]>
Commit | Line | Data |
---|---|---|
419214d7 TC |
1 | <?php |
2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | |
4 | ||
7f559418 | 5 | use Psr\Log\LoggerInterface; |
419214d7 | 6 | use Symfony\Component\DomCrawler\Crawler; |
7f559418 JB |
7 | use GuzzleHttp\Client; |
8 | use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; | |
419214d7 | 9 | |
156bf627 JB |
10 | class DownloadImages |
11 | { | |
7f559418 JB |
12 | const REGENERATE_PICTURES_QUALITY = 80; |
13 | ||
14 | private $client; | |
15 | private $baseFolder; | |
419214d7 | 16 | private $logger; |
7f559418 | 17 | private $mimeGuesser; |
41ada277 | 18 | private $wallabagUrl; |
419214d7 | 19 | |
7f559418 | 20 | public function __construct(Client $client, $baseFolder, LoggerInterface $logger) |
156bf627 | 21 | { |
7f559418 JB |
22 | $this->client = $client; |
23 | $this->baseFolder = $baseFolder; | |
419214d7 | 24 | $this->logger = $logger; |
7f559418 JB |
25 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); |
26 | ||
27 | $this->setFolder(); | |
419214d7 TC |
28 | } |
29 | ||
41ada277 JB |
30 | /** |
31 | * Since we can't inject CraueConfig service because it'll generate a circular reference when injected in the subscriber | |
32 | * we use a different way to inject the current wallabag url. | |
33 | * | |
34 | * @param string $url Usually from `$config->get('wallabag_url')` | |
35 | */ | |
36 | public function setWallabagUrl($url) | |
37 | { | |
38 | $this->wallabagUrl = rtrim($url, '/'); | |
39 | } | |
40 | ||
7f559418 JB |
41 | /** |
42 | * Setup base folder where all images are going to be saved. | |
43 | */ | |
44 | private function setFolder() | |
156bf627 | 45 | { |
419214d7 | 46 | // if folder doesn't exist, attempt to create one and store the folder name in property $folder |
7f559418 JB |
47 | if (!file_exists($this->baseFolder)) { |
48 | mkdir($this->baseFolder, 0777, true); | |
419214d7 | 49 | } |
419214d7 TC |
50 | } |
51 | ||
7f559418 JB |
52 | /** |
53 | * Process the html and extract image from it, save them to local and return the updated html. | |
54 | * | |
55 | * @param string $html | |
56 | * @param string $url Used as a base path for relative image and folder | |
57 | * | |
58 | * @return string | |
59 | */ | |
60 | public function processHtml($html, $url) | |
156bf627 | 61 | { |
7f559418 | 62 | $crawler = new Crawler($html); |
419214d7 TC |
63 | $result = $crawler |
64 | ->filterXpath('//img') | |
65 | ->extract(array('src')); | |
66 | ||
7f559418 JB |
67 | $relativePath = $this->getRelativePath($url); |
68 | ||
419214d7 TC |
69 | // download and save the image to the folder |
70 | foreach ($result as $image) { | |
7f559418 JB |
71 | $imagePath = $this->processSingleImage($image, $url, $relativePath); |
72 | ||
73 | if (false === $imagePath) { | |
74 | continue; | |
75 | } | |
76 | ||
77 | $html = str_replace($image, $imagePath, $html); | |
419214d7 TC |
78 | } |
79 | ||
7f559418 | 80 | return $html; |
419214d7 TC |
81 | } |
82 | ||
7f559418 JB |
83 | /** |
84 | * Process a single image: | |
85 | * - retrieve it | |
86 | * - re-saved it (for security reason) | |
87 | * - return the new local path. | |
88 | * | |
89 | * @param string $imagePath Path to the image to retrieve | |
90 | * @param string $url Url from where the image were found | |
91 | * @param string $relativePath Relative local path to saved the image | |
92 | * | |
93 | * @return string Relative url to access the image from the web | |
94 | */ | |
95 | public function processSingleImage($imagePath, $url, $relativePath = null) | |
156bf627 | 96 | { |
7f559418 JB |
97 | if (null == $relativePath) { |
98 | $relativePath = $this->getRelativePath($url); | |
419214d7 TC |
99 | } |
100 | ||
7f559418 | 101 | $folderPath = $this->baseFolder.'/'.$relativePath; |
419214d7 | 102 | |
7f559418 JB |
103 | // build image path |
104 | $absolutePath = $this->getAbsoluteLink($url, $imagePath); | |
105 | if (false === $absolutePath) { | |
48656e0e | 106 | $this->logger->log('error', 'Can not determine the absolute path for that image, skipping.'); |
