inc/functions.php

   1 <?php
   2 /**
   3  * poche, a read it later open source system
   4  *
   5  * @category   poche
   6  * @author     Nicolas Lœuillet <support@inthepoche.com>
   7  * @copyright  2013
   8  * @license    http://www.wtfpl.net/ see COPYING file
   9  */
  10
  11 /**
  12  * Permet de générer l'URL de poche pour le bookmarklet
  13  */
  14 function get_poche_url()
  15 {
  16     $protocol = "http";
  17     if(isset($_SERVER['HTTPS'])) {
  18         if($_SERVER['HTTPS'] != "off" && $_SERVER['HTTPS'] != "") {
  19             $protocol = "https";
  20         }
  21     }
  22
  23     return $protocol . "://" . $_SERVER['HTTP_HOST'] . $_SERVER['REQUEST_URI'];
  24 }
  25
  26 // function define to retrieve url content
  27 function get_external_file($url)
  28 {
  29     $timeout = 15;
  30     // spoofing FireFox 18.0
  31     $useragent="Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
  32
  33     if  (in_array  ('curl', get_loaded_extensions())) {
  34         // Fetch feed from URL
  35         $curl = curl_init();
  36         curl_setopt($curl, CURLOPT_URL, $url);
  37         curl_setopt($curl, CURLOPT_TIMEOUT, $timeout);
  38         curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
  39         curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
  40         curl_setopt($curl, CURLOPT_HEADER, false);
  41
  42                 // FOR SSL do not verified certificate
  43         curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
  44                 curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
  45
  46         // FeedBurner requires a proper USER-AGENT...
  47         curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
  48         curl_setopt($curl, CURLOPT_ENCODING, "gzip, deflate");
  49         curl_setopt($curl, CURLOPT_USERAGENT, $useragent);
  50
  51         $data = curl_exec($curl);
  52
  53         $httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
  54
  55         $httpcodeOK = isset($httpcode) and ($httpcode == 200 or $httpcode == 301);
  56
  57         curl_close($curl);
  58     } else {
  59
  60         // create http context and add timeout and user-agent
  61         $context = stream_context_create(array(
  62                                                                 'http'=>array('timeout' => $timeout,
  63                                                                                 'header'=> "User-Agent: ".$useragent,   /*spoot Mozilla Firefox*/
  64                                                                                 'follow_location' => true),
  65                                                                 // FOR SSL do not verified certificate
  66                                                                 'ssl' => array('verify_peer' => false,
  67                                                                                 'allow_self_signed' => true)
  68                                                                 )
  69                                                 );
  70
  71         // only download page lesser than 4MB
  72         $data = @file_get_contents($url, false, $context, -1, 4000000); // We download at most 4 MB from source.
  73
  74         if(isset($http_response_header) and isset($http_response_header[0])) {
  75             $httpcodeOK = isset($http_response_header) and isset($http_response_header[0]) and ((strpos($http_response_header[0], '200 OK') !== FALSE) or (strpos($http_response_header[0], '301 Moved Permanently') !== FALSE));
  76         }
  77     }
  78
  79     // if response is not empty and response is OK
  80     if (isset($data) and isset($httpcodeOK) and $httpcodeOK ) {
  81
  82         // take charset of page and get it
  83         preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
  84
  85         // if meta tag is found
  86         if (!empty($meta[0])) {
  87             // retrieve encoding in $enc
  88             preg_match('#charset="?(.*)"#si', $meta[0], $enc);
  89
  90             // if charset is found set it otherwise, set it to utf-8
  91             $html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
  92
  93         } else {
  94             $html_charset = 'utf-8';
  95             $enc[1] = '';
  96         }
  97
  98         // replace charset of url to charset of page
  99         $data = str_replace('charset='.$enc[1], 'charset='.$html_charset, $data);
 100
 101         return $data;
 102     }
 103     else {
 104         return FALSE;
 105     }
 106 }
 107
 108 /**
 109  * Préparation de l'URL avec récupération du contenu avant insertion en base
 110  */
 111 function prepare_url($url)
 112 {
 113     global $msg;
 114
 115     $parametres = array();
 116     $url        = html_entity_decode(trim($url));
 117
 118     // We remove the annoying parameters added by FeedBurner and GoogleFeedProxy (?utm_source=...)
