GoogleTransformer::extractAllGoogleRefs() - Code Metrics - Inspection of "Style & logic (scrutiniz)" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 5eccc7...1faa52 )

by Dispositif

created 2020-04-18 20:24 UTC

GoogleTransformer::extractAllGoogleRefs() A

↳ Parent: GoogleTransformer

Complexity

Conditions	2
Paths	2

Size

Total Lines	15
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	6

Importance

Changes

Metric	Value
cc	2
eloc	7
nc	2
nop	1
dl	0
loc	15
ccs	0
cts	8
cp	0
crap	6
rs	10
c	0
b	0
f	0

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

        return $text;
    }

    /**
     * todo move
     *
     * @param string $text
     *
     * @return array
     */
    public function extractGoogleExternal(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;
    }

    /**
     * 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? -->' ?
            }
        }

        $optimizer = new OuvrageOptimize($ouvrage, null, new Logger());
        $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 = 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\|) ?('.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;
    }

}


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 = $this->extractGoogleExternal($text);
75		if (!empty($links)) {
76		$text = $this->processExternLinks($text, $links);
77		}
78
79		return $text;
80		}
81
82		/**
83		* todo move
84		*
85		* @param string $text
86		*
87		* @return array
88		*/
89	1	public function extractGoogleExternal(string $text): array
90		{
91		// match "* https://books.google.fr/..."
92	1	if (preg_match_all(
93	1	'#^\* ('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^ <{\]}\n\r]+) $#im',
94	1	$text,
95	1	$matches,
96	1	PREG_SET_ORDER
97		)
98		) {
99	1	return $matches;
100		}
101
102		return [];
103		}
104
105		/**
106		* TODO Duplication du dessus...
107		*
108		* @param string $text
109		* @param array $links
110		*
111		* @return string\|string[]
112		* @throws Throwable
113		*/
114		private function processExternLinks(string $text, array $links)
115		{
116		foreach ($links as $pattern) {
117		if ($this->quota->isQuotaReached()) {
118		throw new DomainException('Quota Google atteint');
119		}
120		try {
121		$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($pattern[1]));
122		} catch (Exception $e) {
123		echo "Exception ".$e->getMessage();
124		continue;
125		}
126
127		// todo : ajout point final pour référence ???
128		$citation .= '.';
129
130		$newRef = str_replace($pattern[1], $citation, $pattern[0]);
131		echo $newRef."\n";
132
133		$text = str_replace($pattern[0], $newRef, $text);
134
135		echo "sleep ".self::SLEEP_GOOGLE_API_INTERVAL."\n";
136		sleep(self::SLEEP_GOOGLE_API_INTERVAL);
137		}
138
139		return $text;
140		}
141
142		/**
143		* TODO : extract
144		* Convert GoogleBooks URL to wiki-template {ouvrage} citation.
145		*
146		* @param string $url GoogleBooks URL
147		*
148		* @return string {{ouvrage}}
149		* @throws Exception
150		* @throws Throwable
151		*/
152		public function convertGBurl2OuvrageCitation(string $url): string
153		{
154		if (!GoogleBooksUtil::isGoogleBookURL($url)) {
155		throw new DomainException('Pas de URL Google Books');
156		}
157
158		$gooDat = GoogleBooksUtil::parseGoogleBookQuery($url);
159		if (empty($gooDat['id'])) {
160		throw new DomainException('Pas de ID Google Books');
161		}
162
163		try {
164		$ouvrage = $this->generateOuvrageFromGoogleData($gooDat['id']);
165		} catch (Throwable $e) {
166		// ID n'existe pas sur Google Books
167		if (strpos($e->getMessage(), '"message": "The volume ID could n')) {
168		return sprintf(
169		'{{lien brisé \|url= %s \|titre= %s \|brisé le=%s}}',
170		$url,
171		'Ouvrage inexistant sur Google Books',
