GoogleLivresTemplate::isTrackingUrl() - Code Metrics - Inspection of "Style + maxlag, ISBN_SKIP" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — dev ( eba0dc...b8ced8 )

by Dispositif

created 2020-04-04 13:13 UTC

GoogleLivresTemplate::isTrackingUrl() A

↳ Parent: GoogleLivresTemplate

Complexity

Conditions	3
Paths	3

Size

Total Lines	10
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	3
eloc	5
c	0
b	0
f	0
nc	3
nop	1
dl	0
loc	10
rs	10

<?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\.]{2,6}/(?:books)?(?:books/[^\?]+\.html)?(?:/reader)?\?(?:[a-zA-Z=&]+&)?id=';

    const TRACKING_PARAMETERS
        = [
            'xtor',
            'ved',
            'ots',
            'sig',
            'source',
            'utm_source',
            'utm_medium',
            'utm_campaign',
            'utm_term',
            'utm_content',
        ];
    /**
     * @var array
     */

    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);
    }

    public static function isTrackingUrl(string $url): bool
    {
        $data = self::parseGoogleBookQuery($url);
        foreach ($data as $param => $value) {
            if (in_array($param, self::TRACKING_PARAMETERS)) {
                return true;
            }
        }

        return false;
    }

    /**
     * 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', '.au'], ['.co.ma', '.co.uk', '.com.au'], $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\.]{2,6}/(?:books)?(?:books/[^\?]+\.html)?(?:/reader)?\?(?:[a-zA-Z=&]+&)?id=';
41
42			const TRACKING_PARAMETERS
43			= [
44			'xtor',
45			'ved',
46			'ots',
47			'sig',
48			'source',
49			'utm_source',
50			'utm_medium',
51			'utm_campaign',
52			'utm_term',
53			'utm_content',
54			];
55			/**
56			* @var array
57			*/
58
59			protected $parametersByOrder
60			= ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne'];
61
62			/**
63			* Create {Google Book} from URL.
64			* See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB
65			* https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d.
66			*
67			* @param string $url
68			*
69			* @return GoogleLivresTemplate\|null
70			* @throws Exception
71			*/
72			public static function createFromURL(string $url): ?self
73			{
74			if (!self::isGoogleBookURL($url)) {
75			throw new DomainException('not a Google Book URL');
76			}
77			$gooDat = self::parseGoogleBookQuery($url);
78
79			if (empty($gooDat['id'])) {
80			throw new DomainException("no GoogleBook 'id' in URL");
81			}
82			if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
83			throw new DomainException("GoogleBook 'id' malformed [0-9A-Za-z_\-]{12}");
84			}
85
86			$data = self::mapGooData($gooDat);
87
88			$templ = new self();
89			$templ->hydrate($data);
90
91			return $templ;
92			}
93
94			/**
95			* Check google URL pattern.
96			*
97			* @param string $text
98			*
99			* @return bool
100			*/
101			public static function isGoogleBookURL(string $text): bool
102			{
103			if (preg_match('#^'.self::GOOGLEBOOK_URL_PATTERN.'[^>\]} \n]+$#i', $text) > 0) {
104			return true;
105			}
106
107			return false;
108			}
109
110			/**
111			* Parse URL argument from ?query and #fragment.
112			*
113			* @param string $url
114			*
115			* @return array
116			*/
117			public static function parseGoogleBookQuery(string $url): array
118			{
119			// Note : Also datas in URL after the '#' !!! (URL fragment)
120			$queryData = parse_url($url, PHP_URL_QUERY); // after ?
121			$fragmentData = parse_url($url, PHP_URL_FRAGMENT); // after #
122			// queryData precedence over fragmentData
123			parse_str(implode('&', [$fragmentData, $queryData]), $val);
124
125			return self::arrayKeysToLower($val);
126			}
127
128			public static function isTrackingUrl(string $url): bool
129			{
130			$data = self::parseGoogleBookQuery($url);
131			foreach ($data as $param => $value) {
132			if (in_array($param, self::TRACKING_PARAMETERS)) {
133			return true;
134			}
135			}
136
137			return false;
138			}
139
140			/**
141			* Mapping Google URL data to {Google Livres} data.
142			*
143			* @param array $gooData
144			*
145			* @return array
146			*/
147			private static function mapGooData(array $gooData): array
148			{
149			$data = [];
150			$data['id'] = $gooData['id'];
151
152			// show cover ?
153			if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) {
154			$data['couv'] = '1';
155			}
156
157			// page number
158			if (!empty($gooData['pg'])) {
159			$data['page autre'] = $gooData['pg'];
160
161			// pg=PAx => "page=x"
162			if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) {
163			$data['page'] = $matches[1];
164			unset($data['page autre']);
165			}
166			// pg=PRx => "page=x\|romain=1"
167			if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) {
168			$data['page'] = $matches[1];
169			$data['romain'] = '1';
170			unset($data['page autre']);
171			}
172			}
173			// q : keywords search / dq : quoted phrase search
174			// affichage Google : dq ignoré si q existe