 119     // from shaarli, by sebsauvage
 120     $i=strpos($url,'&utm_source='); if ($i!==false) $url=substr($url,0,$i);
 121     $i=strpos($url,'?utm_source='); if ($i!==false) $url=substr($url,0,$i);
 122     $i=strpos($url,'#xtor=RSS-'); if ($i!==false) $url=substr($url,0,$i);
 123
 124     $title = $url;
 125         $html = Encoding::toUTF8(get_external_file($url,15));
 126         // If get_external_file if not able to retrieve HTTPS content try the same URL with HTTP protocol
 127         if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
 128                         $url = 'http://' . $url;
 129                         $html = Encoding::toUTF8(get_external_file($url,15));
 130         }
 131
 132     if (isset($html) and strlen($html) > 0)
 133     {
 134         $r = new Readability($html, $url);
 135
 136         $r->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
 137                 $r->debug=true;
 138                 $r->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
 139
 140         if($r->init())
 141         {
 142             $content = $r->articleContent->innerHTML;
 143             $parametres['title'] = $r->articleTitle->innerHTML;
 144             $parametres['content'] = $content;
 145             return $parametres;
 146         }
 147     }
 148
 149     $msg->add('e', 'error during url preparation');
 150     logm('error during url preparation');
 151     return FALSE;
 152 }
 153
 154 /**
 155  * On modifie les URLS des images dans le corps de l'article
 156  */
 157 function filtre_picture($content, $url, $id)
 158 {
 159     $matches = array();
 160     preg_match_all('#<\s*(img)[^>]+src="([^"]*)"[^>]*>#Si', $content, $matches, PREG_SET_ORDER);
 161     foreach($matches as $i => $link)
 162     {
 163         $link[1] = trim($link[1]);
 164         if (!preg_match('#^(([a-z]+://)|(\#))#', $link[1]) )
 165         {
 166             $absolute_path = get_absolute_link($link[2],$url);
 167             $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
 168             $directory = create_assets_directory($id);
 169             $fullpath = $directory . '/' . $filename;
 170             download_pictures($absolute_path, $fullpath);
 171             $content = str_replace($matches[$i][2], $fullpath, $content);
 172         }
 173
 174     }
 175
 176     return $content;
 177 }
 178
 179 /**
 180  * Retourne le lien absolu
 181  */
 182 function get_absolute_link($relative_link, $url)
 183 {
 184     /* return if already absolute URL */
 185     if (parse_url($relative_link, PHP_URL_SCHEME) != '') return $relative_link;
 186
 187     /* queries and anchors */
 188     if ($relative_link[0]=='#' || $relative_link[0]=='?') return $url . $relative_link;
 189
 190     /* parse base URL and convert to local variables:
 191        $scheme, $host, $path */
 192     extract(parse_url($url));
 193
 194     /* remove non-directory element from path */
 195     $path = preg_replace('#/[^/]*$#', '', $path);
 196
 197     /* destroy path if relative url points to root */
 198     if ($relative_link[0] == '/') $path = '';
 199
 200     /* dirty absolute URL */
 201     $abs = $host . $path . '/' . $relative_link;
 202
 203     /* replace '//' or '/./' or '/foo/../' with '/' */
 204     $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
 205     for($n=1; $n>0; $abs=preg_replace($re, '/', $abs, -1, $n)) {}
 206
 207     /* absolute URL is ready! */
 208     return $scheme.'://'.$abs;
 209 }
 210
 211 /**
 212  * Téléchargement des images
 213  */
 214
 215 function download_pictures($absolute_path, $fullpath)
 216 {
 217     $rawdata = get_external_file($absolute_path);
 218
 219     if(file_exists($fullpath)) {
 220         unlink($fullpath);
 221     }
 222     $fp = fopen($fullpath, 'x');
 223     fwrite($fp, $rawdata);
 224     fclose($fp);
 225 }
 226
 227 /**
 228  * Crée un répertoire de médias pour l'article
 229  */
 230 function create_assets_directory($id)
 231 {
 232     $assets_path = ABS_PATH;
 233     if(!is_dir($assets_path)) {
 234         mkdir($assets_path, 0705);
 235     }
 236
 237     $article_directory = $assets_path . $id;
 238     if(!is_dir($article_directory)) {
 239         mkdir($article_directory, 0705);
 240     }
 241
 242     return $article_directory;
 243 }
 244
 245 /**
 246  * Suppression du répertoire d'images
 247  */
 248 function remove_directory($directory)
 249 {
 250     if(is_dir($directory)) {
