utils/update.php

   1 #!/usr/bin/php -Cq
   2 <?php
   3
   4         require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
   5         ini_set('memory_limit', '800M');
   6
   7         $aCMDOptions = array(
   8                 "Import / update / index osm data",
   9                 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
  10                 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
  11                 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
  12
  13                 array('max-load', '', 0, 1, 1, 1, 'float', 'Maximum load average - indexing is paused if this is exceeded'),
  14                 array('max-blocking', '', 0, 1, 1, 1, 'int', 'Maximum blocking processes - indexing is aborted / paused if this is exceeded'),
  15
  16                 array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
  17                 array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
  18                 array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'),
  19                 array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
  20
  21                 array('import-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'),
  22
  23                 array('import-hourly', '', 0, 1, 0, 0, 'bool', 'Import hourly diffs'),
  24                 array('import-daily', '', 0, 1, 0, 0, 'bool', 'Import daily diffs'),
  25                 array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
  26
  27                 array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
  28                 array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
  29                 array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
  30
  31                 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
  32                 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
  33                 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
  34                 array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
  35
  36                 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
  37                 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
  38                 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
  39                 array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
  40
  41                 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
  42         );
  43         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
  44
  45         if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
  46         {
  47                 showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
  48         }
  49
  50         if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
  51         if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
  52
  53 /*
  54         // Lock to prevent multiple copies running
  55         if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
  56         {
  57                 echo "Copy already running\n";
  58                 exit;
  59         }
  60         if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
  61         if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
  62         if (getBlockingProcesses() > $aResult['max-blocking'])
  63         {
  64                 echo "Too many blocking processes for import\n";
  65                 exit;
  66         }
  67 */
  68
  69         // Assume osm2pgsql is in the folder above
  70         $sBasePath = dirname(dirname(__FILE__));
  71
  72         date_default_timezone_set('Etc/UTC');
  73
  74         $oDB =& getDB();
  75
  76         $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
  77         if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
  78
  79         // cache memory to be used by osm2pgsql, should not be more than the available memory
  80         $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
  81         if ($iCacheMemory + 500 > getTotalMemoryMB())
  82         {
  83                 $iCacheMemory = getCacheMemoryMB();
  84                 echo "WARNING: resetting cache memory to $iCacheMemory\n";
  85         }
  86         $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
  87         if (!is_null(CONST_Osm2pgsql_Flatnode_File))
  88         {
  89                 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
  90         }
  91
  92
  93         $bFirst = true;
  94         $bContinue = $aResult['import-all'];
  95         while ($bContinue || $bFirst)
  96         {
  97                 $bFirst = false;
  98
  99                 if ($aResult['import-hourly'])
 100                 {
 101                         // Mirror the hourly diffs
 102                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
 103                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
 104                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
 105                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
 106                 }
 107
 108                 if ($aResult['import-daily'])
 109                 {
 110                         // Mirror the daily diffs
 111                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
 112                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
 113                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
 114                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
 115                 }
 116
 117                 if (isset($aResult['import-diff']))
 118                 {
 119                         // import diff directly (e.g. from osmosis --rri)
 120                         $sNextFile = $aResult['import-diff'];
 121                         if (!file_exists($sNextFile))
 122                         {
 123                                 echo "Cannot open $sNextFile\n";
 124                                 exit;
 125                         }
 126                         // Don't update the import status - we don't know what this file contains
 127                         $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
 128                 }
 129
 130                 // Missing file is not an error - it might not be created yet
 131                 if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
 132                 {
 133                         // Import the file
 134                         $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
 135                         echo $sCMD."\n";
 136                         exec($sCMD, $sJunk, $iErrorLevel);
 137
 138                         if ($iErrorLevel)
 139                         {
 140                                 echo "Error from osm2pgsql, $iErrorLevel\n";
 141                                 exit;
 142                         }
 143
 144                         // Move the date onwards
 145                         $oDB->query($sUpdateSQL);
 146                 }
 147                 else
 148                 {
 149                         $bContinue = false;
 150                 }
 151         }
 152
 153         $bModifyXML = false;
 154         $sModifyXMLstr = '';
 155         $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
 156         if (isset($aResult['import-file']) && $aResult['import-file'])
 157         {
 158                 $bModifyXML = true;
 159         }
 160         if (isset($aResult['import-node']) && $aResult['import-node'])
 161         {
 162                 $bModifyXML = true;
 163                 if ($bUseOSMApi)
 164                 {
 165                         $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
 166                 }
 167                 else
 168                 {
 169                         $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
 170                 }
 171         }
