ExternConverterTrait::convertInstitutionnel() - Code Metrics - Inspection of "Add ExternConverterTest::stripEmailAdress() + test" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( aa70d7...319840 )

by Dispositif

created 2020-08-16 15:20 UTC

ExternConverterTrait::convertInstitutionnel() A

↳ Parent: ExternConverterTrait

Complexity

Conditions	5
Paths	2

Size

Total Lines	9
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	5

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	5
eloc	4
nc	2
nop	1
dl	0
loc	9
ccs	3
cts	3
cp	1
crap	5
rs	9.6111
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;
    }

    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 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	protected function convertURLaccess($data): ?string
29		{
30		// NYT, Figaro
31	5	if (isset($data['isAccessibleForFree'])) {
32	2	return $data['isAccessibleForFree'] ? 'ouvert' : 'limité';
33		}
34	5	if (isset($data['DC.rights'])) {
35	1	return (in_array($data['DC.rights'], ['free', 'public domain'])) ? 'ouvert' : 'limité';
36		}
37	4	if (isset($data['og:article:content_tier'])) {
38	1	return ($data['og:article:content_tier'] === 'free') ? 'ouvert' : 'limité';
39		}
40
41	4	return null;
42		}
43
44		/**
45		* Réduit le nombre d'auteurs si > 3.
46		* En $modeEtAll=true vérification pour "et al.=oui".
47		* TODO : wikifyPressAgency()
48		*
49		* @param string\|null $authors
50		* @param bool $modeEtAl
51		*
52		* @return string\|null
53		*/
54	5	protected function authorsEtAl(?string $authors, $modeEtAl = false): ?string
55		{
56	5	if (empty($authors)) {
57	3	return null;
58		}
59		// conserve juste les 2 premiers auteurs TODO : refactor
60		// Bob, Martin ; Yul, Bar ; ... ; ...
61	2	if (preg_match('#([^;]+;[^;]+);[^;]+;.+#', $authors, $matches)) {
62	1	return ($modeEtAl) ? 'oui' : $matches[1];
63		}
64		// Bob Martin, Yul Bar, ..., ...,...
65	1	if (preg_match('#([^,]+,[^,]+),[^,]+,.+#', $authors, $matches)) {
66	1	return ($modeEtAl) ? 'oui' : $matches[1];
67		}
68
69		return ($modeEtAl) ? null : $authors;
70		}
71
72	5	protected function convertDCpage(array $meta): ?string
73		{
74	5	if (isset($meta['citation_firstpage'])) {
75	2	$page = $meta['citation_firstpage'];
76	2	if (isset($meta['citation_lastpage'])) {
77	1	$page .= '–'.$meta['citation_lastpage'];
78		}
79
80	2	return (string)$page;
81		}
82
83	3	return null;
84		}
85
86		public function cleanAuthor(?string $str = null): ?string
87	5	{
88		if ($str === null) {
89	5	return null;
90	5	}
91		$str = $this->clean($str);
92	5	// "https://www.facebook.com/search/top/?q=..."
93	5	if (preg_match('#^https?://.+#i', $str)) {
94		return null;
95	5	}
96		// "Par Bob"
97		if (preg_match('#^Par (.+)$#i', $str, $matches)) {
98		return $matches[1];
99		}
100
101		return $str;
102		}
103	5
104		/**
105		* Note : à appliquer AVANT wikification (sinon bug sur \| )
106	5	*
107		* @param string\|null $str
108		*
109	5	* @return string\|null
110		*/
111	5	public function clean(?string $str = null): ?string
112		{
113		if ($str === null) {
114		return null;
115	5	}
116		$str = $this->stripEmailAdress($str);
117
118	5	$str = str_replace(
119		['\|', "\n", "\t", ''', ''', ''', ''', "\n", " ", "é", '©'],
120		[
121		'/',
122	5	' ',
123		' ',
124		"’",
125		"'",
126		"'",
127		"'",
128		'',
129		' ',
130		"é",
131	5	'',
132		],
133	5	$str
134	2	);
135
136		return html_entity_decode($str);
137	3	}
138	2
139		public function stripEmailAdress(?string $str = null): ?string
140		{
141	1	if ($str === null) {
142		return null;
143		}
144	3
145		return preg_replace('# ?[^ ]+@[^ ]+\.[A-Z]+#i', '', $str);
146		}
147	3
148		protected function convertOGtype2format(?string $ogType): ?string
149		{
150		if (empty($ogType)) {
151		return null;
152	3	}
153	3	// og:type = default: website / video.movie / video.tv_show video.other / article, book, profile
154	3	if (strpos($ogType, 'video') !== false) {
155		return 'vidéo';
156	3	}
157		if (strpos($ogType, 'book') !== false) {
158		return 'livre';
159		}
160
161		return null;
