QueryRegexFinder::loadRules() - Code Metrics - ArthurHoaro/web-thumbnailer - Measure and Improve Code Quality continuously with Scrutinizer

QueryRegexFinder::loadRules() A
last analyzed 2022-11-24 19:40 UTC

↳ Parent: QueryRegexFinder

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	3
nc	1
nop	1
dl	0
loc	5
rs	10
c	0
b	0
f	0

<?php

declare(strict_types=1);

namespace WebThumbnailer\Finder;

use WebThumbnailer\Application\ConfigManager;
use WebThumbnailer\Application\WebAccess\WebAccess;
use WebThumbnailer\Application\WebAccess\WebAccessCUrl;
use WebThumbnailer\Application\WebAccess\WebAccessFactory;
use WebThumbnailer\Exception\BadRulesException;
use WebThumbnailer\Utils\FinderUtils;

/**
 * Generic Finder using regex rules on remote web content.
 * It will use regex rules to resolve a thumbnail in web a page.
 *
 * Mandatory rules:
 *   - image_regex
 *   - thumbnail_url
 *
 * Example:
 *   1. `http://domain.tld/page` content will be downloaded.
 *   2. `image_regex` will be apply on the content
 *   3. Matches will be use to generate `thumbnail_url`.
 */
class QueryRegexFinder extends FinderCommon
{
    /** @var WebAccess instance. */
    protected $webAccess;

    /** @var string thumbnail_url rule. */
    protected $thumbnailUrlFormat;

    /** @var string Regex to apply on provided URL. */
    protected $urlRegex;

    /**
     * @inheritdoc
     * @param mixed[]|null $rules   All existing rules loaded from JSON files.
     * @param mixed[]|null $options Options provided by the user to retrieve a thumbnail.
     *
     * @throws BadRulesException
     */
    public function __construct(string $domain, string $url, ?array $rules, ?array $options)
    {
        $this->webAccess = WebAccessFactory::getWebAccess($url);
        $this->url = $url;
        $this->domain = $domain;
        $this->loadRules($rules);
        $this->finderOptions = $options;
    }

    /**
     * This finder downloads target URL page, and apply the regex given in rules on its content
     * to extract the thumbnail image.
     * The thumb URL must include ${number} to be replaced from the regex match.
     * Also replace eventual URL options.
     *
     * @inheritdoc
     *
     * @throws BadRulesException
     */
    public function find()
    {
        $thumbnail = $content = null;
        $callback = $this->webAccess instanceof WebAccessCUrl
            ? $this->getCurlCallback($content, $thumbnail)
            : null;
        list($headers, $content) = $this->webAccess->getContent(
            $this->url,
            (int) ConfigManager::get('settings.default.timeout', 30),
            (int) ConfigManager::get('settings.default.max_img_dl', 16777216),
            $callback,
            $content
        );
        if (
            empty($content)
            || empty($headers)
            || (empty($thumbnail) && strpos($headers[0], '200') === false)
        ) {
            return false;
        }

        // With curl, the thumb is extracted during the download
        if ($this->webAccess instanceof WebAccessCUrl && ! empty($thumbnail)) {
            return $thumbnail;
        }

        return $this->extractThumbContent($content);
    }

    /**
     * Get a callback for curl write function.
     *
     * @param string|null $content   A variable reference in which the downloaded content should be stored.
     * @param string|null $thumbnail A variable reference in which extracted thumb URL should be stored.
     *
     * @return callable CURLOPT_WRITEFUNCTION callback
     */
    protected function getCurlCallback(?string &$content, ?string &$thumbnail): callable
    {
        $isRedirected = false;

        /**
         * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
         *
         * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
         * Then we extract the title and the charset and stop the download when it's done.
         *
         * Note that when using CURLOPT_WRITEFUNCTION, we have to manually handle the content retrieved,
         * hence the $content reference variable.
         *
         * @param resource $ch   cURL resource
         * @param string   $data chunk of data being downloaded
         *
         * @return int|false length of $data or false if we need to stop the download
         */
        return function ($ch, $data) use (&$content, &$thumbnail, &$isRedirected) {
            $content .= $data;
            $responseCode = curl_getinfo($ch, CURLINFO_RESPONSE_CODE);

