X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/c186a4ead0b6fc903f2a6682315a85d676715bc7..fb95c35977f5bb772533c7977969e6b915ea0a92:/utils/setup.php

diff --git a/utils/setup.php b/utils/setup.php
index cf465b15..2542bd9e 100755
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -18,17 +18,23 @@
 		array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
 		array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
 		array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
+		array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
 		array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
+		array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
+		array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
 		array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
 		array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
 		array('create-partitions', '', 0, 1, 0, 0, 'bool', 'Create required partition tables and triggers'),
+		array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
 		array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
+		array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
 		array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
 		array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
 		array('create-roads', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
 		array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
 		array('osmosis-init-date', '', 0, 1, 1, 1, 'string', 'Generate default osmosis configuration'),
 		array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
+		array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
 		array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
 		array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
 		array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
@@ -37,7 +43,27 @@
 
 	$bDidSomething = false;
 
-	// This is a pretty hard core defult - the number of processors in the box - 1
+	// Check if osm-file is set and points to a valid file if --all or --import-data is given
+	if ($aCMDResult['import-data'] || $aCMDResult['all'])
+	{
+		if (!isset($aCMDResult['osm-file']))
+		{
+			fail('missing --osm-file for data import');
+		}
+
+		if (!file_exists($aCMDResult['osm-file']))
+		{
+			fail('the path supplied to --osm-file does not exist');
+		}
+
+		if (!is_readable($aCMDResult['osm-file']))
+		{
+			fail('osm-file "'.$aCMDResult['osm-file'].'" not readable');
+		}
+	}
+
+
+	// This is a pretty hard core default - the number of processors in the box - 1
 	$iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
 	if ($iInstances < 1)
 	{
@@ -49,6 +75,15 @@
 		$iInstances = getProcessorCount();
 		echo "WARNING: resetting threads to $iInstances\n";
 	}
+
+	// Assume we can steal all the cache memory in the box (unless told otherwise)
+	$iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB());
+	if ($iCacheMemory > getTotalMemoryMB())
+	{
+		$iCacheMemory = getCacheMemoryMB();
+		echo "WARNING: resetting cache memory to $iCacheMemory\n";
+	}
+
 	if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
 	{
 		$sBaseFile = basename($aCMDResult['osm-file']);
@@ -71,18 +106,18 @@
 		{
 			fail('database already exists ('.CONST_Database_DSN.')');
 		}
-		passthru('createdb '.$aDSNInfo['database']);
+		passthru('createdb -E UTF-8 '.$aDSNInfo['database']);
 	}
 
-	if ($aCMDResult['create-db'] || $aCMDResult['all'])
+	if ($aCMDResult['setup-db'] || $aCMDResult['all'])
 	{
-		echo "Create DB (2)\n";
+		echo "Setup DB\n";
 		$bDidSomething = true;
 		// TODO: path detection, detection memory, etc.
 
 		$oDB =& getDB();
 		passthru('createlang plpgsql '.$aDSNInfo['database']);
-        $pgver = (float) CONST_Postgresql_Version;
+		$pgver = (float) CONST_Postgresql_Version;
 		if ($pgver < 9.1) {
 			pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
 		} else {
@@ -106,12 +141,23 @@
 		$bDidSomething = true;
 
 		$osm2pgsql = CONST_Osm2pgsql_Binary;
-		if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql");
-		passthru($osm2pgsql.' -lsc -O gazetteer -C 12000 --hstore -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']);
+		if (!file_exists($osm2pgsql))
+		{
+			echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+			fail("osm2pgsql not found in '$osm2pgsql'");
+		}
+		$osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
+		$osm2pgsql .= ' -lsc -O gazetteer --hstore';
+		$osm2pgsql .= ' -C 16000';
+		$osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
+		passthruCheckReturn($osm2pgsql);
 
 		$oDB =& getDB();
 		$x = $oDB->getRow('select * from place limit 1');
-		if (!$x || PEAR::isError($x)) fail('No Data');
+		if (PEAR::isError($x)) {
+			fail($x->getMessage());
+		}
+		if (!$x) fail('No Data');
 	}
 
 	if ($aCMDResult['create-functions'] || $aCMDResult['all'])
@@ -120,7 +166,9 @@
 		$bDidSomething = true;
 		if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
 		$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-		$sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
+		$sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
+		if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
+		if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
 		pgsqlRunScript($sTemplate);
 	}
 
