GoogleTransformer::processRef() - Code Metrics - Inspection of "Refactor NotificationWorker" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 26e53d...a73e13 )

by Dispositif

created 2020-09-02 16:44 UTC

GoogleTransformer::processRef() A

↳ Parent: GoogleTransformer

Complexity

Conditions	4
Paths	5

Size

Total Lines	27
Code Lines	16

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	20

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	4
eloc	16
c	1
b	0
f	0
nc	5
nop	2
dl	0
loc	27
ccs	0
cts	12
cp	0
crap	20
rs	9.7333

<?php
/**
 * This file is part of dispositif/wikibot application (@github)
 * 2019/2020 © 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\OuvrageTemplate;
use App\Domain\Publisher\GoogleBookMapper;
use App\Domain\Publisher\GoogleBooksUtil;
use App\Domain\Utils\NumberUtil;
use App\Domain\Utils\WikiTextUtil;
use App\Infrastructure\GoogleApiQuota;
use App\Infrastructure\GoogleBooksAdapter;
use App\Infrastructure\Logger;
use DomainException;
use Exception;
use Throwable;

/**
 * TODO REFAC : duplicate, extract methods in trait or in RefBotWorker + ExternBotWorker
 * --
 * Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage|...}}.</ref>
 * in an article wikitext.
 * Class GoogleTransformer
 *
 * @package App\Domain
 */
class GoogleTransformer
{
    const SLEEP_GOOGLE_API_INTERVAL = 5;

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

    /**
     * GoogleTransformer constructor.
     * todo dependency injection
     */
    public function __construct()
    {
        $this->quota = new GoogleApiQuota();
    }

    /**
     * process page wikitext. Return wikitext with the <ref> converted.
     *
     * @param string $text Page wikitext
     *
     * @return string New wikitext
     * @throws Throwable
     */
    public function process(string $text): string
    {
        if ($this->quota->isQuotaReached()) {
            throw new DomainException('Quota Google atteint');
        }

        $refsData = $this->extractAllGoogleRefs($text);
        if (!empty($refsData)) {
            $text = $this->processRef($text, $refsData);
        }

        $links = $this->extractGoogleExternalBullets($text);
        if (!empty($links)) {
            $text = $this->processExternLinks($text, $links);
        }

        return $text;
    }

    /**
     * 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\|) ?('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^>\]} \n]+) ?(?:</ref>|}})#i',
            $text,
            $matches,
            PREG_SET_ORDER
        )
        ) {
            return $matches;
        }

        return [];
    }

    private function processRef(string $text, array $refsData): string
    {
        foreach ($refsData as $ref) {
            if ($this->quota->isQuotaReached()) {
                throw new DomainException('Quota Google atteint');
            }
            try {
                $citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1]));
                sleep(2);
            } catch (Throwable $e) {
                echo "Exception ".$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 $url): string
    {
        if (!GoogleBooksUtil::isGoogleBookURL($url)) {
            throw new DomainException('Pas de URL Google Books');
        }

        $gooDat = GoogleBooksUtil::parseGoogleBookQuery($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(), '"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;
        }

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

        // Si titre absent
        if (!$ouvrage->hasParamValue('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? -->' ?
            }
        }

        $optimizer = OptimizerFactory::fromTemplate($ouvrage, null, new Logger());
        $optimizer->doTasks();
        $ouvrage2 = $optimizer->getOptiTemplate();

        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 = WikiTemplateFactory::create('ouvrage');
        $ouvrage->hydrate($data);

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

        return $ouvrage;
    }

    /**
     * todo move
     *
     * @param string $text
     *
     * @return array
     */
    public function extractGoogleExternalBullets(string $text): array
    {
        // match "* https://books.google.fr/..."
        if (preg_match_all(
            '#^\* *('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^ <{\]}\n\r]+) *$#im',
            $text,
            $matches,
            PREG_SET_ORDER
        )
        ) {
            return $matches;
        }

        return [];
    }

    /**
     * TODO Duplication du dessus...
     *
     * @param string $text
     * @param array  $links
     *
     * @return string|string[]
     * @throws Throwable
     */
    private function processExternLinks(string $text, array $links)
    {
        foreach ($links as $pattern) {
            if ($this->quota->isQuotaReached()) {
                throw new DomainException('Quota Google atteint');
            }
            try {
                $citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($pattern[1]));
            } catch (Exception $e) {
                echo "Exception ".$e->getMessage();
                continue;
            }

