GoogleLivresTemplate::createFromURL() - Code Metrics - Inspection of "Doc" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( a19dc4...d676b1 )

by Dispositif

created 2019-12-31 14:54 UTC

GoogleLivresTemplate::createFromURL() A

↳ Parent: GoogleLivresTemplate

Complexity

Conditions	3
Paths	3

Size

Total Lines	17
Code Lines	9

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	3
eloc	9
c	1
b	0
f	0
nc	3
nop	1
dl	0
loc	17
rs	9.9666

<?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 DomainException;
use Exception;

/**
 * https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres
 * Le premier paramètre (ou id) est obligatoire. L
 * 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
{
    const DEFAULT_GOOGLEBOOK_URL = 'https://books.google.com/books';

    const ALLOW_USER_ORDER = false;

    const MODEL_NAME = 'Google Livres';

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

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

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

    /**
     * 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);
    }

    /**
     * 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");
        }

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

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

        return $templ;
    }

    /**
     * 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']);
            }
        }

        if (!empty($gooData['dq']) || !empty($gooData['q'])) {
            $data['surligne'] = $gooData['dq'] ?? $gooData['q'];
            $data['surligne'] = self::googleUrlEncode($data['surligne']);
        }

        return $data;
    }

    /**
     * Instead of url_encode(). No UTF-8 encoding.
     */
    private 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
     */
    public static function simplifyGoogleUrl(string $url): string
    {
        if (!self::isGoogleBookURL($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");
        }

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

        $googleURL = self::DEFAULT_GOOGLEBOOK_URL;

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

        return $googleURL.'?'.http_build_query($dat);
    }

    /**
     * Parse URL argument from ?query and #fragment.
     *
     * @param string $url
     *
     * @return array
     */
    private 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 $val;
    }

    /**
     * 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) {
            return $matches[0]; // .fr
        }

        return null;
    }

    /**
     * Check google URL pattern.
     *
     * @param string $text
     *
     * @return bool
     */
    public static function isGoogleBookURL(string $text): bool
    {
        if (preg_match('#^https?://(books|play)\.google\.[a-z]{2,3}/(books)?(/reader)?\?id=#i', $text) > 0) {
            return true;
        }

