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'] = $argv[1];
  34
  35 chdir(str_replace('/scripts/tz-update.php','/htdocs',$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 $new_zones = 0;
  48 $modified_zones = 0;
  49 $added_aliases = 0;
  50
  51
  52 function fetch_remote_list($base_url ) {
  53   global $request;
  54   $result = array();
  55   $list_url = $base_url . '?action=list';
  56   printf( "Fetching timezone list\n", $list_url );
  57   $raw_xml = file_get_contents($list_url);
  58   $xml_parser = xml_parser_create_ns('UTF-8');
  59   $xml_tags = array();
  60   xml_parser_set_option ( $xml_parser, XML_OPTION_SKIP_WHITE, 1 );
  61   xml_parser_set_option ( $xml_parser, XML_OPTION_CASE_FOLDING, 0 );
  62   $rc = xml_parse_into_struct( $xml_parser, $raw_xml, $xml_tags );
  63   if ( $rc == false ) {
  64     dbg_error_log( 'ERROR', 'XML parsing error: %s at line %d, column %d', xml_error_string(xml_get_error_code($xml_parser)),
  65                 xml_get_current_line_number($xml_parser), xml_get_current_column_number($xml_parser) );
  66     $request->PreconditionFailed(400, 'invalid-xml',
  67                 sprintf('XML parsing error: %s at line %d, column %d', xml_error_string(xml_get_error_code($xml_parser)),
  68                 xml_get_current_line_number($xml_parser), xml_get_current_column_number($xml_parser) ));
  69   }
  70   xml_parser_free($xml_parser);
  71   $position = 0;
  72   return BuildXMLTree( $xml_tags, $position);
  73 }
  74
  75 function fetch_remote_zone( $base_url, $tzid ) {
  76   $tz_url = $base_url . '?action=get&tzid=' .$tzid;
  77   printf( "Fetching zone for %s from %s\n", $tzid, $tz_url );
  78   dbg_error_log( 'tz/updatecheck', "Fetching zone for %s from %s\n", $tzid, $tz_url );
  79   $vtimezone = file_get_contents( $tz_url );
  80   return $vtimezone;
  81 }
  82
  83 function fetch_db_zone( $tzid ) {
  84   $tzrow = null;
  85   $qry = new AwlQuery('SELECT * FROM timezones WHERE tzid=:tzid', array(':tzid' => $tzid));
  86   if ( $qry->Exec('tz/update',__LINE__,__FILE__) && $qry->rows() > 0 ) {
  87     $tzrow = $qry->Fetch();
  88   }
  89   return $tzrow;
  90 }
  91
  92 function write_updated_zone( $vtimezone, $tzid ) {
  93   global $new_zones, $modified_zones;
  94   if ( empty($vtimezone) ) {
  95     dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no data from server', $tzid );
  96     return;
  97   }
  98   $tzrow = fetch_db_zone($tzid);
  99   if ( isset($tzrow) && $vtimezone == $tzrow->vtimezone ) {
 100     dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no change', $tzid );
 101     return;
 102   }
 103   $vtz = new vCalendar($vtimezone);
 104   $last_modified = $vtz->GetPValue('LAST-MODIFIED');
 105   if ( empty($last_modified) ) {
 106     $last_modified = gmdate('Ymd\THis\Z');
 107     // Then it was probably that way when we last updated the data, too :-(
 108     if ( !empty($tzrow) ) {
 109       $old_vtz = new vCalendar($tzrow->vtimezone);
 110       $old_vtz->ClearProperties('LAST-MODIFIED');
 111       // We need to add & remove this property so the Render is equivalent.