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

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

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

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

        return $text;
    }

}


1		<?php
2		/**
3		* This file is part of dispositif/wikibot application (@github)
4		* 2019/2020 © 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\OuvrageTemplate;
13		use App\Domain\Publisher\GoogleBookMapper;
14		use App\Domain\Publisher\GoogleBooksUtil;
15		use App\Domain\Utils\NumberUtil;
16		use App\Domain\Utils\WikiTextUtil;
17		use App\Infrastructure\GoogleApiQuota;
18		use App\Infrastructure\GoogleBooksAdapter;
19		use App\Infrastructure\Logger;
20		use DomainException;
21		use Exception;
22		use Throwable;
23
24		/**
25		* TODO REFAC : duplicate, extract methods in trait or in RefBotWorker + ExternBotWorker
26		* --
27		* Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage\|...}}.</ref>
28		* in an article wikitext.
29		* Class GoogleTransformer
30		*
31		* @package App\Domain
32		*/
33		class GoogleTransformer
34		{
35		const SLEEP_GOOGLE_API_INTERVAL = 5;
36
37		/**
38		* @var array OuvrageTemplate[]
39		*/
40		private $cacheOuvrageTemplate = [];
41		/**
42		* @var GoogleApiQuota
43		*/
44		private $quota;
45
46		/**
47		* GoogleTransformer constructor.
48		* todo dependency injection
49		*/
50	1	public function __construct()
51		{
52	1	$this->quota = new GoogleApiQuota();
53	1	}
54
55		/**
56		* process page wikitext. Return wikitext with the <ref> converted.
57		*
58		* @param string $text Page wikitext
59		*
60		* @return string New wikitext
61		* @throws Throwable
62		*/
63		public function process(string $text): string
64		{
65		if ($this->quota->isQuotaReached()) {
66		throw new DomainException('Quota Google atteint');
67		}
68
69		$refsData = $this->extractAllGoogleRefs($text);
70		if (!empty($refsData)) {
71		$text = $this->processRef($text, $refsData);
72		}
73
74		$links = $this->extractGoogleExternalBullets($text);
75		if (!empty($links)) {
76		$text = $this->processExternLinks($text, $links);
77		}
78
79		return $text;
80		}
81
82		/**
83		* Extract all <ref>/{ref} with only GoogleBooks URL.
84		* Todo : supprimer point final URL
85		*
86		* @param string $text Page wikitext
87		*
88		* @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...]
89	1	*/
90		private function extractAllGoogleRefs(string $text): array
91		{
92	1	// <ref>...</ref> or {{ref\|...}}
93	1	// GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN
94	1	if (preg_match_all(
95	1	'#(?:<ref[^>]*>\|{{ref\\|) ?('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^>\]} \n]+) ?(?:</ref>\|}})#i',
96	1	$text,
97		$matches,
98		PREG_SET_ORDER
99	1	)
100		) {
101		return $matches;
102		}
103
104		return [];
105		}
106
107		private function processRef(string $text, array $refsData): string
108		{
109		foreach ($refsData as $ref) {
110		if ($this->quota->isQuotaReached()) {
111		throw new DomainException('Quota Google atteint');
112		}
113		try {
114		$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1]));
115		sleep(2);
116		} catch (Throwable $e) {
117		echo "Exception ".$e->getMessage();
118		continue;
119		}
120
121		// ajout point final pour référence
122		$citation .= '.';
123
124		$newRef = str_replace($ref[1], $citation, $ref[0]);
125		echo $newRef."\n";
126
127		$text = str_replace($ref[0], $newRef, $text);
128
129		echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
130		sleep(self::SLEEP_GOOGLE_API_INTERVAL);
131		}
132
133		return $text;
134		}
135
136		/**
137		* TODO : extract
138		* Convert GoogleBooks URL to wiki-template {ouvrage} citation.
139		*
140		* @param string $url GoogleBooks URL
141		*
142		* @return string {{ouvrage}}
143		* @throws Exception
144		* @throws Throwable
145		*/
146		public function convertGBurl2OuvrageCitation(string $url): string
147		{
148		if (!GoogleBooksUtil::isGoogleBookURL($url)) {
149		throw new DomainException('Pas de URL Google Books');
150		}
151
152		$gooDat = GoogleBooksUtil::parseGoogleBookQuery($url);
153		if (empty($gooDat['id'])) {
154		throw new DomainException('Pas de ID Google Books');
155		}
156
157		try {