 172         if (isset($aResult['import-way']) && $aResult['import-way'])
 173         {
 174                 $bModifyXML = true;
 175                 if ($bUseOSMApi)
 176                 {
 177                         $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
 178                 }
 179                 else
 180                 {
 181                         $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
 182                 }
 183                 $sModifyXMLstr = file_get_contents($sCmd);
 184         }
 185         if (isset($aResult['import-relation']) && $aResult['import-relation'])
 186         {
 187                 $bModifyXML = true;
 188                 if ($bUseOSMApi)
 189                 {
 190                         $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
 191                 }
 192                 else
 193                 {
 194                         $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
 195                 }
 196         }
 197         if ($bModifyXML)
 198         {
 199                 // derive change from normal osm file with osmosis
 200                 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 201                 if (isset($aResult['import-file']) && $aResult['import-file'])
 202                 {
 203                         $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 204                         echo $sCMD."\n";
 205                         exec($sCMD, $sJunk, $iErrorLevel);
 206                         if ($iErrorLevel)
 207                         {
 208                                 echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n";
 209                                 exit;
 210                         }
 211                 }
 212                 else
 213                 {
 214                         $aSpec = array(
 215                                 0 => array("pipe", "r"),  // stdin
 216                                 1 => array("pipe", "w"),  // stdout
 217                                 2 => array("pipe", "w") // stderr
 218                         );
 219                         $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 220                         echo $sCMD."\n";
 221                         $hProc = proc_open($sCMD, $aSpec, $aPipes);
 222                         if (!is_resource($hProc))
 223                         {
 224                                 echo "Error converting osm to osc, osmosis failed\n";
 225                                 exit;
 226                         }
 227                         fwrite($aPipes[0], $sModifyXMLstr);
 228                         fclose($aPipes[0]);
 229                         $sOut = stream_get_contents($aPipes[1]);
 230                         if ($aResult['verbose']) echo $sOut;
 231                         fclose($aPipes[1]);
 232                         $sErrors = stream_get_contents($aPipes[2]);
 233                         if ($aResult['verbose']) echo $sErrors;
 234                         fclose($aPipes[2]);
 235                         if ($iError = proc_close($hProc))
 236                         {
 237                                 echo "Error converting osm to osc, osmosis returned: $iError\n";
 238                                 echo $sOut;
 239                                 echo $sErrors;
 240                                 exit;
 241                         }
 242                 }
 243
 244                 // import generated change file
 245                 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
 246                 echo $sCMD."\n";
 247                 exec($sCMD, $sJunk, $iErrorLevel);
 248                 if ($iErrorLevel)
 249                 {
 250                         echo "osm2pgsql exited with error level $iErrorLevel\n";
 251                         exit;
 252                 }
 253         }
 254
 255         if ($aResult['deduplicate'])
 256         {
 257
 258                 $pgver = (float) CONST_Postgresql_Version;
 259                 if ($pgver < 9.3) {
 260                         echo "ERROR: deduplicate is only currently supported in postgresql 9.3";
 261                         exit;
 262                 }
 263
 264                 $oDB =& getDB();
 265                 $sSQL = 'select partition from country_name order by country_code';
 266                 $aPartitions = $oDB->getCol($sSQL);
 267                 if (PEAR::isError($aPartitions))
 268                 {
 269                         fail($aPartitions->getMessage());
 270                 }
 271                 $aPartitions[] = 0;
 272
 273                 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' ' and class is null and type is null and country_code is null group by word_token having count(*) > 1 order by word_token";
 274                 $aDuplicateTokens = $oDB->getAll($sSQL);
 275                 foreach($aDuplicateTokens as $aToken)
 276                 {
 277                         if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
 278                         echo "Deduping ".$aToken['word_token']."\n";
 279                         $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
 280                         $aTokenSet = $oDB->getAll($sSQL);
 281                         if (PEAR::isError($aTokenSet))
 282                         {
 283                                 var_dump($aTokenSet, $sSQL);
 284                                 exit;
 285                         }
 286
 287                         $aKeep = array_shift($aTokenSet);
 288                         $iKeepID = $aKeep['word_id'];
 289
 290                         foreach($aTokenSet as $aRemove)
 291                         {
 292                                 $sSQL = "update search_name set";
 293                                 $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),";
 294                                 $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
 295                                 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 296                                 $x = $oDB->query($sSQL);
 297                                 if (PEAR::isError($x))
 298                                 {
 299                                         var_dump($x);
 300                                         exit;
 301                                 }
 302
 303                                 $sSQL = "update search_name set";
 304                                 $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
 305                                 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
 306                                 $x = $oDB->query($sSQL);
 307                                 if (PEAR::isError($x))
 308                                 {
 309                                         var_dump($x);
 310                                         exit;
 311                                 }
 312
 313                                 $sSQL = "update location_area_country set";
 314                                 $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
 315                                 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 316                                 $x = $oDB->query($sSQL);
 317                                 if (PEAR::isError($x))
 318                                 {
 319                                         var_dump($x);
 320                                         exit;
 321                                 }
 322
 323                                 foreach ($aPartitions as $sPartition)
 324                                 {
 325                                         $sSQL = "update search_name_".$sPartition." set";
 326                                         $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")";
