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-rank'])) $aResult['index-rank'] = 0;
  51
  52 /*
  53         // Lock to prevent multiple copies running
  54         if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
  55         {
  56                 echo "Copy already running\n";
  57                 exit;
  58         }
  59         if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
  60         if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
  61         if (getBlockingProcesses() > $aResult['max-blocking'])
  62         {
  63                 echo "Too many blocking processes for import\n";
  64                 exit;
  65         }
  66 */
  67
  68         // Assume osm2pgsql is in the folder above
  69         $sBasePath = dirname(dirname(__FILE__));
  70
  71         date_default_timezone_set('Etc/UTC');
  72
  73         $oDB =& getDB();
  74
  75         $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
  76         if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
  77
  78         // cache memory to be used by osm2pgsql, should not be more than the available memory
  79         $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
  80         if ($iCacheMemory + 500 > getTotalMemoryMB())
  81         {
  82                 $iCacheMemory = getCacheMemoryMB();
  83                 echo "WARNING: resetting cache memory to $iCacheMemory\n";
  84         }
  85         $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
  86         if (!is_null(CONST_Osm2pgsql_Flatnode_File))
  87         {
  88                 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
  89         }
  90
  91
  92         $bFirst = true;
  93         $bContinue = $aResult['import-all'];
  94         while ($bContinue || $bFirst)
  95         {
  96                 $bFirst = false;
  97
  98                 if ($aResult['import-hourly'])
  99                 {
 100                         // Mirror the hourly diffs
 101                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
 102                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
 103                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
 104                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
 105                 }
 106
 107                 if ($aResult['import-daily'])
 108                 {
 109                         // Mirror the daily diffs
 110                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
 111                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
 112                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
 113                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
 114                 }
 115
 116                 if (isset($aResult['import-diff']))
 117                 {
 118                         // import diff directly (e.g. from osmosis --rri)
 119                         $sNextFile = $aResult['import-diff'];
 120                         if (!file_exists($sNextFile))
 121                         {
 122                                 echo "Cannot open $sNextFile\n";
 123                                 exit;
 124                         }
 125                         // Don't update the import status - we don't know what this file contains
 126                         $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
 127                 }
 128
 129                 // Missing file is not an error - it might not be created yet
 130                 if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
 131                 {
 132                         // Import the file
 133                         $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
 134                         echo $sCMD."\n";
 135                         exec($sCMD, $sJunk, $iErrorLevel);
 136
 137                         if ($iErrorLevel)
 138                         {
 139                                 echo "Error from osm2pgsql, $iErrorLevel\n";
 140                                 exit;
 141                         }
 142
 143                         // Move the date onwards
 144                         $oDB->query($sUpdateSQL);
 145                 }
 146                 else
 147                 {
 148                         $bContinue = false;
 149                 }
 150         }
 151
 152         $bModifyXML = false;
 153         $sModifyXMLstr = '';
 154         $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
 155         if (isset($aResult['import-file']) && $aResult['import-file'])
 156         {
 157                 $bModifyXML = true;
 158         }
 159         if (isset($aResult['import-node']) && $aResult['import-node'])
 160         {
 161                 $bModifyXML = true;
 162                 if ($bUseOSMApi)
 163                 {
 164                         $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
 165                 }
 166                 else
 167                 {
 168                         $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
 169                 }
 170         }
 171         if (isset($aResult['import-way']) && $aResult['import-way'])
 172         {
 173                 $bModifyXML = true;
 174                 if ($bUseOSMApi)
 175                 {
 176                         $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
 177                 }
 178                 else
 179                 {
 180                         $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
 181                 }
 182                 $sModifyXMLstr = file_get_contents($sCmd);
 183         }
 184         if (isset($aResult['import-relation']) && $aResult['import-relation'])
 185         {
 186                 $bModifyXML = true;
 187                 if ($bUseOSMApi)
 188                 {
 189                         $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
 190                 }
 191                 else
 192                 {
 193                         $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
 194                 }
 195         }
 196         if ($bModifyXML)
 197         {
 198                 // derive change from normal osm file with osmosis
 199                 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 200                 if (isset($aResult['import-file']) && $aResult['import-file'])
 201                 {
 202                         $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 203                         echo $sCMD."\n";
 204                         exec($sCMD, $sJunk, $iErrorLevel);
 205                         if ($iErrorLevel)
 206                         {
 207                                 echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n";
 208                                 exit;
 209                         }
 210                 }
 211                 else
 212                 {
 213                         $aSpec = array(
 214                                 0 => array("pipe", "r"),  // stdin
 215                                 1 => array("pipe", "w"),  // stdout
