OuvrageCompleteWorker::isIsbnSkipped() - Code Metrics - Inspection of "Style AbstractWikiTemplate" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Branch — dev (8d7b92)

by Dispositif

created 2020-04-14 15:47 UTC

OuvrageCompleteWorker::isIsbnSkipped() A

↳ Parent: OuvrageCompleteWorker

Complexity

Conditions	4
Paths	2

Size

Total Lines	10
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	4
eloc	5
nc	2
nop	2
dl	0
loc	10
rs	10
c	0
b	0
f	0

<?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\Application;

use App\Domain\Models\Wiki\OuvrageTemplate;
use App\Domain\OuvrageComplete;
use App\Domain\OuvrageFactory;
use App\Domain\OuvrageOptimize;
use App\Domain\Publisher\Wikidata2Ouvrage;
use App\Domain\Utils\TemplateParser;
use App\Infrastructure\Memory;
use App\Infrastructure\WikidataAdapter;
use Exception;
use GuzzleHttp\Client;
use Normalizer;
use Throwable;

/**
 * Class OuvrageCompleteWorker
 *
 * @package App\Application
 */
class OuvrageCompleteWorker
{
    /**
     * Exclusion requête BnF/Google/etc
     * Format EAN ou ISBN10 sans tiret.
     */
    const ISBN_EAN_SKIP
        = [
            '9782918758440', // Profils de lignes du réseau ferré français vol.2
            '9782918758341', // Profils de lignes du réseau ferré français vol.1
        ];
    /**
     * @var bool
     */
    public $verbose = false;

    /**
     * @var QueueInterface
     */
    private $queueAdapter;
    /**
     * @var string
     */
    private $raw = '';
    private $page; // article title

    private $log = [];
    private $notCosmetic = false;
    private $major = false;
    /**
     * @var OuvrageTemplate
     */
    private $ouvrage;

    public function __construct(QueueInterface $queueAdapter, ?bool $verbose = false)
    {
        $this->queueAdapter = $queueAdapter;
        $this->verbose = (bool)$verbose;
    }

    public function run(?int $limit = 10000)
    {
        $memory = new Memory();
        while ($limit > 0) {
            $limit--;
            sleep(1);
            $row = $this->getNewRow2Complete();
            $this->raw = $row['raw'];
            $this->page = $row['page'];

            echo sprintf(
                "-------------------------------\n%s [%s]\n%s\n%s\n",
                date("Y-m-d H:i:s"),
                WikiBotConfig::getGitVersion() ?? '',
                $this->page,
                $this->raw
            );
            if ($this->verbose) {
                $memory->echoMemory(true);
            }

            // initialise variables
            $this->log = [];
            $this->ouvrage = null;
            $this->notCosmetic = false;
            $this->major = false;


            try {
                $parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw);
                $origin = $parse['ouvrage'][0]['model'] ?? null;
            } catch (Throwable $e) {
                echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
                continue;
            }

            if (!$origin instanceof OuvrageTemplate) {
                echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
                continue;
            }

            // Final optimizing (with online predictions)
            $optimizer = new OuvrageOptimize($origin, $this->page);
            $optimizer->doTasks();
            $this->ouvrage = $optimizer->getOuvrage();
            $this->log = array_merge($this->log, $optimizer->getLog());
            $this->notCosmetic = ($optimizer->notCosmetic || $this->notCosmetic);

            /**
             * RECHERCHE ONLINE
             */
            $isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN
            $isbn10 = $origin->getParam('isbn2') ?? $origin->getParam('isbn10') ?? null;
            if (!empty($isbn)
                && !$origin->hasParamValue('isbn invalide')
                && !$origin->hasParamValue('isbn erroné')
            ) {
                $this->onlineIsbnSearch($isbn, $isbn10);
            }

            $this->sendCompleted();
            unset($optimizer);
            unset($parse);
            unset($origin);
        } // END WHILE

        return true;
    }

    /**
     * Get array (title+raw strings) to complete from AMQP queue, SQL Select or file reading.
     *
     * @return array
     * @throws Exception
     */
    private function getNewRow2Complete(): array
    {
        $row = $this->queueAdapter->getNewRaw();
        if (empty($row) || empty($row['raw'])) {
            echo "STOP: no more queue to process \n";
            throw new Exception('no more queue to process');
        }

        return $row;
    }

    /**
     * @param string      $isbn
     * @param string|null $isbn10
     *
     * @return bool
     */
    private function isIsbnSkipped(string $isbn, ?string $isbn10 = null): bool
    {
        if (in_array(str_replace('-', '', $isbn), self::ISBN_EAN_SKIP)
            || ($isbn10 !== null
                && in_array(str_replace('-', '', $isbn10), self::ISBN_EAN_SKIP))
        ) {
            return true;
        }

        return false;
    }

    private function onlineIsbnSearch(string $isbn, ?string $isbn10 = null)
    {
        if ($this->isIsbnSkipped($isbn, $isbn10)) {
            echo "*** SKIP THAT ISBN ***\n";

