RefGoogleBook::convertGBurl2OuvrageCitation() - Code Metrics - Inspection of "Refac (dependency)" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — dev ( 043eb4...bf3609 )

by Dispositif

created 2020-03-24 10:28 UTC

RefGoogleBook::convertGBurl2OuvrageCitation() C

↳ Parent: RefGoogleBook

Complexity

Conditions	13
Paths	24

Size

Total Lines	66
Code Lines	35

Duplication

Lines	0
Ratio	0 %

Importance

Changes	3
Bugs	0	Features	0

Metric	Value
cc	13
eloc	35
c	3
b	0
f	0
nc	24
nop	1
dl	0
loc	66
rs	6.6166

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;

use App\Domain\Models\Wiki\GoogleLivresTemplate;
use App\Domain\Models\Wiki\OuvrageTemplate;
use App\Domain\Publisher\GoogleBookMapper;
use App\Domain\Utils\NumberUtil;
use App\Domain\Utils\WikiTextUtil;
use App\Infrastructure\GoogleBooksAdapter;

/**
 * Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage|...}}.</ref>
 * in an article wikitext.
 * Class RefGoogleBook
 *
 * @package App\Domain
 */
class RefGoogleBook
{
    const SLEEP_GOOGLE_API_INTERVAL = 8;

    /**
     * @var array OuvrageTemplate[]
     */
    private $cacheOuvrageTemplate = [];

    /**
     * RefGoogleBook constructor.
     * todo dependency injection
     */
    public function __construct() { }

    /**
     * Process page wikitext. Return wikitext with the <ref> converted.
     *
     * @param string $text Page wikitext
     *
     * @return string New wikitext
     * @throws \Throwable
     */
    public function process(string $text): string
    {
        $refsData = $this->extractAllGoogleRefs($text);
        if (empty($refsData)) {
            echo "Pas d'URL GB trouvée\n";

            return $text;
        }

        foreach ($refsData as $ref) {
            try {
                $citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1]));
            } catch (\Exception $e) {
                echo "Exception ".$e->getMessage();
                continue;
            }

            // ajout point final pour référence
            $citation .= '.';

            $newRef = str_replace($ref[1], $citation, $ref[0]);
            echo $newRef."\n";

            $text = str_replace($ref[0], $newRef, $text);

            echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
            sleep(self::SLEEP_GOOGLE_API_INTERVAL);
        }

        return $text;
    }

    /**
     * TODO : extract
     * Convert GoogleBooks URL to wiki-template {ouvrage} citation.
     *
     * @param string $url GoogleBooks URL
     *
     * @return string {{ouvrage}}
     * @throws \Exception
     * @throws \Throwable
     */
    public function convertGBurl2OuvrageCitation(string $url): string
    {
        if (!GoogleLivresTemplate::isGoogleBookURL($url)) {
            throw new \DomainException('Pas de URL Google Books');
        }

        $gooDat = GoogleLivresTemplate::parseGoogleBookQuery($url);
        if (empty($gooDat['id'])) {
            throw new \DomainException('Pas de ID Google Books');
        }

        try {
            $ouvrage = $this->generateOuvrageFromGoogleData($gooDat['id']);
        } catch (\Throwable $e) {
            // ID n'existe pas sur Google Books
            if (strpos($e->getMessage(), '404 Not Found')
                && strpos($e->getMessage(), '"message": "The volume ID could n')
            ) {
                return sprintf(
                    '{{lien brisé |url= %s |titre= %s |brisé le=%s}}',
                    $url,
                    'Ouvrage inexistant sur Google Books',
                    date('d-m-Y')
                );
            }
            throw $e;
        }


        $cleanUrl = GoogleLivresTemplate::simplifyGoogleUrl($url);
        $ouvrage->unsetParam('présentation en ligne');
        $ouvrage->setParam('lire en ligne', $cleanUrl);
        $ouvrage->userSeparator = ' |';

        // Si titre absent
        if (empty($ouvrage->getParam('titre'))) {
            throw new \DomainException("Ouvrage sans titre (data Google?)");
        }

        // Google page => 'passage'
        if (!empty($gooDat['pg'])) {
            if (preg_match('#(?:PA|PT)([0-9]+)$#', $gooDat['pg'], $matches)) {
                // Exclusion de page=1, page=2 (vue par défaut sur Google Book)
                if (intval($matches[1]) >= 3) {
                    $page = $matches[1];
                }
            }
            // conversion chiffres Romain pour PR
            if (preg_match('#PR([0-9]+)$#', $gooDat['pg'], $matches)) {
                // Exclusion de page=1, page=2 (vue par défaut sur Google Book)
                if (intval($matches[1]) >= 3) {
                    $page = NumberUtil::arab2roman(intval($matches[1]), true);
                }
            }

            if (!empty($page)) {
                $ouvrage->setParam('passage', $page);
                // ajout commentaire '<!-- utile? -->' ?
            }
        }