 112       $vtz->AddProperty('LAST-MODIFIED',$last_modified);
 113       $vtz->ClearProperties('LAST-MODIFIED');
 114       if ( $vtz->Render() == $old_vtz->Render() ) {
 115         dbg_error_log('tz/updatecheck', 'Skipping zone "%s" - no change', $tzid );
 116         return;
 117       }
 118     }
 119     $vtz->AddProperty('LAST-MODIFIED',$last_modified);
 120   }
 121   dbg_error_log('tz/updatecheck', 'Writing %s zone for "%s"', (empty($tzrow)?"new":"updated"), $tzid );
 122   printf("Writing %s zone for '%s'\n", (empty($tzrow)?"new":"updated"), $tzid );
 123   $params = array(
 124             ':tzid'           => $tzid,
 125             ':olson_name'     => $tzid,
 126             ':vtimezone'      => $vtz->Render(),
 127             ':last_modified'  => $last_modified,
 128             ':etag'                     => md5($vtz->Render())
 129   );
 130   if ( empty($tzrow) ) {
 131     $new_zones++;
 132     $sql = 'INSERT INTO timezones(tzid,active,olson_name,last_modified,etag,vtimezone) ';
 133     $sql .= 'VALUES(:tzid,TRUE,:olson_name,:last_modified,:etag,:vtimezone)';
 134   }
 135   else {
 136     $modified_zones++;
 137     $sql = 'UPDATE timezones SET active=TRUE, olson_name=:olson_name, last_modified=:last_modified, ';
 138     $sql .= 'etag=:etag, vtimezone=:vtimezone WHERE tzid=:tzid';
 139   }
 140   $qry = new AwlQuery($sql,$params);
 141   $qry->Exec('tz/update',__LINE__,__FILE__);
 142 }
 143
 144 function write_zone_aliases( $tzid, $aliases ) {
 145   global $added_aliases;
 146   foreach( $aliases AS $alias_node ) {
 147     $alias = $alias_node->GetContent();
 148     $params = array(':tzid' => $tzid, ':alias' => $alias );
 149     $qry = new AwlQuery('SELECT * FROM tz_aliases JOIN timezones USING(our_tzno) WHERE tzid=:tzid AND tzalias=:alias', $params);
 150     if ( $qry->Exec('tz/update', __LINE__, __FILE__) && $qry->rows() < 1 ) {
 151       $qry->QDo('INSERT INTO tz_aliases(our_tzno,tzalias) SELECT our_tzno, :alias FROM timezones WHERE tzid = :tzid', $params);
 152       $added_aliases++;
 153     }
 154   }
 155 }
 156
 157
 158 if ( empty($c->tzsource) ) $c->tzsource = '../zonedb/vtimezones';
 159 if ( preg_match('{^http}', $c->tzsource ) ) {
 160
 161   $changesince = null;
 162   $qry = new AwlQuery("SELECT tzid, to_char(last_modified,'YYYY-MM-DD\"T\"HH24:MI:SS\"Z\"') AS last_modified FROM timezones");
 163   $current_zones = array();
 164   if ( $qry->Exec('tz/updatecheck',__LINE__,__FILE__) && $qry->rows() > 0 ) {
 165     while( $row = $qry->Fetch() )
 166       $current_zones[$row->tzid] = new RepeatRuleDateTime($row->last_modified);
 167   }
 168
 169   $xmltree = fetch_remote_list($c->tzsource);
 170   $zones = $xmltree->GetElements('urn:ietf:params:xml:ns:timezone-service:summary');
 171   foreach( $zones AS $zone ) {
 172     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:tzid');
 173     $tzid = $elements[0]->GetContent();
 174     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:last-modified');
 175     $last_modified = new RepeatRuleDateTime($elements[0]->GetContent());
 176     if ( !isset($current_zones[$tzid]) || $last_modified > $current_zones[$tzid] ) {
 177       printf("Found timezone %s needs updating\n", $tzid );
 178       $vtimezone = fetch_remote_zone($c->tzsource,$tzid);
 179       write_updated_zone($vtimezone, $tzid);
 180     }
 181     $elements = $zone->GetElements('urn:ietf:params:xml:ns:timezone-service:alias');
 182     write_zone_aliases($tzid, $elements);
 183   }
 184
 185 }
 186 else if ( file_exists($c->tzsource) ) {
 187   /**
 188    * Find all files recursively within the diectory given.
 189    * @param string $dirname The directory to find files in
 190    * @return array of filenames with full path
 191    */
 192   function recursive_files( $dirname ) {
 193     $d = opendir($dirname);
 194     $result = array();
 195     while( $fn = readdir($d) ) {
 196       if ( substr($fn,0,1) == '.' ) continue;
 197       if ( substr($fn,0,14) == 'primary-source' ) continue;
 198       $fn = $dirname.'/'.$fn;
 199       if ( is_dir($fn) ) {
 200         $result = array_merge($result,recursive_files($fn));
 201       }
 202       else {
 203         $result[] = $fn;
 204       }
 205     }
 206     return $result;
 207   }
 208
 209   $qry = new AwlQuery();
 210   foreach( recursive_files($c->tzsource) AS $filename ) {
 211     $tzid = str_replace('.ics', '', str_replace($c->tzsource.'/', '', $filename));
 212     $vtimezone = file_get_contents( $filename, false );
 213     write_updated_zone($vtimezone, $tzid);
 214   }
 215 }
 216 else {
 217   dbg_error_log('ERROR', '$c->tzsource is not configured to a good source of timezone data');
 218 }
 219
 220 printf("Added %d new zones, updated data for %d zones and added %d new aliases\n",
 221            $new_zones, $modified_zones, $added_aliases);
 222
 223 exit(0);