            // Vérifier logique return
            return;
        }

        online:
        if ($this->verbose) {
            echo "sleep 10...\n";
        }
        sleep(10);

        try {
            if ($this->verbose) {
                dump('BIBLIO NAT FRANCE...');
            }
            // BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM !
            $bnfOuvrage = null;
            if ($isbn10) {
                $bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn10);
                sleep(2);
            }
            if (!$isbn10 || empty($bnfOuvrage) || empty($bnfOuvrage->getParam('titre'))) {
                $bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn);
            }
            if (isset($bnfOuvrage) and $bnfOuvrage instanceof OuvrageTemplate) {
                $this->completeOuvrage($bnfOuvrage);

                // Wikidata requests from $infos (ISBN/ISNI)
                if (!empty($bnfOuvrage->getInfos())) {
                    if ($this->verbose) {
                        dump('WIKIDATA...');
                    }
                    // TODO move to factory
                    $wikidataAdapter = new WikidataAdapter(
                        new Client(['timeout' => 5, 'headers' => ['User-Agent' => getenv('USER_AGENT')]])
                    );
                    $wdComplete = new Wikidata2Ouvrage($wikidataAdapter, clone $bnfOuvrage, $this->page);
                    $this->completeOuvrage($wdComplete->getOuvrage());
                }
            }
        } catch (Throwable $e) {
            echo sprintf(
                "*** ERREUR BnF Isbn Search %s %s %s \n",
                $e->getMessage(),
                $e->getFile(),
                $e->getLine()
            );
        }

        if (!isset($bnfOuvrage) || !$this->skipGoogle($bnfOuvrage)) {
            try {
                if ($this->verbose) {
                    dump('GOOGLE...');
                }
                $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
                $this->completeOuvrage($googleOuvrage);
            } catch (Throwable $e) {
                echo "*** ERREUR GOOGLE Isbn Search ***".$e->getMessage()."\n";
                if( strpos($e->getMessage(), 'Could not resolve host: www.googleapis.com') === false) {
                    throw $e;
                }
                unset($e);
            }
        }

        if (!isset($bnfOuvrage) && !isset($googleOuvrage)) {
            try {
                if ($this->verbose) {
                    dump('OpenLibrary...');
                }
                $openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn);
                if (!empty($openLibraryOuvrage)) {
                    $this->completeOuvrage($openLibraryOuvrage);
                }
            } catch (Throwable $e) {
                echo '**** ERREUR OpenLibrary Isbn Search';
            }
        }
    }

    //    private function onlineQuerySearch(string $query)
    //    {
    //        echo "sleep 40...";
    //        sleep(20);
    //        onlineQuerySearch:
    //
    //        try {
    //            dump('GOOGLE SEARCH...');
    //            //            $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
    //            $adapter = new GoogleBooksAdapter();
    //            $data = $adapter->search('blabla');
    //            dump($data);
    //            //die;
    //            //            return $import->getOuvrage();
    //            //            $this->completeOuvrage($googleOuvrage);
    //        } catch (Throwable $e) {
    //            echo "*** ERREUR GOOGLE QuerySearch *** ".$e->getMessage()."\n";
    //            echo "sleep 30min";
    //            sleep(60 * 30);
    //            echo "Wake up\n";
    //            goto onlineQuerySearch;
    //        }
    //    }

    private function completeOuvrage(OuvrageTemplate $onlineOuvrage)
    {
        if ($this->verbose) {
            dump($onlineOuvrage->serialize(true));
        }
        $optimizer = new OuvrageOptimize($onlineOuvrage, $this->page);
        $onlineOptimized = ($optimizer)->doTasks()->getOuvrage();

        $completer = new OuvrageComplete($this->ouvrage, $onlineOptimized);
        $this->ouvrage = $completer->getResult();