            if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
                $isRedirected = true;
                return strlen($data);
            }
            if (!empty($responseCode) && $responseCode !== 200) {
                return false;
            }
            // After a redirection, the content type will keep the previous request value
            // until it finds the next content-type header.
            if (! $isRedirected || strpos(strtolower($data), 'content-type') !== false) {
                $contentType = curl_getinfo($ch, CURLINFO_CONTENT_TYPE);
            }
            if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
                return false;
            }
            if (empty($thumbnail)) {
                $thumbnail = $this->extractThumbContent($data);
            }
            // We got everything we want, stop the download.
            if (!empty($responseCode) && !empty($contentType) && !empty($thumbnail)) {
                return false;
            }

            return strlen($data);
        };
    }

    /**
     * @param string $content to extract thumb from
     *
     * @return string|false Thumbnail URL or false if not found
     *
     * @throws BadRulesException
     */
    public function extractThumbContent(string $content)
    {
        $thumbnailUrl = $this->thumbnailUrlFormat;
        if (preg_match($this->urlRegex, $content, $matches) !== 0) {
            $total = count($matches);
            for ($i = 1; $i < $total; $i++) {
                $thumbnailUrl = str_replace('${' . $i . '}', $matches[$i], $thumbnailUrl);
            }

            // Match only options (not ${number})
            if (preg_match_all('/\${((?!\d)\w+?)}/', $thumbnailUrl, $optionsMatch, PREG_PATTERN_ORDER)) {
                foreach ($optionsMatch[1] as $value) {
                    $thumbnailUrl = $this->replaceOption($thumbnailUrl, $value);
                }
            }
            return $thumbnailUrl;
        }

        return false;
    }

    /** @inheritdoc */
    public function checkRules(?array $rules): bool
    {
        if (count($rules ?? []) > 0 && !FinderUtils::checkMandatoryRules($rules, ['image_regex', 'thumbnail_url'])) {
            throw new BadRulesException();
        }

        return true;
    }

    /**
     * @inheritdoc
     *
     * @throws BadRulesException
     */
    public function loadRules(?array $rules): void
    {
        $this->checkRules($rules);
        $this->urlRegex = FinderUtils::buildRegex($rules['image_regex'], 'im');
        $this->thumbnailUrlFormat = $rules['thumbnail_url'];
    }

