GoogleLivresTemplate::simplifyGoogleUrl() - Code Metrics - Inspection of "Fix LanguageData 'turc' => 'tr'" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — dev ( 4ef148...eba0dc )

by Dispositif

created 2020-03-29 15:26 UTC

GoogleLivresTemplate::simplifyGoogleUrl() B

↳ Parent: GoogleLivresTemplate

Complexity

Conditions	10
Paths	21

Size

Total Lines	68
Code Lines	22

Duplication

Lines	0
Ratio	0 %

Importance

Changes	2
Bugs	0	Features	0

Metric	Value
cc	10
eloc	22
c	2
b	0
f	0
nc	21
nop	1
dl	0
loc	68
rs	7.6666

How to fix Long Method Complexity

<?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\Models\Wiki;

use App\Domain\Utils\ArrayProcessTrait;
use DomainException;
use Exception;

/**
 * https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres
 * Le premier paramètre (ou id) est obligatoire.
 * Le deuxième (ou titre) est requis si on ne veut pas fabriquer le lien brut (inclusion {{ouvrage}} 'Lire en ligne')
 * Class GoogleLivresTemplate.
 */
class GoogleLivresTemplate extends AbstractWikiTemplate
{
    use ArrayProcessTrait;

    const DEFAULT_GOOGLEBOOK_URL = 'https://books.google.com/books';

    const MODEL_NAME = 'Google Livres';

    const REQUIRED_PARAMETERS = ['id' => ''];

    const PARAM_ALIAS
        = [
            '1' => 'id',
            '2' => 'titre',
            'surligné' => 'surligne',
            'BuchID' => 'id',
        ];

    const GOOGLEBOOK_URL_PATTERN = 'https?://(?:books|play)\.google\.[a-z\.]{2,5}/(?:books)?(?:books/[^\?]+\.html)?(?:/reader)?\?(?:[a-zA-Z=&]+&)?id=';

    protected $parametersByOrder
        = ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne'];

    /**
     * Create {Google Book} from URL.
     * See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB
     * https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d.
     *
     * @param string $url
     *
     * @return GoogleLivresTemplate|null
     * @throws Exception
     */
    public static function createFromURL(string $url): ?self
    {
        if (!self::isGoogleBookURL($url)) {
            throw new DomainException('not a Google Book URL');
        }
        $gooDat = self::parseGoogleBookQuery($url);

        if (empty($gooDat['id'])) {
            throw new DomainException("no GoogleBook 'id' in URL");
        }
        if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
            throw new DomainException("GoogleBook 'id' malformed [0-9A-Za-z_\-]{12}");
        }

        $data = self::mapGooData($gooDat);

        $templ = new self();
        $templ->hydrate($data);

        return $templ;
    }

    /**
     * Check google URL pattern.
     *
     * @param string $text
     *
     * @return bool
     */
    public static function isGoogleBookURL(string $text): bool
    {
        if (preg_match('#^'.self::GOOGLEBOOK_URL_PATTERN.'[^>\]} \n]+$#i', $text) > 0) {
            return true;
        }

        return false;
    }

    /**
     * Parse URL argument from ?query and #fragment.
     *
     * @param string $url
     *
     * @return array
     */
    public static function parseGoogleBookQuery(string $url): array
    {
        // Note : Also datas in URL after the '#' !!! (URL fragment)
        $queryData = parse_url($url, PHP_URL_QUERY); // after ?
        $fragmentData = parse_url($url, PHP_URL_FRAGMENT); // after #
        // queryData precedence over fragmentData
        parse_str(implode('&', [$fragmentData, $queryData]), $val);

        return self::arrayKeysToLower($val);
    }

    /**
     * Mapping Google URL data to {Google Livres} data.
     *
     * @param array $gooData
     *
     * @return array
     */
    private static function mapGooData(array $gooData): array
    {
        $data = [];
        $data['id'] = $gooData['id'];

        // show cover ?
        if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) {
            $data['couv'] = '1';
        }

        // page number
        if (!empty($gooData['pg'])) {
            $data['page autre'] = $gooData['pg'];

            //  pg=PAx => "page=x"
            if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) {
                $data['page'] = $matches[1];
                unset($data['page autre']);
            }
            //  pg=PRx => "page=x|romain=1"
            if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) {
                $data['page'] = $matches[1];
                $data['romain'] = '1';
                unset($data['page autre']);
            }
        }
        // q : keywords search / dq : quoted phrase search
        // affichage Google : dq ignoré si q existe
        if (!empty($gooData['dq']) || !empty($gooData['q'])) {
            $data['surligne'] = $gooData['q'] ?? $gooData['dq']; // q prévaut
            $data['surligne'] = self::googleUrlEncode($data['surligne']);
        }

        return $data;
    }

    /**
     * Instead of url_encode(). No UTF-8 encoding.
     *
     * @param string $str
     *
     * @return string
     */
    public static function googleUrlEncode(string $str): string
    {
        return str_replace(' ', '+', trim(urldecode($str)));
    }