@@ -194,9 +242,37 @@
 		pgsqlRunScript($sTemplate);
 	}
 
+	if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all'])
+	{
+		$bDidSomething = true;
+		$sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin';
+		$sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin';
+		if (file_exists($sWikiArticlesFile))
+		{
+			echo "Importing wikipedia articles...";
+			pgsqlRunDropAndRestore($sWikiArticlesFile);
+			echo "...done\n";
+		}
+		else
+		{
+			echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+		}
+		if (file_exists($sWikiRedirectsFile))
+		{
+			echo "Importing wikipedia redirects...";
+			pgsqlRunDropAndRestore($sWikiRedirectsFile);
+			echo "...done\n";
+		}
+		else
+		{
+			echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+		}
+	}
+
+
 	if ($aCMDResult['load-data'] || $aCMDResult['all'])
 	{
-		echo "Load Data\n";
+		echo "Drop old Data\n";
 		$bDidSomething = true;
 
 		$oDB =& getDB();
@@ -219,6 +295,31 @@
 		if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
 		echo '.';
 
+		$sSQL = 'select partition from country_name order by country_code';
+		$aPartitions = $oDB->getCol($sSQL);
+		if (PEAR::isError($aPartitions))
+		{
+			fail($aPartitions->getMessage());
+		}
+		$aPartitions[] = 0;
+		foreach($aPartitions as $sPartition)
+		{
+			if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
+			echo '.';
+		}
+
+		// used by getorcreate_word_id to ignore frequent partial words
+		if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection));
+		echo ".\n";
+
+		// pre-create the word list
+		if (!$aCMDResult['disable-token-precalc'])
+		{
+			echo "Loading word list\n";
+			pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
+		}
+
+		echo "Load Data\n";
 		$aDBInstances = array();
 		for($i = 0; $i < $iInstances; $i++)
 		{
@@ -241,6 +342,7 @@
 			echo '.';
 		}
 		echo "\n";
+		pgsqlRunScript('ALTER TABLE place SET TABLESPACE "data"');
 		echo "Reanalysing database...\n";
 		pgsqlRunScript('ANALYSE');
 	}
@@ -255,7 +357,7 @@
 		{
 			$aDBInstances[$i] =& getDB(true);
 			if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
-			$sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, country_code, geometry) from ';
+			$sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, calculated_country_code, geometry) from ';
 			$sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
 			if ($aCMDResult['verbose']) echo "$sSQL\n";
 			if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
@@ -278,6 +380,8 @@
 	{
 		$bDidSomething = true;
 
+		pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+
 		$aDBInstances = array();
 		for($i = 0; $i < $iInstances; $i++)
 		{
@@ -313,7 +417,7 @@
 			}
 
 			fclose($hFile);
-	
+
 			$bAnyBusy = true;
 			while($bAnyBusy)
 			{
@@ -326,6 +430,9 @@
 			}
 			echo "\n";
 		}
+
+		echo "Creating indexes\n";
+		pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
 	}
 
 	if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
@@ -333,33 +440,48 @@
 		$bDidSomething = true;
 		$oDB =& getDB();
 		if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
-		$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
-		$sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,country_code,";
-		$sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,postcode,";
+		$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
+		$sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,";
+		$sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,";
 		$sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-		$sSQL .= "from placex where postcode is not null group by country_code,postcode) as x";
+		$sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
 		if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
-		$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+		$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calcuclated_country_code,geometry) ";
 		$sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
 		$sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
 		if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 	}
 
-	if ($aCMDResult['osmosis-init'] && isset($aCMDResult['osmosis-init-date']))
+	if (($aCMDResult['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
 	{
 		$bDidSomething = true;
+		$oDB =& getDB();
 
 		if (!file_exists(CONST_Osmosis_Binary)) fail("please download osmosis");
 		if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n";
-		else passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
+		else
+		{
+			passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
+			// server layout changed afer license change, fix path to minutely diffs
+			passthru("sed -i 's:minute-replicate:replication/minute:' ".CONST_BasePath.'/settings/configuration.txt');
+		}
 