419214d7 TC |
107 | |
108 | return false; | |
109 | } | |
110 | ||
48656e0e JB |
111 | try { |
112 | $res = $this->client->get($absolutePath); | |
113 | } catch (\Exception $e) { | |
114 | $this->logger->log('error', 'Can not retrieve image, skipping.', ['exception' => $e]); | |
115 | ||
116 | return false; | |
117 | } | |
7f559418 JB |
118 | |
119 | $ext = $this->mimeGuesser->guess($res->getHeader('content-type')); | |
120 | $this->logger->log('debug', 'Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]); | |
48656e0e JB |
121 | if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) { |
122 | $this->logger->log('error', 'Processed image with not allowed extension. Skipping '.$imagePath); | |
419214d7 TC |
123 | |
124 | return false; | |
125 | } | |
7f559418 JB |
126 | $hashImage = hash('crc32', $absolutePath); |
127 | $localPath = $folderPath.'/'.$hashImage.'.'.$ext; | |
128 | ||
129 | try { | |
130 | $im = imagecreatefromstring($res->getBody()); | |
131 | } catch (\Exception $e) { | |
132 | $im = false; | |
133 | } | |
419214d7 | 134 | |
48656e0e | 135 | if (false === $im) { |
7f559418 | 136 | $this->logger->log('error', 'Error while regenerating image', ['path' => $localPath]); |
419214d7 TC |
137 | |
138 | return false; | |
139 | } | |
140 | ||
7f559418 JB |
141 | switch ($ext) { |
142 | case 'gif': | |
143 | $result = imagegif($im, $localPath); | |
156bf627 | 144 | $this->logger->log('debug', 'Re-creating gif'); |
419214d7 | 145 | break; |
7f559418 JB |
146 | case 'jpeg': |
147 | case 'jpg': | |
148 | $result = imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY); | |
156bf627 | 149 | $this->logger->log('debug', 'Re-creating jpg'); |
419214d7 | 150 | break; |
7f559418 JB |
151 | case 'png': |
152 | $result = imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9)); | |
156bf627 | 153 | $this->logger->log('debug', 'Re-creating png'); |
419214d7 | 154 | } |
7f559418 | 155 | |
419214d7 TC |
156 | imagedestroy($im); |
157 | ||
41ada277 | 158 | return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext; |
419214d7 TC |
159 | } |
160 | ||
7f559418 JB |
161 | /** |
162 | * Generate the folder where we are going to save images based on the entry url. | |
163 | * | |
164 | * @param string $url | |
165 | * | |
166 | * @return string | |
167 | */ | |
168 | private function getRelativePath($url) | |
419214d7 | 169 | { |
7f559418 JB |
170 | $hashUrl = hash('crc32', $url); |
171 | $relativePath = $hashUrl[0].'/'.$hashUrl[1].'/'.$hashUrl; | |
172 | $folderPath = $this->baseFolder.'/'.$relativePath; | |
419214d7 | 173 | |
7f559418 JB |
174 | if (!file_exists($folderPath)) { |
175 | mkdir($folderPath, 0777, true); | |
419214d7 TC |
176 | } |
177 | ||
7f559418 | 178 | $this->logger->log('debug', 'Folder used for that url', ['folder' => $folderPath, 'url' => $url]); |
419214d7 | 179 | |
7f559418 JB |
180 | return $relativePath; |
181 | } | |
419214d7 | 182 | |
7f559418 JB |
183 | /** |
184 | * Make an $url absolute based on the $base. | |
185 | * | |
186 | * @see Graby->makeAbsoluteStr | |
187 | * | |
188 | * @param string $base Base url | |
189 | * @param string $url Url to make it absolute | |
190 | * | |
191 | * @return false|string | |
192 | */ | |
193 | private function getAbsoluteLink($base, $url) | |
194 | { | |
195 | if (preg_match('!^https?://!i', $url)) { | |
196 | // already absolute | |
197 | return $url; | |
419214d7 TC |
198 | } |
199 | ||
7f559418 | 200 | $base = new \SimplePie_IRI($base); |
419214d7 | 201 | |
7f559418 JB |
202 | // remove '//' in URL path (causes URLs not to resolve properly) |
203 | if (isset($base->ipath)) { | |
204 | $base->ipath = preg_replace('!//+!', '/', $base->ipath); | |
419214d7 TC |
205 | } |
206 | ||
7f559418 JB |
207 | if ($absolute = \SimplePie_IRI::absolutize($base, $url)) { |
208 | return $absolute->get_uri(); | |
94654765 | 209 | } |
156bf627 | 210 | |
48656e0e JB |
211 | $this->logger->log('error', 'Can not make an absolute link', ['base' => $base, 'url' => $url]); |
212 | ||
7f559418 | 213 | return false; |
94654765 | 214 | } |
419214d7 | 215 | } |