]>
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; | |
e0597476 | 9 | use Symfony\Component\Finder\Finder; |
419214d7 | 10 | |
156bf627 JB |
11 | class DownloadImages |
12 | { | |
7f559418 JB |
13 | const REGENERATE_PICTURES_QUALITY = 80; |
14 | ||
15 | private $client; | |
16 | private $baseFolder; | |
419214d7 | 17 | private $logger; |
7f559418 | 18 | private $mimeGuesser; |
41ada277 | 19 | private $wallabagUrl; |
419214d7 | 20 | |
e0597476 | 21 | public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger) |
156bf627 | 22 | { |
7f559418 JB |
23 | $this->client = $client; |
24 | $this->baseFolder = $baseFolder; | |
e0597476 | 25 | $this->wallabagUrl = rtrim($wallabagUrl, '/'); |
419214d7 | 26 | $this->logger = $logger; |
7f559418 JB |
27 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); |
28 | ||
29 | $this->setFolder(); | |
419214d7 TC |
30 | } |
31 | ||
7f559418 JB |
32 | /** |
33 | * Setup base folder where all images are going to be saved. | |
34 | */ | |
35 | private function setFolder() | |
156bf627 | 36 | { |
419214d7 | 37 | // if folder doesn't exist, attempt to create one and store the folder name in property $folder |
7f559418 | 38 | if (!file_exists($this->baseFolder)) { |
e044d27f | 39 | mkdir($this->baseFolder, 0755, true); |
419214d7 | 40 | } |
419214d7 TC |
41 | } |
42 | ||
7f559418 JB |
43 | /** |
44 | * Process the html and extract image from it, save them to local and return the updated html. | |
45 | * | |
e0597476 | 46 | * @param int $entryId ID of the entry |
7f559418 | 47 | * @param string $html |
e0597476 | 48 | * @param string $url Used as a base path for relative image and folder |
7f559418 JB |
49 | * |
50 | * @return string | |
51 | */ | |
e0597476 | 52 | public function processHtml($entryId, $html, $url) |
156bf627 | 53 | { |
7f559418 | 54 | $crawler = new Crawler($html); |
419214d7 TC |
55 | $result = $crawler |
56 | ->filterXpath('//img') | |
9bf7752f | 57 | ->extract(['src']); |
419214d7 | 58 | |
e0597476 | 59 | $relativePath = $this->getRelativePath($entryId); |
7f559418 | 60 | |
419214d7 TC |
61 | // download and save the image to the folder |
62 | foreach ($result as $image) { | |
e0597476 | 63 | $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath); |
7f559418 JB |
64 | |
65 | if (false === $imagePath) { | |
66 | continue; | |
67 | } | |
68 | ||
9bf7752f | 69 | // if image contains "&" and we can't find it in the html it might be because it's encoded as & |
fcad69a4 JB |
70 | if (false !== stripos($image, '&') && false === stripos($html, $image)) { |
71 | $image = str_replace('&', '&', $image); | |
72 | } | |
73 | ||
7f559418 | 74 | $html = str_replace($image, $imagePath, $html); |
419214d7 TC |
75 | } |
76 | ||
7f559418 | 77 | return $html; |
419214d7 TC |
78 | } |
79 | ||
7f559418 JB |
80 | /** |
81 | * Process a single image: | |
82 | * - retrieve it | |
83 | * - re-saved it (for security reason) | |
84 | * - return the new local path. | |
85 | * | |
e0597476 | 86 | * @param int $entryId ID of the entry |
7f559418 JB |
87 | * @param string $imagePath Path to the image to retrieve |
88 | * @param string $url Url from where the image were found | |
89 | * @param string $relativePath Relative local path to saved the image | |
90 | * | |
91 | * @return string Relative url to access the image from the web | |
92 | */ | |
e0597476 | 93 | public function processSingleImage($entryId, $imagePath, $url, $relativePath = null) |
156bf627 | 94 | { |
e0597476 JB |
95 | if (null === $relativePath) { |
96 | $relativePath = $this->getRelativePath($entryId); | |
419214d7 TC |
97 | } |
98 | ||
e0597476 JB |
99 | $this->logger->debug('DownloadImages: working on image: '.$imagePath); |
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) { | |
e0597476 | 106 | $this->logger->error('DownloadImages: 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) { | |
e0597476 | 114 | $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]); |
48656e0e JB |
115 | |
116 | return false; | |
117 | } | |
7f559418 JB |
118 | |