        $optimizer = new OuvrageOptimize($ouvrage);
        $optimizer->doTasks();
        $ouvrage2 = $optimizer->getOuvrage();

        return $ouvrage2->serialize();
    }

    /**
     * todo: move (injection) to other class.
     * Generate wiki-template {ouvrage} from GoogleBook ID.
     *
     * @param string $id GoogleBooks ID
     *
     * @return OuvrageTemplate
     * @throws \Exception
     */
    private function generateOuvrageFromGoogleData(string $id): OuvrageTemplate
    {
        // return cached OuvrageTemplate
        if (isset($this->cacheOuvrageTemplate[$id])) {
            return clone $this->cacheOuvrageTemplate[$id];
        }

        // Get Google data by ID ZvhBAAAAcAAJ
        $adapter = new GoogleBooksAdapter();
        $volume = $adapter->getDataByGoogleId($id);

        $mapper = new GoogleBookMapper();
        $mapper->mapLanguageData(true);
        $data = $mapper->process($volume);

        // Generate wiki-template {ouvrage}
        $ouvrage = \App\Domain\WikiTemplateFactory::create('ouvrage');
        $ouvrage->hydrate($data);

        // cache
        $this->cacheOuvrageTemplate[$id] = clone $ouvrage;

        return $ouvrage;
    }

    /**
     * Extract all <ref>/{ref} with only GoogleBooks URL.
     * Todo : supprimer point final URL
     *
     * @param string $text Page wikitext
     *
     * @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...]
     */
    private function extractAllGoogleRefs(string $text): array
    {
        // <ref>...</ref> or {{ref|...}}
        // GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN
        if (preg_match_all(
            '#(?:<ref[^>]*>|{{ref\|) ?('.GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN.'[^>\]} \n]+) ?(?:</ref>|}})#i',
            $text,
            $matches,
            PREG_SET_ORDER
        )
        ) {
            return $matches;
        }