        // todo move that optimizing in OuvrageComplete ?
        $optimizer = new OuvrageOptimize($this->ouvrage, $this->page);
        $this->ouvrage = $optimizer->doTasks()->getOuvrage();

        if ($this->verbose) {
            dump($completer->getLog());
        }
        if ($completer->major) {
            $this->major = true;
        }
        $this->notCosmetic = ($completer->notCosmetic || $this->notCosmetic);
        $this->log = array_merge($this->log, $completer->getLog());
        unset($optimizer);
        unset($completer);
    }

    private function sendCompleted()
    {
        $isbn13 = $this->ouvrage->getParam('isbn') ?? null;

        $finalData = [
            //    'page' =>
            'raw' => $this->raw,
            'opti' => $this->serializeFinalOpti(),
            'optidate' => date("Y-m-d H:i:s"),
            'modifs' => mb_substr(implode(',', $this->log), 0, 250),
            'notcosmetic' => ($this->notCosmetic) ? 1 : 0,
            'major' => ($this->major) ? 1 : 0,
            'isbn' => substr($isbn13, 0, 20),
            'version' => WikiBotConfig::getGitVersion() ?? null,
        ];
        if ($this->verbose) {
            dump($finalData);
        }
        // Json ?
        $result = $this->queueAdapter->sendCompletedData($finalData);

        dump($result); // bool
    }

    /**
     * Final serialization of the completed OuvrageTemplate.
     *
     * @return string
     */
    private function serializeFinalOpti(): string
    {
        //        // Améliore style compact : plus espacé
        //        if ('|' === $this->ouvrage->userSeparator) {
        //            $this->ouvrage->userSeparator = ' |';
        //        }
        $finalOpti = $this->ouvrage->serialize(true);
        $finalOpti = Normalizer::normalize($finalOpti);

