X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/9e6fc8f073a59f38225a6daf72be2d2fc8710fc8..433017b9907c8348a48ef1a9fa7669e3d9523f1b:/lib/setup/SetupClass.php diff --git a/lib/setup/SetupClass.php b/lib/setup/SetupClass.php index a26b7dae..fd2b83de 100755 --- a/lib/setup/SetupClass.php +++ b/lib/setup/SetupClass.php @@ -2,7 +2,8 @@ namespace Nominatim\Setup; -require_once(CONST_BasePath.'/lib/setup/AddressLevelParser.php'); +require_once(CONST_LibDir.'/setup/AddressLevelParser.php'); +require_once(CONST_LibDir.'/Shell.php'); class SetupFunctions { @@ -10,11 +11,13 @@ class SetupFunctions protected $iInstances; protected $sModulePath; protected $aDSNInfo; + protected $bQuiet; protected $bVerbose; protected $sIgnoreErrors; protected $bEnableDiffUpdates; protected $bEnableDebugStatements; protected $bNoPartitions; + protected $bDrop; protected $oDB = null; public function __construct(array $aCMDResult) @@ -29,23 +32,27 @@ class SetupFunctions warn('resetting threads to '.$this->iInstances); } - // Assume we can steal all the cache memory in the box (unless told otherwise) if (isset($aCMDResult['osm2pgsql-cache'])) { $this->iCacheMemory = $aCMDResult['osm2pgsql-cache']; + } elseif (getSetting('FLATNODE_FILE')) { + // When flatnode files are enabled then disable cache per default. + $this->iCacheMemory = 0; } else { + // Otherwise: Assume we can steal all the cache memory in the box. $this->iCacheMemory = getCacheMemoryMB(); } - $this->sModulePath = CONST_Database_Module_Path; + $this->sModulePath = getSetting('DATABASE_MODULE_PATH', CONST_InstallDir.'/module'); info('module path: ' . $this->sModulePath); // parse database string - $this->aDSNInfo = \Nominatim\DB::parseDSN(CONST_Database_DSN); + $this->aDSNInfo = \Nominatim\DB::parseDSN(getSetting('DATABASE_DSN')); if (!isset($this->aDSNInfo['port'])) { $this->aDSNInfo['port'] = 5432; } // setting member variables based on command line options stored in $aCMDResult + $this->bQuiet = isset($aCMDResult['quiet']) && $aCMDResult['quiet']; $this->bVerbose = $aCMDResult['verbose']; //setting default values which are not set by the update.php array @@ -69,6 +76,8 @@ class SetupFunctions } else { $this->bEnableDiffUpdates = false; } + + $this->bDrop = isset($aCMDResult['drop']) && $aCMDResult['drop']; } public function createDB() @@ -76,34 +85,34 @@ class SetupFunctions info('Create DB'); $oDB = new \Nominatim\DB; - if ($oDB->databaseExists()) { - fail('database already exists ('.CONST_Database_DSN.')'); + if ($oDB->checkConnection()) { + fail('database already exists ('.getSetting('DATABASE_DSN').')'); } - $sCreateDBCmd = 'createdb -E UTF-8 -p '.$this->aDSNInfo['port'].' '.$this->aDSNInfo['database']; + $oCmd = (new \Nominatim\Shell('createdb')) + ->addParams('-E', 'UTF-8') + ->addParams('-p', $this->aDSNInfo['port']); + if (isset($this->aDSNInfo['username'])) { - $sCreateDBCmd .= ' -U '.$this->aDSNInfo['username']; + $oCmd->addParams('-U', $this->aDSNInfo['username']); + } + if (isset($this->aDSNInfo['password'])) { + $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); } - if (isset($this->aDSNInfo['hostspec'])) { - $sCreateDBCmd .= ' -h '.$this->aDSNInfo['hostspec']; + $oCmd->addParams('-h', $this->aDSNInfo['hostspec']); } + $oCmd->addParams($this->aDSNInfo['database']); - $result = $this->runWithPgEnv($sCreateDBCmd); - if ($result != 0) fail('Error executing external command: '.$sCreateDBCmd); - } - - public function connect() - { - $this->oDB = new \Nominatim\DB(); - $this->oDB->connect(); + $result = $oCmd->run(); + if ($result != 0) fail('Error executing external command: '.$oCmd->escapedCmd()); } public function setupDB() { info('Setup DB'); - $fPostgresVersion = $this->oDB->getPostgresVersion(); + $fPostgresVersion = $this->db()->getPostgresVersion(); echo 'Postgres version found: '.$fPostgresVersion."\n"; if ($fPostgresVersion < 9.03) { @@ -113,7 +122,7 @@ class SetupFunctions $this->pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); $this->pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis'); - $fPostgisVersion = $this->oDB->getPostgisVersion(); + $fPostgisVersion = $this->db()->getPostgisVersion(); echo 'Postgis version found: '.