X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6382b1c111a49b592c269dec424aacd83495a184..7879ad44cd1e36ae1235c3dcdfd9e477491813e4:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index 65c3d88a..4ed343df 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -1,7 +1,8 @@ #!/usr/bin/php -Cq getTotalMemoryMB()) + if (isset($aCMDResult['osm2pgsql-cache'])) + { + $iCacheMemory = $aCMDResult['osm2pgsql-cache']; + } + else { $iCacheMemory = getCacheMemoryMB(); - echo "WARNING: resetting cache memory to $iCacheMemory\n"; } $aDSNInfo = DB::parseDSN(CONST_Database_DSN); @@ -132,7 +135,7 @@ pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); } else { - pgsqlRunScript('CREATE EXTENSION postgis'); + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis'); } if ($fPostgisVersion < 2.1) { // Function was renamed in 2.1 and throws an annoying deprecation warning @@ -191,7 +194,15 @@ { $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; } - $osm2pgsql .= ' -lsc -O gazetteer --hstore'; + if (CONST_Tablespace_Osm2pgsql_Data) + $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data; + if (CONST_Tablespace_Osm2pgsql_Index) + $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index; + if (CONST_Tablespace_Place_Data) + $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data; + if (CONST_Tablespace_Place_Index) + $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index; + $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1'; $osm2pgsql .= ' -C '.$iCacheMemory; $osm2pgsql .= ' -P '.$aDSNInfo['port']; $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; @@ -209,13 +220,14 @@ { echo "Functions\n"; $bDidSomething = true; - if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built"); + if (!file_exists(CONST_InstallPath.'/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_InstallPath.'/module', $sTemplate); if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate); if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate); if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate); pgsqlRunScript($sTemplate); + if ($fPostgisVersion < 2.0) { echo "Helper functions for postgis < 2.0\n"; $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql'); @@ -258,13 +270,30 @@ if ($aCMDResult['create-tables'] || $aCMDResult['all']) { - echo "Tables\n"; $bDidSomething = true; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql'); + + echo "Tables\n"; + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-data}', + CONST_Tablespace_Address_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-data}', + CONST_Tablespace_Search_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); // re-run the functions + echo "Functions\n"; $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); - $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate); + $sTemplate = str_replace('{modulepath}', + CONST_InstallPath.'/module', $sTemplate); pgsqlRunScript($sTemplate); } @@ -282,6 +311,18 @@ if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); + $sTemplate = replace_tablespace('{ts:address-data}', + CONST_Tablespace_Address_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-data}', + CONST_Tablespace_Search_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); foreach($aMatches as $aMatch) { @@ -363,6 +404,8 @@ echo '.'; if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection)); echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($oDB->connection)); + echo '.'; if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection)); echo '.'; if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection)); @@ -404,65 +447,85 @@ echo "Load Data\n"; $aDBInstances = array(); - for($i = 0; $i < $iInstances; $i++) - { - $aDBInstances[$i] =& getDB(true); + + $aQueriesPlacex = array(); + $aQueriesOsmline = array(); + // the query is divided into parcels, so that the work between the processes, i.e. the DBInstances, will be evenly distributed + $iNumberOfParcels = 100; + for($i = 0; $i < $iNumberOfParcels; $i++) + { $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, '; - $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i; - if ($aCMDResult['verbose']) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); - } - $bAnyBusy = true; - while($bAnyBusy) - { - $bAnyBusy = false; - for($i = 0; $i < $iInstances; $i++) - { - if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; - } - sleep(1); - echo '.'; - } - echo "\n"; - echo "Reanalysing database...\n"; - pgsqlRunScript('ANALYSE'); - } - - if ($aCMDResult['create-roads']) - { - $bDidSomething = true; - - $oDB =& getDB(); - $aDBInstances = array(); + $sSQL .= 'geometry) select * from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and not '; + $sSQL .