X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/e591269fd8c357b922f2809d21f726a2469788b3..f1c4615c32532cdb03845c017f660fa5c5a06331:/utils/setup.php?ds=sidebyside diff --git a/utils/setup.php b/utils/setup.php index fb16254d..a31fbadf 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -22,6 +22,7 @@ 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('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'), 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-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'), @@ -88,6 +89,8 @@ $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $fPostgisVersion = (float) CONST_Postgis_Version; + if ($aCMDResult['create-db'] || $aCMDResult['all']) { echo "Create DB\n"; @@ -109,7 +112,7 @@ $oDB =& getDB(); $sVersionString = $oDB->getOne('select version()'); - preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[.]([0-9]+) #', $sVersionString, $aMatches); + preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches); if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2]) { echo "ERROR: PostgreSQL version is not correct. Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; @@ -125,13 +128,16 @@ pgsqlRunScript('CREATE EXTENSION hstore'); } - $fPostgisVersion = (float) CONST_Postgis_Version; if ($fPostgisVersion < 2.0) { pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); } else { pgsqlRunScript('CREATE EXTENSION postgis'); } + if ($fPostgisVersion < 2.1) { + // Function was renamed in 2.1 and throws an annoying deprecation warning + pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint'); + } $sVersionString = $oDB->getOne('select postgis_full_version()'); preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches); if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2]) @@ -143,7 +149,15 @@ pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql'); - pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql'); + if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) + { + pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz'); + } + else + { + echo "WARNING: external UK postcode table not found.\n"; + } pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); @@ -152,6 +166,13 @@ { pgsqlRunScript('update country_name set partition = 0'); } + + // the following will be needed by create_functions later but + // is only defined in the subsequently called create_tables. + // Create dummies here that will be overwritten by the proper + // versions in create-tables. + pgsqlRunScript('CREATE TABLE place_boundingbox ()'); + pgsqlRunScript('create type wikipedia_article_match as ()'); } if ($aCMDResult['import-data'] || $aCMDResult['all']) @@ -170,8 +191,16 @@ { $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; } + 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'; - $osm2pgsql .= ' -C '.$iCacheMemory; + $osm2pgsql .= ' -C 25000'; $osm2pgsql .= ' -P '.$aDSNInfo['port']; $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; passthruCheckReturn($osm2pgsql); @@ -193,6 +222,16 @@ $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); + 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'); + } else { + echo "Helper functions for postgis >= 2.0\n"; + $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_20_aux.sql'); + } pgsqlRunScript($sTemplate); } @@ -228,13 +267,30 @@ 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_BasePath.'/module', $sTemplate); pgsqlRunScript($sTemplate); } @@ -252,6 +308,18 @@ 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) { @@ -496,7 +564,7 @@ $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 calculated_country_code,postcode) as x"; + $sSQL .= "from placex where postcode is not null and calculated_country_code not in ('ie') group by calculated_country_code,postcode) as x"; if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) "; @@ -543,49 +611,59 @@ // Search for the correct state file - uses file timestamps so need to sort by date descending $sRepURL = CONST_Replication_Url."/"; - $sRep = file_get_contents($sRepURL."?C=M;O=D"); + $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1"); // download.geofabrik.de: 000/26-Feb-2013 11:53 - // planet.openstreetmap.org: 273/ 22-Mar-2013 07:41 - - preg_match_all('#([0-9]{3}/).*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER); - $aPrevRepMatch = false; - foreach($aRepMatches as $aRepMatch) + // 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, $aRepMatches, PREG_SET_ORDER); + if ($aRepMatches) { - if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; - $aPrevRepMatch = $aRepMatch; - } - if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; + $aPrevRepMatch = false; + foreach($aRepMatches as $aRepMatch) + { + if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; + $aPrevRepMatch = $aRepMatch; + } + if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; - $sRepURL .