RefGoogleBook::convertGBurl2OuvrageCitation() - Code Metrics - Inspection of "Add refGoo2ouvrageWorker" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Test Setup Failed

Push — master ( c5c2a9...00574e )

by Dispositif

created 2020-03-11 18:32 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	2
Bugs	0	Features	0

Metric	Value
cc	13
eloc	35
c	2
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\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
     */
    public function process(string $text): string
    {
        $refsData = $this->extractAllGoogleRefs($text);
        if (empty($refsData)) {
            echo "Pas d'URL GB trouvée";

            return $text;
        }

        foreach ($refsData as $ref) {
            try {
                $citation = $this->convertGBurl2OuvrageCitation($this->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;
    }

    /**
     * Strip the final point (".") as in <ref> ending.
     * TODO move to WikiRef or TextUtil class
     *
     * @param string $str
     *
     * @return string
     */
    private function stripFinalPoint(string $str): string
    {
        if (substr($str, -1, 1) === '.') {
            return substr($str, 0, strlen($str) - 1);
        }

        return $str;
    }

    /**
     * Convert GoogleBooks URL to wiki-template {ouvrage} citation.
     *
     * @param string $url GoogleBooks URL
     *
     * @return string {{ouvrage}}
     * @throws \Exception
     * @throws \Throwable
     */
    private 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 |CodexBot=1}}',
                    $url,
                    'ID introuvable 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 = new OuvrageTemplate();
        $ouvrage->hydrate($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|...}}
        if (preg_match_all(
            '#(?:<ref[^>]*>|{{ref\|) ?(https?://(?:books|play)\.google\.[a-z]{2,3}/(?:books)?(?:/reader)?\?id=[^>\]} \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\Infrastructure\GoogleBooksAdapter;
17
18			/**
19			* Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage\|...}}.</ref>
20			* in an article wikitext.
21			* Class RefGoogleBook
22			*
23			* @package App\Domain
24			*/
25			class RefGoogleBook
26			{
27			const SLEEP_GOOGLE_API_INTERVAL = 8;
28
29			/**
30			* @var array OuvrageTemplate[]
31			*/
32			private $cacheOuvrageTemplate = [];
33
34			/**
35			* RefGoogleBook constructor.
36			* todo dependency injection
37			*/
38			public function __construct() { }
39
40			/**
41			* Process page wikitext. Return wikitext with the <ref> converted.
42			*
43			* @param string $text Page wikitext
44			*
45			* @return string New wikitext
46			*/
47			public function process(string $text): string
48			{
49			$refsData = $this->extractAllGoogleRefs($text);
50			if (empty($refsData)) {
51			echo "Pas d'URL GB trouvée";
52
53			return $text;
54			}
55
56			foreach ($refsData as $ref) {
57			try {
58			$citation = $this->convertGBurl2OuvrageCitation($this->stripFinalPoint($ref[1]));
59			} catch (\Exception $e) {
60			echo "Exception ".$e->getMessage();
61			continue;
62			}
63
64			// ajout point final pour référence
65			$citation .= '.';
66
67			$newRef = str_replace($ref[1], $citation, $ref[0]);
68			echo $newRef."\n";
69
70			$text = str_replace($ref[0], $newRef, $text);
71
72			echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
73			sleep(self::SLEEP_GOOGLE_API_INTERVAL);
74			}
75
76			return $text;
77			}
78
79			/**
80			* Strip the final point (".") as in <ref> ending.
81			* TODO move to WikiRef or TextUtil class
82			*
83			* @param string $str
84			*
85			* @return string
86			*/
87			private function stripFinalPoint(string $str): string
88			{
89			if (substr($str, -1, 1) === '.') {
90			return substr($str, 0, strlen($str) - 1);
91			}
92
93			return $str;
94			}
95
96			/**
97			* Convert GoogleBooks URL to wiki-template {ouvrage} citation.
98			*
99			* @param string $url GoogleBooks URL
100			*
101			* @return string {{ouvrage}}
