QueryRegexFinder::find() - Code Metrics - Inspection of "Support redirection in cURL download callback" - ArthurHoaro/web-thumbnailer - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( e15538...7938c9 )

by Arthur

created 2018-05-01 14:50 UTC

QueryRegexFinder::find() C

↳ Parent: QueryRegexFinder

Complexity

Conditions	8
Paths	6

Size

Total Lines	26
Code Lines	17

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	8
eloc	17
nc	6
nop	0
dl	0
loc	26
rs	5.3846
c	0
b	0
f	0

<?php

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;

/**
 * Class QueryRegexFinder
 *
 * 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`.
 *
 * @package WebThumbnailer\Finder
 */
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
     *
     * @throws BadRulesException
     */
    public function __construct($domain, $url, $rules, $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 $content   A variable reference in which the downloaded content should be stored.
     * @param string $thumbnail A variable reference in which extracted thumb URL should be stored.
     *
     * @return \Closure CURLOPT_WRITEFUNCTION callback
     */
    protected function getCurlCallback(&$content, &$thumbnail)
    {
        $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|bool 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 $content
     * @return bool|mixed|string
     * @throws BadRulesException
     */
    public function extractThumbContent($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($rules)
    {
        if (! FinderUtils::checkMandatoryRules($rules, [
            'image_regex',
            'thumbnail_url'
        ])) {
            throw new BadRulesException();
        }
    }

