plugins/cache_starred_images/init.php

   1 <?php
   2 class Cache_Starred_Images extends Plugin implements IHandler {
   3
   4         /* @var PluginHost $host */
   5         private $host;
   6         private $cache_dir;
   7     private $max_cache_attempts = 5; // per-article
   8
   9         function about() {
  10                 return array(1.0,
  11                         "Automatically cache Starred articles' images and HTML5 video files",
  12                         "fox",
  13                         true);
  14         }
  15
  16         /**
  17          * @SuppressWarnings(PHPMD.UnusedFormalParameter)
  18          */
  19         function csrf_ignore($method) {
  20                 return false;
  21         }
  22
  23         /**
  24          * @SuppressWarnings(PHPMD.UnusedFormalParameter)
  25          */
  26         function before($method) {
  27                 return true;
  28         }
  29
  30         function after() {
  31                 return true;
  32         }
  33
  34         function init($host) {
  35                 $this->host = $host;
  36
  37                 $this->cache_dir = CACHE_DIR . "/starred-images/";
  38
  39                 if (!is_dir($this->cache_dir)) {
  40                         mkdir($this->cache_dir);
  41                 }
  42
  43                 if (is_dir($this->cache_dir)) {
  44
  45                         if (!is_writable($this->cache_dir))
  46                                 chmod($this->cache_dir, 0777);
  47
  48                         if (is_writable($this->cache_dir)) {
  49                                 $host->add_hook($host::HOOK_UPDATE_TASK, $this);
  50                                 $host->add_hook($host::HOOK_HOUSE_KEEPING, $this);
  51                                 $host->add_hook($host::HOOK_SANITIZE, $this);
  52                                 $host->add_handler("public", "cache_starred_images_getimage", $this);
  53
  54                         } else {
  55                                 user_error("Starred cache directory is not writable.", E_USER_WARNING);
  56                         }
  57
  58                 } else {
  59                         user_error("Unable to create starred cache directory.", E_USER_WARNING);
  60                 }
  61         }
  62
  63         function cache_starred_images_getimage() {
  64                 ob_end_clean();
  65
  66                 $hash = basename($_REQUEST["hash"]);
  67
  68                 if ($hash) {
  69
  70                         $filename = $this->cache_dir . "/" . basename($hash);
  71
  72                         if (file_exists($filename)) {
  73                                 header("Content-Disposition: attachment; filename=\"$hash\"");
  74
  75                                 send_local_file($filename);
  76                         } else {
  77                                 header($_SERVER["SERVER_PROTOCOL"]." 404 Not Found");
  78                                 echo "File not found.";
  79                         }
  80                 }
  81         }
  82
  83         /**
  84          * @SuppressWarnings(PHPMD.UnusedLocalVariable)
  85          */
  86         function hook_house_keeping() {
  87                 $files = glob($this->cache_dir . "/*.{png,mp4,status}", GLOB_BRACE);
  88
  89                 $last_article_id = 0;
  90                 $article_exists = 1;
  91
  92                 foreach ($files as $file) {
  93                         list ($article_id, $hash) = explode("-", basename($file));
  94
  95                         if ($article_id != $last_article_id) {
  96                                 $last_article_id = $article_id;
  97
  98                                 $sth = $this->pdo->prepare("SELECT id FROM ttrss_entries WHERE id = ?");
  99                                 $sth->execute([$article_id]);
 100
 101                                 $article_exists = $sth->fetch();
 102                         }
 103
 104                         if (!$article_exists) {
 105                                 unlink($file);
 106                         }
 107                 }
 108         }
 109
 110         /**
 111          * @SuppressWarnings(PHPMD.UnusedFormalParameter)
 112          */
 113         function hook_sanitize($doc, $site_url, $allowed_elements, $disallowed_attributes, $article_id) {
 114                 $xpath = new DOMXpath($doc);
 115
 116                 if ($article_id) {
 117                         $entries = $xpath->query('(//img[@src])|(//video/source[@src])');
 118
 119                         foreach ($entries as $entry) {
 120                                 if ($entry->hasAttribute('src')) {
 121                                         $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
 122
 123                                         $extension = $entry->tagName == 'source' ? '.mp4' : '.png';
 124                                         $local_filename = $this->cache_dir . $article_id . "-" . sha1($src) . $extension;
 125
 126                                         if (file_exists($local_filename)) {
 127                                                 $entry->setAttribute("src", get_self_url_prefix() .
