X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/a46c739ffb1801c83a86b4776c70a3e9cb0440c6..f1cbca788531e9eefe5fbbadd7ac0d218f6aa076:/utils/setup.php?ds=sidebyside diff --git a/utils/setup.php b/utils/setup.php index 55e0038c..734f669f 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -22,10 +22,12 @@ 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'), array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'), + array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"), 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)'), @@ -124,7 +126,17 @@ pgsqlRunScript('CREATE EXTENSION hstore'); } - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); + $fPostgisVersion = (float) CONST_Postgis_Version; + if ($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]) @@ -133,14 +145,33 @@ exit; } - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); 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'); + + if ($aCMDResult['no-partitions']) + { + pgsqlRunScript('update country_name set partition = 0'); + } + + // the following will be needed by create_functions later but + // is only defined in the subsequently called create_tables. + // Create dummies here that will be overwritten by the proper + // versions in create-tables. + pgsqlRunScript('CREATE TABLE place_boundingbox ()'); + pgsqlRunScript('create type wikipedia_article_match as ()'); } if ($aCMDResult['import-data'] || $aCMDResult['all']) @@ -161,7 +192,7 @@ } $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 .= ' -C 18000'; $osm2pgsql .= ' -P '.$aDSNInfo['port']; $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; passthruCheckReturn($osm2pgsql); @@ -183,6 +214,7 @@ $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); } @@ -233,13 +265,13 @@ echo "Partition Tables\n"; $bDidSomething = true; $oDB =& getDB(); - $sSQL = 'select partition from country_name order by country_code'; + $sSQL = 'select distinct partition from country_name'; $aPartitions = $oDB->getCol($sSQL); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); } - $aPartitions[] = 0; + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -262,13 +294,13 @@ echo "Partition Functions\n"; $bDidSomething = true; $oDB =& getDB(); - $sSQL = 'select partition from country_name order by country_code'; + $sSQL = 'select distinct partition from country_name'; $aPartitions = $oDB->getCol($sSQL); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); } - $aPartitions[] = 0; + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -338,13 +370,13 @@ if (!pg_query($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'; + $sSQL = 'select distinct partition from country_name'; $aPartitions = $oDB->getCol($sSQL); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); } - $aPartitions[] = 0; + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; foreach($aPartitions as $sPartition) { if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection)); @@ -503,75 +535,80 @@ if (!file_exists(CONST_Osmosis_Binary)) { echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; - fail("osmosis not found in '".CONST_Osmosis_Binary."'"); - } - if (file_exists(CONST_BasePath.'/settings/configuration.txt')) - { - echo "settings/configuration.txt already exists\n"; + if (!$aCMDResult['all']) + { + fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + } } else { - passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/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'); - } - - // Find the last node in the DB - $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes"); - - // 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"; - $sLastNodeXML = file_get_contents($sLastNodeURL); - preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate); - $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60); - + 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'); + // 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'); + } - // 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"); - // 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 $aRepMatch) - { - if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; - $aPrevRepMatch = $aRepMatch; - } - if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; + // Find the last node in the DB + $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes"); + + // 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"; + $sLastNodeXML = file_get_contents($sLastNodeURL); + preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate); + $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60); + + // 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"); + // 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 $aRepMatch) + { + if (strtotime($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"); + 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"); - 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 $aRepMatch) - { - if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; - $aPrevRepMatch = $aRepMatch; + $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 $aRepMatch) + { + if (strtotime($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, $sSQL); } - if ($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, $sSQL); } if ($aCMDResult['index'] || $aCMDResult['all']) @@ -592,13 +629,13 @@ echo "Search indices\n"; $bDidSomething = true; $oDB =& getDB(); - $sSQL = 'select partition from country_name order by country_code'; + $sSQL = 'select distinct partition from country_name'; $aPartitions = $oDB->getCol($sSQL); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); } - $aPartitions[] = 0; + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql'); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -657,10 +694,29 @@ // 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') ); @@ -668,7 +724,6 @@ $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); if (!is_resource($hProcess)) fail('unable to start pgsql'); - fclose($ahPipes[0]); // TODO: error checking while(!feof($ahPipes[1])) @@ -677,15 +732,28 @@ } 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) { + 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']) + $sCMD .= ' -v ON_ERROR_STOP=1'; $aDescriptors = array( 0 => array('pipe', 'r'), 1 => STDOUT, @@ -698,10 +766,15 @@ while(strlen($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 ($iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } } function pgsqlRunRestoreData($sDumpFile) @@ -729,7 +802,11 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } } function pgsqlRunDropAndRestore($sDumpFile) @@ -757,7 +834,11 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } } function passthruCheckReturn($cmd)