ImportSitemapCommand::execute() - Code Metrics - Inspection of "Merge pull request #16 from brevia-ai/feat/78-link..." - brevia-ai/brevia-bedita - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — main ( 27fe00...4991e0 )

by Stefano

created 2024-07-24 08:27 UTC

ImportSitemapCommand::execute() D

↳ Parent: ImportSitemapCommand

Complexity

Conditions	16
Paths	180

Size

Total Lines	80
Code Lines	56

Duplication

Lines	0
Ratio	0 %

Importance

Changes	10
Bugs	2	Features	0

Metric	Value
cc	16
eloc	56
c	10
b	2
f	0
nc	180
nop	2
dl	0
loc	80
rs	4.9

How to fix Long Method Complexity

<?php
declare(strict_types=1);

/**
 * BEdita Brevia plugin
 *
 * Copyright 2024 Atlas Srl
 */
namespace Brevia\BEdita\Command;

use BEdita\Core\Utility\LoggedUser;
use Brevia\BEdita\Client\BreviaClient;
use Brevia\BEdita\Utility\ReadCSVTrait;
use Cake\Command\Command;
use Cake\Console\Arguments;
use Cake\Console\ConsoleIo;
use Cake\Console\ConsoleOptionParser;
use Cake\Log\LogTrait;
use Cake\ORM\Table;
use Cake\Utility\Hash;

/**
 * Import links from sitemap and create links
 *
 * @property \BEdita\Core\Model\Table\ObjectsTable $Collections
 */
class ImportSitemapCommand extends Command
{
    use LogTrait;
    use ReadCSVTrait;

    /**
     * Brevia API client
     *
     * @var \Brevia\BEdita\Client\BreviaClient
     */
    protected BreviaClient $client;

    /**
     * Links Table
     *
     * @var \Cake\ORM\Table
     */
    protected Table $Links;

    /**
     * @inheritDoc
     */
    public $defaultTable = 'Collections';

    /**
     * @inheritDoc
     */
    protected function buildOptionParser(ConsoleOptionParser $parser): ConsoleOptionParser
    {
        return $parser->addOption('sitemap', [
                'help' => 'File path or URL of sitemap to import',
                'short' => 's',
                'required' => true,
            ])
            ->addOption('prefix', [
                'help' => 'Optional path prefix of URLs to import',
                'short' => 'p',
                'required' => false,
            ])
            ->addOption('black-list', [
                'help' => 'Path to a black list file containing URLs to exclude - txt file with one URL per line',
                'short' => 'b',
                'required' => false,
            ])
            ->addOption('collection', [
                'help' => 'Collection used to index (use the unique collection name)',
                'short' => 'c',
                'required' => true,
            ]);
    }

    /**
     * @inheritDoc
     */
    public function initialize(): void
    {
        $this->client = new BreviaClient();
        $this->Links = $this->fetchTable('Links');
    }

    /**
     * @inheritDoc
     */
    public function execute(Arguments $args, ConsoleIo $io)
    {
        $sitemap = $args->getOption('sitemap');
        $content = '';
        if (!empty($sitemap)) {
            if (strpos($sitemap, 'http://') !== 0 && strpos($sitemap, 'https://') !== 0 && !file_exists($sitemap)) {
                $io->abort(sprintf('File not found: %s', $sitemap));
            }
            $content = file_get_contents($sitemap);
            if ($content === false) {
                $io->abort(sprintf('Error reading sitemap URL: %s', $sitemap));
            }
        }

        $name = $args->getOption('collection');
        $response = $this->client->get('/collections', compact('name'));
        $collectionId = Hash::get($response->getJson(), '0.cmetadata.id');
        if (empty($collectionId)) {
            $io->abort(sprintf('Collection not found: %s', $name));
        }
        $collection = $this->Collections->get($collectionId, ['contain' => ['HasDocuments']]);
        $currentUrls = array_filter(array_map(function ($link) {
                $link = $link->getTable()->get($link->id);

                return $link->get('url');
        },
            (array)$collection->get('has_documents')));
        $prefix = $args->getOption('prefix');

        $blackListPath = (string)$args->getOption('black-list');
        $blackList = [];
        if (!empty($blackListPath)) {
            if (!file_exists($blackListPath)) {
                $io->abort(sprintf('Blacklist file not found: %s', $blackListPath));
            }
            $blackList = (array)file($blackListPath, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
        }