= '(class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\');'; + array_push($aQueriesPlacex, $sSQL); + $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, '; + $sSQL .= 'geometry) from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and '; + $sSQL .= 'class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\''; + array_push($aQueriesOsmline, $sSQL); + } + for($i = 0; $i < $iInstances; $i++) { $aDBInstances[$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, 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)); } - $bAnyBusy = true; - while($bAnyBusy) + // now execute the query blocks, in the first round for placex, then for osmline, + // because insert_osmline depends on the placex table + echo 'Inserting from place to placex.'; + $aQueries = $aQueriesPlacex; + for($j = 0; $j < 2; $j++) { - $bAnyBusy = false; - for($i = 0; $i < $iInstances; $i++) + $bAnyBusy = true; + while($bAnyBusy) + { + $bAnyBusy = false; + + for($i = 0; $i < $iInstances; $i++) + { + if (pg_connection_busy($aDBInstances[$i]->connection)) + { + $bAnyBusy = true; + } + else if (count($aQueries) > 0) + { + $query = array_pop($aQueries); + if (!pg_send_query($aDBInstances[$i]->connection, $query)) + { + fail(pg_last_error($oDB->connection)); + } + else + { + pg_get_result($aDBInstances[$i]->connection); + $bAnyBusy = true; + } + } + } + sleep(1); + echo '.'; + } + echo "\n"; + if ($j == 0) //for the second round with osmline { - if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; + echo 'Inserting from place to osmline.'; + $aQueries = $aQueriesOsmline; } - sleep(1); - echo '.'; } - echo "\n"; + + echo "Reanalysing database...\n"; + pgsqlRunScript('ANALYSE'); } if ($aCMDResult['import-tiger-data']) { $bDidSomething = true; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql'); + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); $aDBInstances = array(); for($i = 0; $i < $iInstances; $i++) @@ -470,7 +533,7 @@ $aDBInstances[$i] =& getDB(true); } - foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile) + foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) { echo $sFile.': '; $hFile = fopen($sFile, "r"); @@ -514,7 +577,13 @@ } echo "Creating indexes\n"; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql'); + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); } if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) @@ -528,14 +597,13 @@ $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y "; $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); - $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_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'] || $aCMDResult['all']) + if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables { $bDidSomething = true; $oDB =& getDB(); @@ -550,20 +618,20 @@ } else { - if (file_exists(CONST_BasePath.'/settings/configuration.txt')) + if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) { echo "settings/configuration.txt already exists\n"; } else { - passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings'); + passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings'); // update osmosis configuration.txt with our settings - passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_BasePath.'/settings/configuration.txt'); - passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_BasePath.'/settings/configuration.txt'); + passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt'); + passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt'); } // Find the last node in the DB - $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes"); + $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'"); // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed) $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1"; @@ -577,7 +645,6 @@ // download.geofabrik.de: 000/26-Feb-2013 11:53 // planet.openstreetmap.org: 273/ 2013-03-11 07:41 - preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER); - var_dump($aRepMatches); if ($aRepMatches) { $aPrevRepMatch = false; @@ -591,7 +658,6 @@ $sRepURL .= $aRepMatch[1]; $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1"); preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER); - var_dump($aRepMatches); $aPrevRepMatch = false; foreach($aRepMatches as $aRepMatch) { @@ -603,7 +669,6 @@ $sRepURL .= $aRepMatch[1]; $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1"); preg_match_all('#([0-9]{3}).state.txt\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER); - var_dump($aRepMatches); $aPrevRepMatch = false; foreach($aRepMatches as $aRepMatch) { @@ -616,7 +681,7 @@ echo "Getting state file: $sRepURL\n"; $sStateFile = file_get_contents($sRepURL); if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file"); - file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile); + file_put_contents(CONST_InstallPath.'