X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/5f97080c738f59d58d9ec334b31be300bf6d04a8..f156da6b63e4a3e6a65ec49d2943b88255adb4c6:/utils/update.php diff --git a/utils/update.php b/utils/update.php index 7fc864c5..c5fc14a6 100755 --- a/utils/update.php +++ b/utils/update.php @@ -15,6 +15,8 @@ array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'), array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'), + array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'), + array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'), array('import-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'), @@ -24,10 +26,12 @@ array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'), array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'), + array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'), array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'), array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'), array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'), + array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'), array('index', '', 0, 1, 0, 0, 'bool', 'Index'), array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'), @@ -44,6 +48,8 @@ } if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1; + if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0; + /* // Lock to prevent multiple copies running if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1) @@ -63,8 +69,27 @@ // Assume osm2pgsql is in the folder above $sBasePath = dirname(dirname(__FILE__)); + date_default_timezone_set('Etc/UTC'); + $oDB =& getDB(); + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + + // cache memory to be used by osm2pgsql, should not be more than the available memory + $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000); + if ($iCacheMemory + 500 > getTotalMemoryMB()) + { + $iCacheMemory = getCacheMemoryMB(); + echo "WARNING: resetting cache memory to $iCacheMemory\n"; + } + $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port']; + if (!is_null(CONST_Osm2pgsql_Flatnode_File)) + { + $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; + } + + $bFirst = true; $bContinue = $aResult['import-all']; while ($bContinue || $bFirst) @@ -93,9 +118,9 @@ { // import diff directly (e.g. from osmosis --rri) $sNextFile = $aResult['import-diff']; - if (!file_exists($nextFile)) + if (!file_exists($sNextFile)) { - echo "Cannot open $nextFile\n"; + echo "Cannot open $sNextFile\n"; exit; } // Don't update the import status - we don't know what this file contains @@ -103,10 +128,10 @@ } // Missing file is not an error - it might not be created yet - if (($aResult['import-hourly'] || $aResult['import-daily']) && file_exists($sNextFile)) + if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile)) { // Import the file - $sCMD = $sBasePath.'/osm2pgsql/osm2pgsql -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sNextFile; + $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile; echo $sCMD."\n"; exec($sCMD, $sJunk, $iErrorLevel); @@ -125,67 +150,117 @@ } } - $sModifyXML = false; + $bModifyXML = false; + $sModifyXMLstr = ''; + $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api']; if (isset($aResult['import-file']) && $aResult['import-file']) { - $sModifyXML = file_get_contents($aResult['import-file']); + $bModifyXML = true; } if (isset($aResult['import-node']) && $aResult['import-node']) { - $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']); + $bModifyXML = true; + if ($bUseOSMApi) + { + $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']); + } + else + { + $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;'); + } } if (isset($aResult['import-way']) && $aResult['import-way']) { - $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full'); + $bModifyXML = true; + if ($bUseOSMApi) + { + $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full'; + } + else + { + $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;'; + } + $sModifyXMLstr = file_get_contents($sCmd); } if (isset($aResult['import-relation']) && $aResult['import-relation']) { - $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full'); + $bModifyXML = true; + if ($bUseOSMApi) + { + $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full'); + } + else + { + $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;'); + } } - if ($sModifyXML) + if ($bModifyXML) { - // Hack into a modify request - $sModifyXML = str_replace('', - '', $sModifyXML); - $sModifyXML = str_replace('', '', $sModifyXML); - - // Outputing this is too verbose - if ($aResult['verbose'] && false) var_dump($sModifyXML); - - $sDatabaseName = 'nominatim'; - $aSpec = array( - 0 => array("pipe", "r"), // stdin - 1 => array("pipe", "w"), // stdout - 2 => array("pipe", "w") // stderr - ); - $aPipes = array(); - $sCMD = $sBasePath.'/osm2pgsql/osm2pgsql -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' -'; - echo $sCMD."