 		$sDate = $aCMDResult['osmosis-init-date'];
-		$sURL = 'http://toolserver.org/~mazder/replicate-sequences/?'.$sDate;
+		$aDate = date_parse_from_format("Y-m-d\TH-i", $sDate);
+		$sURL = 'http://toolserver.org/~mazder/replicate-sequences/?';
+		$sURL .= 'Y='.$aDate['year'].'&m='.$aDate['month'].'&d='.$aDate['day'];
+		$sURL .= '&H='.$aDate['hour'].'&i='.$aDate['minute'].'&s=0';
+		$sURL .= '&stream=minute';
 		echo "Getting state file: $sURL\n";
 		$sStateFile = file_get_contents($sURL);
 		if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
 		file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
+		echo "Updating DB status\n";
+		pg_query($oDB->connection, 'TRUNCATE import_status');
+		$sSQL = "INSERT INTO import_status VALUES('".$sDate."')";
+		pg_query($oDB->connection, $sSQL);
+
 	}
 
 	if ($aCMDResult['index'] || $aCMDResult['all'])
@@ -368,11 +490,11 @@
 		$sOutputFile = '';
 		if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
 		$sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$iInstances.$sOutputFile;
-		passthru($sBaseCmd.' -R 4');
-		pgsqlRunScript('ANALYSE');
-		passthru($sBaseCmd.' -r 5 -R 25');
-		pgsqlRunScript('ANALYSE');
-		passthru($sBaseCmd.' -r 26');
+		passthruCheckReturn($sBaseCmd.' -R 4');
+		if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+		passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
+		if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+		passthruCheckReturn($sBaseCmd.' -r 26');
 	}
 
 	if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
@@ -407,7 +529,12 @@
 	{
 		$bDidSomething = true;
 		$sTargetDir = $aCMDResult['create-website'];
-		if (!is_dir($sTargetDir)) fail('please specify a directory to setup');
+		if (!is_dir($sTargetDir))
+		{
+			echo "You must create the website directory before calling this function.\n";
+			fail("Target directory does not exist.");
+		}
+
 		@symlink(CONST_BasePath.'/website/details.php', $sTargetDir.'/details.php');
 		@symlink(CONST_BasePath.'/website/reverse.php', $sTargetDir.'/reverse.php');
 		@symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/search.php');
@@ -426,9 +553,10 @@
 	{
 		if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
-		// Convert database DSN to psql paramaters
+		// Convert database DSN to psql parameters
 		$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
-		$sCMD = 'psql -f '.$sFilename.' '.$aDSNInfo['database'];
+		if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+		$sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -f '.$sFilename;
 
 		$aDescriptors = array(
 			0 => array('pipe', 'r'),
@@ -453,10 +581,10 @@
 
 	function pgsqlRunScript($sScript)
 	{
-		// Convert database DSN to psql paramaters
+		// Convert database DSN to psql parameters
 		$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 		if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
-		$sCMD = 'psql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database'];
+		$sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
 		$aDescriptors = array(
 			0 => array('pipe', 'r'),
 			1 => STDOUT, 
@@ -474,3 +602,66 @@
 		fclose($ahPipes[0]);
 		proc_close($hProcess);
 	}
+
+	function pgsqlRunRestoreData($sDumpFile)
+	{
+		// Convert database DSN to psql parameters
+		$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+		if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+		$sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc -a '.$sDumpFile;
+
+		$aDescriptors = array(
+			0 => array('pipe', 'r'),
+			1 => array('pipe', 'w'),
+			2 => array('file', '/dev/null', 'a')
+		);
+		$ahPipes = null;
+		$hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+		if (!is_resource($hProcess)) fail('unable to start pg_restore');
+
+		fclose($ahPipes[0]);
+
+		// TODO: error checking
+		while(!feof($ahPipes[1]))
+		{
+			echo fread($ahPipes[1], 4096);
+		}
+		fclose($ahPipes[1]);
+
+		proc_close($hProcess);
+	}
+
+	function pgsqlRunDropAndRestore($sDumpFile)
+	{
+		// Convert database DSN to psql parameters
+		$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+		if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+		$sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile;
+
+		$aDescriptors = array(
+			0 => array('pipe', 'r'),
+			1 => array('pipe', 'w'),
+			2 => array('file', '/dev/null', 'a')
+		);
+		$ahPipes = null;
+		$hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+		if (!is_resource($hProcess)) fail('unable to start pg_restore');
+
+		fclose($ahPipes[0]);
+
+		// TODO: error checking
+		while(!feof($ahPipes[1]))
+		{
+			echo fread($ahPipes[1], 4096);
+		}
+		fclose($ahPipes[1]);
+
+		proc_close($hProcess);
+	}
+
+	function passthruCheckReturn($cmd)
+	{
+		$result = -1;
+		passthru($cmd, $result);
+		if ($result != 0) fail('Error executing external command: '.$cmd);
+	}