repository/url/lib.php

   1 <?php
   2
   3 // This file is part of Moodle - http://moodle.org/
   4 //
   5 // Moodle is free software: you can redistribute it and/or modify
   6 // it under the terms of the GNU General Public License as published by
   7 // the Free Software Foundation, either version 3 of the License, or
   8 // (at your option) any later version.
   9 //
  10 // Moodle is distributed in the hope that it will be useful,
  11 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 // GNU General Public License for more details.
  14 //
  15 // You should have received a copy of the GNU General Public License
  16 // along with Moodle.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 /**
  19  * This plugin is used to access files by providing an url
  20  *
  21  * @since Moodle 2.0
  22  * @package    repository_url
  23  * @copyright  2010 Dongsheng Cai {@link http://dongsheng.org}
  24  * @license    http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
  25  */
  26 require_once($CFG->dirroot . '/repository/lib.php');
  27 require_once(__DIR__.'/locallib.php');
  28
  29 /**
  30  * repository_url class
  31  * A subclass of repository, which is used to download a file from a specific url
  32  *
  33  * @since Moodle 2.0
  34  * @package    repository_url
  35  * @copyright  2009 Dongsheng Cai {@link http://dongsheng.org}
  36  * @license    http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
  37  */
  38 class repository_url extends repository {
  39     /** @var int Maximum time of recursion. */
  40     const MAX_RECURSION_TIME = 5;
  41     /** @var int Maximum number of CSS imports. */
  42     const MAX_CSS_IMPORTS = 10;
  43     /** @var int CSS import counter. */
  44     var $cssimportcounter = 0;
  45     var $processedfiles = array();
  46     /** @var int Recursion counter. */
  47     var $recursioncounter = 0;
  48
  49     /**
  50      * @param int $repositoryid
  51      * @param object $context
  52      * @param array $options
  53      */
  54     public function __construct($repositoryid, $context = SYSCONTEXTID, $options = array()){
  55         global $CFG;
  56         parent::__construct($repositoryid, $context, $options);
  57         $this->file_url = optional_param('file', '', PARAM_RAW);
  58         $this->file_url = $this->escape_url($this->file_url);
  59     }
  60
  61     public function check_login() {
  62         if (!empty($this->file_url)) {
  63             return true;
  64         } else {
  65             return false;
  66         }
  67     }
  68     /**
  69      * @return mixed
  70      */
  71     public function print_login() {
  72         $strdownload = get_string('download', 'repository');
  73         $strname     = get_string('rename', 'repository_url');
  74         $strurl      = get_string('url', 'repository_url');
  75         if ($this->options['ajax']) {
  76             $url = new stdClass();
  77             $url->label = $strurl.': ';
  78             $url->id   = 'fileurl';
  79             $url->type = 'text';
  80             $url->name = 'file';
  81
  82             $ret['login'] = array($url);
  83             $ret['login_btn_label'] = get_string('download', 'repository_url');
  84             $ret['allowcaching'] = true; // indicates that login form can be cached in filepicker.js
  85             return $ret;
  86         } else {
  87             echo <<<EOD
  88 <table>
  89 <tr>
  90 <td>{$strurl}: </td><td><input name="file" type="text" /></td>
  91 </tr>
  92 </table>
  93 <input type="submit" value="{$strdownload}" />
  94 EOD;
  95
  96         }
  97     }
  98
  99     /**
 100      * @param mixed $path
 101      * @param string $search
 102      * @return array
 103      */
 104     public function get_listing($path='', $page='') {
 105         $ret = array();
 106         $ret['list'] = array();
 107         $ret['nosearch'] = true;
 108         $ret['norefresh'] = true;
 109         $ret['nologin'] = true;
 110
 111         $this->file_url = clean_param($this->file_url, PARAM_URL);
 112         if (empty($this->file_url)) {
 113             throw new repository_exception('validfiletype', 'repository_url');
 114         }
 115
 116         $this->parse_file(null, $this->file_url, $ret, true);
 117         return $ret;
 118     }
 119
 120     /**
 121      * Parses one file (either html or css)
 122      *
 123      * @param string $baseurl (optional) URL of the file where link to this file was found
 124      * @param string $relativeurl relative or absolute link to the file
 125      * @param array $list
 126      * @param bool $mainfile true only for main HTML false and false for all embedded/linked files
 127      */
 128     protected function parse_file($baseurl, $relativeurl, &$list, $mainfile = false) {
 129         if (preg_match('/([\'"])(.*)\1/', $relativeurl, $matches)) {
 130             $relativeurl = $matches[2];
 131         }
 132         if (empty($baseurl)) {
 133             $url = $relativeurl;
 134         } else {
 135             $url = htmlspecialchars_decode(url_to_absolute($baseurl, $relativeurl));
 136         }
 137         if (in_array($url, $this->processedfiles)) {
 138             // Avoid endless recursion for the same URL with same parameters.
 139             return;
 140         }
 141         // Remove the query string and anchors before check.
 142         $recursioncheckurl = (new moodle_url($url))->out_omit_querystring();
 143         if (in_array($recursioncheckurl, $this->processedfiles)) {
 144             $this->recursioncounter++;
 145         }
 146         if ($this->recursioncounter >= self::MAX_RECURSION_TIME) {
 147             // Avoid endless recursion for the same URL with different parameters.