\n"; - $hProc = proc_open($sCMD, $aSpec, $aPipes); - if (!is_resource($hProc)) + // derive change from normal osm file with osmosis + $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc'; + if (isset($aResult['import-file']) && $aResult['import-file']) + { + $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile; + echo $sCMD."\n"; + exec($sCMD, $sJunk, $iErrorLevel); + if ($iErrorLevel) + { + echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n"; + exit; + } + } + else { - echo "$sBasePath/osm2pgsql failed\n"; - exit; + $aSpec = array( + 0 => array("pipe", "r"), // stdin + 1 => array("pipe", "w"), // stdout + 2 => array("pipe", "w") // stderr + ); + $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile; + echo $sCMD."\n"; + $hProc = proc_open($sCMD, $aSpec, $aPipes); + if (!is_resource($hProc)) + { + echo "Error converting osm to osc, osmosis failed\n"; + exit; + } + fwrite($aPipes[0], $sModifyXMLstr); + fclose($aPipes[0]); + $sOut = stream_get_contents($aPipes[1]); + if ($aResult['verbose']) echo $sOut; + fclose($aPipes[1]); + $sErrors = stream_get_contents($aPipes[2]); + if ($aResult['verbose']) echo $sErrors; + fclose($aPipes[2]); + if ($iError = proc_close($hProc)) + { + echo "Error converting osm to osc, osmosis returned: $iError\n"; + echo $sOut; + echo $sErrors; + exit; + } } - fwrite($aPipes[0], $sModifyXML); - fclose($aPipes[0]); - $sOut = stream_get_contents($aPipes[1]); - if ($aResult['verbose']) echo $sOut; - fclose($aPipes[1]); - $sErrors = stream_get_contents($aPipes[2]); - if ($aResult['verbose']) echo $sErrors; - fclose($aPipes[2]); - if ($iError = proc_close($hProc)) + + // import generated change file + $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile; + echo $sCMD."\n"; + exec($sCMD, $sJunk, $iErrorLevel); + if ($iErrorLevel) { - echo "osm2pgsql existed with error level $iError\n"; - echo $sOut; - echo $sErrors; + echo "osm2pgsql exited with error level $iErrorLevel\n"; exit; } } if ($aResult['deduplicate']) { + + $pgver = (float) CONST_Postgresql_Version; + if ($pgver < 9.3) { + echo "ERROR: deduplicate is only currently supported in postgresql 9.3"; + exit; + } + $oDB =& getDB(); $sSQL = 'select partition from country_name order by country_code'; $aPartitions = $oDB->getCol($sSQL); @@ -208,15 +283,15 @@ var_dump($aTokenSet, $sSQL); exit; } - + $aKeep = array_shift($aTokenSet); $iKeepID = $aKeep['word_id']; foreach($aTokenSet as $aRemove) { $sSQL = "update search_name set"; - $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.","; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),"; + $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) @@ -226,7 +301,7 @@ } $sSQL = "update search_name set"; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) @@ -236,7 +311,7 @@ } $sSQL = "update location_area_country set"; - $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) @@ -248,8 +323,7 @@ foreach ($aPartitions as $sPartition) { $sSQL = "update search_name_".$sPartition." set"; - $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.","; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) @@ -258,18 +332,8 @@ exit; } - $sSQL = "update search_name_".$sPartition." set"; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; - $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]"; - $x = $oDB->query($sSQL); - if (PEAR::isError($x)) - { - var_dump($x); - exit; - } - $sSQL = "update location_area_country set"; - $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) @@ -293,19 +357,27 @@ if ($aResult['index']) { - index($aResult, $sDatabaseDSN); + passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']); } if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) { + + if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) { + echo "Error: Update interval too low for download.geofabrik.de. Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n"; + exit; + } + $sImportFile = CONST_BasePath.'/data/osmosischange.osc'; - $sOsmosisCMD = CONST_BasePath.'/osmosis-0.38/bin/osmosis'; + $sOsmosisCMD = CONST_Osmosis_Binary; $sOsmosisConfigDirectory = CONST_BasePath.'/settings'; - $sDatabaseName = 'nominatim'; $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile; - $sCMDImport = $sBasePath.'/osm2pgsql/osm2pgsql -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sImportFile; - $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -t 15 -F '; -// $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -t 15 '; + $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory; + $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile; + $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances']; + if (!$aResult['no-npi']) { + $sCMDIndex .