        $xml = simplexml_load_string($content);
        $json = json_encode($xml);
        $data = (array)json_decode($json, true);
        $urls = Hash::extract($data, 'url.{n}.loc');
        if (empty($urls)) {
            $io->abort('No URLs found in sitemap');
        }
        $entities = [];
        LoggedUser::setUserAdmin();
        foreach ($urls as $url) {
            if (
                in_array($url, $currentUrls) ||
                in_array(urldecode($url), $currentUrls) ||
                in_array($url, $blackList) ||
                ($prefix && strpos($url, $prefix) !== 0)
            ) {
                continue;
            }
            $io->info('Adding link: ' . $url);
            $data = [
                'status' => 'on',
                'title' => $url,
                'url' => $url,
                'extra' => [
                    'brevia' => [
                        'metadata' => [
                            'type' => 'links',
                            'url' => $url,
                        ],
                        'options' => $this->linkOptions($url, (array)$collection->get('link_load_options')),
                    ],
                ],
            ];
            $entity = $this->Links->newEntity($data);
            $entities[] = $this->Links->saveOrFail($entity);
        }
        // @phpstan-ignore-next-line
        $this->Collections->addRelated($collection, 'has_documents', $entities);

        $io->out('Done. Link added successfully: ' . count($entities));

        return null;
    }

    /**
     * Get link options
     *
     * @param string $url URL
     * @param array $linkLoadOptions Link load options
     * @return array
     */
    protected function linkOptions(string $url, array $linkLoadOptions): array
    {
        $options = array_filter($linkLoadOptions, function ($o) use ($url) {
            return $o['url'] === $url;
        });
        $selector = Hash::get($options, '0.selector');
        if (!empty($selector)) {
            return compact('selector');
        }
        $options = array_filter($linkLoadOptions, function ($o) use ($url) {
            return strpos($url, $o['url']) === 0;
        });