$fPostgisVersion."\n"; if ($fPostgisVersion < 2.2) { @@ -121,34 +130,35 @@ class SetupFunctions exit(1); } - $i = $this->oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'"); + $sPgUser = getSetting('DATABASE_WEBUSER'); + $i = $this->db()->getOne("select count(*) from pg_user where usename = '$sPgUser'"); if ($i == 0) { - echo "\nERROR: Web user '".CONST_Database_Web_User."' does not exist. Create it with:\n"; - echo "\n createuser ".CONST_Database_Web_User."\n\n"; + echo "\nERROR: Web user '".$sPgUser."' does not exist. Create it with:\n"; + echo "\n createuser ".$sPgUser."\n\n"; exit(1); } // Try accessing the C module, so we know early if something is wrong - checkModulePresence(); // raises exception on failure + $this->checkModulePresence(); // raises exception on failure - if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) { + if (!file_exists(CONST_DataDir.'/data/country_osm_grid.sql.gz')) { echo 'Error: you need to download the country_osm_grid first:'; - echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz https://www.nominatim.org/data/country_grid.sql.gz\n"; + echo "\n wget -O ".CONST_DataDir."/data/country_osm_grid.sql.gz https://www.nominatim.org/data/country_grid.sql.gz\n"; exit(1); } - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); - $this->pgsqlRunScriptFile(CONST_ExtraDataPath.'/country_osm_grid.sql.gz'); - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql'); - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode_table.sql'); + $this->pgsqlRunScriptFile(CONST_DataDir.'/data/country_name.sql'); + $this->pgsqlRunScriptFile(CONST_DataDir.'/data/country_osm_grid.sql.gz'); + $this->pgsqlRunScriptFile(CONST_DataDir.'/data/gb_postcode_table.sql'); + $this->pgsqlRunScriptFile(CONST_DataDir.'/data/us_postcode_table.sql'); - $sPostcodeFilename = CONST_BasePath.'/data/gb_postcode_data.sql.gz'; + $sPostcodeFilename = CONST_DataDir.'/data/gb_postcode_data.sql.gz'; if (file_exists($sPostcodeFilename)) { $this->pgsqlRunScriptFile($sPostcodeFilename); } else { warn('optional external GB postcode table file ('.$sPostcodeFilename.') not found. Skipping.'); } - $sPostcodeFilename = CONST_BasePath.'/data/us_postcode_data.sql.gz'; + $sPostcodeFilename = CONST_DataDir.'/data/us_postcode_data.sql.gz'; if (file_exists($sPostcodeFilename)) { $this->pgsqlRunScriptFile($sPostcodeFilename); } else { @@ -158,56 +168,66 @@ class SetupFunctions if ($this->bNoPartitions) { $this->pgsqlRunScript('update country_name set partition = 0'); } - - // the following will be needed by createFunctions later but - // is only defined in the subsequently called createTables - // Create dummies here that will be overwritten by the proper - // versions in create-tables. - $this->pgsqlRunScript('CREATE TABLE IF NOT EXISTS place_boundingbox ()'); - $this->pgsqlRunScript('CREATE TYPE wikipedia_article_match AS ()', false); } public function importData($sOSMFile) { info('Import data'); - $osm2pgsql = CONST_Osm2pgsql_Binary; - if (!file_exists($osm2pgsql)) { - echo "Check CONST_Osm2pgsql_Binary in your local settings file.\n"; + if (!file_exists(getOsm2pgsqlBinary())) { + echo "Check NOMINATIM_OSM2PGSQL_BINARY in your local .env file.\n"; echo "Normally you should not need to set this manually.\n"; - fail("osm2pgsql not found in '$osm2pgsql'"); + fail("osm2pgsql not found in '".getOsm2pgsqlBinary()."'"); } - $osm2pgsql .= ' -S '.CONST_Import_Style; + $oCmd = new \Nominatim\Shell(getOsm2pgsqlBinary()); + $oCmd->addParams('--style', getImportStyle()); - if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { - $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; + if (getSetting('FLATNODE_FILE')) { + $oCmd->addParams('--flat-nodes', getSetting('FLATNODE_FILE')); + } + if (getSetting('TABLESPACE_OSM_DATA')) { + $oCmd->addParams('--tablespace-slim-data', getSetting('TABLESPACE_OSM_DATA')); + } + if (getSetting('TABLESPACE_OSM_INDEX')) { + $oCmd->addParams('--tablespace-slim-index', getSetting('TABLESPACE_OSM_INDEX')); + } + if (getSetting('TABLESPACE_PLACE_DATA')) { + $oCmd->addParams('--tablespace-main-data', getSetting('TABLESPACE_PLACE_DATA')); } + if (getSetting('TABLESPACE_PLACE_INDEX')) { + $oCmd->addParams('--tablespace-main-index', getSetting('TABLESPACE_PLACE_INDEX')); + } + $oCmd->addParams('--latlong', '--slim', '--create'); + $oCmd->addParams('--output', 'gazetteer'); + $oCmd->addParams('--hstore'); + $oCmd->addParams('--number-processes', 1); + $oCmd->addParams('--with-forward-dependencies', 'false'); + $oCmd->addParams('--log-progress', 'true'); + $oCmd->addParams('--cache', $this->iCacheMemory); + $oCmd->addParams('--port', $this->aDSNInfo['port']); - 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'; - $osm2pgsql .= ' -C '.$this->iCacheMemory; - $osm2pgsql .= ' -P '.$this->aDSNInfo['port']; if (isset($this->aDSNInfo['username'])) { - $osm2pgsql .= ' -U '.$this->aDSNInfo['username']; + $oCmd->addParams('--username', $this->aDSNInfo['username']); + } + if (isset($this->aDSNInfo['password'])) { + $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); } if (isset($this->aDSNInfo['hostspec'])) { - $osm2pgsql .= ' -H '.$this->aDSNInfo['hostspec']; + $oCmd->addParams('--host', $this->aDSNInfo['hostspec']); } - $osm2pgsql .= ' -d '.$this->aDSNInfo['database'].' '.$sOSMFile; - - $this->runWithPgEnv($osm2pgsql); + $oCmd->addParams('--database', $this->aDSNInfo['database']); + $oCmd->addParams($sOSMFile); + $oCmd->run(); - if (!$this->sIgnoreErrors && !$this->oDB->getRow('select * from place limit 1')) { + if (!$this->sIgnoreErrors && !$this->db()->getRow('select * from place limit 1')) { fail('No Data'); } + + if ($this->bDrop) { + $this->dropTable('planet_osm_nodes'); + $this->removeFlatnodeFile(); + } } public function createFunctions() @@ -215,7 +235,7 @@ class SetupFunctions info('Create Functions'); // Try accessing the C module, so we know early if something is wrong - checkModulePresence(); // raises exception on failure + $this->checkModulePresence(); // raises exception on failure $this->createSqlFunctions(); } @@ -224,89 +244,35 @@ class SetupFunctions { info('Create Tables'); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql'); - $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:address-data}', - CONST_Tablespace_Address_Data, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:address-index}', - CONST_Tablespace_Address_Index, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:search-data}', - CONST_Tablespace_Search_Data, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:search-index}', - CONST_Tablespace_Search_Index, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:aux-data}', - CONST_Tablespace_Aux_Data, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:aux-index}', - CONST_Tablespace_Aux_Index, - $sTemplate - ); + $sTemplate = file_get_contents(CONST_DataDir.'/sql/tables.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); $this->pgsqlRunScript($sTemplate, false); if ($bReverseOnly) { - $this->pgExec('DROP TABLE search_name'); + $this->dropTable('search_name'); } - $oAlParser = new AddressLevelParser(CONST_Address_Level_Config); - $oAlParser->createTable($this->oDB, 'address_levels'); + $oAlParser = new AddressLevelParser(getSettingConfig('ADDRESS_LEVEL_CONFIG', 'address-levels.json')); + $oAlParser->createTable($this->db(), 'address_levels'); } - public function createPartitionTables() + public function createTableTriggers() { - info('Create Partition Tables'); - - $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); - $sTemplate = $this->replaceTablespace( - '{ts:address-data}', - CONST_Tablespace_Address_Data, - $sTemplate - ); - - $sTemplate = $this->replaceTablespace( - '{ts:address-index}', - CONST_Tablespace_Address_Index, - $sTemplate - ); + info('Create Tables'); - $sTemplate = $this->replaceTablespace( - '{ts:search-data}', - CONST_Tablespace_Search_Data, - $sTemplate - ); + $sTemplate = file_get_contents(CONST_DataDir.'/sql/table-triggers.