 128                                                         "/public.php?op=cache_starred_images_getimage&method=image&hash=" .
 129                                                         $article_id . "-" . sha1($src) . $extension);
 130                                         }
 131
 132                                 }
 133                         }
 134                 }
 135
 136                 return $doc;
 137         }
 138
 139         function hook_update_task() {
 140                 $res = $this->pdo->query("SELECT content, ttrss_user_entries.owner_uid, link, site_url, ttrss_entries.id, plugin_data
 141                         FROM ttrss_entries, ttrss_user_entries LEFT JOIN ttrss_feeds ON
 142                                 (ttrss_user_entries.feed_id = ttrss_feeds.id)
 143                         WHERE ref_id = ttrss_entries.id AND
 144                                 marked = true AND
 145                                 (UPPER(content) LIKE '%<IMG%' OR UPPER(content) LIKE '%<VIDEO%') AND
 146                                 site_url != '' AND
 147                                 plugin_data NOT LIKE '%starred_cache_images%'
 148                         ORDER BY ".sql_random_function()." LIMIT 100");
 149
 150                 $usth = $this->pdo->prepare("UPDATE ttrss_entries SET plugin_data = ? WHERE id = ?");
 151
 152                 while ($line = $res->fetch()) {
 153                         if ($line["site_url"]) {
 154                                 $success = $this->cache_article_images($line["content"], $line["site_url"], $line["owner_uid"], $line["id"]);
 155
 156                                 if ($success) {
 157                                         $plugin_data = "starred_cache_images,${line['owner_uid']}:" . $line["plugin_data"];
 158
 159                                         $usth->execute([$plugin_data, $line['id']]);
 160                                 }
 161                         }
 162                 }
 163         }
 164
 165         /**
 166          * @SuppressWarnings(PHPMD.UnusedFormalParameter)
 167          */
 168         function cache_article_images($content, $site_url, $owner_uid, $article_id) {
 169                 libxml_use_internal_errors(true);
 170
 171                 $status_filename = $this->cache_dir . $article_id . "-" . sha1($site_url) . ".status";
 172
 173                 //_debug("status: $status_filename"); return;
 174
 175         if (file_exists($status_filename))
 176             $status = json_decode(file_get_contents($status_filename), true);
 177         else
 178             $status = [];
 179
 180         $status["attempt"] += 1;
 181
 182         // only allow several download attempts for article
 183         if ($status["attempt"] > $this->max_cache_attempts) {
 184             _debug("too many attempts for $site_url");
 185             return;
 186         }
 187
 188         if (!file_put_contents($status_filename, json_encode($status))) {
 189             user_error("unable to write status file: $status_filename", E_USER_WARNING);
 190             return;
 191         }
 192
 193                 $charset_hack = '<head>
 194                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 195                 </head>';
 196
 197                 $doc = new DOMDocument();
 198                 $doc->loadHTML($charset_hack . $content);
 199                 $xpath = new DOMXPath($doc);
 200
 201                 $entries = $xpath->query('(//img[@src])|(//video/source[@src])');
 202
 203                 $success = false;
 204                 $has_images = false;
 205
 206                 foreach ($entries as $entry) {
 207
 208                         if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
 209
 210                                 $has_images = true;
 211                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
 212
 213                                 $extension = $entry->tagName == 'source' ? '.mp4' : '.png';
 214
 215                                 $local_filename = $this->cache_dir . $article_id . "-" . sha1($src) . $extension;
 216
 217                                 //_debug("cache_images: downloading: $src to $local_filename");
 218
 219                                 if (!file_exists($local_filename)) {
 220                                         $file_content = fetch_file_contents(["url" => $src, "max_size" => MAX_CACHE_FILE_SIZE]);
 221
 222                                         if ($file_content) {
 223                         if (strlen($file_content) > MIN_CACHE_FILE_SIZE) {
 224                             file_put_contents($local_filename, $file_content);
 225                         }
 226
 227                                                 $success = true;
 228                                         }
 229                                 } else {
 230                                         $success = true;
 231                                 }
 232                         }
 233                 }
 234
 235                 return $success || !$has_images;
 236         }
 237
 238         function api_version() {
 239                 return 2;
 240         }
 241 }