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
  30                 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
  31                 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
  32                 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
  33
  34                 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
  35                 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
  36                 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
  37                 array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
  38
  39                 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
  40         );
  41         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
  42
  43         if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
  44         {
  45                 showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
  46         }
  47
  48         if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
  49 /*
  50         // Lock to prevent multiple copies running
  51         if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
  52         {
  53                 echo "Copy already running\n";
  54                 exit;
  55         }
  56         if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
  57         if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
  58         if (getBlockingProcesses() > $aResult['max-blocking'])
  59         {
  60                 echo "Too many blocking processes for import\n";
  61                 exit;
  62         }
  63 */
  64
  65         // Assume osm2pgsql is in the folder above
  66         $sBasePath = dirname(dirname(__FILE__));
  67
  68         $oDB =& getDB();
  69
  70         $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
  71
  72         $bFirst = true;
  73         $bContinue = $aResult['import-all'];
  74         while ($bContinue || $bFirst)
  75         {
  76                 $bFirst = false;
  77
  78                 if ($aResult['import-hourly'])
  79                 {
  80                         // Mirror the hourly diffs
  81                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
  82                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
  83                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
  84                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
  85                 }
  86
  87                 if ($aResult['import-daily'])
  88                 {
  89                         // Mirror the daily diffs
  90                         exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
  91                         $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
  92                         $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
  93                         $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
  94                 }
  95
  96                 if (isset($aResult['import-diff']))
  97                 {
  98                         // import diff directly (e.g. from osmosis --rri)
  99                         $sNextFile = $aResult['import-diff'];
 100                         if (!file_exists($sNextFile))
 101                         {
 102                                 echo "Cannot open $nextFile\n";
 103                                 exit;
 104                         }
 105                         // Don't update the import status - we don't know what this file contains
 106                         $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
 107                 }
 108
 109                 // Missing file is not an error - it might not be created yet
 110                 if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
 111                 {
 112                         // Import the file
 113                         $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;
 114                         echo $sCMD."\n";
 115                         exec($sCMD, $sJunk, $iErrorLevel);
 116
 117                         if ($iErrorLevel)
 118                         {
 119                                 echo "Error from osm2pgsql, $iErrorLevel\n";
 120                                 exit;
 121                         }
 122
 123                         // Move the date onwards
 124                         $oDB->query($sUpdateSQL);
 125                 }
 126                 else
 127                 {
 128                         $bContinue = false;
 129                 }
 130         }
 131
 132         $bModifyXML = false;
 133         $sModifyXMLstr = '';
 134         if (isset($aResult['import-file']) && $aResult['import-file'])
 135         {
 136                 $bModifyXML = true;
 137         }
 138         if (isset($aResult['import-node']) && $aResult['import-node'])
 139         {
 140                 $bModifyXML = true;
 141                 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
 142         }
 143         if (isset($aResult['import-way']) && $aResult['import-way'])
 144         {
 145                 $bModifyXML = true;
 146                 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full');
 147         }
 148         if (isset($aResult['import-relation']) && $aResult['import-relation'])
 149         {
 150                 $bModifyXML = true;
 151                 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
 152         }
 153         if ($bModifyXML)
 154         {
 155                 // derive change from normal osm file with osmosis
 156                 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 157                 if ($aResult['import-file'])
 158                 {
 159                         $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 160                         echo $sCMD."\n";
 161                         exec($sCMD, $sJunk, $iErrorLevel);
 162                         if ($iErrorLevel)
 163                         {
 164                                 echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n";
 165                                 exit;
 166                         }
 167                 }
 168                 else
 169                 {
 170                         $aSpec = array(
 171                                 0 => array("pipe", "r"),  // stdin
 172                                 1 => array("pipe", "w"),  // stdout
 173                                 2 => array("pipe", "w") // stderr
 174                         );
 175                         $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
 176                         echo $sCMD."\n";
 177                         $hProc = proc_open($sCMD, $aSpec, $aPipes);
 178                         if (!is_resource($hProc))
 179                         {
 180                                 echo "Error converting osm to osc, osmosis failed\n";
 181                                 exit;
 182                         }
 183                         fwrite($aPipes[0], $sModifyXMLstr);
 184                         fclose($aPipes[0]);
 185                         $sOut = stream_get_contents($aPipes[1]);
 186                         if ($aResult['verbose']) echo $sOut;
 187                         fclose($aPipes[1]);
 188                         $sErrors = stream_get_contents($aPipes[2]);
 189                         if ($aResult['verbose']) echo $sErrors;
 190                         fclose($aPipes[2]);
 191                         if ($iError = proc_close($hProc))
 192                         {
 193                                 echo "Error converting osm to osc, osmosis returned: $iError\n";
 194                                 echo $sOut;
 195                                 echo $sErrors;
 196                                 exit;
 197                         }
 198                 }
 199
 200                 // import generated change file
 201                 $aPipes = array();
 202                 $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
 203                 echo $sCMD."\n";
 204                 exec($sCMD, $sJunk, $iErrorLevel);
 205                 if ($iErrorLevel)
 206                 {
 207                         echo "osm2pgsql exited with error level $iErrorLevel\n";
 208                         exit;
 209                 }
 210         }
 211
 212         if ($aResult['deduplicate'])
 213         {
 214                 $oDB =& getDB();
 215                 $sSQL = 'select partition from country_name order by country_code';
 216                 $aPartitions = $oDB->getCol($sSQL);
 217                 if (PEAR::isError($aPartitions))
 218                 {
 219                         fail($aPartitions->getMessage());
 220                 }
 221                 $aPartitions[] = 0;
 222
 223                 $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";
 224                 $aDuplicateTokens = $oDB->getAll($sSQL);
 225                 foreach($aDuplicateTokens as $aToken)
 226                 {
 227                         if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