 216                                 2 => array("pipe", "w") // stderr
 217                         );
 218                         $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 219                         echo $sCMD."\n";
 220                         $hProc = proc_open($sCMD, $aSpec, $aPipes);
 221                         if (!is_resource($hProc))
 222                         {
 223                                 echo "Error converting osm to osc, osmosis failed\n";
 224                                 exit;
 225                         }
 226                         fwrite($aPipes[0], $sModifyXMLstr);
 227                         fclose($aPipes[0]);
 228                         $sOut = stream_get_contents($aPipes[1]);
 229                         if ($aResult['verbose']) echo $sOut;
 230                         fclose($aPipes[1]);
 231                         $sErrors = stream_get_contents($aPipes[2]);
 232                         if ($aResult['verbose']) echo $sErrors;
 233                         fclose($aPipes[2]);
 234                         if ($iError = proc_close($hProc))
 235                         {
 236                                 echo "Error converting osm to osc, osmosis returned: $iError\n";
 237                                 echo $sOut;
 238                                 echo $sErrors;
 239                                 exit;
 240                         }
 241                 }
 242
 243                 // import generated change file
 244                 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
 245                 echo $sCMD."\n";
 246                 exec($sCMD, $sJunk, $iErrorLevel);
 247                 if ($iErrorLevel)
 248                 {
 249                         echo "osm2pgsql exited with error level $iErrorLevel\n";
 250                         exit;
 251                 }
 252         }
 253
 254         if ($aResult['deduplicate'])
 255         {
 256                 $oDB =& getDB();
 257                 $sSQL = 'select partition from country_name order by country_code';
 258                 $aPartitions = $oDB->getCol($sSQL);
 259                 if (PEAR::isError($aPartitions))
 260                 {
 261                         fail($aPartitions->getMessage());
 262                 }
 263                 $aPartitions[] = 0;
 264
 265                 $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";
 266                 $aDuplicateTokens = $oDB->getAll($sSQL);
 267                 foreach($aDuplicateTokens as $aToken)
 268                 {
 269                         if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
 270                         echo "Deduping ".$aToken['word_token']."\n";
 271                         $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
 272                         $aTokenSet = $oDB->getAll($sSQL);
 273                         if (PEAR::isError($aTokenSet))
 274                         {
 275                                 var_dump($aTokenSet, $sSQL);
 276                                 exit;
 277                         }
 278
 279                         $aKeep = array_shift($aTokenSet);
 280                         $iKeepID = $aKeep['word_id'];
 281
 282                         foreach($aTokenSet as $aRemove)
 283                         {
 284                                 $sSQL = "update search_name set";
 285                                 $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
 286                                 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 287                                 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 288                                 $x = $oDB->query($sSQL);
 289                                 if (PEAR::isError($x))
 290                                 {
 291                                         var_dump($x);
 292                                         exit;
 293                                 }
 294
 295                                 $sSQL = "update search_name set";
 296                                 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 297                                 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
 298                                 $x = $oDB->query($sSQL);
 299                                 if (PEAR::isError($x))
 300                                 {
 301                                         var_dump($x);
 302                                         exit;
 303                                 }
 304
 305                                 $sSQL = "update location_area_country set";
 306                                 $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
 307                                 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 308                                 $x = $oDB->query($sSQL);
 309                                 if (PEAR::isError($x))
 310                                 {
 311                                         var_dump($x);
 312                                         exit;
 313                                 }
 314
 315                                 foreach ($aPartitions as $sPartition)
 316                                 {
 317                                         $sSQL = "update search_name_".$sPartition." set";
 318                                         $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
 319                                         $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 320                                         $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 321                                         $x = $oDB->query($sSQL);
 322                                         if (PEAR::isError($x))
 323                                         {
 324                                                 var_dump($x);
 325                                                 exit;
 326                                         }
 327
 328                                         $sSQL = "update search_name_".$sPartition." set";
 329                                         $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 330                                         $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
 331                                         $x = $oDB->query($sSQL);
 332                                         if (PEAR::isError($x))
 333                                         {
 334                                                 var_dump($x);
 335                                                 exit;
 336                                         }
 337
 338                                         $sSQL = "update location_area_country set";
 339                                         $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
 340                                         $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 341                                         $x = $oDB->query($sSQL);
 342                                         if (PEAR::isError($x))
