html_entity_decode($data['author']);
250		}
251
252		// author ['name'=>'Bob','@type'=>'Person']
253		if (0 === $indice
254		&& isset($data['author'])
255		&& isset($data['author']['name'])
256		&& (!isset($data['author']['@type'])
257		\|\| 'Person' === $data['author']['@type'])
258		) {
259		if (is_string($data['author']['name'])) {
260		return html_entity_decode($data['author']['name']);
261		}
262
263		return html_entity_decode($data['author']['name'][0]);
264		}
265
266		// author [ 0 => ['name'=>'Bob'], 1=> ...]
267		if (isset($data['author']) && isset($data['author'][$indice])
268		&& (!isset($data['author'][$indice]['@type'])
269		\|\| 'Person' === $data['author'][$indice]['@type'])
270		) {
271		if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
272		return html_entity_decode($data['author'][$indice]['name']);
273		}
274
275		// "author" => [ "@type" => "Person", "name" => [] ]
276		if (isset($data['author'][$indice]['name'][0])) {
277		return html_entity_decode($data['author'][$indice]['name'][0]);
278		}
279		}
280
281		return null;
282		}
283
284		protected function convertInstitutionnel($data): ?string
285		{
286		if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
287		&& 'Person' !== $data['author'][0]['@type']
288		) {
289		return html_entity_decode($data['author'][0]['name']);
290		}
291
292		return null;
293		}
294
295		/**
296		* todo move to generalize as utility
297		* @throws Exception
298		*/
299		protected function convertDate(?string $str): ?string
300		{
301		if (empty($str)) {
302		return null;
303		}
304		$str = str_replace(' 00:00:00', '', $str);
305		$str = str_replace('/', '-', $str);
306
307		// "2012"
308		if (preg_match('#^[12]\d{3}$#', $str)) {
309		return $str;
310		}
311		// "1775-1783" (Gallica)
312		if (preg_match('#^[12]\d{3}-[12]\d{3}$#', $str)) {
313		return $str;
314		}
315
316		try {
317		$date = new DateTime($str);
318		} catch (Exception $e) {
319		// 23/11/2015 00:00:00
320		if (isset($this->log) && method_exists($this->log, 'notice')) {
321		$this->log->notice('EXCEPTION DATE');
322		}
323
324		return '<!-- ' . $str . ' -->';
325		}
326
327		return $date->format('d-m-Y');
328		}
329
330		/**
331		* Wikification des noms/acronymes d'agences de presse.
332		* Note : utiliser APRES clean() et cleanAuthor() sinon bug "\|"
333		*/
334		protected function wikifyPressAgency(?string $str): ?string
335		{
336		if (empty($str)) {
337		return null;
338		}
339		// skip potential wikilinks
340		if (strpos($str, '[') !== false) {
341		return $str;
342		}
343		$str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse\|AFP]]', $str);
344		$str = str_replace('Reuters', '[[Reuters]]', $str);
345		$str = str_replace('Associated Press', '[[Associated Press]]', $str);
346		$str = preg_replace('#\b(PA)\b#', '[[Press Association\|PA]]', $str);
347		$str = preg_replace('#\b(AP)\b#', '[[Associated Press\|AP]]', $str);
348		$str = str_replace('Xinhua', '[[Xinhua]]', $str);
349		$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse\|ATS]]', $str);
350
351		return preg_replace('#\b(PC\|CP)\b#', '[[La Presse canadienne\|PC]]', $str);
352		}
353		}
354

Dispositif / Wikibot

Push — master ( dff8a4...2556d0 )

ExternConverterTrait::cleanSEOTitle() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like