= $aRepMatch[1]; - $sRep = file_get_contents($sRepURL."?C=M;O=D"); - preg_match_all('#([0-9]{3}/).*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER); - $aPrevRepMatch = false; - foreach($aRepMatches as $aRepMatch) - { - if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; - $aPrevRepMatch = $aRepMatch; - } - if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; + $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); + $aPrevRepMatch = false; + foreach($aRepMatches as $aRepMatch) + { + if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; + $aPrevRepMatch = $aRepMatch; + } + if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; - $sRepURL .= $aRepMatch[1]; - $sRep = file_get_contents($sRepURL."?C=M;O=D"); - preg_match_all('#([0-9]{3}).state.txt.*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER); - $aPrevRepMatch = false; - foreach($aRepMatches as $aRepMatch) + $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); + $aPrevRepMatch = false; + foreach($aRepMatches as $aRepMatch) + { + if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; + $aPrevRepMatch = $aRepMatch; + } + if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; + + $sRepURL .= $aRepMatch[1].'.state.txt'; + 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', $sStateFile); + echo "Updating DB status\n"; + pg_query($oDB->connection, 'TRUNCATE import_status'); + $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')"; + pg_query($oDB->connection, $sSQL); + } + else { - if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; - $aPrevRepMatch = $aRepMatch; + if (!$aCMDResult['all']) + { + fail("Cannot read state file directory."); + } } - if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; - - $sRepURL .= $aRepMatch[1].'.state.txt'; - 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', $sStateFile); - echo "Updating DB status\n"; - pg_query($oDB->connection, 'TRUNCATE import_status'); - $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')"; - pg_query($oDB->connection, $sSQL); } } @@ -616,6 +694,12 @@ if (!$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 $aMatch) { @@ -664,6 +748,10 @@ { showUsage($aCMDOptions, true); } + else + { + echo "Setup finished.\n"; + } function pgsqlRunScriptFile($sFilename) { @@ -672,10 +760,29 @@ // 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 '.$aDSNInfo['database'].' -f '.$sFilename; + $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database']; + + $ahGzipPipes = null; + if (preg_match('/\\.gz$/', $sFilename)) + { + $aDescriptors = array( + 0 => array('pipe', 'r'), + 1 => array('pipe', 'w'), + 2 => array('file', '/dev/null', 'a') + ); + $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes); + if (!is_resource($hGzipProcess)) fail('unable to start zcat'); + $aReadPipe = $ahGzipPipes[1]; + fclose($ahGzipPipes[0]); + } + else + { + $sCMD .= ' -f '.$sFilename; + $aReadPipe = array('pipe', 'r'); + } $aDescriptors = array( - 0 => array('pipe', 'r'), + 0 => $aReadPipe, 1 => array('pipe', 'w'), 2 => array('file', '/dev/null', 'a') ); @@ -683,7 +790,6 @@ $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); if (!is_resource($hProcess)) fail('unable to start pgsql'); - fclose($ahPipes[0]); // TODO: error checking while(!feof($ahPipes[1])) @@ -692,15 +798,28 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } + if ($ahGzipPipes) + { + fclose($ahGzipPipes[1]); + proc_close($hGzipProcess); + } + } - function pgsqlRunScript($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 '.$aDSNInfo['database']; + if ($bfatal && !$aCMDResult['ignore-errors']) + $sCMD .= ' -v ON_ERROR_STOP=1'; $aDescriptors = array( 0 => array('pipe', 'r'), 1 => STDOUT, @@ -713,10 +832,15 @@ while(strlen($sScript)) { $written = fwrite($ahPipes[0], $sScript); + if ($written <= 0) break; $sScript = substr($sScript, $written); } fclose($ahPipes[0]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($bfatal && $iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } } function pgsqlRunRestoreData($sDumpFile) @@ -744,7 +868,7 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); } function pgsqlRunDropAndRestore($sDumpFile) @@ -772,7 +896,7 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); } function passthruCheckReturn($cmd) @@ -781,3 +905,15 @@ passthru($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; + } +