X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/090e9e7fb6823d00cb12a11f2bb57a1b7cc99611..b03be15442502aa2e52bacff70fb983cf4397d70:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index 9e946035..caaf0f91 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); @@ -108,40 +110,38 @@ $oDB =& getDB(); - $sVersionString = $oDB->getOne('select version()'); - preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[.]([0-9]+) #', $sVersionString, $aMatches); - if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2]) + $fPostgresVersion = getPostgresVersion($oDB); + echo 'Postgres version found: '.$fPostgresVersion."\n"; + + if ($fPostgresVersion < 9.1) { - echo "ERROR: PostgreSQL version is not correct. Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; - exit; + fail("Minimum supported version of Postgresql is 9.1."); } - 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'); - } + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis'); - $fPostgisVersion = (float) CONST_Postgis_Version; - if ($fPostgisVersion < 2.0) { - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); - } else { - pgsqlRunScript('CREATE EXTENSION postgis'); + // For extratags and namedetails the hstore_to_json converter is + // needed which is only available from Postgresql 9.3+. For older + // versions add a dummy function that returns nothing. + $iNumFunc = $oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'"); + if (PEAR::isError($iNumFunc)) + { + fail("Cannot query stored procedures.", $iNumFunc); } - if ($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'); + if ($iNumFunc == 0) + { + pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable"); + echo "WARNING: Postgresql is too old. extratags and namedetails API not available."; } - $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]) + + $fPostgisVersion = getPostgisVersion($oDB); + echo 'Postgis version found: '.$fPostgisVersion."\n"; + + if ($fPostgisVersion < 2.1) { - echo "ERROR: PostGIS version is not correct. Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; - exit; + // 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'); } pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); @@ -156,9 +156,10 @@ { 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 (CONST_Use_Extra_US_Postcodes) + { + pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); + } if ($aCMDResult['no-partitions']) { @@ -189,7 +190,15 @@ { $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']; @@ -207,55 +216,34 @@ { echo "Functions\n"; $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', $sTemplate); - if ($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 (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built"); + create_sql_functions($aCMDResult); } - if ($aCMDResult['create-minimal-tables']) + if ($aCMDResult['create-tables'] || $aCMDResult['all']) { - echo "Minimal Tables\n"; $bDidSomething = true; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tables-minimal.sql'); - - $sScript = ''; - - // Backstop the import process - easliest possible import id - $sScript .= "insert into import_npi_log values (18022);\n"; - $hFile = @fopen(CONST_BasePath.'/settings/partitionedtags.def', "r"); - if (!$hFile) fail('unable to open list of partitions: '.CONST_BasePath.'/settings/partitionedtags.def'); - - while (($sLine = fgets($hFile, 4096)) !== false && $sLine && substr($sLine,0,1) !='#') - { - list($sClass, $sType) = explode(' ', trim($sLine)); - $sScript .= "create table place_classtype_".$sClass."_".$sType." as "; - $sScript .= "select place_id as place_id,geometry as centroid from placex limit 0;\n"; - - $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_centroid "; - $sScript .= "ON place_classtype_".$sClass."_".$sType." USING GIST (centroid);\n"; - - $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id "; - $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n"; - } - fclose($hFile); - pgsqlRunScript($sScript); - } - - if ($aCMDResult['create-tables'] || $aCMDResult['all']) - { echo "Tables\n"; - $bDidSomething = true; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql'); + $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 - $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); - $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate); - pgsqlRunScript($sTemplate); + echo "Functions\n"; + create_sql_functions($aCMDResult); } if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) @@ -272,6 +260,18 @@ if (!$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 $aMatch) { @@ -353,6 +353,8 @@ echo '.'; if (!pg_query($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)); @@ -394,65 +396,85 @@ 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) { - if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; + $bAnyBusy = false; + + for($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 + { + 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++) @@ -460,7 +482,7 @@ $aDBInstances[$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"); @@ -504,7 +526,13 @@ } 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']) @@ -519,13 +547,16 @@ $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; if (!pg_query($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 (CONST_Use_Extra_US_Postcodes) + { + $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(); @@ -540,20 +571,20 @@ } 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"; @@ -563,49 +594,59 @@ // 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/