X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/c325e1b0398ad786ab46f98707a641fbcd67739a..71883fc3410127b97ae0524ca07a187e125d7801:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index 4de5953f..a1d8ae5d 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -22,19 +22,21 @@ 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-partitions', '', 0, 1, 0, 0, 'bool', 'Create required partition tables and triggers'), + 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)'), 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('create-roads', '', 0, 1, 0, 0, 'bool', ''), 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 opertions during index (EXPERT)'), + 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'), @@ -43,7 +45,27 @@ $bDidSomething = false; - // This is a pretty hard core defult - the number of processors in the box - 1 + // Check if osm-file is set and points to a valid file if --all or --import-data is given + if ($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) { @@ -64,16 +86,6 @@ echo "WARNING: resetting cache memory to $iCacheMemory\n"; } - if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date'])) - { - $sBaseFile = basename($aCMDResult['osm-file']); - if (preg_match('#^planet-([0-9]{2})([0-9]{2})([0-9]{2})[.]#', $sBaseFile, $aMatch)) - { - $iTime = mktime(0, 0, 0, $aMatch[2], $aMatch[3], '20'.$aMatch[1]); - $iTime -= (60*60*24); - $aCMDResult['osmosis-init-date'] = date('Y-m-d', $iTime).'T22:00:00Z'; - } - } $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; @@ -86,33 +98,80 @@ { fail('database already exists ('.CONST_Database_DSN.')'); } - passthru('createdb -E UTF-8 '.$aDSNInfo['database']); + passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$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']); + + $sVersionString = $oDB->getOne('select version()'); + 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"; + exit; + } + + passthru('createlang plpgsql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']); $pgver = (float) CONST_Postgresql_Version; if ($pgver < 9.1) { pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/sql/hstore_compatability_9_0.sql'); } else { pgsqlRunScript('CREATE EXTENSION hstore'); } - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.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]) + { + echo "ERROR: PostGIS version is not correct. Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; + exit; + } + 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'); - pgsqlRunScriptFile(CONST_BasePath.'/data/worldboundaries.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'] || $aCMDResult['all']) @@ -126,10 +185,16 @@ echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; fail("osm2pgsql not found in '$osm2pgsql'"); } + + if (!is_null(CONST_Osm2pgsql_Flatnode_File)) + { + $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; + } $osm2pgsql .= ' -lsc -O gazetteer --hstore'; $osm2pgsql .= ' -C '.$iCacheMemory; + $osm2pgsql .= ' -P '.$aDSNInfo['port']; $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; - passthru($osm2pgsql); + passthruCheckReturn($osm2pgsql); $oDB =& getDB(); $x = $oDB->getRow('select * from place limit 1'); @@ -148,6 +213,7 @@ $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); } @@ -193,20 +259,49 @@ pgsqlRunScript($sTemplate); } - if ($aCMDResult['create-partitions'] || $aCMDResult['all']) + if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) + { + echo "Partition Tables\n"; + $bDidSomething = true; + $oDB =& getDB(); + $sSQL = 'select distinct partition from country_name'; + $aPartitions = $oDB->getCol($sSQL); + if (PEAR::isError($aPartitions)) + { + fail($aPartitions->getMessage()); + } + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; + + $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); + preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); + foreach($aMatches as $aMatch) + { + $sResult = ''; + foreach($aPartitions as $sPartitionName) + { + $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]); + } + $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); + } + + pgsqlRunScript($sTemplate); + } + + + if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { - echo "Partitions\n"; + 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/partitions.src.sql'); + $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); foreach($aMatches as $aMatch) { @@ -251,7 +346,7 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) { - echo "Load Data\n"; + echo "Drop old Data\n"; $bDidSomething = true; $oDB =& getDB(); @@ -274,36 +369,37 @@ if (!pg_query($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)); 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']) { - if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct svals(name) as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection)); - echo '.'; - if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct postcode as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection)); - echo '.'; - if (!pg_query($oDB->connection, 'select count(getorcreate_housenumber_id(v)) from (select distinct housenumber as v from place where housenumber is not null) as w;')) fail(pg_last_error($oDB->connection)); - echo '.'; + echo "Loading word list\n"; + pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql'); } + echo "Load Data\n"; $aDBInstances = array(); for($i = 0; $i < $iInstances; $i++) { $aDBInstances[$i] =& getDB(true); $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; - $sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, '; + $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)); @@ -334,7 +430,7 @@ { $aDBInstances[$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)); @@ -357,6 +453,8 @@ { $bDidSomething = true; + pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql'); + $aDBInstances = array(); for($i = 0; $i < $iInstances; $i++) { @@ -405,6 +503,9 @@ } echo "\n"; } + + echo "Creating indexes\n"; + pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql'); } if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) @@ -412,43 +513,101 @@ $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,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'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date'])) + if ($aCMDResult['osmosis-init'] || $aCMDResult['all']) { $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'); - - $sDate = $aCMDResult['osmosis-init-date']; - $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, $sSQL); + if (!file_exists(CONST_Osmosis_Binary)) + { + echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + if (!$aCMDResult['all']) + { + fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + } + } + else + { + 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'); + } + + // 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/