 327                                         $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 328                                         $x = $oDB->query($sSQL);
 329                                         if (PEAR::isError($x))
 330                                         {
 331                                                 var_dump($x);
 332                                                 exit;
 333                                         }
 334
 335                                         $sSQL = "update location_area_country set";
 336                                         $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
 337                                         $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 338                                         $x = $oDB->query($sSQL);
 339                                         if (PEAR::isError($x))
 340                                         {
 341                                                 var_dump($x);
 342                                                 exit;
 343                                         }
 344                                 }
 345
 346                                 $sSQL = "delete from word where word_id = ".$aRemove['word_id'];
 347                                 $x = $oDB->query($sSQL);
 348                                 if (PEAR::isError($x))
 349                                 {
 350                                         var_dump($x);
 351                                         exit;
 352                                 }
 353                         }
 354
 355                 }
 356         }
 357
 358         if ($aResult['index'])
 359         {
 360                 passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
 361         }
 362
 363         if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
 364         {
 365
 366                 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
 367                         echo "Error: Update interval too low for download.geofabrik.de.  Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n";
 368                         exit;
 369                 }
 370
 371                 $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
 372                 $sOsmosisCMD = CONST_Osmosis_Binary;
 373                 $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
 374                 $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
 375                 $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
 376                 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
 377                 $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
 378                 if (!$aResult['no-npi']) {
 379                         $sCMDIndex .= '-F ';
 380                 }
 381                 while(true)
 382                 {
 383                         $fStartTime = time();
 384                         $iFileSize = 1001;
 385
 386                         // Logic behind this is that osm2pgsql locks the database quite a bit
 387                         // So it is better to import lots of small files
 388                         // But indexing works most efficiently on large amounts of data
 389                         // So do lots of small imports and a BIG index
 390
 391 //                      while($aResult['import-osmosis-all'] && $iFileSize > 1000)
 392 //                      {
 393                                 if (!file_exists($sImportFile))
 394                                 {
 395                                         // First check if there are new updates published (except for minutelies - there's always new diffs to process)
 396                                         if ( CONST_Replication_Update_Interval > 60 )
 397                                         {
 398
 399                                                 unset($aReplicationLag);
 400                                                 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
 401                                                 while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
 402                                                 {
 403                                                         if ($iErrorLevel)
 404                                                         {
 405                                                                 echo "Error: $iErrorLevel. ";
 406                                                                 echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
 407                                                         }
 408                                                         else
 409                                                         {
 410                                                                 echo ".";
 411                                                         }
 412                                                         sleep(CONST_Replication_Recheck_Interval);
 413                                                         unset($aReplicationLag);
 414                                                         exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
 415                                                 }
 416                                                 // There are new replication files - use osmosis to download the file
 417                                                 echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
 418                                         }
 419                                         $fStartTime = time();
 420                                         $fCMDStartTime = time();
 421                                         echo $sCMDDownload."\n";
 422                                         exec($sCMDDownload, $sJunk, $iErrorLevel);
 423                                         while ($iErrorLevel > 0)
 424                                         {
 425                                                 echo "Error: $iErrorLevel\n";
 426                                                 sleep(60);
 427                                                 echo 'Re-trying: '.$sCMDDownload."\n";
 428                                                 exec($sCMDDownload, $sJunk, $iErrorLevel);
 429                                         }
 430                                         $iFileSize = filesize($sImportFile);
 431                                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 432                                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
 433                                         var_Dump($sSQL);
 434                                         $oDB->query($sSQL);
 435                                         echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 436                                 }
 437
 438                                 $iFileSize = filesize($sImportFile);
 439                                 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 440
 441                                 // Import the file
 442                                 $fCMDStartTime = time();
 443                                 echo $sCMDImport."\n";
 444                                 exec($sCMDImport, $sJunk, $iErrorLevel);
 445                                 if ($iErrorLevel)
 446                                 {
 447                                         echo "Error: $iErrorLevel\n";
 448                                         exit($iErrorLevel);
 449                                 }
 450                                 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
 451                                 var_Dump($sSQL);
 452                                 $oDB->query($sSQL);
 453                                 echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 454
 455                                 // Archive for debug?