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:search-index}', - CONST_Tablespace_Search_Index, - $sTemplate - ); + $this->pgsqlRunScript($sTemplate, false); + } - $sTemplate = $this->replaceTablespace( - '{ts:aux-data}', - CONST_Tablespace_Aux_Data, - $sTemplate - ); + public function createPartitionTables() + { + info('Create Partition Tables'); - $sTemplate = $this->replaceTablespace( - '{ts:aux-index}', - CONST_Tablespace_Aux_Index, - $sTemplate - ); + $sTemplate = file_get_contents(CONST_DataDir.'/sql/partition-tables.src.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); $this->pgsqlRunPartitionScript($sTemplate); } @@ -315,25 +281,21 @@ class SetupFunctions { info('Create Partition Functions'); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); + $sTemplate = file_get_contents(CONST_DataDir.'/sql/partition-functions.src.sql'); $this->pgsqlRunPartitionScript($sTemplate); } public function importWikipediaArticles() { - $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin'; - $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin'; + $sWikiArticlePath = getSetting('WIKIPEDIA_DATA_PATH', CONST_DataDir.'/data'); + $sWikiArticlesFile = $sWikiArticlePath.'/wikimedia-importance.sql.gz'; if (file_exists($sWikiArticlesFile)) { - info('Importing wikipedia articles'); - $this->pgsqlRunDropAndRestore($sWikiArticlesFile); - } else { - warn('wikipedia article dump file not found - places will have default importance'); - } - if (file_exists($sWikiRedirectsFile)) { - info('Importing wikipedia redirects'); - $this->pgsqlRunDropAndRestore($sWikiRedirectsFile); + info('Importing wikipedia articles and redirects'); + $this->dropTable('wikipedia_article'); + $this->dropTable('wikipedia_redirect'); + $this->pgsqlRunScriptFile($sWikiArticlesFile); } else { - warn('wikipedia redirect dump file not found - some place importance values may be missing'); + warn('wikipedia importance dump file not found - places will have default importance'); } } @@ -341,48 +303,48 @@ class SetupFunctions { info('Drop old Data'); - $this->pgExec('TRUNCATE word'); - echo '.'; - $this->pgExec('TRUNCATE placex'); + $oDB = $this->db(); + + $oDB->exec('TRUNCATE word'); echo '.'; - $this->pgExec('TRUNCATE location_property_osmline'); + $oDB->exec('TRUNCATE placex'); echo '.'; - $this->pgExec('TRUNCATE place_addressline'); + $oDB->exec('TRUNCATE location_property_osmline'); echo '.'; - $this->pgExec('TRUNCATE place_boundingbox'); + $oDB->exec('TRUNCATE place_addressline'); echo '.'; - $this->pgExec('TRUNCATE location_area'); + $oDB->exec('TRUNCATE location_area'); echo '.'; if (!$this->dbReverseOnly()) { - $this->pgExec('TRUNCATE search_name'); + $oDB->exec('TRUNCATE search_name'); echo '.'; } - $this->pgExec('TRUNCATE search_name_blank'); + $oDB->exec('TRUNCATE search_name_blank'); echo '.'; - $this->pgExec('DROP SEQUENCE seq_place'); + $oDB->exec('DROP SEQUENCE seq_place'); echo '.'; - $this->pgExec('CREATE SEQUENCE seq_place start 100000'); + $oDB->exec('CREATE SEQUENCE seq_place start 100000'); echo '.'; $sSQL = 'select distinct partition from country_name'; - $aPartitions = $this->oDB->getCol($sSQL); + $aPartitions = $oDB->getCol($sSQL); if (!$this->bNoPartitions) $aPartitions[] = 0; foreach ($aPartitions as $sPartition) { - $this->pgExec('TRUNCATE location_road_'.$sPartition); + $oDB->exec('TRUNCATE location_road_'.$sPartition); echo '.'; } // used by getorcreate_word_id to ignore frequent partial words $sSQL = 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS '; - $sSQL .= '$$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE'; - $this->pgExec($sSQL); + $sSQL .= '$$ SELECT '.getSetting('MAX_WORD_FREQUENCY').' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE'; + $oDB->exec($sSQL); echo ".\n"; // pre-create the word list if (!$bDisableTokenPrecalc) { info('Loading word list'); - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql'); + $this->pgsqlRunScriptFile(CONST_DataDir.'