 228                         echo "Deduping ".$aToken['word_token']."\n";
 229                         $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
 230                         $aTokenSet = $oDB->getAll($sSQL);
 231                         if (PEAR::isError($aTokenSet))
 232                         {
 233                                 var_dump($aTokenSet, $sSQL);
 234                                 exit;
 235                         }
 236
 237                         $aKeep = array_shift($aTokenSet);
 238                         $iKeepID = $aKeep['word_id'];
 239
 240                         foreach($aTokenSet as $aRemove)
 241                         {
 242                                 $sSQL = "update search_name set";
 243                                 $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
 244                                 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 245                                 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 246                                 $x = $oDB->query($sSQL);
 247                                 if (PEAR::isError($x))
 248                                 {
 249                                         var_dump($x);
 250                                         exit;
 251                                 }
 252
 253                                 $sSQL = "update search_name set";
 254                                 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 255                                 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
 256                                 $x = $oDB->query($sSQL);
 257                                 if (PEAR::isError($x))
 258                                 {
 259                                         var_dump($x);
 260                                         exit;
 261                                 }
 262
 263                                 $sSQL = "update location_area_country set";
 264                                 $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
 265                                 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 266                                 $x = $oDB->query($sSQL);
 267                                 if (PEAR::isError($x))
 268                                 {
 269                                         var_dump($x);
 270                                         exit;
 271                                 }
 272
 273                                 foreach ($aPartitions as $sPartition)
 274                                 {
 275                                         $sSQL = "update search_name_".$sPartition." set";
 276                                         $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
 277                                         $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 278                                         $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
 279                                         $x = $oDB->query($sSQL);
 280                                         if (PEAR::isError($x))
 281                                         {
 282                                                 var_dump($x);
 283                                                 exit;
 284                                         }
 285
 286                                         $sSQL = "update search_name_".$sPartition." set";
 287                                         $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
 288                                         $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
 289                                         $x = $oDB->query($sSQL);
 290                                         if (PEAR::isError($x))
 291                                         {
 292                                                 var_dump($x);
 293                                                 exit;
 294                                         }
 295
 296                                         $sSQL = "update location_area_country set";
 297                                         $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
 298                                         $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
 299                                         $x = $oDB->query($sSQL);
 300                                         if (PEAR::isError($x))
 301                                         {
 302                                                 var_dump($x);
 303                                                 exit;
 304                                         }
 305                                 }
 306
 307                                 $sSQL = "delete from word where word_id = ".$aRemove['word_id'];
 308                                 $x = $oDB->query($sSQL);
 309                                 if (PEAR::isError($x))
 310                                 {
 311                                         var_dump($x);
 312                                         exit;
 313                                 }
 314                         }
 315
 316                 }
 317         }
 318
 319         if ($aResult['index'])
 320         {
 321                 index($aResult, $sDatabaseDSN);
 322         }
 323
 324         if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
 325         {
 326                 $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
 327                 $sOsmosisCMD = CONST_Osmosis_Binary;
 328                 $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
 329                 $sDatabaseName = 'nominatim';
 330                 $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
 331                 $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sImportFile;
 332                 $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -t '.$aResult['index-instances'];
 333                 if (!$aResult['no-npi']) {
 334                         $sCMDIndex .= '-F ';
 335                 }
 336                 while(true)
 337                 {
 338                         $fStartTime = time();
 339                         $iFileSize = 1001;
 340
 341                         // Logic behind this is that osm2pgsql locks the database quite a bit
 342                         // So it is better to import lots of small files
 343                         // But indexing works most efficiently on large amounts of data
 344                         // So do lots of small imports and a BIG index
 345
 346 //                      while($aResult['import-osmosis-all'] && $iFileSize > 1000)
 347 //                      {
 348                                 if (!file_exists($sImportFile))
 349                                 {
 350                                         // Use osmosis to download the file
 351                                         $fCMDStartTime = time();
 352                                         echo $sCMDDownload."\n";
 353                                         exec($sCMDDownload, $sJunk, $iErrorLevel);
 354                                         while ($iErrorLevel == 1)
 355                                         {
 356                                                 echo "Error: $iErrorLevel\n";
 357                                                 sleep(60);
 358                                                 echo 'Re-trying: '.$sCMDDownload."\n";
 359                                                 exec($sCMDDownload, $sJunk, $iErrorLevel);
 360                                         }
 361                                         $iFileSize = filesize($sImportFile);
 362                                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 363                                         echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 364                                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
 365                                         $oDB->query($sSQL);
 366                                 }
 367
 368                                 $iFileSize = filesize($sImportFile);
 369                                 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 370
 371                                 // Import the file
 372                                 $fCMDStartTime = time();
 373                                 echo $sCMDImport."\n";
 374                                 exec($sCMDImport, $sJunk, $iErrorLevel);
 375                                 if ($iErrorLevel)
 376                                 {
 377                                         echo "Error: $iErrorLevel\n";
 378                                         exit;
 379                                 }
 380                                 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 381                                 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
 382                                 var_Dump($sSQL);
 383                                 $oDB->query($sSQL);
 384
 385                                 // Archive for debug?