162		}
163
164		/**
165		* https://developers.facebook.com/docs/internationalization#locales
166	3	* @param string\|null $lang
167	2	*
168	3	* @return string\|null
169		*/
170	2	protected function convertLangue(?string $lang = null): ?string
171	2	{
172		if (empty($lang)) {
173		return null;
174		}
175		// en_GB
176		if (preg_match('#^([a-z]{2})_[A-Z]{2}$#', $lang, $matches)) {
177		return $matches[1];
178	3	}
179
180		return Language::all2wiki($lang);
181	3	}
182
183	3	protected function convertAuteur($data, $indice)
184	3	{
185		// author=Bob
186		if (isset($data['author']) && is_string($data['author']) && $indice === 1) {
187		return html_entity_decode($data['author']);
188		}
189	3
190		// author ['name'=>'Bob','@type'=>'Person']
191		if (0 === $indice
192		&& isset($data['author'])
193		&& isset($data['author']['name'])
194		&& (!isset($data['author']['@type'])
195		\|\| 'Person' === $data['author']['@type'])
196		) {
197		if (is_string($data['author']['name'])) {
198	5	return html_entity_decode($data['author']['name']);
199		}
200	5
201		return html_entity_decode($data['author']['name'][0]);
202		}
203
204		// author [ 0 => ['name'=>'Bob'], 1=> ...]
205	5	if (isset($data['author']) && isset($data['author'][$indice])
206		&& (!isset($data['author'][$indice]['@type'])
207		\|\| 'Person' === $data['author'][$indice]['@type'])
208		) {
209		if (isset($data['author'][$indice]['name']) && is_string($data['author'][$indice]['name'])) {
210	5	return html_entity_decode($data['author'][$indice]['name']);
211		}
212
213		// "author" => [ "@type" => "Person", "name" => [] ]
214		if (isset($data['author'][$indice]['name'][0])) {
215		return html_entity_decode($data['author'][$indice]['name'][0]);
216		}
217	5	}
218
219		return null;
220		}
221
222		protected function convertInstitutionnel($data)
223		{
224		if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type'])
225		&& 'Person' !== $data['author'][0]['@type']
226		) {
227	5	return html_entity_decode($data['author'][0]['name']);
228		}
229	5
230	3	return null;
231		}
232
233	4	/**
234		* @param string $str
235		*
236	4	* @return string
237	4	* @throws Exception
238	4	*/
239	4	protected function convertDate(?string $str): ?string
240	4	{
241	4	if (empty($str)) {
242	4	return null;
243	4	}
244		$str = str_replace(' 00:00:00', '', $str);
245	4	$str = str_replace('/', '-', $str);
246
247		// "2012"
248		if (preg_match('#^[12][0-9]{3}$#', $str)) {
249		return $str;
250		}
251		// "1775-1783" (Gallica)
252		if (preg_match('#^[12][0-9]{3}-[12][0-9]{3}$#', $str)) {
253		return $str;
254		}
255
256		try {
257		$date = new DateTime($str);
258		} catch (Exception $e) {
259		// 23/11/2015 00:00:00
260		dump('EXCEPTION DATE');
261
262		return '<!-- '.$str.' -->';
263		}
264
265		return $date->format('d-m-Y');
266		}
267
268		/**
269		* Wikification des noms/acronymes d'agences de presse.
270		* Note : utiliser APRES clean() et cleanAuthor() sinon bug "\|"
271		*
272		* @param string $str
273		*
274		* @return string
275		*/
276		protected function wikifyPressAgency(?string $str): ?string
277		{
278		if (empty($str)) {
279		return null;
280		}
281		// skip potential wikilinks
282		if (strpos($str, '[') !== false) {
283		return $str;
284		}
285		$str = preg_replace('#\b(AFP)\b#i', '[[Agence France-Presse\|AFP]]', $str);
286		$str = str_replace('Reuters', '[[Reuters]]', $str);
287		$str = str_replace('Associated Press', '[[Associated Press]]', $str);
288		$str = preg_replace('#\b(PA)\b#', '[[Press Association\|PA]]', $str);
289		$str = preg_replace('#\b(AP)\b#', '[[Associated Press\|AP]]', $str);
290		$str = str_replace('Xinhua', '[[Xinhua]]', $str);
291		$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse\|ATS]]', $str);
292		$str = preg_replace('#\b(PC\|CP)\b#', '[[La Presse canadienne\|PC]]', $str);
293
294		return $str;
295		}
296
297		}
298

Dispositif / Wikibot

Push — master ( aa70d7...319840 )

ExternConverterTrait::convertInstitutionnel() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like