 251         $files = array_diff(scandir($directory), array('.','..'));
 252         foreach ($files as $file) {
 253             (is_dir("$directory/$file")) ? remove_directory("$directory/$file") : unlink("$directory/$file");
 254         }
 255         return rmdir($directory);
 256     }
 257 }
 258
 259 function display_view($view, $id = 0, $full_head = 'yes')
 260 {
 261     global $tpl, $store, $msg;
 262
 263     switch ($view)
 264     {
 265         case 'export':
 266             $entries = $store->retrieveAll();
 267             $tpl->assign('export', myTool::renderJson($entries));
 268             $tpl->draw('export');
 269             logm('export view');
 270             break;
 271         case 'config':
 272             $tpl->assign('load_all_js', 0);
 273             $tpl->draw('head');
 274             $tpl->draw('home');
 275             $tpl->draw('config');
 276             $tpl->draw('js');
 277             $tpl->draw('footer');
 278             logm('config view');
 279             break;
 280         case 'view':
 281             $entry = $store->retrieveOneById($id);
 282
 283             if ($entry != NULL) {
 284                 $tpl->assign('id', $entry['id']);
 285                 $tpl->assign('url', $entry['url']);
 286                 $tpl->assign('title', $entry['title']);
 287                 $tpl->assign('content', $entry['content']);
 288                 $tpl->assign('is_fav', $entry['is_fav']);
 289                 $tpl->assign('is_read', $entry['is_read']);
 290                 $tpl->assign('load_all_js', 0);
 291                 $tpl->draw('view');
 292             }
 293             else {
 294                 logm('error in view call : entry is NULL');
 295             }
 296
 297             logm('view link #' . $id);
 298             break;
 299         default: # home view
 300             $entries = $store->getEntriesByView($view);
 301
 302             $tpl->assign('entries', $entries);
 303
 304             if ($full_head == 'yes') {
 305                 $tpl->assign('load_all_js', 1);
 306                 $tpl->draw('head');
 307                 $tpl->draw('home');
 308             }
 309
 310             $tpl->draw('entries');
 311
 312             if ($full_head == 'yes') {
 313                 $tpl->draw('js');
 314                 $tpl->draw('footer');
 315             }
 316             break;
 317     }
 318 }
 319
 320 /**
 321  * Appel d'une action (mark as fav, archive, delete)
 322  */
 323 function action_to_do($action, $url, $id = 0)
 324 {
 325     global $store, $msg;
 326
 327     switch ($action)
 328     {
 329         case 'add':
 330             if ($url == '')
 331                 continue;
 332
 333             if (MyTool::isUrl($url)) {
 334                 if($parametres_url = prepare_url($url)) {
 335                     $store->add($url, $parametres_url['title'], $parametres_url['content']);
 336                     $last_id = $store->getLastId();
 337                     if (DOWNLOAD_PICTURES) {
 338                         $content = filtre_picture($parametres_url['content'], $url, $last_id);
 339                     }
 340                     $msg->add('s', 'the link has been added successfully');
 341                 }
 342             }
 343             else {
 344                 $msg->add('e', 'the link has been added successfully');
 345                 logm($url . ' is not a valid url');
 346             }
 347
 348             logm('add link ' . $url);
 349             break;
 350         case 'delete':
 351             remove_directory(ABS_PATH . $id);
 352             $store->deleteById($id);
 353             $msg->add('s', 'the link has been deleted successfully');
 354             logm('delete link #' . $id);
 355             break;
 356         case 'toggle_fav' :
 357             $store->favoriteById($id);
 358             $msg->add('s', 'the favorite toggle has been done successfully');
 359             logm('mark as favorite link #' . $id);
 360             break;
 361         case 'toggle_archive' :
 362             $store->archiveById($id);
 363             $msg->add('s', 'the archive toggle has been done successfully');
 364             logm('archive link #' . $id);
 365             break;
 366         default:
 367             break;
 368     }
 369 }
 370
 371 function logm($message)
 372 {
 373     $t = strval(date('Y/m/d_H:i:s')).' - '.$_SERVER["REMOTE_ADDR"].' - '.strval($message)."\n";
 374     file_put_contents('./log.txt',$t,FILE_APPEND);
 375 }