119 | $ext = $this->mimeGuesser->guess($res->getHeader('content-type')); | |
e0597476 | 120 | $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]); |
48656e0e | 121 | if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) { |
fcad69a4 | 122 | $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$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) { |
e0597476 | 136 | $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]); |
419214d7 TC |
137 | |
138 | return false; | |
139 | } | |
140 | ||
7f559418 JB |
141 | switch ($ext) { |
142 | case 'gif': | |
001cc716 | 143 | imagegif($im, $localPath); |
e0597476 | 144 | $this->logger->debug('DownloadImages: Re-creating gif'); |
419214d7 | 145 | break; |
7f559418 JB |
146 | case 'jpeg': |
147 | case 'jpg': | |
001cc716 | 148 | imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY); |
e0597476 | 149 | $this->logger->debug('DownloadImages: Re-creating jpg'); |
419214d7 | 150 | break; |
7f559418 | 151 | case 'png': |
7a3260ae KD |
152 | imagealphablending($im, false); |
153 | imagesavealpha($im, true); | |
001cc716 | 154 | imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9)); |
e0597476 | 155 | $this->logger->debug('DownloadImages: Re-creating png'); |
419214d7 | 156 | } |
7f559418 | 157 | |
419214d7 TC |
158 | imagedestroy($im); |
159 | ||
41ada277 | 160 | return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext; |
419214d7 TC |
161 | } |
162 | ||
e0597476 JB |
163 | /** |
164 | * Remove all images for the given entry id. | |
165 | * | |
166 | * @param int $entryId ID of the entry | |
167 | */ | |
168 | public function removeImages($entryId) | |
169 | { | |
170 | $relativePath = $this->getRelativePath($entryId); | |
171 | $folderPath = $this->baseFolder.'/'.$relativePath; | |
172 | ||
173 | $finder = new Finder(); | |
174 | $finder | |
175 | ->files() | |
176 | ->ignoreDotFiles(true) | |
177 | ->in($folderPath); | |
178 | ||
179 | foreach ($finder as $file) { | |
180 | @unlink($file->getRealPath()); | |
181 | } | |
182 | ||
183 | @rmdir($folderPath); | |
184 | } | |
185 | ||
7f559418 JB |
186 | /** |
187 | * Generate the folder where we are going to save images based on the entry url. | |
188 | * | |
e0597476 | 189 | * @param int $entryId ID of the entry |
7f559418 JB |
190 | * |
191 | * @return string | |
192 | */ | |
e0597476 | 193 | private function getRelativePath($entryId) |
419214d7 | 194 | { |
e0597476 JB |
195 | $hashId = hash('crc32', $entryId); |
196 | $relativePath = $hashId[0].'/'.$hashId[1].'/'.$hashId; | |
7f559418 | 197 | $folderPath = $this->baseFolder.'/'.$relativePath; |
419214d7 | 198 | |
7f559418 JB |
199 | if (!file_exists($folderPath)) { |
200 | mkdir($folderPath, 0777, true); | |
419214d7 TC |
201 | } |
202 | ||
e0597476 | 203 | $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]); |
419214d7 | 204 | |
7f559418 JB |
205 | return $relativePath; |
206 | } | |
419214d7 | 207 | |
7f559418 JB |
208 | /** |
209 | * Make an $url absolute based on the $base. | |
210 | * | |
211 | * @see Graby->makeAbsoluteStr | |
212 | * | |
213 | * @param string $base Base url | |
214 | * @param string $url Url to make it absolute | |
215 | * | |
216 | * @return false|string | |
217 | */ | |
218 | private function getAbsoluteLink($base, $url) | |
219 | { | |
220 | if (preg_match('!^https?://!i', $url)) { | |
221 | // already absolute | |
222 | return $url; | |
419214d7 TC |
223 | } |
224 | ||
7f559418 | 225 | $base = new \SimplePie_IRI($base); |
419214d7 | 226 | |
7f559418 JB |
227 | // remove '//' in URL path (causes URLs not to resolve properly) |
228 | if (isset($base->ipath)) { | |
229 | $base->ipath = preg_replace('!//+!', '/', $base->ipath); | |
419214d7 TC |
230 | } |
231 | ||
7f559418 JB |
232 | if ($absolute = \SimplePie_IRI::absolutize($base, $url)) { |
233 | return $absolute->get_uri(); | |
94654765 | 234 | } |
156bf627 | 235 | |
e0597476 | 236 | $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]); |
48656e0e | 237 | |
7f559418 | 238 | return false; |
94654765 | 239 | } |
419214d7 | 240 | } |