/data/words.sql'); } info('Load Data'); @@ -392,7 +354,7 @@ class SetupFunctions $iLoadThreads = max(1, $this->iInstances - 1); for ($i = 0; $i < $iLoadThreads; $i++) { // https://secure.php.net/manual/en/function.pg-connect.php - $DSN = CONST_Database_DSN; + $DSN = getSetting('DATABASE_DSN'); $DSN = preg_replace('/^pgsql:/', '', $DSN); $DSN = preg_replace('/;/', ' ', $DSN); $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW); @@ -412,7 +374,7 @@ class SetupFunctions // last thread for interpolation lines // https://secure.php.net/manual/en/function.pg-connect.php - $DSN = CONST_Database_DSN; + $DSN = getSetting('DATABASE_DSN'); $DSN = preg_replace('/^pgsql:/', '', $DSN); $DSN = preg_replace('/;/', ' ', $DSN); $aDBInstances[$iLoadThreads] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW); @@ -453,43 +415,36 @@ class SetupFunctions info('Reanalysing database'); $this->pgsqlRunScript('ANALYSE'); - $sDatabaseDate = getDatabaseDate($this->oDB); - $this->oDB->exec('TRUNCATE import_status'); + $sDatabaseDate = getDatabaseDate($oDB); + $oDB->exec('TRUNCATE import_status'); if (!$sDatabaseDate) { warn('could not determine database date.'); } else { $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')"; - $this->oDB->exec($sSQL); + $oDB->exec($sSQL); echo "Latest data imported from $sDatabaseDate.\n"; } } - public function importTigerData() + public function importTigerData($sTigerPath) { info('Import Tiger data'); - $aFilenames = glob(CONST_Tiger_Data_Path.'/*.sql'); - info('Found '.count($aFilenames).' SQL files in path '.CONST_Tiger_Data_Path); - if (empty($aFilenames)) return; + $aFilenames = glob($sTigerPath.'/*.sql'); + info('Found '.count($aFilenames).' SQL files in path '.$sTigerPath); + if (empty($aFilenames)) { + warn('Tiger data import selected but no files found in path '.$sTigerPath); + return; + } + $sTemplate = file_get_contents(CONST_DataDir.'/sql/tiger_import_start.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql'); - $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:aux-data}', - CONST_Tablespace_Aux_Data, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:aux-index}', - CONST_Tablespace_Aux_Index, - $sTemplate - ); $this->pgsqlRunScript($sTemplate, false); $aDBInstances = array(); for ($i = 0; $i < $this->iInstances; $i++) { // https://secure.php.net/manual/en/function.pg-connect.php - $DSN = CONST_Database_DSN; + $DSN = getSetting('DATABASE_DSN'); $DSN = preg_replace('/^pgsql:/', '', $DSN); $DSN = preg_replace('/;/', ' ', $DSN); $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW | PGSQL_CONNECT_ASYNC); @@ -535,25 +490,16 @@ class SetupFunctions } info('Creating indexes on Tiger data'); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql'); - $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:aux-data}', - CONST_Tablespace_Aux_Data, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:aux-index}', - CONST_Tablespace_Aux_Index, - $sTemplate - ); + $sTemplate = file_get_contents(CONST_DataDir.'/sql/tiger_import_finish.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate, false); } public function calculatePostcodes($bCMDResultAll) { info('Calculate Postcodes'); - $this->pgExec('TRUNCATE location_postcode'); + $this->db()->exec('TRUNCATE location_postcode'); $sSQL = 'INSERT INTO location_postcode'; $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; @@ -564,7 +510,7 @@ class SetupFunctions $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'"; $sSQL .= ' AND geometry IS NOT null'; $sSQL .= ' GROUP BY country_code, pc'; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); // only add postcodes that are not yet available in OSM $sSQL = 'INSERT INTO location_postcode'; @@ -574,7 +520,7 @@ class SetupFunctions $sSQL .= ' FROM us_postcode WHERE postcode NOT IN'; $sSQL .= ' (SELECT postcode FROM location_postcode'; $sSQL .= " WHERE country_code = 'us')"; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); // add missing postcodes for GB (if available) $sSQL = 'INSERT INTO location_postcode'; @@ -583,80 +529,104 @@ class SetupFunctions $sSQL .= ' FROM gb_postcode WHERE postcode NOT IN'; $sSQL .= ' (SELECT postcode FROM location_postcode'; $sSQL .= " WHERE country_code = 'gb')"; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); if (!$bCMDResultAll) { $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'"; $sSQL .