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

Passed

Push — dev ( b8ced8...f7ea32 )

by Dispositif

created 2020-04-11 02:33 UTC

CompleteProcess::skipGoogle() A

↳ Parent: CompleteProcess

Complexity

Conditions	5
Paths	2

Size

Total Lines	11
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	5
eloc	6
nc	2
nop	1
dl	0
loc	11
rs	9.6111
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\WikidataAdapter;
use GuzzleHttp\Client;
use Normalizer;
use Throwable;

/**
 * Class CompleteProcess
 */
class CompleteProcess
{
    /**
     * 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 string|null
     * @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 ($this->verbose) {
                dump('BIBLIO NAT FRANCE...');
            }
            // BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM !
            if ($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\WikidataAdapter;
19			use GuzzleHttp\Client;
20			use Normalizer;
21			use Throwable;
22
23			/**
24			* Class CompleteProcess
25			*/
26			class CompleteProcess
27			{
28			/**
29			* Exclusion requête BnF/Google/etc
30			* Format EAN ou ISBN10 sans tiret.
31			*/
32			const ISBN_EAN_SKIP
33			= [
34			'9782918758440', // Profils de lignes du réseau ferré français vol.2
35			'9782918758341', // Profils de lignes du réseau ferré français vol.1
36			];
37			/**
38			* @var bool
39			*/
40			public $verbose = false;
41
42			/**
43			* @var QueueInterface
44			*/
45			private $queueAdapter;
46			/**
47			* @var string
48			*/
49			private $raw = '';
50			private $page; // article title
51
52			private $log = [];
53			private $notCosmetic = false;
54			private $major = false;
55			/**
56			* @var OuvrageTemplate
57			*/
58			private $ouvrage;
59
60			public function __construct(QueueInterface $queueAdapter, ?bool $verbose = false)
61			{
62			$this->queueAdapter = $queueAdapter;
63			$this->verbose = (bool)$verbose;
64			}
65
66			public function run(?int $limit = 10000)
67			{
68			$memory = new Memory();
69			while ($limit > 0) {
70			$limit--;
71			sleep(1);
72			$row = $this->getNewRow2Complete();
73			$this->raw = $row['raw'];
74			$this->page = $row['page'];
75
76			echo sprintf(
77			"-------------------------------\n%s [%s]\n%s\n%s\n",
78			date("Y-m-d H:i:s"),
79			WikiBotConfig::getGitVersion() ?? '',
80			$this->page,
81			$this->raw
82			);
83			if ($this->verbose) {
84			$memory->echoMemory(true);
85			}
86
87			// initialise variables
88			$this->log = [];
89			$this->ouvrage = null;
90			$this->notCosmetic = false;
91			$this->major = false;
92
93
94			try {
95			$parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw);
96			$origin = $parse['ouvrage'][0]['model'] ?? null;
97			} catch (Throwable $e) {
98			echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
99			continue;
100			}
101
102			if (!$origin instanceof OuvrageTemplate) {
103			echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw);
104			continue;
105			}
106
107			// Final optimizing (with online predictions)
108			$optimizer = new OuvrageOptimize($origin, $this->page);
109			$optimizer->doTasks();
110			$this->ouvrage = $optimizer->getOuvrage();
111			$this->log = array_merge($this->log, $optimizer->getLog());
112			$this->notCosmetic = ($optimizer->notCosmetic \|\| $this->notCosmetic);
113
114			/**
115			* RECHERCHE ONLINE
116			*/
117			$isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN
118			$isbn10 = $origin->getParam('isbn2') ?? $origin->getParam('isbn10') ?? null;
119			if (!empty($isbn)
120			&& !$origin->hasParamValue('isbn invalide')
121			&& !$origin->hasParamValue('isbn erroné')