 456                                 unlink($sImportFile);
 457 //                      }
 458
 459                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 460
 461                         // Index file
 462                         $sThisIndexCmd = $sCMDIndex;
 463                         $fCMDStartTime = time();
 464
 465                         if (!$aResult['no-npi'])
 466                         {
 467                                 $iFileID = $oDB->getOne('select nextval(\'file\')');
 468                                 if (PEAR::isError($iFileID))
 469                                 {
 470                                         echo $iFileID->getMessage()."\n";
 471                                         exit(-1);
 472                                 }
 473                                 $sFileDir = CONST_BasePath.'/export/diff/';
 474                                 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 475                                 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 476
 477                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 478                                 $sThisIndexCmd .= $sFileDir;
 479                                 $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
 480                                 $sThisIndexCmd .= ".npi.out";
 481
 482                                 preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
 483                                 $sFileDir = CONST_BasePath.'/export/index/';
 484                                 $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
 485
 486                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 487                                 file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
 488                         }
 489
 490                         if (!$aResult['no-index'])
 491                         {
 492                                 echo "$sThisIndexCmd\n";
 493                                 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 494                                 if ($iErrorLevel)
 495                                 {
 496                                         echo "Error: $iErrorLevel\n";
 497                                         exit($iErrorLevel);
 498                                 }
 499
 500                                 if (!$aResult['no-npi'])
 501                                 {
 502                                         $sFileDir = CONST_BasePath.'/export/diff/';
 503                                         $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 504                                         $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 505
 506                                         $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
 507                                         echo "$sThisIndexCmd\n";
 508                                         exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 509                                         if ($iErrorLevel)
 510                                         {
 511                                                 echo "Error: $iErrorLevel\n";
 512                                                 exit($iErrorLevel);
 513                                         }
 514
 515                                         rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
 516                                                 $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
 517                                 }
 518                         }
 519
 520                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
 521                         var_Dump($sSQL);
 522                         $oDB->query($sSQL);
 523                         echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 524
 525                         $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
 526                         $oDB->query($sSQL);
 527
 528                         $fDuration = time() - $fStartTime;
 529                         echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n";
 530                         if (!$aResult['import-osmosis-all']) exit;
 531
 532                         if ( CONST_Replication_Update_Interval > 60 )
 533                         {
 534                                 $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
 535                         }
 536                         else
 537                         {
 538                                 $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
 539                         }
 540                         echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
 541                         sleep($iSleep);
 542                 }
 543
 544         }
 545
 546         if ($aResult['import-npi-all'])
 547         {
 548                 $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
 549                 if (PEAR::isError($iNPIID))
 550                 {
 551                         var_dump($iNPIID);
 552                         exit;
 553                 }
 554                 $sConfigDirectory = CONST_BasePath.'/settings';
 555                 $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
 556                 while(true)
 557                 {
 558                         $fStartTime = time();
 559
 560                         $iNPIID++;
 561
 562                         $sImportFile = CONST_BasePath.'/export/diff/';
 563                         $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
 564                         $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 565                         $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
 566                         $sImportFile .= ".npi";
 567                         while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
 568                         {
 569                                 echo "sleep (waiting for $sImportFile)\n";
 570                                 sleep(10);
 571                         }
 572                         if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
 573
 574                         $iFileSize = filesize($sImportFile);
 575
 576                         // Import the file
 577                         $fCMDStartTime = time();
 578                         $sCMDImport = $sCMDImportTemplate . $sImportFile;
 579                         echo $sCMDImport."\n";
 580                         exec($sCMDImport, $sJunk, $iErrorLevel);
 581                         if ($iErrorLevel)
 582                         {
 583                                 echo "Error: $iErrorLevel\n";
 584                                 exit;
 585                         }
 586                         $sBatchEnd = $iNPIID;
 587                         echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 588                         $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
 589                         var_Dump($sSQL);
 590                         $oDB->query($sSQL);
 591                 }
 592
 593         }
 594
 595         function getosmosistimestamp($sOsmosisConfigDirectory)
 596         {
 597                 $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
 598                 preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
 599                 return str_replace('\:',':',$aResult[1]);
 600         }