    /**
     * Clean the google book URL from optional&tracking data.
     *
     * @param string $url
     *
     * @return string URL
     * @throws Exception
     */
    public static function simplifyGoogleUrl(string $url): string
    {
        if (!self::isGoogleBookURL($url)) {
            // not DomainException for live testing with OuvrageOptimize
            throw new Exception('not a Google Book URL');
        }

        $gooDat = self::parseGoogleBookQuery($url);
        if (empty($gooDat['id'])) {
            throw new DomainException("no GoogleBook 'id' in URL");
        }
        if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
            throw new DomainException("GoogleBook 'id' malformed");
        }

        // clean encoding q= dq=
        //                if(isset($gooDat['q'])) {
        //                    $gooDat['q'] = self::googleUrlEncode($gooDat['q']);
        //                }
        //                if(isset($gooDat['dq'])) {
        //                    $gooDat['dq'] = self::googleUrlEncode($gooDat['dq']);
        //                }

        $dat = [];
        // keep only a few parameters (+'q' ?)
        // q : keywords search / dq : quoted phrase search
        $keeps = ['id', 'pg', 'printsec', 'q', 'dq'];
        foreach ($keeps as $keep) {
            if (!empty($gooDat[$keep])) {
                $dat[$keep] = $gooDat[$keep];
            }
        }

        // 1 exemple : https://fr.wikipedia.org/w/index.php?title=Foudre_de_Catatumbo&diff=next&oldid=168721836&diffmode=source
        // 1. mettre URL &dq= pour final
        //
        // 2. si q!=dq (changement ultérieur formulaire recherche) alors q= prévaut pour résultat final
        // 2. mettre URL &q pour final
        //
        // 3. Recherche global sur http://books.google.fr => pg= dq= (#q= avec q==dq)
        // 3. dans ce cas (q==dq), url final avec seulement dq= donne résultat OK
        //
        // 4 . if you use a url without &redir_esc=y#v=onepage for a book with "Preview" available,
        // usually &dq shows the highlighted text in full page view whereas &q shows the snippet view (so you have to
        // click on the snippet to see the full page).
        // &dq allows highlighting in books where there is "Preview" available and &pg=PTx is in the URL
        //
        // #v=onepage ou #v=snippet
        if (isset($dat['q']) && isset($dat['dq'])) {
            // si q==dq alors dq prévaut pour affichage (sinon affichage différent avec url seulement q=)
            if ($dat['q'] === $dat['dq']) {
                unset($dat['q']);
            } // si q!=dq (exemple : nouveaux mots clés dans formulaire recherche) alors q= prévaut pour résultat final
            else {
                unset($dat['dq']);
            }
        }

        $googleURL = self::DEFAULT_GOOGLEBOOK_URL;

        // domain .com .fr
        $gooDomain = self::parseGoogleDomain($url);
        if ($gooDomain) {
            $googleURL = str_replace('.com', $gooDomain, $googleURL);
        }

        // todo http_build_query process an urlencode, but a not encoded q= value ("fu+bar") is beautiful
        return $googleURL.'?'.http_build_query($dat);
    }

    /**
     * return '.fr' or '.com'.
     *
     * @param string $url
     *
     * @return string|null
     */
    private static function parseGoogleDomain(string $url): ?string
    {
        $host = parse_url($url, PHP_URL_HOST);
        if (!empty($host) && preg_match('#\.[a-z]{2,3}$#', $host, $matches) > 0) {
            // Maroc : google.co.ma (sous-domaine!!)
            return str_replace(['.ma', '.uk'], ['.co.ma', '.co.uk'], $matches[0]); // .fr
        }

        return null;
    }

    /**
     * Check if Google URL or wiki {Google Books} template.
     *
     * @param string $text
     *
     * @return bool
     */
    public static function isGoogleBookValue(string $text): bool
    {
        if (true === self::isGoogleBookURL($text)) {
            return true;
        }
        if (preg_match('#^{{[ \n]*Google (Livres|Books)[^}]+}}$#i', $text) > 0) {
            return true;
        }

        return false;
    }

    /**
     * Serialize the wiki-template.
     * Improvement : force param order : id/titre/...
     *
     * @param bool|null $cleanOrder
     *
     * @return string
     */
    public function serialize(?bool $cleanOrder = true): string
    {
        $text = parent::serialize();

        // Documentation suggère non affichage de ces 2 paramètres
        return str_replace(['id=', 'titre='], '', $text);
    }
}