= 'and word NOT IN (SELECT postcode FROM location_postcode)'; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); } $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM '; $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p'; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); } public function index($bIndexNoanalyse) { - $sOutputFile = ''; - $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$this->aDSNInfo['database'].' -P ' - .$this->aDSNInfo['port'].' -t '.$this->iInstances.$sOutputFile; + $this->checkModulePresence(); // raises exception on failure + + $oBaseCmd = (new \Nominatim\Shell(CONST_DataDir.'/nominatim/nominatim.py')) + ->addParams('--database', $this->aDSNInfo['database']) + ->addParams('--port', $this->aDSNInfo['port']) + ->addParams('--threads', $this->iInstances); + + if (!$this->bQuiet) { + $oBaseCmd->addParams('-v'); + } + if ($this->bVerbose) { + $oBaseCmd->addParams('-v'); + } if (isset($this->aDSNInfo['hostspec'])) { - $sBaseCmd .= ' -H '.$this->aDSNInfo['hostspec']; + $oBaseCmd->addParams('--host', $this->aDSNInfo['hostspec']); } if (isset($this->aDSNInfo['username'])) { - $sBaseCmd .= ' -U '.$this->aDSNInfo['username']; + $oBaseCmd->addParams('--user', $this->aDSNInfo['username']); + } + if (isset($this->aDSNInfo['password'])) { + $oBaseCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); } info('Index ranks 0 - 4'); - $iStatus = $this->runWithPgEnv($sBaseCmd.' -R 4'); + $oCmd = (clone $oBaseCmd)->addParams('--maxrank', 4); + echo $oCmd->escapedCmd(); + + $iStatus = $oCmd->run(); if ($iStatus != 0) { fail('error status ' . $iStatus . ' running nominatim!'); } if (!$bIndexNoanalyse) $this->pgsqlRunScript('ANALYSE'); + info('Index administrative boundaries'); + $oCmd = (clone $oBaseCmd)->addParams('-b'); + $iStatus = $oCmd->run(); + if ($iStatus != 0) { + fail('error status ' . $iStatus . ' running nominatim!'); + } + info('Index ranks 5 - 25'); - $iStatus = $this->runWithPgEnv($sBaseCmd.' -r 5 -R 25'); + $oCmd = (clone $oBaseCmd)->addParams('--minrank', 5, '--maxrank', 25); + $iStatus = $oCmd->run(); if ($iStatus != 0) { fail('error status ' . $iStatus . ' running nominatim!'); } + if (!$bIndexNoanalyse) $this->pgsqlRunScript('ANALYSE'); info('Index ranks 26 - 30'); - $iStatus = $this->runWithPgEnv($sBaseCmd.' -r 26'); + $oCmd = (clone $oBaseCmd)->addParams('--minrank', 26); + $iStatus = $oCmd->run(); if ($iStatus != 0) { fail('error status ' . $iStatus . ' running nominatim!'); } info('Index postcodes'); $sSQL = 'UPDATE location_postcode SET indexed_status = 0'; - $this->pgExec($sSQL); + $this->db()->exec($sSQL); } public function createSearchIndices() { info('Create Search indices'); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql'); + $sSQL = 'SELECT relname FROM pg_class, pg_index '; + $sSQL .= 'WHERE pg_index.indisvalid = false AND pg_index.indexrelid = pg_class.oid'; + $aInvalidIndices = $this->db()->getCol($sSQL); + + foreach ($aInvalidIndices as $sIndexName) { + info("Cleaning up invalid index $sIndexName"); + $this->db()->exec("DROP INDEX $sIndexName;"); + } + + $sTemplate = file_get_contents(CONST_DataDir.'/sql/indices.src.sql'); + if (!$this->bDrop) { + $sTemplate .= file_get_contents(CONST_DataDir.'/sql/indices_updates.src.sql'); + } if (!$this->dbReverseOnly()) { - $sTemplate .= file_get_contents(CONST_BasePath.'/sql/indices_search.src.sql'); + $sTemplate .= file_get_contents(CONST_DataDir.'/sql/indices_search.src.sql'); } - $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:address-index}', - CONST_Tablespace_Address_Index, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:search-index}', - CONST_Tablespace_Search_Index, - $sTemplate - ); - $sTemplate = $this->replaceTablespace( - '{ts:aux-index}', - CONST_Tablespace_Aux_Index, - $sTemplate - ); + $sTemplate = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate); } @@ -670,10 +640,11 @@ class SetupFunctions $this->pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(name->'name'), country_code) from country_name where name ? 'name') as x"); $sSQL = 'select count(*) from (select getorcreate_country(make_standard_name(v),' .'country_code) from (select country_code, skeys(name) as k, svals(name) as v from country_name) x where k '; - if (CONST_Languages) { + $sLanguages = getSetting('LANGUAGES'); + if ($sLanguages) { $sSQL .