= '-F '; + } while(true) { $fStartTime = time(); @@ -320,11 +392,35 @@ // { if (!file_exists($sImportFile)) { - // Use osmosis to download the file + // First check if there are new updates published (except for minutelies - there's always new diffs to process) + if ( CONST_Replication_Update_Interval > 60 ) + { + + unset($aReplicationLag); + exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); + while ($iErrorLevel > 0 || $aReplicationLag[0] < 1) + { + if ($iErrorLevel) + { + echo "Error: $iErrorLevel. "; + echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n"; + } + else + { + echo "."; + } + sleep(CONST_Replication_Recheck_Interval); + unset($aReplicationLag); + exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); + } + // There are new replication files - use osmosis to download the file + echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n"; + } + $fStartTime = time(); $fCMDStartTime = time(); echo $sCMDDownload."\n"; exec($sCMDDownload, $sJunk, $iErrorLevel); - while ($iErrorLevel == 1) + while ($iErrorLevel > 0) { echo "Error: $iErrorLevel\n"; sleep(60); @@ -333,9 +429,10 @@ } $iFileSize = filesize($sImportFile); $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory); - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')"; + var_Dump($sSQL); $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; } $iFileSize = filesize($sImportFile); @@ -348,36 +445,39 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')"; var_Dump($sSQL); $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; // Archive for debug? unlink($sImportFile); +// } - $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory); + $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory); - // Index file - $fCMDStartTime = time(); + // Index file + $sThisIndexCmd = $sCMDIndex; + $fCMDStartTime = time(); + + if (!$aResult['no-npi']) + { $iFileID = $oDB->getOne('select nextval(\'file\')'); if (PEAR::isError($iFileID)) { - echo $oDB->getMessage()."\n"; - exit; + echo $iFileID->getMessage()."\n"; + exit(-1); } $sFileDir = CONST_BasePath.'/export/diff/'; $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT); $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT); if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true); - $sThisIndexCmd = $sCMDIndex; $sThisIndexCmd .= $sFileDir; $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT); $sThisIndexCmd .= ".npi.out"; - echo "$sThisIndexCmd\n"; preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch); $sFileDir = CONST_BasePath.'/export/index/'; @@ -385,46 +485,62 @@ if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true); file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND); + } + if (!$aResult['no-index']) + { + echo "$sThisIndexCmd\n"; exec($sThisIndexCmd, $sJunk, $iErrorLevel); if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } - $sFileDir = CONST_BasePath.'/export/diff/'; - $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT); - $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT); - - $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out"; - echo "$sThisIndexCmd\n"; - exec($sThisIndexCmd, $sJunk, $iErrorLevel); - if ($iErrorLevel) + if (!$aResult['no-npi']) { - echo "Error: $iErrorLevel\n"; - exit; - } + $sFileDir = CONST_BasePath.'/export/diff/'; + $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT); + $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT); + + $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out"; + echo "$sThisIndexCmd\n"; + exec($sThisIndexCmd, $sJunk, $iErrorLevel); + if ($iErrorLevel) + { + echo "Error: $iErrorLevel\n"; + exit($iErrorLevel); + } - rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2", - $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2"); + rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2", + $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2"); + } + } - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; - $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')"; - $oDB->query($sSQL); + $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')"; + var_Dump($sSQL); + $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; - $sSQL = "update import_status set lastimportdate = '$sBatchEnd'"; - $oDB->query($sSQL); + $sSQL = "update import_status set lastimportdate = '$sBatchEnd'"; + $oDB->query($sSQL); + $fDuration = time() - $fStartTime; + echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n"; + if (!$aResult['import-osmosis-all']) exit; - $fDuration = time() - $fStartTime; - echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n"; - if (!$aResult['import-osmosis-all']) exit; -// } - echo "Sleeping ".max(0,60-$fDuration)." seconds\n"; - sleep(max(0,60-$fDuration)); + if ( CONST_Replication_Update_Interval > 60 ) + { + $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time())); + } + else + { + $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration); + } + echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n"; + sleep($iSleep); } - + } if ($aResult['import-npi-all'])