scripts/tz-update.php

   1 #!/usr/bin/env php
   2 <?php
   3 /**
   4 * DAViCal Timezone Service handler - update timezones
   5 *
   6 * @package   davical
   7 * @subpackage   tzservice
   8 * @author    Andrew McMillan <andrew@morphoss.com>
   9 * @copyright Morphoss Ltd
  10 * @license   http://gnu.org/copyleft/gpl.html GNU GPL v3 or later
  11 */
  12 $script_file = __FILE__;
  13 if ( $argc < 2 ) {
  14
  15   echo <<<USAGE
  16 Usage:
  17
  18         $script_file davical.example.com [timezone_source]
  19
  20 Where 'davical.example.com' is the hostname of your DAViCal server and the
  21 optional 'timezone_source' is the source of the timezone data.  If not specified
  22 this will default to the value of the $$c->tzsource configuration value, with
  23 a further default to the zonedb/vtimezone directory relative to the root of the
  24 DAViCal installation.
  25
  26 This script can be used to initialise or update the timezone information in
  27 DAViCal used for the in-built timezone service.
  28
  29 USAGE;
  30   exit(1);
  31 }
  32
  33 $_SERVER['SERVER_NAME'] = $argv[1];
  34 $original_dir = getcwd();
  35 chdir(str_replace('/scripts/tz-update.php','/htdocs',$script_file));
  36
  37 require_once("./always.php");
  38
  39 if ( isset($argv[2]) ) {
  40   $c->tzsource = $argv[2];
  41 }
  42
  43 require_once('vCalendar.php');
  44 require_once('XMLDocument.php');
  45 require_once('RRule-v2.php');
  46
  47 chdir($original_dir);
  48
  49 $new_zones = 0;
  50 $modified_zones = 0;
  51 $added_aliases = 0;
  52
  53
  54 function fetch_remote_list($base_url ) {
  55   global $request;
  56   $result = array();
  57   $list_url = $base_url . '?action=list';
  58   printf( "Fetching timezone list\n", $list_url );
  59   $raw_xml = file_get_contents($list_url);
  60   $xml_parser = xml_parser_create_ns('UTF-8');
  61   $xml_tags = array();
  62   xml_parser_set_option ( $xml_parser, XML_OPTION_SKIP_WHITE, 1 );
  63   xml_parser_set_option ( $xml_parser, XML_OPTION_CASE_FOLDING, 0 );
  64   $rc = xml_parse_into_struct( $xml_parser, $raw_xml, $xml_tags );
  65   if ( $rc == false ) {
  66     dbg_error_log( 'ERROR', 'XML parsing error: %s at line %d, column %d', xml_error_string(xml_get_error_code($xml_parser)),
  67                 xml_get_current_line_number($xml_parser), xml_get_current_column_number($xml_parser) );
  68     $request->PreconditionFailed(400, 'invalid-xml',
  69                 sprintf('XML parsing error: %s at line %d, column %d', xml_error_string(xml_get_error_code($xml_parser)),
  70                 xml_get_current_line_number($xml_parser), xml_get_current_column_number($xml_parser) ));
  71   }
  72   xml_parser_free($xml_parser);
  73   $position = 0;
  74   return BuildXMLTree( $xml_tags, $position);
  75 }
  76
  77 function fetch_remote_zone( $base_url, $tzid ) {
  78   $tz_url = $base_url . '?action=get&tzid=' .$tzid;
  79   printf( "Fetching zone for %s from %s\n", $tzid, $tz_url );
  80   dbg_error_log( 'tz/updatecheck', "Fetching zone for %s from %s\n", $tzid, $tz_url );
  81   $vtimezone = file_get_contents( $tz_url );
  82   return $vtimezone;
  83 }
  84
  85 function fetch_db_zone( $tzid ) {
  86   $tzrow = null;
  87   $qry = new AwlQuery('SELECT * FROM timezones WHERE tzid=:tzid', array(':tzid' => $tzid));
  88   if ( $qry->Exec('tz/update',__LINE__,__FILE__) && $qry->rows() > 0 ) {
  89     $tzrow = $qry->Fetch();
  90   }
  91   return $tzrow;
  92 }
  93
  94 function write_updated_zone( $vtimezone, $tzid ) {
  95   global $new_zones, $modified_zones;
  96   if ( empty($vtimezone) ) {
  97     dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no data from server', $tzid );
  98     return;
  99   }
 100   $tzrow = fetch_db_zone($tzid);
 101   if ( isset($tzrow) && $vtimezone == $tzrow->vtimezone ) {
 102     dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no change', $tzid );
 103     return;
 104   }
 105   $vtz = new vCalendar($vtimezone);
 106   $last_modified = $vtz->GetPValue('LAST-MODIFIED');
 107   if ( empty($last_modified) ) {
 108     $last_modified = gmdate('Ymd\THis\Z');
 109     // Then it was probably that way when we last updated the data, too :-(
 110     if ( !empty($tzrow) ) {
 111       $old_vtz = new vCalendar($tzrow->vtimezone);
 112       $old_vtz->ClearProperties('LAST-MODIFIED');
 113       // We need to add & remove this property so the Render is equivalent.