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\Models\Wiki;
11
12			use App\Domain\Utils\ArrayProcessTrait;
13			use DomainException;
14			use Exception;
15
16			/**
17			* https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres
18			* Le premier paramètre (ou id) est obligatoire.
19			* Le deuxième (ou titre) est requis si on ne veut pas fabriquer le lien brut (inclusion {{ouvrage}} 'Lire en ligne')
20			* Class GoogleLivresTemplate.
21			*/
22			class GoogleLivresTemplate extends AbstractWikiTemplate
23			{
24			use ArrayProcessTrait;
25
26			const DEFAULT_GOOGLEBOOK_URL = 'https://books.google.com/books';
27
28			const MODEL_NAME = 'Google Livres';
29
30			const REQUIRED_PARAMETERS = ['id' => ''];
31
32			const PARAM_ALIAS
33			= [
34			'1' => 'id',
35			'2' => 'titre',
36			'surligné' => 'surligne',
37			'BuchID' => 'id',
38			];
39
40			const GOOGLEBOOK_URL_PATTERN = 'https?://(?:books\|play)\.google\.[a-z\.]{2,5}/(?:books)?(?:books/[^\?]+\.html)?(?:/reader)?\?(?:[a-zA-Z=&]+&)?id=';
41
42			protected $parametersByOrder
43			= ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne'];
44
45			/**
46			* Create {Google Book} from URL.
47			* See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB
48			* https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d.
49			*
50			* @param string $url
51			*
52			* @return GoogleLivresTemplate\|null
53			* @throws Exception
54			*/
55			public static function createFromURL(string $url): ?self
56			{
57			if (!self::isGoogleBookURL($url)) {
58			throw new DomainException('not a Google Book URL');
59			}
60			$gooDat = self::parseGoogleBookQuery($url);
61
62			if (empty($gooDat['id'])) {
63			throw new DomainException("no GoogleBook 'id' in URL");
64			}
65			if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
66			throw new DomainException("GoogleBook 'id' malformed [0-9A-Za-z_\-]{12}");
67			}
68
69			$data = self::mapGooData($gooDat);
70
71			$templ = new self();
72			$templ->hydrate($data);
73
74			return $templ;
75			}
76
77			/**
78			* Check google URL pattern.
79			*
80			* @param string $text
81			*
82			* @return bool
83			*/
84			public static function isGoogleBookURL(string $text): bool
85			{
86			if (preg_match('#^'.self::GOOGLEBOOK_URL_PATTERN.'[^>\]} \n]+$#i', $text) > 0) {
87			return true;
88			}
89
90			return false;
91			}
92
93			/**
94			* Parse URL argument from ?query and #fragment.
95			*
96			* @param string $url
97			*
98			* @return array
99			*/
100			public static function parseGoogleBookQuery(string $url): array
101			{
102			// Note : Also datas in URL after the '#' !!! (URL fragment)
103			$queryData = parse_url($url, PHP_URL_QUERY); // after ?
104			$fragmentData = parse_url($url, PHP_URL_FRAGMENT); // after #
105			// queryData precedence over fragmentData
106			parse_str(implode('&', [$fragmentData, $queryData]), $val);
107
108			return self::arrayKeysToLower($val);
109			}
110
111			/**
112			* Mapping Google URL data to {Google Livres} data.
113			*
114			* @param array $gooData
115			*
116			* @return array
117			*/
118			private static function mapGooData(array $gooData): array
119			{
120			$data = [];
121			$data['id'] = $gooData['id'];
122
123			// show cover ?
124			if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) {
125			$data['couv'] = '1';
126			}
127
128			// page number
129			if (!empty($gooData['pg'])) {
130			$data['page autre'] = $gooData['pg'];
131
132			// pg=PAx => "page=x"
133			if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) {
134			$data['page'] = $matches[1];
135			unset($data['page autre']);
136			}
137			// pg=PRx => "page=x\|romain=1"
138			if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) {
139			$data['page'] = $matches[1];
140			$data['romain'] = '1';
141			unset($data['page autre']);
142			}
143			}
144			// q : keywords search / dq : quoted phrase search
145			// affichage Google : dq ignoré si q existe
146			if (!empty($gooData['dq']) \|\| !empty($gooData['q'])) {
147			$data['surligne'] = $gooData['q'] ?? $gooData['dq']; // q prévaut
148			$data['surligne'] = self::googleUrlEncode($data['surligne']);
149			}
150
151			return $data;
152			}
153
154			/**
155			* Instead of url_encode(). No UTF-8 encoding.
156			*
157			* @param string $str
158			*
159			* @return string
160			*/
161			public static function googleUrlEncode(string $str): string
162			{
163			return str_replace(' ', '+', trim(urldecode($str)));
164			}
165
166			/**
