filecrawler/lib.php

   1 <?php
   2 /*
   3     Filecrawler plugin for Elgg
   4     Initial author: Sven Edge <sven@elgg.net> September 2006
   5
   6     Plugin to crawl user-uploaded files, and
   7     - for files that exist on the filesystem but not in the db, either add to the user's root filestore folder or delete them from the filesystem
   8     - for files with entries in the db but not on the filesystem, purge them from the database
   9
  10     Config options:
  11
  12     $CFG->filecrawler_add_fs_orphans_to_db = true;
  13     $CFG->filecrawler_purge_fs_orphans = false;
  14     $CFG->filecrawler_purge_db_orphans = true;
  15     $CFG->filecrawler_fix_fs_permissions = true;
  16
  17 */
  18
  19     function filecrawler_pagesetup() {
  20
  21     } // end function filecrawler_pagesetup
  22
  23
  24
  25     function filecrawler_init() {
  26
  27         // global $function;
  28         global $CFG;
  29         if (isset($CFG->filecrawler_purge_db_orphans) && $CFG->filecrawler_purge_db_orphans == false) {
  30             $CFG->filecrawler_purge_db_orphans = false;
  31         } else {
  32             $CFG->filecrawler_purge_db_orphans = true;
  33         }
  34
  35         if (isset($CFG->filecrawler_fix_fs_permissions) && $CFG->filecrawler_fix_fs_permissions == false) {
  36             $CFG->filecrawler_fix_fs_permissions = false;
  37         } else {
  38             $CFG->filecrawler_fix_fs_permissions = true;
  39         }
  40
  41         // default to adding, but purge if explicitly told to.
  42         if (isset($CFG->filecrawler_add_fs_orphans_to_db) && $CFG->filecrawler_add_fs_orphans_to_db == false) {
  43             if (empty($CFG->filecrawler_purge_fs_orphans)) {
  44                 $CFG->filecrawler_purge_fs_orphans = false;
  45             } else {
  46                 $CFG->filecrawler_purge_fs_orphans = true;
  47             }
  48         } else {
  49             $CFG->filecrawler_add_fs_orphans_to_db = true;
  50             $CFG->filecrawler_purge_fs_orphans = false;
  51         }
  52
  53     } // end function filecrawler_init
  54
  55
  56
  57     // called by cron.php
  58     function filecrawler_cron() {
  59
  60         if (!empty($CFG->filecrawler_lastcron) && (time() - (60*60*24)) < $CFG->filecrawler_lastcron) {
  61             return true;
  62         } else {
  63             filecrawler_crawl();
  64         }
  65
  66         set_config('filecrawler_lastcron',time());
  67
  68     } // end function filecrawler_cron
  69
  70
  71
  72     // perambulate the user filestore, or just one directory
  73     function filecrawler_crawl($limitinitial = "") {
  74
  75         global $CFG;
  76         if (is_dir($CFG->dataroot . 'files/')) {
  77             chdir($CFG->dataroot . 'files/');
  78
  79             $limitinitial = substr($limitinitial, 0, 1);
  80             if ($limitinitial) {
  81                 $initials = array($limitinitial);
  82             } else {
  83                 $initials = glob("*", GLOB_ONLYDIR); // GLOB_ONLYDIR only on windows php > 4.3.3
  84             }
  85
  86             $userdirs = array();
  87             if (is_array($initials)) {
  88                 foreach ($initials as $aninitial) {
  89                     chdir($CFG->dataroot . 'files/' . $aninitial);
  90                     $userdirs = array_merge($userdirs , glob("*", GLOB_ONLYDIR)); // GLOB_ONLYDIR only on windows php > 4.3.3
  91                 }
  92             }
  93
  94             foreach ($userdirs as $adir) {
  95                 filecrawler_process_dir($adir);
  96             }
  97         }
  98
  99     } // end function filecrawler_crawl
 100
 101
 102
 103     // process a (user's) directory's files.