 114       $vtz->AddProperty('LAST-MODIFIED',$last_modified);
 115       $vtz->ClearProperties('LAST-MODIFIED');
 116       if ( $vtz->Render() == $old_vtz->Render() ) {
 117         dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no change', $tzid );
 118         return;
 119       }
 120     }
 121     $vtz->AddProperty('LAST-MODIFIED',$last_modified);
 122   }
 123   dbg_error_log('tz/updatecheck', 'Writing %s zone for "%s"', (empty($tzrow)?"new":"updated"), $tzid );
 124   printf("Writing %s zone for '%s'\n", (empty($tzrow)?"new":"updated"), $tzid );
 125   $params = array(
 126             ':tzid'           => $tzid,
 127             ':olson_name'     => $tzid,
 128             ':vtimezone'      => $vtz->Render(),
 129             ':last_modified'  => $last_modified,
 130             ':etag'                     => md5($vtz->Render())
 131   );
 132   if ( empty($tzrow) ) {
 133     $new_zones++;
 134     $sql = 'INSERT INTO timezones(tzid,active,olson_name,last_modified,etag,vtimezone) ';
 135     $sql .= 'VALUES(:tzid,TRUE,:olson_name,:last_modified,:etag,:vtimezone)';
 136   }
 137   else {
 138     $modified_zones++;
 139     $sql = 'UPDATE timezones SET active=TRUE, olson_name=:olson_name, last_modified=:last_modified, ';
 140     $sql .= 'etag=:etag, vtimezone=:vtimezone WHERE tzid=:tzid';
 141   }
 142   $qry = new AwlQuery($sql,$params);
 143   $qry->Exec('tz/update',__LINE__,__FILE__);
 144 }
 145
 146 function write_zone_aliases( $tzid, $aliases ) {
 147   global $added_aliases;
 148   foreach( $aliases AS $alias_node ) {
 149     $alias = $alias_node->GetContent();
 150     $params = array(':tzid' => $tzid, ':alias' => $alias );
 151     $qry = new AwlQuery('SELECT * FROM tz_aliases JOIN timezones USING(our_tzno) WHERE tzid=:tzid AND tzalias=:alias', $params);
 152     if ( $qry->Exec('tz/update', __LINE__, __FILE__) && $qry->rows() < 1 ) {
 153       $qry->QDo('INSERT INTO tz_aliases(our_tzno,tzalias) SELECT our_tzno, :alias FROM timezones WHERE tzid = :tzid', $params);
 154       $added_aliases++;
 155     }
 156   }
 157 }
 158
 159
 160 if ( empty($c->tzsource) ) $c->tzsource = '../zonedb/vtimezones';
 161 if ( preg_match('{^http}', $c->tzsource ) ) {
 162
 163   $changesince = null;
 164   $qry = new AwlQuery("SELECT tzid, to_char(last_modified,'YYYY-MM-DD\"T\"HH24:MI:SS\"Z\"') AS last_modified FROM timezones");
 165   $current_zones = array();
 166   if ( $qry->Exec('tz/updatecheck',__LINE__,__FILE__) && $qry->rows() > 0 ) {
 167     while( $row = $qry->Fetch() )
 168       $current_zones[$row->tzid] = new RepeatRuleDateTime($row->last_modified);
 169   }
 170
 171   $xmltree = fetch_remote_list($c->tzsource);
 172   $zones = $xmltree->GetElements('urn:ietf:params:xml:ns:timezone-service:summary');
 173   foreach( $zones AS $zone ) {
 174     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:tzid');
 175     $tzid = $elements[0]->GetContent();
 176     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:last-modified');
 177     $last_modified = new RepeatRuleDateTime($elements[0]->GetContent());
 178     if ( !isset($current_zones[$tzid]) || $last_modified > $current_zones[$tzid] ) {
 179       printf("Found timezone %s needs updating\n", $tzid );
 180       $vtimezone = fetch_remote_zone($c->tzsource,$tzid);
 181       write_updated_zone($vtimezone, $tzid);
 182     }
 183     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:alias');
 184     write_zone_aliases($tzid, $elements);
 185   }
 186
 187 }
 188 else if ( file_exists($c->tzsource) ) {
 189   /**
 190    * Find all files recursively within the diectory given.
 191    * @param string $dirname The directory to find files in
 192    * @return array of filenames with full path
 193    */
 194   function recursive_files( $dirname ) {
 195     $d = opendir($dirname);
 196     $result = array();
 197     while( $fn = readdir($d) ) {
 198       if ( substr($fn,0,1) == '.' ) continue;
 199       if ( substr($fn,0,14) == 'primary-source' ) continue;
 200       $fn = $dirname.'/'.$fn;
 201       if ( is_dir($fn) ) {
 202         $result = array_merge($result,recursive_files($fn));
 203       }
 204       else {
 205         $result[] = $fn;
 206       }
 207     }
 208     return $result;
 209   }
 210
 211   $qry = new AwlQuery();
 212   foreach( recursive_files($c->tzsource) AS $filename ) {
 213     $tzid = str_replace('.ics', '', str_replace($c->tzsource.'/', '', $filename));
 214     $vtimezone = file_get_contents( $filename, false );
 215     write_updated_zone($vtimezone, $tzid);
 216   }
 217 }
 218 else {
 219   dbg_error_log('ERROR', '$c->tzsource is not configured to a good source of timezone data');
 220 }
 221
 222 printf("Added %d new zones, updated data for %d zones and added %d new aliases\n",
 223            $new_zones, $modified_zones, $added_aliases);
 224
 225 exit(0);