172		date('d-m-Y')
173		);
174		}
175		throw $e;
176		}
177
178		$cleanUrl = GoogleBooksUtil::simplifyGoogleUrl($url);
179		$ouvrage->unsetParam('présentation en ligne');
180		$ouvrage->setParam('lire en ligne', $cleanUrl);
181		$ouvrage->userSeparator = ' \|';
182
183		// Si titre absent
184		if (!$ouvrage->hasParamValue('titre')) {
185		throw new DomainException("Ouvrage sans titre (data Google?)");
186		}
187
188		// Google page => 'passage'
189		if (!empty($gooDat['pg'])) {
190		if (preg_match('#(?:PA\|PT)([0-9]+)$#', $gooDat['pg'], $matches)) {
191		// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
192		if (intval($matches[1]) >= 3) {
193		$page = $matches[1];
194		}
195		}
196		// conversion chiffres Romain pour PR
197		if (preg_match('#PR([0-9]+)$#', $gooDat['pg'], $matches)) {
198		// Exclusion de page=1, page=2 (vue par défaut sur Google Book)
199		if (intval($matches[1]) >= 3) {
200		$page = NumberUtil::arab2roman(intval($matches[1]), true);
201		}
202		}
203
204		if (!empty($page)) {
205		$ouvrage->setParam('passage', $page);
206		// ajout commentaire '<!-- utile? -->' ?
207		}
208		}
209
210		$optimizer = new OuvrageOptimize($ouvrage, null, new Logger());
211		$optimizer->doTasks();
212		$ouvrage2 = $optimizer->getOuvrage();
213
214		return $ouvrage2->serialize();
215		}
216
217		/**
218		* todo: move (injection) to other class.
219		* Generate wiki-template {ouvrage} from GoogleBook ID.
220		*
221		* @param string $id GoogleBooks ID
222		*
223		* @return OuvrageTemplate
224		* @throws Exception
225		*/
226		private function generateOuvrageFromGoogleData(string $id): OuvrageTemplate
227		{
228		// return cached OuvrageTemplate
229		if (isset($this->cacheOuvrageTemplate[$id])) {
230		return clone $this->cacheOuvrageTemplate[$id];
231		}
232
233		// Get Google data by ID ZvhBAAAAcAAJ
234		$adapter = new GoogleBooksAdapter();
235		$volume = $adapter->getDataByGoogleId($id);
236
237		$mapper = new GoogleBookMapper();
238		$mapper->mapLanguageData(true);
239		$data = $mapper->process($volume);
240
241		// Generate wiki-template {ouvrage}
242		$ouvrage = WikiTemplateFactory::create('ouvrage');
243		$ouvrage->hydrate($data);
244
245		// cache
246		$this->cacheOuvrageTemplate[$id] = clone $ouvrage;
247
248		return $ouvrage;
249		}
250
251		/**
252		* Extract all <ref>/{ref} with only GoogleBooks URL.
253		* Todo : supprimer point final URL
254		*
255		* @param string $text Page wikitext
256		*
257		* @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...]
258		*/
259		private function extractAllGoogleRefs(string $text): array
260		{
261		// <ref>...</ref> or {{ref\|...}}
262		// GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN
263		if (preg_match_all(
264		'#(?:<ref[^>]*>\|{{ref\\|) ?('.GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN.'[^>\]} \n]+) ?(?:</ref>\|}})#i',
265		$text,
266		$matches,
267		PREG_SET_ORDER
268		)
269		) {
270		return $matches;
271		}
272
273		return [];
274		}
275
276		private function processRef(string $text, array $refsData): string
277		{
278		foreach ($refsData as $ref) {
279		if ($this->quota->isQuotaReached()) {
280		throw new DomainException('Quota Google atteint');
281		}
282		try {
283		$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1]));
284		sleep(2);
285		} catch (\Throwable $e) {
286		echo "Exception ".$e->getMessage();
287		continue;
288		}
289
290		// ajout point final pour référence
291		$citation .= '.';
292
293		$newRef = str_replace($ref[1], $citation, $ref[0]);
294		echo $newRef."\n";
295
296		$text = str_replace($ref[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 ( 5eccc7...1faa52 )

GoogleTransformer::extractAllGoogleRefs() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like