        return false;
    }

    /**
     * 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;
    }
}


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 DomainException;
13			use Exception;
14
15			/**
16			* https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres
17			* Le premier paramètre (ou id) est obligatoire. L
18			* Le deuxième (ou titre) est requis si on ne veut pas fabriquer le lien brut (inclusion {{ouvrage}} 'Lire en ligne')
19			* Class GoogleLivresTemplate.
20			*/
21			class GoogleLivresTemplate extends AbstractWikiTemplate
22			{
23			const DEFAULT_GOOGLEBOOK_URL = 'https://books.google.com/books';
24
25			const ALLOW_USER_ORDER = false;
26
27			const MODEL_NAME = 'Google Livres';
28
29			const REQUIRED_PARAMETERS = ['id' => ''];
30
31			const PARAM_ALIAS
32			= [
33			'1' => 'id',
34			'2' => 'titre',
35			'surligné' => 'surligne',
36			'BuchID' => 'id',
37			];
38
39			protected $parametersByOrder
40			= ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne'];
41
42			/**
43			* Serialize the wiki-template.
44			* Improvement : force param order : id/titre/...
45			*
46			* @param bool\|null $cleanOrder
47			*
48			* @return string
49			*/
50			public function serialize(?bool $cleanOrder = true): string
51			{
52			$text = parent::serialize();
53
54			// Documentation suggère non affichage de ces 2 paramètres
55			return str_replace(['id=', 'titre='], '', $text);
56			}
57
58			/**
59			* Create {Google Book} from URL.
60			* See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB
61			* https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d.
62			*
63			* @param string $url
64			*
65			* @return GoogleLivresTemplate\|null
66			* @throws Exception
67			*/
68			public static function createFromURL(string $url): ?self
69			{
70			if (!self::isGoogleBookURL($url)) {
71			throw new DomainException('not a Google Book URL');
72			}
73			$gooDat = self::parseGoogleBookQuery($url);
74
75			if (empty($gooDat['id'])) {
76			throw new DomainException("no GoogleBook 'id' in URL");
77			}
78
79			$data = self::mapGooData($gooDat);
80
81			$templ = new self();
82			$templ->hydrate($data);
83
84			return $templ;
85			}
86
87			/**
88			* Mapping Google URL data to {Google Livres} data.
89			*
90			* @param array $gooData
91			*
92			* @return array
93			*/
94			private static function mapGooData(array $gooData): array
95			{
96			$data = [];
97			$data['id'] = $gooData['id'];
98
99			// show cover ?
100			if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) {
101			$data['couv'] = '1';
102			}
103
104			// page number
105			if (!empty($gooData['pg'])) {
106			$data['page autre'] = $gooData['pg'];
107
108			// pg=PAx => "page=x"
109			if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) {
110			$data['page'] = $matches[1];
111			unset($data['page autre']);
112			}
113			// pg=PRx => "page=x\|romain=1"
114			if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) {
115			$data['page'] = $matches[1];
116			$data['romain'] = '1';
117			unset($data['page autre']);
118			}
119			}
120
121			if (!empty($gooData['dq']) \|\| !empty($gooData['q'])) {
122			$data['surligne'] = $gooData['dq'] ?? $gooData['q'];
123			$data['surligne'] = self::googleUrlEncode($data['surligne']);
124			}
125
126			return $data;
127			}
128
129			/**
130			* Instead of url_encode(). No UTF-8 encoding.
131			*/
132			private static function googleUrlEncode(string $str): string
133			{
134			return str_replace(' ', '+', trim(urldecode($str)));
135			}
136
137			/**
138			* Clean the google book URL from optional&tracking data.
139			*
140			* @param string $url
141			*
142			* @return string URL
143			*/
144			public static function simplifyGoogleUrl(string $url): string
145			{
146			if (!self::isGoogleBookURL($url)) {
147			throw new DomainException('not a Google Book URL');
148			}
149
150			$gooDat = self::parseGoogleBookQuery($url);
151			if (empty($gooDat['id'])) {
152			throw new DomainException("no GoogleBook 'id' in URL");
153			}
154
155			$dat = [];
156			// keep only a few parameters (+'q' ?)
157			$keeps = ['id', 'pg', 'printsec', 'dq'];
158			foreach ($keeps as $keep) {
159			if (!empty($gooDat[$keep])) {
160			$dat[$keep] = $gooDat[$keep];
161			}
162			}
163
164			$googleURL = self::DEFAULT_GOOGLEBOOK_URL;
165
166			// domain .com .fr
167			$gooDomain = self::parseGoogleDomain($url);
168			if ($gooDomain) {
169			$googleURL = str_replace('.com', $gooDomain, $googleURL);
170			}
171
172			return $googleURL.'?'.http_build_query($dat);
173			}
174
175			/**
176			* Parse URL argument from ?query and #fragment.
177			*
178			* @param string $url
179			*
180			* @return array
181			*/
182			private static function parseGoogleBookQuery(string $url): array
183			{
184			// Note : Also datas in URL after the '#' !!! (URL fragment)
185			$queryData = parse_url($url, PHP_URL_QUERY); // after ?
186			$fragmentData = parse_url($url, PHP_URL_FRAGMENT); // after #
187			// queryData precedence over fragmentData
188			parse_str(implode('&', [$fragmentData, $queryData]), $val);
189
190			return $val;
191			}
192
193			/**
194			* return '.fr' or '.com'.
195			*
196			* @param string $url
197			*
198			* @return string\|null
199			*/
200			private static function parseGoogleDomain(string $url): ?string
201			{
202			$host = parse_url($url, PHP_URL_HOST);
203			if (!empty($host) && preg_match('#\.[a-z]{2,3}$#', $host, $matches) > 0) {
204			return $matches[0]; // .fr
205			}
206
207			return null;
208			}
209
210			/**
211			* Check google URL pattern.
212			*
213			* @param string $text
214			*
215			* @return bool
216			*/
217			public static function isGoogleBookURL(string $text): bool
218			{
219			if (preg_match('#^https?://(books\|play)\.google\.[a-z]{2,3}/(books)?(/reader)?\?id=#i', $text) > 0) {
220			return true;
221			}
222
223			return false;
224			}
225
226			/**
227			* Check if Google URL or wiki {Google Books} template.
228			*
229			* @param string $text
230			*
231			* @return bool
232			*/
233			public static function isGoogleBookValue(string $text): bool
234			{
235			if (true === self::isGoogleBookURL($text)) {
236			return true;
237			}
238			if (preg_match('#^{{[ \n]*Google (Livres\|Books)[^}]+}}$#i', $text) > 0) {
239			return true;
240			}
241
242			return false;
243			}
244			}
245

Dispositif / Wikibot

Push — master ( a19dc4...d676b1 )

GoogleLivresTemplate::createFromURL() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like