122			) {
123			$this->onlineIsbnSearch($isbn, $isbn10);
124			}
125
126			$this->sendCompleted();
127			unset($optimizer);
128			unset($parse);
129			unset($origin);
130			} // END WHILE
131
132			return true;
133			}
134
135			/**
136			* Get array (title+raw strings) to complete from AMQP queue, SQL Select or file reading.
137			*
138			* @return string\|null
139			* @throws \Exception
140			*/
141			private function getNewRow2Complete(): ?array
142			{
143			$row = $this->queueAdapter->getNewRaw();
144			if (empty($row) \|\| empty($row['raw'])) {
145			echo "STOP: no more queue to process \n";
146			throw new \Exception('no more queue to process');
147			}
148
149			return $row;
			1 ignored issue – show Bug Best Practice introduced 2020-04-11 02:40 UTC by Report Bug Copy Issue Report The expression `return $row` returns the type `array` which is incompatible with the documented return type `null\|string`. Loading history...
150			}
151
152			/**
153			* @param string $isbn
154			* @param string\|null $isbn10
155			*
156			* @return bool
157			*/
158			private function isIsbnSkipped(string $isbn, ?string $isbn10 = null): bool
159			{
160			if (in_array(str_replace('-', '', $isbn), self::ISBN_EAN_SKIP)
161			\|\| ($isbn10 !== null
162			&& in_array(str_replace('-', '', $isbn10), self::ISBN_EAN_SKIP))
163			) {
164			return true;
165			}
166
167			return false;
168			}
169
170			private function onlineIsbnSearch(string $isbn, ?string $isbn10 = null)
171			{
172			if ($this->isIsbnSkipped($isbn, $isbn10)) {
173			echo "* SKIP THAT ISBN *\n";
174
175			// Vérifier logique return
176			return;
177			}
178
179			online:
180			if ($this->verbose) {
181			echo "sleep 10...\n";
182			}
183			sleep(10);
184
185			try {
186			if ($this->verbose) {
187			dump('BIBLIO NAT FRANCE...');
188			}
189			// BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM !
190			if ($isbn10) {
191			$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn10);
192			sleep(2);
193			}
194			if (!$isbn10 \|\| empty($bnfOuvrage) \|\| empty($bnfOuvrage->getParam('titre'))) {
195			$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn);
196			}
197			if (isset($bnfOuvrage) and $bnfOuvrage instanceof OuvrageTemplate) {
198			$this->completeOuvrage($bnfOuvrage);
199
200			// Wikidata requests from $infos (ISBN/ISNI)
201			if (!empty($bnfOuvrage->getInfos())) {
202			if ($this->verbose) {
203			dump('WIKIDATA...');
204			}
205			// TODO move to factory
206			$wikidataAdapter = new WikidataAdapter(
207			new Client(['timeout' => 5, 'headers' => ['User-Agent' => getenv('USER_AGENT')]])
208			);
209			$wdComplete = new Wikidata2Ouvrage($wikidataAdapter, clone $bnfOuvrage, $this->page);
210			$this->completeOuvrage($wdComplete->getOuvrage());
211			}
212			}
213			} catch (Throwable $e) {
214			echo sprintf(
215			"*** ERREUR BnF Isbn Search %s %s %s \n",
216			$e->getMessage(),
217			$e->getFile(),
218			$e->getLine()
219			);
220			}
221
222			if (!isset($bnfOuvrage) \|\| !$this->skipGoogle($bnfOuvrage)) {
223			try {
224			if ($this->verbose) {
225			dump('GOOGLE...');
226			}
227			$googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
228			$this->completeOuvrage($googleOuvrage);
229			} catch (Throwable $e) {
230			echo "* ERREUR GOOGLE Isbn Search *".$e->getMessage()."\n";
231			if( strpos($e->getMessage(), 'Could not resolve host: www.googleapis.com') === false) {
232			throw $e;
233			}
234			unset($e);
235			}
236			}
237
238			if (!isset($bnfOuvrage) && !isset($googleOuvrage)) {
239			try {
240			if ($this->verbose) {
241			dump('OpenLibrary...');
242			}
243			$openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn);