 343                                         {
 344                                                 var_dump($x);
 345                                                 exit;
 346                                         }
 347                                 }
 348
 349                                 $sSQL = "delete from word where word_id = ".$aRemove['word_id'];
 350                                 $x = $oDB->query($sSQL);
 351                                 if (PEAR::isError($x))
 352                                 {
 353                                         var_dump($x);
 354                                         exit;
 355                                 }
 356                         }
 357
 358                 }
 359         }
 360
 361         if ($aResult['index'])
 362         {
 363                 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
 364                 passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
 365         }
 366
 367         if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
 368         {
 369
 370                 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
 371                         echo "Error: Update interval too low for download.geofabrik.de.  Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n";
 372                         exit;
 373                 }
 374
 375                 $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
 376                 $sOsmosisCMD = CONST_Osmosis_Binary;
 377                 $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
 378                 $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
 379                 $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
 380                 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
 381                 $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
 382                 if (!$aResult['no-npi']) {
 383                         $sCMDIndex .= '-F ';
 384                 }
 385                 while(true)
 386                 {
 387                         $fStartTime = time();
 388                         $iFileSize = 1001;
 389
 390                         // Logic behind this is that osm2pgsql locks the database quite a bit
 391                         // So it is better to import lots of small files
 392                         // But indexing works most efficiently on large amounts of data
 393                         // So do lots of small imports and a BIG index
 394
 395 //                      while($aResult['import-osmosis-all'] && $iFileSize > 1000)
 396 //                      {
 397                                 if (!file_exists($sImportFile))
 398                                 {
 399                                         // First check if there are new updates published (except for minutelies - there's always new diffs to process)
 400                                         if ( CONST_Replication_Update_Interval > 60 )
 401                                         {
 402
 403                                                 unset($aReplicationLag);
 404                                                 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
 405                                                 while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
 406                                                 {
 407                                                         if ($iErrorLevel)
 408                                                         {
 409                                                                 echo "Error: $iErrorLevel. ";
 410                                                                 echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
 411                                                         }
 412                                                         else
 413                                                         {
 414                                                                 echo ".";
 415                                                         }
 416                                                         sleep(CONST_Replication_Recheck_Interval);
 417                                                         unset($aReplicationLag);
 418                                                         exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
 419                                                 }
 420                                                 // There are new replication files - use osmosis to download the file
 421                                                 echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
 422                                         }
 423                                         $fStartTime = time();
 424                                         $fCMDStartTime = time();
 425                                         echo $sCMDDownload."\n";
 426                                         exec($sCMDDownload, $sJunk, $iErrorLevel);
 427                                         while ($iErrorLevel > 0)
 428                                         {
 429                                                 echo "Error: $iErrorLevel\n";
 430                                                 sleep(60);
 431                                                 echo 'Re-trying: '.$sCMDDownload."\n";
 432                                                 exec($sCMDDownload, $sJunk, $iErrorLevel);
 433                                         }
 434                                         $iFileSize = filesize($sImportFile);
 435                                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 436                                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
 437                                         var_Dump($sSQL);
 438                                         $oDB->query($sSQL);
 439                                         echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 440                                 }
 441
 442                                 $iFileSize = filesize($sImportFile);
 443                                 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 444
 445                                 // Import the file
 446                                 $fCMDStartTime = time();
 447                                 echo $sCMDImport."\n";
 448                                 exec($sCMDImport, $sJunk, $iErrorLevel);
 449                                 if ($iErrorLevel)
 450                                 {
 451                                         echo "Error: $iErrorLevel\n";
 452                                         exit($iErrorLevel);
 453                                 }
 454                                 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
 455                                 var_Dump($sSQL);
 456                                 $oDB->query($sSQL);
 457                                 echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 458
 459                                 // Archive for debug?