        return $finalOpti;
    }

    private function skipGoogle($bnfOuvrage): bool
    {
        if ($bnfOuvrage instanceOf OuvrageTemplate
            && $bnfOuvrage->hasParamValue('titre')
            && ($this->ouvrage->hasParamValue('lire en ligne')
                || $this->ouvrage->hasParamValue('présentation en ligne'))
        ) {
            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\Application;
11
12			use App\Domain\Models\Wiki\OuvrageTemplate;
13			use App\Domain\OuvrageComplete;
14			use App\Domain\OuvrageFactory;
15			use App\Domain\OuvrageOptimize;
16			use App\Domain\Publisher\Wikidata2Ouvrage;
17			use App\Domain\Utils\TemplateParser;
18			use App\Infrastructure\Memory;
19			use App\Infrastructure\WikidataAdapter;
20			use Exception;
21			use GuzzleHttp\Client;
22			use Normalizer;
23			use Throwable;
24
25			/**
26			* Class OuvrageCompleteWorker
27			*
28			* @package App\Application
29			*/
30			class OuvrageCompleteWorker
31			{
32			/**
33			* Exclusion requête BnF/Google/etc
34			* Format EAN ou ISBN10 sans tiret.
35			*/
36			const ISBN_EAN_SKIP
37			= [
38			'9782918758440', // Profils de lignes du réseau ferré français vol.2
39			'9782918758341', // Profils de lignes du réseau ferré français vol.1
40			];
41			/**
42			* @var bool
43			*/
44			public $verbose = false;
45
46			/**
47			* @var QueueInterface
48			*/
49			private $queueAdapter;
50			/**
51			* @var string
52			*/
53			private $raw = '';
54			private $page; // article title
55
56			private $log = [];
57			private $notCosmetic = false;
58			private $major = false;
59			/**
60			* @var OuvrageTemplate
61			*/
62			private $ouvrage;
63
64			public function __construct(QueueInterface $queueAdapter, ?bool $verbose = false)
65			{
66			$this->queueAdapter = $queueAdapter;
67			$this->verbose = (bool)$verbose;
68			}
69
70			public function run(?int $limit = 10000)
71			{
72			$memory = new Memory();
73			while ($limit > 0) {
74			$limit--;
75			sleep(1);
76			$row = $this->getNewRow2Complete();
77			$this->raw = $row['raw'];
78			$this->page = $row['page'];
79
80			echo sprintf(
81			"-------------------------------\n%s [%s]\n%s\n%s\n",
82			date("Y-m-d H:i:s"),
83			WikiBotConfig::getGitVersion() ?? '',
84			$this->page,
85			$this->raw
86			);
87			if ($this->verbose) {
88			$memory->echoMemory(true);
89			}
90
91			// initialise variables
92			$this->log = [];
93			$this->ouvrage = null;
94			$this->notCosmetic = false;
95			$this->major = false;
96
97
98			try {
99			$parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw);
100			$origin = $parse['ouvrage'][0]['model'] ?? null;
101			} catch (Throwable $e) {
102			echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
103			continue;
104			}
105
106			if (!$origin instanceof OuvrageTemplate) {
107			echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
108			continue;
109			}
110
111			// Final optimizing (with online predictions)
112			$optimizer = new OuvrageOptimize($origin, $this->page);
113			$optimizer->doTasks();
114			$this->ouvrage = $optimizer->getOuvrage();
115			$this->log = array_merge($this->log, $optimizer->getLog());
116			$this->notCosmetic = ($optimizer->notCosmetic \|\| $this->notCosmetic);
117
118			/**
119			* RECHERCHE ONLINE
120			*/
121			$isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN
122			$isbn10 = $origin->getParam('isbn2') ?? $origin->getParam('isbn10') ?? null;
123			if (!empty($isbn)
124			&& !$origin->hasParamValue('isbn invalide')
125			&& !$origin->hasParamValue('isbn erroné')
126			) {
127			$this->onlineIsbnSearch($isbn, $isbn10);
128			}
129
130			$this->sendCompleted();
131			unset($optimizer);
132			unset($parse);
133			unset($origin);
134			} // END WHILE
135
136			return true;
137			}
138
139			/**
140			* Get array (title+raw strings) to complete from AMQP queue, SQL Select or file reading.
141			*
142			* @return array
143			* @throws Exception
144			*/
145			private function getNewRow2Complete(): array
146			{
147			$row = $this->queueAdapter->getNewRaw();
148			if (empty($row) \|\| empty($row['raw'])) {
149			echo "STOP: no more queue to process \n";
150			throw new Exception('no more queue to process');
151			}
152
153			return $row;
154			}
155
156			/**
157			* @param string $isbn
158			* @param string\|null $isbn10
159			*
160			* @return bool
161			*/
162			private function isIsbnSkipped(string $isbn, ?string $isbn10 = null): bool
163			{
164			if (in_array(str_replace('-', '', $isbn), self::ISBN_EAN_SKIP)
165			\|\| ($isbn10 !== null
166			&& in_array(str_replace('-', '', $isbn10), self::ISBN_EAN_SKIP))
167			) {
168			return true;
169			}
170
171			return false;
172			}
173
174			private function onlineIsbnSearch(string $isbn, ?string $isbn10 = null)
175			{
176			if ($this->isIsbnSkipped($isbn, $isbn10)) {
177			echo "* SKIP THAT ISBN *\n";
178
179			// Vérifier logique return
180			return;
181			}
182
183			online:
184			if ($this->verbose) {
185			echo "sleep 10...\n";
186			}
187			sleep(10);
188
189			try {
190			if ($this->verbose) {
191			dump('BIBLIO NAT FRANCE...');
192			}
193			// BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM !
194			$bnfOuvrage = null;
195			if ($isbn10) {
196			$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn10);
197			sleep(2);
