Spatie\Crawler\PdoCrawlQueue

Complexity

Total Complexity

13

Size/Duplication

Total Lines	165
Duplicated Lines	18.18 %

Coupling/Cohesion

Components	1
Dependencies	2

Importance

Changes

0

8 Methods

Rating	Name	Duplication	Size	Complexity
A	markAsProcessed()	0	11	1
A	__construct()	0	5	1
B	add()	0	41	3
A	hasPendingUrls()	0	13	1
A	getUrlById()	0	17	2
A	hasAlreadyBeenProcessed()	13	14	1
A	has()	0	17	2
A	getPendingUrl()	17	17	2

<?php

namespace Spatie\Crawler;

use Spatie\Crawler\Exception\UrlNotFoundByIndex;

class PdoCrawlQueue implements CrawlQueue
{
    const STATUS_PENDING = 1;
    const STATUS_PROCESSED = 2;

    /** @var \Pdo */
    private $pdo;

    /** @var string */
    private $table;

    public function __construct(\Pdo $pdo, string $table)
    {
        $this->pdo = $pdo;
        $this->table = $table;
    }

    public function add(CrawlUrl $url): CrawlQueue
    {
        if ($this->has($url)) {
            return $this;
        }

        $data = serialize($url);

        $this->pdo->beginTransaction();

        try {
            $statement = $this->pdo->prepare(
                'insert into '.$this->table.' (url, status, object) values (:url, :status, :object)'
            );
            $statement->execute([
                'url' => $url->url,
                'object' => $data,
                'status' => self::STATUS_PENDING,
            ]);
            $statement->closeCursor();
            $url->setId($this->pdo->lastInsertId());

            $statement = $this->pdo->prepare(
                'update '.$this->table.' set object = :object where id = :id'
            );

            $data = serialize($url);

            $statement->execute([
                'object' => $data,
                'id' => $url->getId(),
            ]);
            $statement->closeCursor();
            $this->pdo->commit();
        } catch (\Exception $exception) {
            $this->pdo->rollBack();
            throw $exception;
        }

        return $this;
    }

    public function hasPendingUrls(): bool
    {
        $statement = $this->pdo->prepare(
            'select url from '.$this->table.' where status = :status limit 1'
        );
        $statement->execute([
            'status' => self::STATUS_PENDING,
        ]);
        $result = $statement->fetch();
        $statement->closeCursor();

        return (bool) $result;
    }

    /**
     * @param int $id
     *
     * @return \Spatie\Crawler\CrawlUrl|null
     */
    public function getUrlById(int $id): CrawlUrl
    {
        $statement = $this->pdo->prepare(
            'select object from '.$this->table.' where id = :id'
        );
        $statement->execute([
            'id' => $id,
        ]);
        $result = $statement->fetch();
        $statement->closeCursor();

        if (! $result) {
            throw new UrlNotFoundByIndex(sprintf('#%d crawl url not found in database', $id));
        }

        return unserialize($result['object']);
    }

    public function hasAlreadyBeenProcessed(CrawlUrl $url)

This method seems to be duplicated in your project.

    {
        $statement = $this->pdo->prepare(
            'select id from '.$this->table.' where url = :url and status = :status'
        );
        $statement->execute([
            'url' => $url->url,
            'status' => self::STATUS_PROCESSED,
        ]);
        $result = $statement->fetch();
        $statement->closeCursor();

        return (bool) $result;
    }

    public function markAsProcessed(CrawlUrl $crawlUrl)
    {
        $statement = $this->pdo->prepare(
            'update  '.$this->table.' set status = :status where url = :url'
        );
        $statement->execute([
            'url' => $crawlUrl->url,
            'status' => self::STATUS_PROCESSED,
        ]);
        $statement->closeCursor();
    }

    /**
     * @param CrawlUrl|Url $crawlUrl
     *
     * @return bool
     */
    public function has($crawlUrl): bool
    {
        if ($crawlUrl instanceof Url) {
            $crawlUrl = CrawlUrl::create($crawlUrl);
        }

        $statement = $this->pdo->prepare(
            'select id from '.$this->table.' where url = :url'
        );

        $statement->execute([
            'url' => $crawlUrl->url,
        ]);
        $result = $statement->fetch();

        return (bool) $result;
    }

    /** @return \Spatie\Crawler\CrawlUrl|null */
    public function getPendingUrl()

This method seems to be duplicated in your project.

    {
        $statement = $this->pdo->prepare(
            'select object from '.$this->table.' where status = :status limit 1'
        );
        $statement->execute([
            'status' => self::STATUS_PENDING,
        ]);
        $result = $statement->fetch();
        $statement->closeCursor();

        if (! $result) {
            return;
        }

        return unserialize($result['object']);
    }
}