 460                                 unlink($sImportFile);
 461 //                      }
 462
 463                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 464
 465                         // Index file
 466                         if (!isset($aResult['index-instances']))
 467                         {
 468                                 if (getLoadAverage() < 24)
 469                                         $iIndexInstances = 2;
 470                                 else
 471                                         $iIndexInstances = 1;
 472                         } else
 473                                 $iIndexInstances = $aResult['index-instances'];
 474
 475                         $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;
 476                         $fCMDStartTime = time();
 477
 478                         if (!$aResult['no-npi'])
 479                         {
 480                                 $iFileID = $oDB->getOne('select nextval(\'file\')');
 481                                 if (PEAR::isError($iFileID))
 482                                 {
 483                                         echo $iFileID->getMessage()."\n";
 484                                         exit(-1);
 485                                 }
 486                                 $sFileDir = CONST_BasePath.'/export/diff/';
 487                                 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 488                                 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 489
 490                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 491                                 $sThisIndexCmd .= $sFileDir;
 492                                 $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
 493                                 $sThisIndexCmd .= ".npi.out";
 494
 495                                 preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
 496                                 $sFileDir = CONST_BasePath.'/export/index/';
 497                                 $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
 498
 499                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 500                                 file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
 501                         }
 502
 503                         if (!$aResult['no-index'])
 504                         {
 505                                 echo "$sThisIndexCmd\n";
 506                                 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 507                                 if ($iErrorLevel)
 508                                 {
 509                                         echo "Error: $iErrorLevel\n";
 510                                         exit($iErrorLevel);
 511                                 }
 512
 513                                 if (!$aResult['no-npi'])
 514                                 {
 515                                         $sFileDir = CONST_BasePath.'/export/diff/';
 516                                         $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 517                                         $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 518
 519                                         $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
 520                                         echo "$sThisIndexCmd\n";
 521                                         exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 522                                         if ($iErrorLevel)
 523                                         {
 524                                                 echo "Error: $iErrorLevel\n";
 525                                                 exit($iErrorLevel);
 526                                         }
 527
 528                                         rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
 529                                                 $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
 530                                 }
 531                         }
 532
 533                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
 534                         var_Dump($sSQL);
 535                         $oDB->query($sSQL);
 536                         echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 537
 538                         $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
 539                         $oDB->query($sSQL);
 540
 541                         $fDuration = time() - $fStartTime;
 542                         echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n";
 543                         if (!$aResult['import-osmosis-all']) exit;
 544
 545                         if ( CONST_Replication_Update_Interval > 60 )
 546                         {
 547                                 $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
 548                         }
 549                         else
 550                         {
 551                                 $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
 552                         }
 553                         echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
 554                         sleep($iSleep);
 555                 }
 556
 557         }
 558
 559         if ($aResult['import-npi-all'])
 560         {
 561                 $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
 562                 if (PEAR::isError($iNPIID))
 563                 {
 564                         var_dump($iNPIID);
 565                         exit;
 566                 }
 567                 $sConfigDirectory = CONST_BasePath.'/settings';
 568                 $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
 569                 while(true)
 570                 {
 571                         $fStartTime = time();
 572
 573                         $iNPIID++;
 574
 575                         $sImportFile = CONST_BasePath.'/export/diff/';
 576                         $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
 577                         $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 578                         $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
 579                         $sImportFile .= ".npi";
 580                         while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
 581                         {
 582                                 echo "sleep (waiting for $sImportFile)\n";
 583                                 sleep(10);
 584                         }
 585                         if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
 586
 587                         $iFileSize = filesize($sImportFile);
 588
 589                         // Import the file
 590                         $fCMDStartTime = time();
 591                         $sCMDImport = $sCMDImportTemplate . $sImportFile;
 592                         echo $sCMDImport."\n";
 593                         exec($sCMDImport, $sJunk, $iErrorLevel);
 594                         if ($iErrorLevel)
 595                         {
 596                                 echo "Error: $iErrorLevel\n";
 597                                 exit;
 598                         }
 599                         $sBatchEnd = $iNPIID;
 600                         echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 601                         $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
 602                         var_Dump($sSQL);
 603                         $oDB->query($sSQL);
 604                 }
 605
 606         }
 607
 608         function getosmosistimestamp($sOsmosisConfigDirectory)
 609         {
 610                 $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
 611                 preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
 612                 return str_replace('\:',':',$aResult[1]);
 613         }