198			}
199			if (!$isbn10 \|\| empty($bnfOuvrage) \|\| empty($bnfOuvrage->getParam('titre'))) {
200			$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn);
201			}
202			if (isset($bnfOuvrage) and $bnfOuvrage instanceof OuvrageTemplate) {
203			$this->completeOuvrage($bnfOuvrage);
204
205			// Wikidata requests from $infos (ISBN/ISNI)
206			if (!empty($bnfOuvrage->getInfos())) {
207			if ($this->verbose) {
208			dump('WIKIDATA...');
209			}
210			// TODO move to factory
211			$wikidataAdapter = new WikidataAdapter(
212			new Client(['timeout' => 5, 'headers' => ['User-Agent' => getenv('USER_AGENT')]])
213			);
214			$wdComplete = new Wikidata2Ouvrage($wikidataAdapter, clone $bnfOuvrage, $this->page);
215			$this->completeOuvrage($wdComplete->getOuvrage());
216			}
217			}
218			} catch (Throwable $e) {
219			echo sprintf(
220			"*** ERREUR BnF Isbn Search %s %s %s \n",
221			$e->getMessage(),
222			$e->getFile(),
223			$e->getLine()
224			);
225			}
226
227			if (!isset($bnfOuvrage) \|\| !$this->skipGoogle($bnfOuvrage)) {
228			try {
229			if ($this->verbose) {
230			dump('GOOGLE...');
231			}
232			$googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
233			$this->completeOuvrage($googleOuvrage);
234			} catch (Throwable $e) {
235			echo "* ERREUR GOOGLE Isbn Search *".$e->getMessage()."\n";
236			if( strpos($e->getMessage(), 'Could not resolve host: www.googleapis.com') === false) {
237			throw $e;
238			}
239			unset($e);
240			}
241			}
242
243			if (!isset($bnfOuvrage) && !isset($googleOuvrage)) {
244			try {
245			if ($this->verbose) {
246			dump('OpenLibrary...');
247			}
248			$openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn);
249			if (!empty($openLibraryOuvrage)) {
250			$this->completeOuvrage($openLibraryOuvrage);
251			}
252			} catch (Throwable $e) {
253			echo '**** ERREUR OpenLibrary Isbn Search';
254			}
255			}
256			}
257
258			// private function onlineQuerySearch(string $query)
259			// {
260			// echo "sleep 40...";
261			// sleep(20);
262			// onlineQuerySearch:
263			//
264			// try {
265			// dump('GOOGLE SEARCH...');
266			// // $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
267			// $adapter = new GoogleBooksAdapter();
268			// $data = $adapter->search('blabla');
269			// dump($data);
270			// //die;
271			// // return $import->getOuvrage();
272			// // $this->completeOuvrage($googleOuvrage);
273			// } catch (Throwable $e) {
274			// echo "* ERREUR GOOGLE QuerySearch * ".$e->getMessage()."\n";
275			// echo "sleep 30min";
276			// sleep(60 * 30);
277			// echo "Wake up\n";
278			// goto onlineQuerySearch;
279			// }
280			// }
281
282			private function completeOuvrage(OuvrageTemplate $onlineOuvrage)
283			{
284			if ($this->verbose) {
285			dump($onlineOuvrage->serialize(true));
286			}
287			$optimizer = new OuvrageOptimize($onlineOuvrage, $this->page);
288			$onlineOptimized = ($optimizer)->doTasks()->getOuvrage();
289
290			$completer = new OuvrageComplete($this->ouvrage, $onlineOptimized);
291			$this->ouvrage = $completer->getResult();
292
293			// todo move that optimizing in OuvrageComplete ?
294			$optimizer = new OuvrageOptimize($this->ouvrage, $this->page);
295			$this->ouvrage = $optimizer->doTasks()->getOuvrage();
296
297			if ($this->verbose) {
298			dump($completer->getLog());
299			}
300			if ($completer->major) {
301			$this->major = true;
302			}
303			$this->notCosmetic = ($completer->notCosmetic \|\| $this->notCosmetic);
304			$this->log = array_merge($this->log, $completer->getLog());
305			unset($optimizer);
306			unset($completer);
307			}
308
309			private function sendCompleted()
310			{
311			$isbn13 = $this->ouvrage->getParam('isbn') ?? null;
312
313			$finalData = [
314			// 'page' =>
315			'raw' => $this->raw,
316			'opti' => $this->serializeFinalOpti(),
317			'optidate' => date("Y-m-d H:i:s"),
318			'modifs' => mb_substr(implode(',', $this->log), 0, 250),
319			'notcosmetic' => ($this->notCosmetic) ? 1 : 0,
320			'major' => ($this->major) ? 1 : 0,
321			'isbn' => substr($isbn13, 0, 20),
322			'version' => WikiBotConfig::getGitVersion() ?? null,
323			];
324			if ($this->verbose) {
325			dump($finalData);
326			}
327			// Json ?
328			$result = $this->queueAdapter->sendCompletedData($finalData);
329
330			dump($result); // bool
331			}
332
333			/**
334			* Final serialization of the completed OuvrageTemplate.
335			*
336			* @return string
337			*/
338			private function serializeFinalOpti(): string
339			{
340			// // Améliore style compact : plus espacé
341			// if ('\|' === $this->ouvrage->userSeparator) {
342			// $this->ouvrage->userSeparator = ' \|';
343			// }
344			$finalOpti = $this->ouvrage->serialize(true);
345			$finalOpti = Normalizer::normalize($finalOpti);
346
347			return $finalOpti;
348			}
349
350			private function skipGoogle($bnfOuvrage): bool
351			{
352			if ($bnfOuvrage instanceOf OuvrageTemplate
353			&& $bnfOuvrage->hasParamValue('titre')
354			&& ($this->ouvrage->hasParamValue('lire en ligne')
355			\|\| $this->ouvrage->hasParamValue('présentation en ligne'))
356			) {
357			return true;
358			}
359
360			return false;
361			}
362			}
363

Dispositif / Wikibot

Branch — dev (8d7b92)

OuvrageCompleteWorker::isIsbnSkipped() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like