/settings/state.txt', $sStateFile); echo "Updating DB status\n"; pg_query($oDB->connection, 'TRUNCATE import_status'); $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')"; @@ -637,7 +702,7 @@ $bDidSomething = true; $sOutputFile = ''; if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output']; - $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile; + $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile; passthruCheckReturn($sBaseCmd.' -R 4'); if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE'); passthruCheckReturn($sBaseCmd.' -r 5 -R 25'); @@ -659,6 +724,12 @@ if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql'); + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); foreach($aMatches as $aMatch) { @@ -683,13 +754,14 @@ 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'); - @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/index.php'); - @symlink(CONST_BasePath.'/website/deletable.php', $sTargetDir.'/deletable.php'); - @symlink(CONST_BasePath.'/website/polygons.php', $sTargetDir.'/polygons.php'); - @symlink(CONST_BasePath.'/website/status.php', $sTargetDir.'/status.php'); + @symlink(CONST_InstallPath.'/website/details.php', $sTargetDir.'/details.php'); + @symlink(CONST_InstallPath.'/website/reverse.php', $sTargetDir.'/reverse.php'); + @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/search.php'); + @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/index.php'); + @symlink(CONST_InstallPath.'/website/lookup.php', $sTargetDir.'/lookup.php'); + @symlink(CONST_InstallPath.'/website/deletable.php', $sTargetDir.'/deletable.php'); + @symlink(CONST_InstallPath.'/website/polygons.php', $sTargetDir.'/polygons.php'); + @symlink(CONST_InstallPath.'/website/status.php', $sTargetDir.'/status.php'); @symlink(CONST_BasePath.'/website/images', $sTargetDir.'/images'); @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js'); @symlink(CONST_BasePath.'/website/css', $sTargetDir.'/css'); @@ -703,6 +775,69 @@ } } + if ($aCMDResult['drop']) + { + // The implementation is potentially a bit dangerous because it uses + // a positive selection of tables to keep, and deletes everything else. + // Including any tables that the unsuspecting user might have manually + // created. USE AT YOUR OWN PERIL. + $bDidSomething = true; + + // tables we want to keep. everything else goes. + $aKeepTables = array( + "*columns", + "import_polygon_*", + "import_status", + "place_addressline", + "location_property*", + "placex", + "search_name", + "seq_*", + "word", + "query_log", + "new_query_log", + "gb_postcode", + "spatial_ref_sys", + "country_name", + "place_classtype_*" + ); + + $oDB =& getDB(); + $aDropTables = array(); + $aHaveTables = $oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'"); + if (PEAR::isError($aHaveTables)) + { + fail($aPartitions->getMessage()); + } + foreach($aHaveTables as $sTable) + { + $bFound = false; + foreach ($aKeepTables as $sKeep) + { + if (fnmatch($sKeep, $sTable)) + { + $bFound = true; + break; + } + } + if (!$bFound) array_push($aDropTables, $sTable); + } + + foreach ($aDropTables as $sDrop) + { + if ($aCMDResult['verbose']) echo "dropping table $sDrop\n"; + @pg_query($oDB->connection, "DROP TABLE $sDrop CASCADE"); + // ignore warnings/errors as they might be caused by a table having + // been deleted already by CASCADE + } + + if (!is_null(CONST_Osm2pgsql_Flatnode_File)) + { + if ($aCMDResult['verbose']) echo "deleting ".CONST_Osm2pgsql_Flatnode_File."\n"; + unlink(CONST_Osm2pgsql_Flatnode_File); + } + } + if (!$bDidSomething) { showUsage($aCMDOptions, true); @@ -770,14 +905,14 @@ } - function pgsqlRunScript($sScript) + function pgsqlRunScript($sScript, $bfatal = true) { global $aCMDResult; // 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'].' -d '.$aDSNInfo['database']; - if (!$aCMDResult['ignore-errors']) + if ($bfatal && !$aCMDResult['ignore-errors']) $sCMD .= ' -v ON_ERROR_STOP=1'; $aDescriptors = array( 0 => array('pipe', 'r'), @@ -796,7 +931,7 @@ } fclose($ahPipes[0]); $iReturn = proc_close($hProcess); - if ($iReturn > 0) + if ($bfatal && $iReturn > 0) { fail("pgsql returned with error code ($iReturn)"); } @@ -864,3 +999,15 @@ passthru($cmd, $result); if ($result != 0) fail('Error executing external command: '.$cmd); } + + function replace_tablespace($sTemplate, $sTablespace, $sSql) + { + if ($sTablespace) + $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"', + $sSql); + else + $sSql = str_replace($sTemplate, '', $sSql); + + return $sSql; + } +