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

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


1			<?php
2
3			namespace WebThumbnailer\Finder;
4
5			use WebThumbnailer\Application\ConfigManager;
6			use WebThumbnailer\Application\WebAccess\WebAccess;
7			use WebThumbnailer\Application\WebAccess\WebAccessCUrl;
8			use WebThumbnailer\Application\WebAccess\WebAccessFactory;
9			use WebThumbnailer\Exception\BadRulesException;
10			use WebThumbnailer\Utils\FinderUtils;
11
12			/**
13			* Class QueryRegexFinder
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			* @package WebThumbnailer\Finder
28			*/
29			class QueryRegexFinder extends FinderCommon
30			{
31			/**
32			* @var WebAccess instance.
33			*/
34			protected $webAccess;
35
36			/**
37			* @var string thumbnail_url rule.
38			*/
39			protected $thumbnailUrlFormat;
40
41			/**
42			* @var string Regex to apply on provided URL.
43			*/
44			protected $urlRegex;
45
46			/**
47			* @inheritdoc
48			*
49			* @throws BadRulesException
50			*/
51			public function __construct($domain, $url, $rules, $options)
52			{
53			$this->webAccess = WebAccessFactory::getWebAccess($url);
54			$this->url = $url;
55			$this->domain = $domain;
56			$this->loadRules($rules);
57			$this->finderOptions = $options;
58			}
59
60			/**
61			* This finder downloads target URL page, and apply the regex given in rules on its content
62			* to extract the thumbnail image.
63			* The thumb URL must include ${number} to be replaced from the regex match.
64			* Also replace eventual URL options.
65			*
66			* @inheritdoc
67			*
68			* @throws BadRulesException
69			*/
70			public function find()
71			{
72			$thumbnail = $content = null;
73			$callback = $this->webAccess instanceof WebAccessCUrl
74			? $this->getCurlCallback($content, $thumbnail)
75			: null;
76			list($headers, $content) = $this->webAccess->getContent(
77			$this->url,
78			(int) ConfigManager::get('settings.default.timeout', 30),
79			(int) ConfigManager::get('settings.default.max_img_dl', 16777216),
80			$callback,
81			$content
82			);
83			if (empty($content)
84			\|\| empty($headers)
85			\|\| (empty($thumbnail) && strpos($headers[0], '200') === false)
86			) {
87			return false;
88			}
89
90			// With curl, the thumb is extracted during the download
91			if ($this->webAccess instanceof WebAccessCUrl && ! empty($thumbnail)) {
92			return $thumbnail;
93			}
94
95			return $this->extractThumbContent($content);
96			}
97
98			/**
99			* Get a callback for curl write function.
100			*
101			* @param string $content A variable reference in which the downloaded content should be stored.
102			* @param string $thumbnail A variable reference in which extracted thumb URL should be stored.
103			*
104			* @return \Closure CURLOPT_WRITEFUNCTION callback
105			*/
106			protected function getCurlCallback(&$content, &$thumbnail)
107			{
108			$isRedirected = false;
109
110			/**
111			* cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
112			*
113			* While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
114			* Then we extract the title and the charset and stop the download when it's done.
115			*
116			* Note that when using CURLOPT_WRITEFUNCTION, we have to manually handle the content retrieved,
117			* hence the $content reference variable.
118			*
119			* @param resource $ch cURL resource
120			* @param string $data chunk of data being downloaded
121			*
122			* @return int\|bool length of $data or false if we need to stop the download
123			*/
124			return function (&$ch, $data) use (&$content, &$thumbnail, &$isRedirected) {
125			$content .= $data;
126			$responseCode = curl_getinfo($ch, CURLINFO_RESPONSE_CODE);
127
128			if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
129			$isRedirected = true;
130			return strlen($data);
131			}
132			if (!empty($responseCode) && $responseCode !== 200) {
133			return false;
134			}
135			// After a redirection, the content type will keep the previous request value
136			// until it finds the next content-type header.
137			if (! $isRedirected \|\| strpos(strtolower($data), 'content-type') !== false) {
138			$contentType = curl_getinfo($ch, CURLINFO_CONTENT_TYPE);
139			}
140			if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
141			return false;
142			}
143			if (empty($thumbnail)) {
144			$thumbnail = $this->extractThumbContent($data);
145			}
146			// We got everything we want, stop the download.
147			if (!empty($responseCode) && !empty($contentType) && !empty($thumbnail)) {
148			return false;
149			}
150
151			return strlen($data);
152			};
153			}
154
155			/**
156			* @param $content
157			* @return bool\|mixed\|string
158			* @throws BadRulesException
159			*/
160			public function extractThumbContent($content)
161			{
162			$thumbnailUrl = $this->thumbnailUrlFormat;
163			if (preg_match($this->urlRegex, $content, $matches) !== 0) {
164			$total = count($matches);
165			for ($i = 1; $i < $total; $i++) {
166			$thumbnailUrl = str_replace('${'. $i . '}', $matches[$i], $thumbnailUrl);
167			}
168
169			// Match only options (not ${number})
170			if (preg_match_all('/\${((?!\d)\w+?)}/', $thumbnailUrl, $optionsMatch, PREG_PATTERN_ORDER)) {
171			foreach ($optionsMatch[1] as $value) {
172			$thumbnailUrl = $this->replaceOption($thumbnailUrl, $value);
173			}
174			}
175			return $thumbnailUrl;
176			}
177			return false;
178			}
179
180			/**
181			* @inheritdoc
182			*/
183			public function checkRules($rules)
184			{
185			if (! FinderUtils::checkMandatoryRules($rules, [
186			'image_regex',
187			'thumbnail_url'
188			])) {
189			throw new BadRulesException();
190			}
191			}
192
193			/**
194			* @inheritdoc
195			*
196			* @throws BadRulesException
197			*/
198			public function loadRules($rules)
199			{
200			$this->checkRules($rules);
201			$this->urlRegex = FinderUtils::buildRegex($rules['image_regex'], 'im');
202			$this->thumbnailUrlFormat = $rules['thumbnail_url'];
203			}
204
205			/**
206			* @inheritdoc
207			*/
208			public function getName()
209			{
210			return 'Query Regex';
211			}
212			}
213

ArthurHoaro / web-thumbnailer

Push — master ( e15538...7938c9 )

QueryRegexFinder::find() C

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like