 104     function filecrawler_process_dir($username) {
 105
 106         global $CFG;
 107         $parentdir = $CFG->dataroot . 'files/';
 108         $username = trim($username);
 109         $initial = substr($username, 0, 1);
 110         $userdir = $initial . '/' . $username . '/';
 111
 112         if ($username && is_dir($parentdir . $userdir) && $userid = user_info_username('ident',$username)) {
 113
 114             if ($CFG->filecrawler_purge_db_orphans == true) {
 115                 $dbfiles = get_records('files', 'files_owner', $userid);
 116                 foreach ($dbfiles as $key => $adbfile) {
 117                     if (!file_exists($CFG->dataroot . $adbfile->location)) {
 118
 119                         mtrace('Deleting files table entry for missing file ' . $adbfile->location . '.');
 120                         // how safe to be...?
 121                         //delete_records('files', 'ident', $adbfile['ident']);
 122                         delete_records('files', 'ident', $adbfile->ident, 'files_owner', $userid);
 123
 124                     }
 125                 }
 126             }
 127
 128             if ($CFG->filecrawler_add_fs_orphans_to_db || $CFG->filecrawler_purge_fs_orphans) {
 129
 130                 // query repetition less efficient but safer
 131                 $dbfiles = get_records('files', 'files_owner', $userid);
 132                 foreach ($dbfiles as $adbfile) {
 133                     $dbfilelocations[] = $CFG->dataroot . $adbfile->location;
 134                 }
 135
 136                 //get list of files
 137                 chdir($parentdir . $userdir);
 138                 $fsfiles = glob("*");
 139                 if (is_array($fsfiles)) {
 140                     foreach ($fsfiles as $afsfile) {
 141                         if (!in_array($parentdir . $userdir . $afsfile, $dbfilelocations)) {
 142                             if ($CFG->filecrawler_add_fs_orphans_to_db) {
 143
 144                                 mtrace('Adding ' . $userdir . $afsfile . ' to files table.');
 145                                 //add it
 146                                 $f = new StdClass;
 147                                 $f->owner = $userid;
 148                                 $f->files_owner = $userid;
 149                                 $f->folder = '-1';
 150                                 $f->originalname = $afsfile;
 151                                 $f->title = 'Auto-added file';
 152                                 $f->description = '';
 153                                 $f->location = 'files/' . $userdir . $afsfile;
 154                                 if ($CFG->default_access == 'PRIVATE') {
 155                                     $f->access = 'user' . $userid;
 156                                 } else {
 157                                     $f->access = $CFG->default_access;
 158                                 }
 159                                 $f->size = filesize($CFG->dataroot . $f->location);
 160                                 $f->time_uploaded = time();
 161                                 $file_id = insert_record('files', $f);
 162                                 if ($file_id) {
 163                                     $rssresult = run("files:rss:publish", array($userid, false));
 164                                     $rssresult = run("profile:rss:publish", array($userid, false));
 165                                 }
 166
 167                             } elseif ($CFG->filecrawler_purge_fs_orphans) {
 168
 169                                 mtrace('Deleting ' . $userdir . $afsfile . '.');
 170                                 unlink($CFG->dataroot . $afsfile);
 171
 172                             }
 173
 174                         } // end if
 175                     } // end foreach
 176                 } // end if
 177             }
 178
 179
 180         } else {
 181             return false;
 182         }
 183
 184         if ($CFG->filecrawler_fix_fs_permissions) {
 185             //mtrace('setting perms on ' . $parentdir . $userdir);
 186             chmod($parentdir . $userdir, $CFG->directorypermissions);
 187             chdir($parentdir . $userdir);
 188             $fsfiles = glob("*");
 189             if (is_array($fsfiles)) {
 190                 foreach ($fsfiles as $afsfile) {
 191                     //mtrace('setting perms on ' . $parentdir . $userdir . $afsfile);
 192                     chmod($parentdir . $userdir . $afsfile, $CFG->filepermissions);
 193                 }
 194             }
 195         }
 196
 197     } // end function filecrawler_process_dir
 198
 199 ?>