X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/e6d983474bc4f1720b2ef4460041b423b02fc172..be268f95e621ec767f1c4f26ca5dfe29a586d6b3:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index bc8f14f7..0026a813 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -1,7 +1,8 @@ #!/usr/bin/php -Cq getProcessorCount()) + { + $iInstances = getProcessorCount(); + echo "WARNING: resetting threads to $iInstances\n"; + } + + // Assume we can steal all the cache memory in the box (unless told otherwise) + if (isset($aCMDResult['osm2pgsql-cache'])) { + $iCacheMemory = $aCMDResult['osm2pgsql-cache']; + } + else + { + $iCacheMemory = getCacheMemoryMB(); + } + + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + + if ($aCMDResult['create-db'] || $aCMDResult['all']) + { + echo "Create DB\n"; + $bDidSomething = true; + $oDB =& DB::connect(CONST_Database_DSN, false); + if (!PEAR::isError($oDB)) + { + fail('database already exists ('.CONST_Database_DSN.')'); + } + passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']); + } + + if ($aCMDResult['setup-db'] || $aCMDResult['all']) + { + echo "Setup DB\n"; $bDidSomething = true; // TODO: path detection, detection memory, etc. -// passthru('createdb nominatim'); - passthru('createlang plpgsql nominatim'); - passthru('psql -f '.CONST_Path_Postgresql_Contrib.'/_int.sql nominatim'); - passthru('psql -f '.CONST_Path_Postgresql_Contrib.'/hstore.sql nominatim'); - passthru('psql -f '.CONST_Path_Postgresql_Postgis.'/postgis.sql nominatim'); - passthru('psql -f '.CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/country_name.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/country_osm_grid.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/gb_postcode.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/us_statecounty.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/us_state.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/data/worldboundaries.sql nominatim'); + + $oDB =& getDB(); + + $fPostgresVersion = getPostgresVersion($oDB); + echo 'Postgres version found: '.$fPostgresVersion."\n"; + + if ($fPostgresVersion < 9.1) + { + fail("Minimum supported version of Postgresql is 9.1."); + } + + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS 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.", $iNumFunc); + } + if ($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."; + } + + $fPostgisVersion = getPostgisVersion($oDB); + echo 'Postgis version found: '.$fPostgisVersion."\n"; + + if ($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'); + } + + 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_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"; + } + if (CONST_Use_Extra_US_Postcodes) + { + 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']) + { + echo "Import\n"; + $bDidSomething = true; + + $osm2pgsql = CONST_Osm2pgsql_Binary; + if (!file_exists($osm2pgsql)) + { + 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; + } + 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'].' '.$aCMDResult['osm-file']; + passthruCheckReturn($osm2pgsql); + + $oDB =& getDB(); + $x = $oDB->getRow('select * from place limit 1'); + if (PEAR::isError($x)) { + fail($x->getMessage()); + } + if (!$x) fail('No Data'); + } + + if ($aCMDResult['create-functions'] || $aCMDResult['all']) + { + echo "Functions\n"; + $bDidSomething = true; + 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_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 ($aCMDResult['create-tables'] || $aCMDResult['all']) + { + $bDidSomething = true; + + 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_InstallPath.'/module', $sTemplate); + pgsqlRunScript($sTemplate); + } + + 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'); + $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) + { + $sResult = ''; + foreach($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 "Partition Functions\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-functions.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['import-wikipedia-articles'] || $aCMDResult['all']) + { + $bDidSomething = true; + $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin'; + $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin'; + if (file_exists($sWikiArticlesFile)) + { + echo "Importing wikipedia articles..."; + pgsqlRunDropAndRestore($sWikiArticlesFile); + echo "...done\n"; + } + else + { + echo "WARNING: wikipedia article dump file not found - places will have default importance\n"; + } + if (file_exists($sWikiRedirectsFile)) + { + echo "Importing wikipedia redirects..."; + pgsqlRunDropAndRestore($sWikiRedirectsFile); + echo "...done\n"; + } + else + { + echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n"; + } + } + + + if ($aCMDResult['load-data'] || $aCMDResult['all']) + { + echo "Drop old Data\n"; + $bDidSomething = true; + + $oDB =& getDB(); + if (!pg_query($oDB->connection, 'TRUNCATE word')) fail(pg_last_error($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE placex')) 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($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($oDB->connection)); + echo '.'; + if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection)); + echo '.'; + + $sSQL = 'select distinct partition from country_name'; + $aPartitions = $oDB->getCol($sSQL); + if (PEAR::isError($aPartitions)) + { + fail($aPartitions->getMessage()); + } + 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']) + { + 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, 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['import-tiger-data']) + { + $bDidSomething = true; + + $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++) + { + $aDBInstances[$i] =& getDB(true); + } + + foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) + { + echo $sFile.': '; + $hFile = fopen($sFile, "r"); + $sSQL = fgets($hFile, 100000); + $iLines = 0; + + while(true) + { + for($i = 0; $i < $iInstances; $i++) + { + if (!pg_connection_busy($aDBInstances[$i]->connection)) + { + while(pg_get_result($aDBInstances[$i]->connection)); + $sSQL = fgets($hFile, 100000); + if (!$sSQL) break 2; + if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + $iLines++; + if ($iLines == 1000) + { + echo "."; + $iLines = 0; + } + } + } + usleep(10); + } + + fclose($hFile); + + $bAnyBusy = true; + while($bAnyBusy) + { + $bAnyBusy = false; + for($i = 0; $i < $iInstances; $i++) + { + if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; + } + usleep(10); + } + echo "\n"; + } + + echo "Creating indexes\n"; + $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']) + { + $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,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 "; + $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; + if (!pg_query($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 (isset($aCMDResult['load-data']) && $aCMDResult['load-data']) + if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables { $bDidSomething = true; - passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['load-data']); - passthru('psql -f '.CONST_BasePath.'/sql/functions.sql nominatim'); - passthru('psql -f '.CONST_BasePath.'/sql/tables.sql nominatim'); + $oDB =& getDB(); + + if (!file_exists(CONST_Osmosis_Binary)) + { + echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + if (!$aCMDResult['all']) + { + fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + } + } + else + { + if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) + { + echo "settings/configuration.txt already exists\n"; + } + else + { + 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_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(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"; + $sLastNodeXML = file_get_contents($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;F=1"); + // download.geofabrik.de: 000/