167			* Clean the google book URL from optional&tracking data.
168			*
169			* @param string $url
170			*
171			* @return string URL
172			* @throws Exception
173			*/
174			public static function simplifyGoogleUrl(string $url): string
175			{
176			if (!self::isGoogleBookURL($url)) {
177			// not DomainException for live testing with OuvrageOptimize
178			throw new Exception('not a Google Book URL');
179			}
180
181			$gooDat = self::parseGoogleBookQuery($url);
182			if (empty($gooDat['id'])) {
183			throw new DomainException("no GoogleBook 'id' in URL");
184			}
185			if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
186			throw new DomainException("GoogleBook 'id' malformed");
187			}
188
189			// clean encoding q= dq=
190			// if(isset($gooDat['q'])) {
191			// $gooDat['q'] = self::googleUrlEncode($gooDat['q']);
192			// }
193			// if(isset($gooDat['dq'])) {
194			// $gooDat['dq'] = self::googleUrlEncode($gooDat['dq']);
195			// }
196
197			$dat = [];
198			// keep only a few parameters (+'q' ?)
199			// q : keywords search / dq : quoted phrase search
200			$keeps = ['id', 'pg', 'printsec', 'q', 'dq'];
201			foreach ($keeps as $keep) {
202			if (!empty($gooDat[$keep])) {
203			$dat[$keep] = $gooDat[$keep];
204			}
205			}
206
207			// 1 exemple : https://fr.wikipedia.org/w/index.php?title=Foudre_de_Catatumbo&diff=next&oldid=168721836&diffmode=source
208			// 1. mettre URL &dq= pour final
209			//
210			// 2. si q!=dq (changement ultérieur formulaire recherche) alors q= prévaut pour résultat final
211			// 2. mettre URL &q pour final
212			//
213			// 3. Recherche global sur http://books.google.fr => pg= dq= (#q= avec q==dq)
214			// 3. dans ce cas (q==dq), url final avec seulement dq= donne résultat OK
215			//
216			// 4 . if you use a url without &redir_esc=y#v=onepage for a book with "Preview" available,
217			// usually &dq shows the highlighted text in full page view whereas &q shows the snippet view (so you have to
218			// click on the snippet to see the full page).
219			// &dq allows highlighting in books where there is "Preview" available and &pg=PTx is in the URL
220			//
221			// #v=onepage ou #v=snippet
222			if (isset($dat['q']) && isset($dat['dq'])) {
223			// si q==dq alors dq prévaut pour affichage (sinon affichage différent avec url seulement q=)
224			if ($dat['q'] === $dat['dq']) {
225			unset($dat['q']);
226			} // si q!=dq (exemple : nouveaux mots clés dans formulaire recherche) alors q= prévaut pour résultat final
227			else {
228			unset($dat['dq']);
229			}
230			}
231
232			$googleURL = self::DEFAULT_GOOGLEBOOK_URL;
233
234			// domain .com .fr
235			$gooDomain = self::parseGoogleDomain($url);
236			if ($gooDomain) {
237			$googleURL = str_replace('.com', $gooDomain, $googleURL);
238			}
239
240			// todo http_build_query process an urlencode, but a not encoded q= value ("fu+bar") is beautiful
241			return $googleURL.'?'.http_build_query($dat);
242			}
243
244			/**
245			* return '.fr' or '.com'.
246			*
247			* @param string $url
248			*
249			* @return string\|null
250			*/
251			private static function parseGoogleDomain(string $url): ?string
252			{
253			$host = parse_url($url, PHP_URL_HOST);
254			if (!empty($host) && preg_match('#\.[a-z]{2,3}$#', $host, $matches) > 0) {
255			// Maroc : google.co.ma (sous-domaine!!)
256			return str_replace(['.ma', '.uk'], ['.co.ma', '.co.uk'], $matches[0]); // .fr
257			}
258
259			return null;
260			}
261
262			/**
263			* Check if Google URL or wiki {Google Books} template.
264			*
265			* @param string $text
266			*
267			* @return bool
268			*/
269			public static function isGoogleBookValue(string $text): bool
270			{
271			if (true === self::isGoogleBookURL($text)) {
272			return true;
273			}
274			if (preg_match('#^{{[ \n]*Google (Livres\|Books)[^}]+}}$#i', $text) > 0) {
275			return true;
276			}
277
278			return false;
279			}
280
281			/**
282			* Serialize the wiki-template.
283			* Improvement : force param order : id/titre/...
284			*
285			* @param bool\|null $cleanOrder
286			*
287			* @return string
288			*/
289			public function serialize(?bool $cleanOrder = true): string
290			{
291			$text = parent::serialize();
292
293			// Documentation suggère non affichage de ces 2 paramètres
294			return str_replace(['id=', 'titre='], '', $text);
295			}
296			}
297

Dispositif / Wikibot

Push — dev ( 4ef148...eba0dc )

GoogleLivresTemplate::simplifyGoogleUrl() B

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like