158		$ouvrage = $this->generateOuvrageFromGoogleData($gooDat['id']);
159		} catch (Throwable $e) {
160		// ID n'existe pas sur Google Books
161		if (strpos($e->getMessage(), '"message": "The volume ID could n')) {
162		return sprintf(
163		'{{lien brisé \|url= %s \|titre=%s \|brisé le=%s}}',
164		$url,
165		'Ouvrage inexistant sur Google Books',
166		date('d-m-Y')
167		);
168		}
169		throw $e;
170		}
171
172		$cleanUrl = GoogleBooksUtil::simplifyGoogleUrl($url);
173		$ouvrage->unsetParam('présentation en ligne');
174		$ouvrage->setParam('lire en ligne', $cleanUrl);
175		$ouvrage->userSeparator = ' \|';
176
177		// Si titre absent
178		if (!$ouvrage->hasParamValue('titre')) {
179		throw new DomainException("Ouvrage sans titre (data Google?)");
180		}
181
182		// Google page => 'passage'
183		if (!empty($gooDat['pg'])) {
184		if (preg_match('#(?:PA\|PT)([0-9]+)$#', $gooDat['pg'], $matches)) {
185		// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
186		if (intval($matches[1]) >= 3) {
187		$page = $matches[1];
188		}
189		}
190		// conversion chiffres Romain pour PR
191		if (preg_match('#PR([0-9]+)$#', $gooDat['pg'], $matches)) {
192		// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
193		if (intval($matches[1]) >= 3) {
194		$page = NumberUtil::arab2roman(intval($matches[1]), true);
195		}
196		}
197
198		if (!empty($page)) {
199		$ouvrage->setParam('passage', $page);
200		// ajout commentaire '<!-- utile? -->' ?
201		}
202		}
203
204		$optimizer = OptimizerFactory::fromTemplate($ouvrage, null, new Logger());
205		$optimizer->doTasks();
206		$ouvrage2 = $optimizer->getOptiTemplate();
207
208		return $ouvrage2->serialize();
209		}
210
211		/**
212		* todo: move (injection) to other class.
213		* Generate wiki-template {ouvrage} from GoogleBook ID.
214		*
215		* @param string $id GoogleBooks ID
216		*
217		* @return OuvrageTemplate
218		* @throws Exception
219		*/
220		private function generateOuvrageFromGoogleData(string $id): OuvrageTemplate
221		{
222		// return cached OuvrageTemplate
223		if (isset($this->cacheOuvrageTemplate[$id])) {
224		return clone $this->cacheOuvrageTemplate[$id];
225		}
226
227		// Get Google data by ID ZvhBAAAAcAAJ
228		$adapter = new GoogleBooksAdapter();
229		$volume = $adapter->getDataByGoogleId($id);
230
231		$mapper = new GoogleBookMapper();
232		$mapper->mapLanguageData(true);
233		$data = $mapper->process($volume);
234
235		// Generate wiki-template {ouvrage}
236		$ouvrage = WikiTemplateFactory::create('ouvrage');
237		$ouvrage->hydrate($data);
238
239		// cache
240		$this->cacheOuvrageTemplate[$id] = clone $ouvrage;
241
242		return $ouvrage;
243		}
244
245		/**
246		* todo move
247		*
248		* @param string $text
249		*
250		* @return array
251		*/
252		public function extractGoogleExternalBullets(string $text): array
253		{
254		// match "* https://books.google.fr/..."
255		if (preg_match_all(
256		'#^\* ('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^ <{\]}\n\r]+) $#im',
257		$text,
258		$matches,
259		PREG_SET_ORDER
260		)
261		) {
262		return $matches;
263		}
264
265		return [];
266		}
267
268		/**
269		* TODO Duplication du dessus...
270		*
271		* @param string $text
272		* @param array $links
273		*
274		* @return string\|string[]
275		* @throws Throwable
276		*/
277		private function processExternLinks(string $text, array $links)
278		{
279		foreach ($links as $pattern) {
280		if ($this->quota->isQuotaReached()) {
281		throw new DomainException('Quota Google atteint');
282		}
283		try {
284		$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($pattern[1]));
285		} catch (Exception $e) {
286		echo "Exception ".$e->getMessage();
287		continue;
288		}
289
290		// todo : ajout point final pour référence ???
291		$citation .= '.';
292
293		$newRef = str_replace($pattern[1], $citation, $pattern[0]);
294		echo $newRef."\n";
295
296		$text = str_replace($pattern[0], $newRef, $text);
297
298		echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
299		sleep(self::SLEEP_GOOGLE_API_INTERVAL);
300		}
301
302		return $text;
303		}
304
305		}
306

Dispositif / Wikibot

Push — master ( 26e53d...a73e13 )

GoogleTransformer::processRef() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like