        return ['selector' => Hash::get($options, '0.selector')];
    }
}


1			<?php
2			declare(strict_types=1);
3
4			/**
5			* BEdita Brevia plugin
6			*
7			* Copyright 2024 Atlas Srl
8			*/
9			namespace Brevia\BEdita\Command;
10
11			use BEdita\Core\Utility\LoggedUser;
12			use Brevia\BEdita\Client\BreviaClient;
13			use Brevia\BEdita\Utility\ReadCSVTrait;
14			use Cake\Command\Command;
15			use Cake\Console\Arguments;
16			use Cake\Console\ConsoleIo;
17			use Cake\Console\ConsoleOptionParser;
18			use Cake\Log\LogTrait;
19			use Cake\ORM\Table;
20			use Cake\Utility\Hash;
21
22			/**
23			* Import links from sitemap and create links
24			*
25			* @property \BEdita\Core\Model\Table\ObjectsTable $Collections
26			*/
27			class ImportSitemapCommand extends Command
28			{
29			use LogTrait;
30			use ReadCSVTrait;
31
32			/**
33			* Brevia API client
34			*
35			* @var \Brevia\BEdita\Client\BreviaClient
36			*/
37			protected BreviaClient $client;
38
39			/**
40			* Links Table
41			*
42			* @var \Cake\ORM\Table
43			*/
44			protected Table $Links;
45
46			/**
47			* @inheritDoc
48			*/
49			public $defaultTable = 'Collections';
50
51			/**
52			* @inheritDoc
53			*/
54			protected function buildOptionParser(ConsoleOptionParser $parser): ConsoleOptionParser
55			{
56			return $parser->addOption('sitemap', [
57			'help' => 'File path or URL of sitemap to import',
58			'short' => 's',
59			'required' => true,
60			])
61			->addOption('prefix', [
62			'help' => 'Optional path prefix of URLs to import',
63			'short' => 'p',
64			'required' => false,
65			])
66			->addOption('black-list', [
67			'help' => 'Path to a black list file containing URLs to exclude - txt file with one URL per line',
68			'short' => 'b',
69			'required' => false,
70			])
71			->addOption('collection', [
72			'help' => 'Collection used to index (use the unique collection name)',
73			'short' => 'c',
74			'required' => true,
75			]);
76			}
77
78			/**
79			* @inheritDoc
80			*/
81			public function initialize(): void
82			{
83			$this->client = new BreviaClient();
84			$this->Links = $this->fetchTable('Links');
85			}
86
87			/**
88			* @inheritDoc
89			*/
90			public function execute(Arguments $args, ConsoleIo $io)
91			{
92			$sitemap = $args->getOption('sitemap');
93			$content = '';
94			if (!empty($sitemap)) {
95			if (strpos($sitemap, 'http://') !== 0 && strpos($sitemap, 'https://') !== 0 && !file_exists($sitemap)) {
96			$io->abort(sprintf('File not found: %s', $sitemap));
97			}
98			$content = file_get_contents($sitemap);
99			if ($content === false) {
100			$io->abort(sprintf('Error reading sitemap URL: %s', $sitemap));
101			}
102			}
103
104			$name = $args->getOption('collection');
105			$response = $this->client->get('/collections', compact('name'));
106			$collectionId = Hash::get($response->getJson(), '0.cmetadata.id');
107			if (empty($collectionId)) {
108			$io->abort(sprintf('Collection not found: %s', $name));
109			}
110			$collection = $this->Collections->get($collectionId, ['contain' => ['HasDocuments']]);
111			$currentUrls = array_filter(array_map(function ($link) {
112			$link = $link->getTable()->get($link->id);
113
114			return $link->get('url');
115			},
116			(array)$collection->get('has_documents')));
117			$prefix = $args->getOption('prefix');
118
119			$blackListPath = (string)$args->getOption('black-list');
120			$blackList = [];
121			if (!empty($blackListPath)) {
122			if (!file_exists($blackListPath)) {
123			$io->abort(sprintf('Blacklist file not found: %s', $blackListPath));
124			}
125			$blackList = (array)file($blackListPath, FILE_IGNORE_NEW_LINES \| FILE_SKIP_EMPTY_LINES);
126			}
127
128			$xml = simplexml_load_string($content);
129			$json = json_encode($xml);
130			$data = (array)json_decode($json, true);
131			$urls = Hash::extract($data, 'url.{n}.loc');
132			if (empty($urls)) {
133			$io->abort('No URLs found in sitemap');
134			}
135			$entities = [];
136			LoggedUser::setUserAdmin();
137			foreach ($urls as $url) {
138			if (
139			in_array($url, $currentUrls) \|\|
140			in_array(urldecode($url), $currentUrls) \|\|
141			in_array($url, $blackList) \|\|
142			($prefix && strpos($url, $prefix) !== 0)
143			) {
144			continue;
145			}
146			$io->info('Adding link: ' . $url);
147			$data = [
148			'status' => 'on',
149			'title' => $url,
150			'url' => $url,
151			'extra' => [
152			'brevia' => [
153			'metadata' => [
154			'type' => 'links',
155			'url' => $url,
156			],
157			'options' => $this->linkOptions($url, (array)$collection->get('link_load_options')),
158			],
159			],
160			];
161			$entity = $this->Links->newEntity($data);
162			$entities[] = $this->Links->saveOrFail($entity);
163			}
164			// @phpstan-ignore-next-line
165			$this->Collections->addRelated($collection, 'has_documents', $entities);
166
167			$io->out('Done. Link added successfully: ' . count($entities));
168
169			return null;
170			}
171
172			/**
173			* Get link options
174			*
175			* @param string $url URL
176			* @param array $linkLoadOptions Link load options
177			* @return array
178			*/
179			protected function linkOptions(string $url, array $linkLoadOptions): array
180			{
181			$options = array_filter($linkLoadOptions, function ($o) use ($url) {
182			return $o['url'] === $url;
183			});
184			$selector = Hash::get($options, '0.selector');
185			if (!empty($selector)) {
186			return compact('selector');
187			}
188			$options = array_filter($linkLoadOptions, function ($o) use ($url) {
189			return strpos($url, $o['url']) === 0;
190			});
191
192			return ['selector' => Hash::get($options, '0.selector')];
193			}
194			}
195

brevia-ai / brevia-bedita

Push — main ( 27fe00...4991e0 )

ImportSitemapCommand::execute() D

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like