 386                                 unlink($sImportFile);
 387 //                      }
 388
 389                         $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 390
 391                         // Index file
 392                         $sThisIndexCmd = $sCMDIndex;
 393
 394                         if (!$aResult['no-npi'])
 395                         {
 396                                 $fCMDStartTime = time();
 397                                 $iFileID = $oDB->getOne('select nextval(\'file\')');
 398                                 if (PEAR::isError($iFileID))
 399                                 {
 400                                         echo $iFileID->getMessage()."\n";
 401                                         exit;
 402                                 }
 403                                 $sFileDir = CONST_BasePath.'/export/diff/';
 404                                 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 405                                 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 406
 407                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 408                                 $sThisIndexCmd .= $sFileDir;
 409                                 $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
 410                                 $sThisIndexCmd .= ".npi.out";
 411
 412                                 preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
 413                                 $sFileDir = CONST_BasePath.'/export/index/';
 414                                 $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
 415
 416                                 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
 417                                 file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
 418                         }
 419
 420                         if (!$aResult['no-index'])
 421                         {
 422                                 echo "$sThisIndexCmd\n";
 423                                 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 424                                 if ($iErrorLevel)
 425                                 {
 426                                         echo "Error: $iErrorLevel\n";
 427                                         exit;
 428                                 }
 429
 430                                 if (!$aResult['no-npi'])
 431                                 {
 432                                         $sFileDir = CONST_BasePath.'/export/diff/';
 433                                         $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
 434                                         $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 435
 436                                         $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
 437                                         echo "$sThisIndexCmd\n";
 438                                         exec($sThisIndexCmd, $sJunk, $iErrorLevel);
 439                                         if ($iErrorLevel)
 440                                         {
 441                                                 echo "Error: $iErrorLevel\n";
 442                                                 exit;
 443                                         }
 444
 445                                         rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
 446                                                 $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
 447                                 }
 448                         }
 449
 450                         echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 451                         $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
 452                         $oDB->query($sSQL);
 453
 454                         $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
 455                         $oDB->query($sSQL);
 456
 457                         $fDuration = time() - $fStartTime;
 458                         echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
 459                         if (!$aResult['import-osmosis-all']) exit;
 460
 461                         echo "Sleeping ".max(0,60-$fDuration)." seconds\n";
 462                         sleep(max(0,60-$fDuration));
 463                 }
 464
 465         }
 466
 467         if ($aResult['import-npi-all'])
 468         {
 469                 $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
 470                 if (PEAR::isError($iNPIID))
 471                 {
 472                         var_dump($iNPIID);
 473                         exit;
 474                 }
 475                 $sConfigDirectory = CONST_BasePath.'/settings';
 476                 $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
 477                 while(true)
 478                 {
 479                         $fStartTime = time();
 480
 481                         $iNPIID++;
 482
 483                         $sImportFile = CONST_BasePath.'/export/diff/';
 484                         $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
 485                         $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 486                         $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
 487                         $sImportFile .= ".npi";
 488                         while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
 489                         {
 490                                 echo "sleep (waiting for $sImportFile)\n";
 491                                 sleep(10);
 492                         }
 493                         if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
 494
 495                         $iFileSize = filesize($sImportFile);
 496
 497                         // Import the file
 498                         $fCMDStartTime = time();
 499                         $sCMDImport = $sCMDImportTemplate . $sImportFile;
 500                         echo $sCMDImport."\n";
 501                         exec($sCMDImport, $sJunk, $iErrorLevel);
 502                         if ($iErrorLevel)
 503                         {
 504                                 echo "Error: $iErrorLevel\n";
 505                                 exit;
 506                         }
 507                         $sBatchEnd = $iNPIID;
 508                         echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 509                         $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
 510                         var_Dump($sSQL);
 511                         $oDB->query($sSQL);
 512                 }
 513
 514         }
 515
 516         function getosmosistimestamp($sOsmosisConfigDirectory)
 517         {
 518                 $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
 519                 preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
 520                 return str_replace('\:',':',$aResult[1]);
 521         }