= 'in '; $sDelim = '('; - foreach (explode(',', CONST_Languages) as $sLang) { + foreach (explode(',', $sLanguages) as $sLang) { $sSQL .= $sDelim."'name:$sLang'"; $sDelim = ','; } @@ -715,7 +686,7 @@ class SetupFunctions ); $aDropTables = array(); - $aHaveTables = $this->oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'"); + $aHaveTables = $this->db()->getListOfTables(); foreach ($aHaveTables as $sTable) { $bFound = false; @@ -728,32 +699,94 @@ class SetupFunctions if (!$bFound) array_push($aDropTables, $sTable); } foreach ($aDropTables as $sDrop) { - if ($this->bVerbose) echo "Dropping table $sDrop\n"; - $this->oDB->exec("DROP TABLE IF EXISTS $sDrop CASCADE"); + $this->dropTable($sDrop); } - if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { - if (file_exists(CONST_Osm2pgsql_Flatnode_File)) { - if ($this->bVerbose) echo 'Deleting '.CONST_Osm2pgsql_Flatnode_File."\n"; - unlink(CONST_Osm2pgsql_Flatnode_File); - } - } + $this->removeFlatnodeFile(); } - private function pgsqlRunDropAndRestore($sDumpFile) + /** + * Setup the directory for the API scripts. + * + * @return null + */ + public function setupWebsite() { - $sCMD = 'pg_restore -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database'].' --no-owner -Fc --clean '.$sDumpFile; - if ($this->oDB->getPostgresVersion() >= 9.04) { - $sCMD .= ' --if-exists'; - } - if (isset($this->aDSNInfo['hostspec'])) { - $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; + if (!is_dir(CONST_InstallDir.'/website')) { + info('Creating directory for website scripts at: '.CONST_InstallDir.'/website'); + mkdir(CONST_InstallDir.'/website'); + } + + $aScripts = array( + 'deletable.php', + 'details.php', + 'lookup.php', + 'polygons.php', + 'reverse.php', + 'search.php', + 'status.php' + ); + + foreach($aScripts as $sScript) { + $rFile = fopen(CONST_InstallDir.'/website/'.$sScript, 'w'); + + fwrite($rFile, "aDSNInfo['username'])) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; + } + + /** + * Return the connection to the database. + * + * @return Database object. + * + * Creates a new connection if none exists yet. Otherwise reuses the + * already established connection. + */ + private function db() + { + if (is_null($this->oDB)) { + $this->oDB = new \Nominatim\DB(); + $this->oDB->connect(); } - $this->runWithPgEnv($sCMD); + return $this->oDB; + } + + private function removeFlatnodeFile() + { + $sFName = getSetting('FLATNODE_FILE'); + if ($sFName && file_exists($sFName)) { + if ($this->bVerbose) echo 'Deleting '.$sFName."\n"; + unlink($sFName); + } } private function pgsqlRunScript($sScript, $bfatal = true) @@ -768,7 +801,22 @@ class SetupFunctions private function createSqlFunctions() { - $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); + $sBasePath = CONST_DataDir.'/sql/functions/'; + $sTemplate = file_get_contents($sBasePath.'utils.sql'); + $sTemplate .= file_get_contents($sBasePath.'normalization.sql'); + $sTemplate .= file_get_contents($sBasePath.'ranking.sql'); + $sTemplate .= file_get_contents($sBasePath.'importance.sql'); + $sTemplate .= file_get_contents($sBasePath.'address_lookup.sql'); + $sTemplate .= file_get_contents($sBasePath.'interpolation.sql'); + if ($this->db()->tableExists('place')) { + $sTemplate .= file_get_contents($sBasePath.'place_triggers.sql'); + } + if ($this->db()->tableExists('placex')) { + $sTemplate .= file_get_contents($sBasePath.'placex_triggers.sql'); + } + if ($this->db()->tableExists('location_postcode')) { + $sTemplate .= file_get_contents($sBasePath.'postcode_triggers.sql'); + } $sTemplate = str_replace('{modulepath}', $this->sModulePath, $sTemplate); if ($this->bEnableDiffUpdates) { $sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate); @@ -776,13 +824,13 @@ class SetupFunctions if ($this->bEnableDebugStatements) { $sTemplate = str_replace('--DEBUG:', '', $sTemplate); } - if (CONST_Limit_Reindexing) { + if (getSettingBool('LIMIT_REINDEXING')) { $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate); } - if (!CONST_Use_US_Tiger_Data) { + if (!getSettingBool('USE_US_TIGER_DATA')) { $sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate); } - if (!CONST_Use_Aux_Location_data) { + if (!getSettingBool('USE_AUX_LOCATION_DATA')) { $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate); } @@ -795,7 +843,7 @@ class SetupFunctions private function pgsqlRunPartitionScript($sTemplate) { $sSQL = 'select distinct partition from country_name'; - $aPartitions = $this->oDB->getCol($sSQL); + $aPartitions = $this->db()->getCol($sSQL); if (!