 148             return;
 149         }
 150         $this->processedfiles[] = $url;
 151         $curl = new curl;
 152         $curl->setopt(array('CURLOPT_FOLLOWLOCATION' => true, 'CURLOPT_MAXREDIRS' => 3));
 153         $msg = $curl->head($url);
 154         $info = $curl->get_info();
 155         if ($info['http_code'] != 200) {
 156             if ($mainfile) {
 157                 $list['error'] = $msg;
 158             }
 159         } else {
 160             $csstoanalyze = '';
 161             if ($mainfile && (strstr($info['content_type'], 'text/html') || empty($info['content_type']))) {
 162                 // parse as html
 163                 $htmlcontent = $curl->get($info['url']);
 164                 $ddoc = new DOMDocument();
 165                 @$ddoc->loadHTML($htmlcontent);
 166                 // extract <img>
 167                 $tags = $ddoc->getElementsByTagName('img');
 168                 foreach ($tags as $tag) {
 169                     $url = $tag->getAttribute('src');
 170                     $this->add_image_to_list($info['url'], $url, $list);
 171                 }
 172                 // analyse embedded css (<style>)
 173                 $tags = $ddoc->getElementsByTagName('style');
 174                 foreach ($tags as $tag) {
 175                     if ($tag->getAttribute('type') == 'text/css') {
 176                         $csstoanalyze .= $tag->textContent."\n";
 177                     }
 178                 }
 179                 // analyse links to css (<link type='text/css' href='...'>)
 180                 $tags = $ddoc->getElementsByTagName('link');
 181                 foreach ($tags as $tag) {
 182                     if ($tag->getAttribute('type') == 'text/css' && strlen($tag->getAttribute('href'))) {
 183                         $this->parse_file($info['url'], $tag->getAttribute('href'), $list);
 184                     }
 185                 }
 186             } else if (strstr($info['content_type'], 'css')) {
 187                 // parse as css
 188                 $csscontent = $curl->get($info['url']);
 189                 $csstoanalyze .= $csscontent."\n";
 190             } else if (strstr($info['content_type'], 'image/')) {
 191                 // download this file
 192                 $this->add_image_to_list($info['url'], $info['url'], $list);
 193             } else {
 194                 $list['error'] = get_string('validfiletype', 'repository_url');
 195             }
 196
 197             // parse all found css styles
 198             if (strlen($csstoanalyze)) {
 199                 $urls = extract_css_urls($csstoanalyze);
 200                 if (!empty($urls['property'])) {
 201                     foreach ($urls['property'] as $url) {
 202                         $this->add_image_to_list($info['url'], $url, $list);
 203                     }
 204                 }
 205                 if (!empty($urls['import'])) {
 206                     foreach ($urls['import'] as $cssurl) {
 207                         // Limit the number of CSS imports to avoid infinite imports.
 208                         if ($this->cssimportcounter >= self::MAX_CSS_IMPORTS) {
 209                             return;
 210                         }
 211                         $this->cssimportcounter++;
 212                         $this->parse_file($info['url'], $cssurl, $list);
 213                     }
 214                 }
 215             }
 216         }
 217     }
 218     protected function add_image_to_list($baseurl, $url, &$list) {
 219         if (empty($list['list'])) {
 220             $list['list'] = array();
 221         }
 222         $src = url_to_absolute($baseurl, htmlspecialchars_decode($url));
 223         foreach ($list['list'] as $image) {
 224             if ($image['source'] == $src) {
 225                 return;
 226             }
 227         }
 228         $list['list'][] = array(
 229             'title'=>$this->guess_filename($url, ''),
 230             'source'=>$src,
 231             'thumbnail'=>$src,
 232             'thumbnail_height'=>84,
 233             'thumbnail_width'=>84
 234         );
 235     }
 236     public function guess_filename($url, $type) {
 237         $pattern = '#\/([\w_\?\-.]+)$#';
 238         $matches = null;
 239         preg_match($pattern, $url, $matches);
 240         if (empty($matches[1])) {
 241             return $url;
 242         } else {
 243             return $matches[1];
 244         }
 245     }
 246
 247     /**
 248      * Escapes a url by replacing spaces with %20.
 249      *
 250      * Note: In general moodle does not automatically escape urls, but for the purposes of making this plugin more user friendly
 251      * and make it consistent with some other areas in moodle (such as mod_url), urls will automatically be escaped.
 252      *
 253      * If moodle_url or PARAM_URL is changed to clean characters that need to be escaped, then this function can be removed
 254      *
 255      * @param string $url An unescaped url.
 256      * @return string The escaped url
 257      */
 258     protected function escape_url($url) {
 259         $url = str_replace('"', '%22', $url);
 260         $url = str_replace('\'', '%27', $url);
 261         $url = str_replace(' ', '%20', $url);
 262         $url = str_replace('<', '%3C', $url);
 263         $url = str_replace('>', '%3E', $url);
 264         return $url;
 265     }
 266
 267     public function supported_returntypes() {
 268         return (FILE_INTERNAL | FILE_EXTERNAL);
 269     }
 270
 271     /**
 272      * Return the source information
 273      *
 274      * @param stdClass $url
 275      * @return string|null
 276      */
 277     public function get_file_source_info($url) {
 278         return $url;
 279     }
 280
 281     /**
 282      * file types supported by url downloader plugin
 283      *
 284      * @return array
 285      */
 286     public function supported_filetypes() {
 287         return array('web_image');
 288     }
 289
 290     /**
 291      * Is this repository accessing private data?
 292      *
 293      * @return bool
 294      */
 295     public function contains_private_data() {
 296         return false;
 297     }
 298 }