        return [];
    }
}


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;
11
12			use App\Domain\Models\Wiki\GoogleLivresTemplate;
13			use App\Domain\Models\Wiki\OuvrageTemplate;
14			use App\Domain\Publisher\GoogleBookMapper;
15			use App\Domain\Utils\NumberUtil;
16			use App\Domain\Utils\WikiTextUtil;
17			use App\Infrastructure\GoogleBooksAdapter;
18
19			/**
20			* Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage\|...}}.</ref>
21			* in an article wikitext.
22			* Class RefGoogleBook
23			*
24			* @package App\Domain
25			*/
26			class RefGoogleBook
27			{
28			const SLEEP_GOOGLE_API_INTERVAL = 8;
29
30			/**
31			* @var array OuvrageTemplate[]
32			*/
33			private $cacheOuvrageTemplate = [];
34
35			/**
36			* RefGoogleBook constructor.
37			* todo dependency injection
38			*/
39			public function __construct() { }
40
41			/**
42			* Process page wikitext. Return wikitext with the <ref> converted.
43			*
44			* @param string $text Page wikitext
45			*
46			* @return string New wikitext
47			* @throws \Throwable
48			*/
49			public function process(string $text): string
50			{
51			$refsData = $this->extractAllGoogleRefs($text);
52			if (empty($refsData)) {
53			echo "Pas d'URL GB trouvée\n";
54
55			return $text;
56			}
57
58			foreach ($refsData as $ref) {
59			try {
60			$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1]));
61			} catch (\Exception $e) {
62			echo "Exception ".$e->getMessage();
63			continue;
64			}
65
66			// ajout point final pour référence
67			$citation .= '.';
68
69			$newRef = str_replace($ref[1], $citation, $ref[0]);
70			echo $newRef."\n";
71
72			$text = str_replace($ref[0], $newRef, $text);
73
74			echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
75			sleep(self::SLEEP_GOOGLE_API_INTERVAL);
76			}
77
78			return $text;
79			}
80
81			/**
82			* TODO : extract
83			* Convert GoogleBooks URL to wiki-template {ouvrage} citation.
84			*
85			* @param string $url GoogleBooks URL
86			*
87			* @return string {{ouvrage}}
88			* @throws \Exception
89			* @throws \Throwable
90			*/
91			public function convertGBurl2OuvrageCitation(string $url): string
92			{
93			if (!GoogleLivresTemplate::isGoogleBookURL($url)) {
94			throw new \DomainException('Pas de URL Google Books');
95			}
96
97			$gooDat = GoogleLivresTemplate::parseGoogleBookQuery($url);
98			if (empty($gooDat['id'])) {
99			throw new \DomainException('Pas de ID Google Books');
100			}
101
102			try {
103			$ouvrage = $this->generateOuvrageFromGoogleData($gooDat['id']);
104			} catch (\Throwable $e) {
105			// ID n'existe pas sur Google Books
106			if (strpos($e->getMessage(), '404 Not Found')
107			&& strpos($e->getMessage(), '"message": "The volume ID could n')
108			) {
109			return sprintf(
110			'{{lien brisé \|url= %s \|titre= %s \|brisé le=%s}}',
111			$url,
112			'Ouvrage inexistant sur Google Books',
113			date('d-m-Y')
114			);
115			}
116			throw $e;
117			}
118
119
120			$cleanUrl = GoogleLivresTemplate::simplifyGoogleUrl($url);
121			$ouvrage->unsetParam('présentation en ligne');
122			$ouvrage->setParam('lire en ligne', $cleanUrl);
123			$ouvrage->userSeparator = ' \|';
124
125			// Si titre absent
126			if (empty($ouvrage->getParam('titre'))) {
127			throw new \DomainException("Ouvrage sans titre (data Google?)");
128			}
129
130			// Google page => 'passage'
131			if (!empty($gooDat['pg'])) {
132			if (preg_match('#(?:PA\|PT)([0-9]+)$#', $gooDat['pg'], $matches)) {
133			// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
134			if (intval($matches[1]) >= 3) {
135			$page = $matches[1];
136			}
137			}
138			// conversion chiffres Romain pour PR
139			if (preg_match('#PR([0-9]+)$#', $gooDat['pg'], $matches)) {
140			// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
141			if (intval($matches[1]) >= 3) {
142			$page = NumberUtil::arab2roman(intval($matches[1]), true);
143			}
144			}
145
146			if (!empty($page)) {
147			$ouvrage->setParam('passage', $page);
148			// ajout commentaire '<!-- utile? -->' ?
149			}
150			}
151
152			$optimizer = new OuvrageOptimize($ouvrage);
153			$optimizer->doTasks();
154			$ouvrage2 = $optimizer->getOuvrage();
155
156			return $ouvrage2->serialize();
157			}
158
159			/**
160			* todo: move (injection) to other class.
161			* Generate wiki-template {ouvrage} from GoogleBook ID.
162			*
163			* @param string $id GoogleBooks ID
164			*
165			* @return OuvrageTemplate
166			* @throws \Exception
167			*/
168			private function generateOuvrageFromGoogleData(string $id): OuvrageTemplate
169			{
170			// return cached OuvrageTemplate
171			if (isset($this->cacheOuvrageTemplate[$id])) {
172			return clone $this->cacheOuvrageTemplate[$id];
173			}
174
175			// Get Google data by ID ZvhBAAAAcAAJ
176			$adapter = new GoogleBooksAdapter();
177			$volume = $adapter->getDataByGoogleId($id);
178
179			$mapper = new GoogleBookMapper();
180			$mapper->mapLanguageData(true);
181			$data = $mapper->process($volume);
182
183			// Generate wiki-template {ouvrage}
184			$ouvrage = \App\Domain\WikiTemplateFactory::create('ouvrage');
185			$ouvrage->hydrate($data);
186
187			// cache
188			$this->cacheOuvrageTemplate[$id] = clone $ouvrage;
189
190			return $ouvrage;
191			}
192
193			/**
194			* Extract all <ref>/{ref} with only GoogleBooks URL.
195			* Todo : supprimer point final URL
196			*
197			* @param string $text Page wikitext
198			*
199			* @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...]
200			*/
201			private function extractAllGoogleRefs(string $text): array
202			{
203			// <ref>...</ref> or {{ref\|...}}
204			// GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN
205			if (preg_match_all(
206			'#(?:<ref[^>]*>\|{{ref\\|) ?('.GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN.'[^>\]} \n]+) ?(?:</ref>\|}})#i',
207			$text,
208			$matches,
209			PREG_SET_ORDER
210			)
211			) {
212			return $matches;
213			}
214
215			return [];
216			}
217			}
218

Dispositif / Wikibot

Push — dev ( 043eb4...bf3609 )

RefGoogleBook::convertGBurl2OuvrageCitation() C

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like