175			if (!empty($gooData['dq']) \|\| !empty($gooData['q'])) {
176			$data['surligne'] = $gooData['q'] ?? $gooData['dq']; // q prévaut
177			$data['surligne'] = self::googleUrlEncode($data['surligne']);
178			}
179
180			return $data;
181			}
182
183			/**
184			* Instead of url_encode(). No UTF-8 encoding.
185			*
186			* @param string $str
187			*
188			* @return string
189			*/
190			public static function googleUrlEncode(string $str): string
191			{
192			return str_replace(' ', '+', trim(urldecode($str)));
193			}
194
195			/**
196			* Clean the google book URL from optional&tracking data.
197			*
198			* @param string $url
199			*
200			* @return string URL
201			* @throws Exception
202			*/
203			public static function simplifyGoogleUrl(string $url): string
204			{
205			if (!self::isGoogleBookURL($url)) {
206			// not DomainException for live testing with OuvrageOptimize
207			throw new Exception('not a Google Book URL');
208			}
209
210			$gooDat = self::parseGoogleBookQuery($url);
211			if (empty($gooDat['id'])) {
212			throw new DomainException("no GoogleBook 'id' in URL");
213			}
214			if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) {
215			throw new DomainException("GoogleBook 'id' malformed");
216			}
217
218			// clean encoding q= dq=
219			// if(isset($gooDat['q'])) {
220			// $gooDat['q'] = self::googleUrlEncode($gooDat['q']);
221			// }
222			// if(isset($gooDat['dq'])) {
223			// $gooDat['dq'] = self::googleUrlEncode($gooDat['dq']);
224			// }
225
226			$dat = [];
227			// keep only a few parameters (+'q' ?)
228			// q : keywords search / dq : quoted phrase search
229			$keeps = ['id', 'pg', 'printsec', 'q', 'dq'];
230			foreach ($keeps as $keep) {
231			if (!empty($gooDat[$keep])) {
232			$dat[$keep] = $gooDat[$keep];
233			}
234			}
235
236			// 1 exemple : https://fr.wikipedia.org/w/index.php?title=Foudre_de_Catatumbo&diff=next&oldid=168721836&diffmode=source
237			// 1. mettre URL &dq= pour final
238			//
239			// 2. si q!=dq (changement ultérieur formulaire recherche) alors q= prévaut pour résultat final
240			// 2. mettre URL &q pour final
241			//
242			// 3. Recherche global sur http://books.google.fr => pg= dq= (#q= avec q==dq)
243			// 3. dans ce cas (q==dq), url final avec seulement dq= donne résultat OK
244			//
245			// 4 . if you use a url without &redir_esc=y#v=onepage for a book with "Preview" available,
246			// usually &dq shows the highlighted text in full page view whereas &q shows the snippet view (so you have to
247			// click on the snippet to see the full page).
248			// &dq allows highlighting in books where there is "Preview" available and &pg=PTx is in the URL
249			//
250			// #v=onepage ou #v=snippet
251			if (isset($dat['q']) && isset($dat['dq'])) {
252			// si q==dq alors dq prévaut pour affichage (sinon affichage différent avec url seulement q=)
253			if ($dat['q'] === $dat['dq']) {
254			unset($dat['q']);
255			} // si q!=dq (exemple : nouveaux mots clés dans formulaire recherche) alors q= prévaut pour résultat final
256			else {
257			unset($dat['dq']);
258			}
259			}
260
261			$googleURL = self::DEFAULT_GOOGLEBOOK_URL;
262
263			// domain .com .fr
264			$gooDomain = self::parseGoogleDomain($url);
265			if ($gooDomain) {
266			$googleURL = str_replace('.com', $gooDomain, $googleURL);
267			}
268
269			// todo http_build_query process an urlencode, but a not encoded q= value ("fu+bar") is beautiful
270			return $googleURL.'?'.http_build_query($dat);
271			}
272
273			/**
274			* return '.fr' or '.com'.
275			*
276			* @param string $url
277			*
278			* @return string\|null
279			*/
280			private static function parseGoogleDomain(string $url): ?string
281			{
282			$host = parse_url($url, PHP_URL_HOST);
283			if (!empty($host) && preg_match('#\.[a-z]{2,3}$#', $host, $matches) > 0) {
284			// Maroc : google.co.ma (sous-domaine!!)
285			return str_replace(['.ma', '.uk', '.au'], ['.co.ma', '.co.uk', '.com.au'], $matches[0]); // .fr
286			}
287
288			return null;
289			}
290
291			/**
292			* Check if Google URL or wiki {Google Books} template.
293			*
294			* @param string $text
295			*
296			* @return bool
297			*/
298			public static function isGoogleBookValue(string $text): bool
299			{
300			if (true === self::isGoogleBookURL($text)) {
301			return true;
302			}
303			if (preg_match('#^{{[ \n]*Google (Livres\|Books)[^}]+}}$#i', $text) > 0) {
304			return true;
305			}
306
307			return false;
308			}
309
310			/**
311			* Serialize the wiki-template.
312			* Improvement : force param order : id/titre/...
313			*
314			* @param bool\|null $cleanOrder
315			*
316			* @return string
317			*/
318			public function serialize(?bool $cleanOrder = true): string
319			{
320			$text = parent::serialize();
321
322			// Documentation suggère non affichage de ces 2 paramètres
323			return str_replace(['id=', 'titre='], '', $text);
324			}
325			}
326

Dispositif / Wikibot

Push — dev ( eba0dc...b8ced8 )

GoogleLivresTemplate::isTrackingUrl() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like