102			* @throws \Exception
103			* @throws \Throwable
104			*/
105			private function convertGBurl2OuvrageCitation(string $url): string
106			{
107			if (!GoogleLivresTemplate::isGoogleBookURL($url)) {
108			throw new \DomainException('Pas de URL Google Books');
109			}
110
111			$gooDat = GoogleLivresTemplate::parseGoogleBookQuery($url);
112			if (empty($gooDat['id'])) {
113			throw new \DomainException('Pas de ID Google Books');
114			}
115
116			try {
117			$ouvrage = $this->generateOuvrageFromGoogleData($gooDat['id']);
118			} catch (\Throwable $e) {
119			// ID n'existe pas sur Google Books
120			if (strpos($e->getMessage(), '404 Not Found')
121			&& strpos($e->getMessage(), '"message": "The volume ID could n')
122			) {
123			return sprintf(
124			'{{lien brisé \|url= %s \|titre= %s \|brisé le=%s \|CodexBot=1}}',
125			$url,
126			'ID introuvable sur Google Books',
127			date('d-m-Y')
128			);
129			}
130			throw $e;
131			}
132
133
134			$cleanUrl = GoogleLivresTemplate::simplifyGoogleUrl($url);
135			$ouvrage->unsetParam('présentation en ligne');
136			$ouvrage->setParam('lire en ligne', $cleanUrl);
137			$ouvrage->userSeparator = ' \|';
138
139			// Si titre absent
140			if (empty($ouvrage->getParam('titre'))) {
141			throw new \DomainException("Ouvrage sans titre (data Google?)");
142			}
143
144			// Google page => 'passage'
145			if (!empty($gooDat['pg'])) {
146			if (preg_match('#(?:PA\|PT)([0-9]+)$#', $gooDat['pg'], $matches)) {
147			// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
148			if (intval($matches[1]) >= 3) {
149			$page = $matches[1];
150			}
151			}
152			// conversion chiffres Romain pour PR
153			if (preg_match('#PR([0-9]+)$#', $gooDat['pg'], $matches)) {
154			// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
155			if (intval($matches[1]) >= 3) {
156			$page = NumberUtil::arab2roman(intval($matches[1]), true);
157			}
158			}
159
160			if (!empty($page)) {
161			$ouvrage->setParam('passage', $page);
162			// ajout commentaire '<!-- utile? -->' ?
163			}
164			}
165
166			$optimizer = new OuvrageOptimize($ouvrage);
167			$optimizer->doTasks();
168			$ouvrage2 = $optimizer->getOuvrage();
169
170			return $ouvrage2->serialize();
171			}
172
173			/**
174			* todo: move (injection) to other class.
175			* Generate wiki-template {ouvrage} from GoogleBook ID.
176			*
177			* @param string $id GoogleBooks ID
178			*
179			* @return OuvrageTemplate
180			* @throws \Exception
181			*/
182			private function generateOuvrageFromGoogleData(string $id): OuvrageTemplate
183			{
184			// return cached OuvrageTemplate
185			if (isset($this->cacheOuvrageTemplate[$id])) {
186			return clone $this->cacheOuvrageTemplate[$id];
187			}
188
189			// Get Google data by ID ZvhBAAAAcAAJ
190			$adapter = new GoogleBooksAdapter();
191			$volume = $adapter->getDataByGoogleId($id);
192
193			$mapper = new GoogleBookMapper();
194			$mapper->mapLanguageData(true);
195			$data = $mapper->process($volume);
196
197			// Generate wiki-template {ouvrage}
198			$ouvrage = new OuvrageTemplate();
199			$ouvrage->hydrate($data);
200
201			// cache
202			$this->cacheOuvrageTemplate[$id] = clone $ouvrage;
203
204			return $ouvrage;
205			}
206
207			/**
208			* Extract all <ref>/{ref} with only GoogleBooks URL.
209			* Todo : supprimer point final URL
210			*
211			* @param string $text Page wikitext
212			*
213			* @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...]
214			*/
215			private function extractAllGoogleRefs(string $text): array
216			{
217			// <ref>...</ref> or {{ref\|...}}
218			if (preg_match_all(
219			'#(?:<ref[^>]*>\|{{ref\\|) ?(https?://(?:books\|play)\.google\.[a-z]{2,3}/(?:books)?(?:/reader)?\?id=[^>\]} \n]+) ?(?:</ref>\|}})#i',
220			$text,
221			$matches,
222			PREG_SET_ORDER
223			)
224			) {
225			return $matches;
226			}
227
228			return [];
229			}
230			}
231

Dispositif / Wikibot

Push — master ( c5c2a9...00574e )

RefGoogleBook::convertGBurl2OuvrageCitation() C

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like