    /** @inheritdoc */
    public function getName(): string
    {
        return 'Query Regex';
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace WebThumbnailer\Finder;
6
7			use WebThumbnailer\Application\ConfigManager;
8			use WebThumbnailer\Application\WebAccess\WebAccess;
9			use WebThumbnailer\Application\WebAccess\WebAccessCUrl;
10			use WebThumbnailer\Application\WebAccess\WebAccessFactory;
11			use WebThumbnailer\Exception\BadRulesException;
12			use WebThumbnailer\Utils\FinderUtils;
13
14			/**
15			* Generic Finder using regex rules on remote web content.
16			* It will use regex rules to resolve a thumbnail in web a page.
17			*
18			* Mandatory rules:
19			* - image_regex
20			* - thumbnail_url
21			*
22			* Example:
23			* 1. `http://domain.tld/page` content will be downloaded.
24			* 2. `image_regex` will be apply on the content
25			* 3. Matches will be use to generate `thumbnail_url`.
26			*/
27			class QueryRegexFinder extends FinderCommon
28			{
29			/** @var WebAccess instance. */
30			protected $webAccess;
31
32			/** @var string thumbnail_url rule. */
33			protected $thumbnailUrlFormat;
34
35			/** @var string Regex to apply on provided URL. */
36			protected $urlRegex;
37
38			/**
39			* @inheritdoc
40			* @param mixed[]\|null $rules All existing rules loaded from JSON files.
41			* @param mixed[]\|null $options Options provided by the user to retrieve a thumbnail.
42			*
43			* @throws BadRulesException
44			*/
45			public function __construct(string $domain, string $url, ?array $rules, ?array $options)
46			{
47			$this->webAccess = WebAccessFactory::getWebAccess($url);
48			$this->url = $url;
49			$this->domain = $domain;
50			$this->loadRules($rules);
51			$this->finderOptions = $options;
52			}
53
54			/**
55			* This finder downloads target URL page, and apply the regex given in rules on its content
56			* to extract the thumbnail image.
57			* The thumb URL must include ${number} to be replaced from the regex match.
58			* Also replace eventual URL options.
59			*
60			* @inheritdoc
61			*
62			* @throws BadRulesException
63			*/
64			public function find()
65			{
66			$thumbnail = $content = null;
67			$callback = $this->webAccess instanceof WebAccessCUrl
68			? $this->getCurlCallback($content, $thumbnail)
69			: null;
70			list($headers, $content) = $this->webAccess->getContent(
71			$this->url,
72			(int) ConfigManager::get('settings.default.timeout', 30),
73			(int) ConfigManager::get('settings.default.max_img_dl', 16777216),
74			$callback,
75			$content
76			);
77			if (
78			empty($content)
79			\|\| empty($headers)
80			\|\| (empty($thumbnail) && strpos($headers[0], '200') === false)
81			) {
82			return false;
83			}
84
85			// With curl, the thumb is extracted during the download
86			if ($this->webAccess instanceof WebAccessCUrl && ! empty($thumbnail)) {
87			return $thumbnail;
88			}
89
90			return $this->extractThumbContent($content);
91			}
92
93			/**
94			* Get a callback for curl write function.
95			*
96			* @param string\|null $content A variable reference in which the downloaded content should be stored.
97			* @param string\|null $thumbnail A variable reference in which extracted thumb URL should be stored.
98			*
99			* @return callable CURLOPT_WRITEFUNCTION callback
100			*/
101			protected function getCurlCallback(?string &$content, ?string &$thumbnail): callable
102			{
103			$isRedirected = false;
104
105			/**
106			* cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
107			*
108			* While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
109			* Then we extract the title and the charset and stop the download when it's done.
110			*
111			* Note that when using CURLOPT_WRITEFUNCTION, we have to manually handle the content retrieved,
112			* hence the $content reference variable.
113			*
114			* @param resource $ch cURL resource
115			* @param string $data chunk of data being downloaded
116			*
117			* @return int\|false length of $data or false if we need to stop the download
118			*/
119			return function ($ch, $data) use (&$content, &$thumbnail, &$isRedirected) {
120			$content .= $data;
121			$responseCode = curl_getinfo($ch, CURLINFO_RESPONSE_CODE);
122
123			if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
124			$isRedirected = true;
125			return strlen($data);
126			}
127			if (!empty($responseCode) && $responseCode !== 200) {
128			return false;
129			}
130			// After a redirection, the content type will keep the previous request value
131			// until it finds the next content-type header.
132			if (! $isRedirected \|\| strpos(strtolower($data), 'content-type') !== false) {
133			$contentType = curl_getinfo($ch, CURLINFO_CONTENT_TYPE);
134			}
135			if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
136			return false;
137			}
138			if (empty($thumbnail)) {
139			$thumbnail = $this->extractThumbContent($data);
140			}
141			// We got everything we want, stop the download.
142			if (!empty($responseCode) && !empty($contentType) && !empty($thumbnail)) {
143			return false;
144			}
145
146			return strlen($data);
147			};
148			}
149
150			/**
151			* @param string $content to extract thumb from
152			*
153			* @return string\|false Thumbnail URL or false if not found
154			*
155			* @throws BadRulesException
156			*/
157			public function extractThumbContent(string $content)
158			{
159			$thumbnailUrl = $this->thumbnailUrlFormat;
160			if (preg_match($this->urlRegex, $content, $matches) !== 0) {
161			$total = count($matches);
162			for ($i = 1; $i < $total; $i++) {
163			$thumbnailUrl = str_replace('${' . $i . '}', $matches[$i], $thumbnailUrl);
164			}
165
166			// Match only options (not ${number})
167			if (preg_match_all('/\${((?!\d)\w+?)}/', $thumbnailUrl, $optionsMatch, PREG_PATTERN_ORDER)) {
168			foreach ($optionsMatch[1] as $value) {
169			$thumbnailUrl = $this->replaceOption($thumbnailUrl, $value);
170			}
171			}
172			return $thumbnailUrl;
173			}
174
175			return false;
176			}
177
178			/** @inheritdoc */
179			public function checkRules(?array $rules): bool
180			{
181			if (count($rules ?? []) > 0 && !FinderUtils::checkMandatoryRules($rules, ['image_regex', 'thumbnail_url'])) {
182			throw new BadRulesException();
183			}
184
185			return true;
186			}
187
188			/**
189			* @inheritdoc
190			*
191			* @throws BadRulesException
192			*/
193			public function loadRules(?array $rules): void
194			{
195			$this->checkRules($rules);
196			$this->urlRegex = FinderUtils::buildRegex($rules['image_regex'], 'im');
197			$this->thumbnailUrlFormat = $rules['thumbnail_url'];
198			}
199
200			/** @inheritdoc */
201			public function getName(): string
202			{
203			return 'Query Regex';
204			}
205			}
206

ArthurHoaro / web-thumbnailer

QueryRegexFinder::loadRules() A last analyzed 2022-11-24 19:40 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

QueryRegexFinder::loadRules() A
last analyzed 2022-11-24 19:40 UTC