X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/410ce603f7e669bb07c14f37efbfff1ef118efae..ccaf7fa47fc7b02927ca4fe0f717e9e5b725ab9d:/utils/update.php

diff --git a/utils/update.php b/utils/update.php
index eadd3851..a561aaa3 100755
--- a/utils/update.php
+++ b/utils/update.php
@@ -26,6 +26,7 @@
 
 		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'),
@@ -46,8 +47,8 @@
 		showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
 	}
 
-	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)
@@ -67,9 +68,26 @@
 	// 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'];
@@ -101,7 +119,7 @@
 			$sNextFile = $aResult['import-diff'];
 			if (!file_exists($sNextFile))
 			{
-				echo "Cannot open $nextFile\n";
+				echo "Cannot open $sNextFile\n";
 				exit;
 			}
 			// Don't update the import status - we don't know what this file contains
@@ -112,7 +130,7 @@
 		if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
 		{
 			// Import the file
-			$sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;
+			$sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
 			echo $sCMD."\n";
 			exec($sCMD, $sJunk, $iErrorLevel);
 
@@ -147,7 +165,7 @@
 		}
 		else
 		{
-			$sModifyXMLstr = file_get_contents('http://overpass.osm.rambler.ru/cgi/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
+			$sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
 		}
 	}
 	if (isset($aResult['import-way']) && $aResult['import-way'])
@@ -159,7 +177,7 @@
 		}
 		else
 		{
-			$sCmd = 'http://overpass.osm.rambler.ru/cgi/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
+			$sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
 		}
 		$sModifyXMLstr = file_get_contents($sCmd);
 	}
@@ -172,7 +190,7 @@
 		}
 		else
 		{
-			$sModifyXMLstr = file_get_contents('http://overpass.osm.rambler.ru/cgi/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
+			$sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
 		}
 	}
 	if ($bModifyXML)
@@ -223,7 +241,7 @@
 		}
 
 		// import generated change file
-		$sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
+		$sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
 		echo $sCMD."\n";
 		exec($sCMD, $sJunk, $iErrorLevel);
 		if ($iErrorLevel)
@@ -235,6 +253,13 @@
 
 	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);
@@ -257,15 +282,15 @@
 				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))
@@ -275,7 +300,7 @@
 				}
 
 				$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))
@@ -285,7 +310,7 @@
 				}
 
 				$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($x))
@@ -297,8 +322,7 @@
 				foreach ($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))
@@ -307,18 +331,8 @@
 						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))
@@ -342,17 +356,25 @@
 
 	if ($aResult['index'])
 	{
-		passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
+		if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
+		passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$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_Osmosis_Binary;
 		$sOsmosisConfigDirectory = CONST_BasePath.'/settings';
 		$sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
-		$sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile;
-		$sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
+		$sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
+		$sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
+		$sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
 		if (!$aResult['no-npi']) {
 			$sCMDIndex .= '-F ';
 		}
@@ -370,11 +392,35 @@
 //			{
 				if (!file_exists($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);
@@ -383,9 +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);
@@ -398,12 +445,12 @@
 				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);
@@ -412,16 +459,25 @@
 			$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 
 			// Index file
-			$sThisIndexCmd = $sCMDIndex;
+			if (!isset($aResult['index-instances']))
+			{
+				if (getLoadAverage() < 24)
+					$iIndexInstances = 2;
+				else
+					$iIndexInstances = 1;
+			} else
+				$iIndexInstances = $aResult['index-instances'];
+
+			$sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;
+			$fCMDStartTime = time();
 
 			if (!$aResult['no-npi'])
 			{
-				$fCMDStartTime = time();
 				$iFileID = $oDB->getOne('select nextval(\'file\')');
 				if (PEAR::isError($iFileID))
 				{
 					echo $iFileID->getMessage()."\n";
-					exit;
+					exit(-1);
 				} 
 				$sFileDir = CONST_BasePath.'/export/diff/';
 				$sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
@@ -447,7 +503,7 @@
 				if ($iErrorLevel)
 				{
 					echo "Error: $iErrorLevel\n";
-					exit;
+					exit($iErrorLevel);
 				}
 
 				if (!$aResult['no-npi'])
@@ -462,7 +518,7 @@
 					if ($iErrorLevel)
 					{
 						echo "Error: $iErrorLevel\n";
-						exit;
+						exit($iErrorLevel);
 					}
 
 					rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
@@ -470,19 +526,28 @@
 				}
 			}
 
-			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')";
+			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);
 
 			$fDuration = time() - $fStartTime;
-			echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
+			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;
 
-			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);
 		}
 
 	}