244			if (!empty($openLibraryOuvrage)) {
245			$this->completeOuvrage($openLibraryOuvrage);
246			}
247			} catch (Throwable $e) {
248			echo '**** ERREUR OpenLibrary Isbn Search';
249			}
250			}
251			}
252
253			// private function onlineQuerySearch(string $query)
254			// {
255			// echo "sleep 40...";
256			// sleep(20);
257			// onlineQuerySearch:
258			//
259			// try {
260			// dump('GOOGLE SEARCH...');
261			// // $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn);
262			// $adapter = new GoogleBooksAdapter();
263			// $data = $adapter->search('blabla');
264			// dump($data);
265			// //die;
266			// // return $import->getOuvrage();
267			// // $this->completeOuvrage($googleOuvrage);
268			// } catch (Throwable $e) {
269			// echo "* ERREUR GOOGLE QuerySearch * ".$e->getMessage()."\n";
270			// echo "sleep 30min";
271			// sleep(60 * 30);
272			// echo "Wake up\n";
273			// goto onlineQuerySearch;
274			// }
275			// }
276
277			private function completeOuvrage(OuvrageTemplate $onlineOuvrage)
278			{
279			if ($this->verbose) {
280			dump($onlineOuvrage->serialize(true));
281			}
282			$optimizer = new OuvrageOptimize($onlineOuvrage, $this->page);
283			$onlineOptimized = ($optimizer)->doTasks()->getOuvrage();
284
285			$completer = new OuvrageComplete($this->ouvrage, $onlineOptimized);
286			$this->ouvrage = $completer->getResult();
287
288			// todo move that optimizing in OuvrageComplete ?
289			$optimizer = new OuvrageOptimize($this->ouvrage, $this->page);
290			$this->ouvrage = $optimizer->doTasks()->getOuvrage();
291
292			if ($this->verbose) {
293			dump($completer->getLog());
294			}
295			if ($completer->major) {
296			$this->major = true;
297			}
298			$this->notCosmetic = ($completer->notCosmetic \|\| $this->notCosmetic);
299			$this->log = array_merge($this->log, $completer->getLog());
300			unset($optimizer);
301			unset($completer);
302			}
303
304			private function sendCompleted()
305			{
306			$isbn13 = $this->ouvrage->getParam('isbn') ?? null;
307
308			$finalData = [
309			// 'page' =>
310			'raw' => $this->raw,
311			'opti' => $this->serializeFinalOpti(),
312			'optidate' => date("Y-m-d H:i:s"),
313			'modifs' => mb_substr(implode(',', $this->log), 0, 250),
314			'notcosmetic' => ($this->notCosmetic) ? 1 : 0,
315			'major' => ($this->major) ? 1 : 0,
316			'isbn' => substr($isbn13, 0, 20),
317			'version' => WikiBotConfig::getGitVersion() ?? null,
318			];
319			if ($this->verbose) {
320			dump($finalData);
321			}
322			// Json ?
323			$result = $this->queueAdapter->sendCompletedData($finalData);
324
325			dump($result); // bool
326			}
327
328			/**
329			* Final serialization of the completed OuvrageTemplate.
330			*
331			* @return string
332			*/
333			private function serializeFinalOpti(): string
334			{
335			// // Améliore style compact : plus espacé
336			// if ('\|' === $this->ouvrage->userSeparator) {
337			// $this->ouvrage->userSeparator = ' \|';
338			// }
339			$finalOpti = $this->ouvrage->serialize(true);
340			$finalOpti = Normalizer::normalize($finalOpti);
341
342			return $finalOpti;
343			}
344
345			private function skipGoogle($bnfOuvrage): bool
346			{
347			if ($bnfOuvrage instanceOf OuvrageTemplate
348			&& $bnfOuvrage->hasParamValue('titre')
349			&& ($this->ouvrage->hasParamValue('lire en ligne')
350			\|\| $this->ouvrage->hasParamValue('présentation en ligne'))
351			) {
352			return true;
353			}
354
355			return false;
356			}
357			}
358

Dispositif / Wikibot

Push — dev ( b8ced8...f7ea32 )

CompleteProcess::skipGoogle() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like