X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/ab6a99677293c9ddb96a717f03b8a009e98ab955..4e5b08053a5186715368d31bca769d5f02b15b0a:/utils/setup.php?ds=inline diff --git a/utils/setup.php b/utils/setup.php index 397eaa1b..80c139f7 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -1,7 +1,7 @@ #!/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) + $iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB()); + if ($iCacheMemory > getTotalMemoryMB()) + { + $iCacheMemory = getCacheMemoryMB(); + echo "WARNING: resetting cache memory to $iCacheMemory\n"; + } + + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($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. + + $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]) + { + 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'); + $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_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/us_statecounty.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); + } + + 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; + } + $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'; + $osm2pgsql .= ' -P '.$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_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:', '', $sTemplate); + pgsqlRunScript($sTemplate); + } + + if ($aCMDResult['create-minimal-tables']) + { + 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'); + + // re-run the functions + $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); + $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate); + pgsqlRunScript($sTemplate); + } + + if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { - $sSQL = 'select distinct country_code from country_name order by country_code'; + echo "Partition Tables\n"; + $bDidSomething = true; + $oDB =& getDB(); + $sSQL = 'select partition from country_name order by country_code'; $aPartitions = $oDB->getCol($sSQL); - $aPartitions[] = 'none'; + if (PEAR::isError($aPartitions)) + { + fail($aPartitions->getMessage()); + } + $aPartitions[] = 0; - $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql'); + $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) { @@ -31,7 +252,510 @@ } $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); } - echo $sTemplate; + + pgsqlRunScript($sTemplate); + } + + + if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) + { + echo "Partition Functions\n"; + $bDidSomething = true; + $oDB =& getDB(); + $sSQL = 'select partition from country_name order by country_code'; + $aPartitions = $oDB->getCol($sSQL); + if (PEAR::isError($aPartitions)) + { + fail($aPartitions->getMessage()); + } + $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($sWikiRedirectsFile); + echo "...done\n"; + } + else + { + echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n"; + } } - showUsage($aCMDOptions, true); + + 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 partition from country_name order by country_code'; + $aPartitions = $oDB->getCol($sSQL); + if (PEAR::isError($aPartitions)) + { + fail($aPartitions->getMessage()); + } + $aPartitions[] = 0; + foreach($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['create-roads']) + { + $bDidSomething = true; + + $oDB =& getDB(); + $aDBInstances = array(); + for($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) + { + $bAnyBusy = false; + for($i = 0; $i < $iInstances; $i++) + { + if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; + } + sleep(1); + echo '.'; + } + echo "\n"; + } + + if ($aCMDResult['import-tiger-data']) + { + $bDidSomething = true; + + pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql'); + + $aDBInstances = array(); + for($i = 0; $i < $iInstances; $i++) + { + $aDBInstances[$i] =& getDB(true); + } + + foreach(glob(CONST_BasePath.'/data/tiger2011/*.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)) $bAnyBusy = true; + } + usleep(10); + } + echo "\n"; + } + + echo "Creating indexes\n"; + pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql'); + } + + if ($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 "; + $sSQL .= "from placex where postcode is not null and calculated_country_code not in ('ie') group by calculated_country_code,postcode) as x"; + if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + + $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) "; + $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']) + { + $bDidSomething = true; + $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"; + fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + } + 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/