$this->bNoPartitions) $aPartitions[] = 0; preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -814,19 +862,21 @@ class SetupFunctions { if (!file_exists($sFilename)) fail('unable to find '.$sFilename); - $sCMD = 'psql -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database']; + $oCmd = (new \Nominatim\Shell('psql')) + ->addParams('--port', $this->aDSNInfo['port']) + ->addParams('--dbname', $this->aDSNInfo['database']); + if (!$this->bVerbose) { - $sCMD .= ' -q'; + $oCmd->addParams('--quiet'); } if (isset($this->aDSNInfo['hostspec'])) { - $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; + $oCmd->addParams('--host', $this->aDSNInfo['hostspec']); } if (isset($this->aDSNInfo['username'])) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; + $oCmd->addParams('--username', $this->aDSNInfo['username']); } - $aProcEnv = null; if (isset($this->aDSNInfo['password'])) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); } $ahGzipPipes = null; if (preg_match('/\\.gz$/', $sFilename)) { @@ -835,12 +885,14 @@ class SetupFunctions 1 => array('pipe', 'w'), 2 => array('file', '/dev/null', 'a') ); - $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes); + $oZcatCmd = new \Nominatim\Shell('zcat', $sFilename); + + $hGzipProcess = proc_open($oZcatCmd->escapedCmd(), $aDescriptors, $ahGzipPipes); if (!is_resource($hGzipProcess)) fail('unable to start zcat'); $aReadPipe = $ahGzipPipes[1]; fclose($ahGzipPipes[0]); } else { - $sCMD .= ' -f '.$sFilename; + $oCmd->addParams('--file', $sFilename); $aReadPipe = array('pipe', 'r'); } $aDescriptors = array( @@ -849,7 +901,8 @@ class SetupFunctions 2 => array('file', '/dev/null', 'a') ); $ahPipes = null; - $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes, null, $aProcEnv); + + $hProcess = proc_open($oCmd->escapedCmd(), $aDescriptors, $ahPipes, null, $oCmd->aEnv); if (!is_resource($hProcess)) fail('unable to start pgsql'); // TODO: error checking while (!feof($ahPipes[1])) { @@ -866,43 +919,41 @@ class SetupFunctions } } - private function replaceTablespace($sTemplate, $sTablespace, $sSql) - { - if ($sTablespace) { - $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"', $sSql); - } else { - $sSql = str_replace($sTemplate, '', $sSql); - } - return $sSql; - } - - private function runWithPgEnv($sCmd) + private function replaceSqlPatterns($sSql) { - if ($this->bVerbose) { - echo "Execute: $sCmd\n"; - } - - $aProcEnv = null; + $sSql = str_replace('{www-user}', getSetting('DATABASE_WEBUSER'), $sSql); + + $aPatterns = array( + '{ts:address-data}' => getSetting('TABLESPACE_ADDRESS_DATA'), + '{ts:address-index}' => getSetting('TABLESPACE_ADDRESS_INDEX'), + '{ts:search-data}' => getSetting('TABLESPACE_SEARCH_DATA'), + '{ts:search-index}' => getSetting('TABLESPACE_SEARCH_INDEX'), + '{ts:aux-data}' => getSetting('TABLESPACE_AUX_DATA'), + '{ts:aux-index}' => getSetting('TABLESPACE_AUX_INDEX') + ); - if (isset($this->aDSNInfo['password'])) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + foreach ($aPatterns as $sPattern => $sTablespace) { + if ($sTablespace) { + $sSql = str_replace($sPattern, 'TABLESPACE "'.$sTablespace.'"', $sSql); + } else { + $sSql = str_replace($sPattern, '', $sSql); + } } - return runWithEnv($sCmd, $aProcEnv); + return $sSql; } /** - * Execute the SQL command on the open database. + * Drop table with the given name if it exists. * - * @param string $sSQL SQL command to execute. + * @param string $sName Name of table to remove. * * @return null - * - * @pre connect() must have been called. */ - private function pgExec($sSQL) + private function dropTable($sName) { - $this->oDB->exec($sSQL); + if ($this->bVerbose) echo "Dropping table $sName\n"; + $this->db()->deleteTable($sName); } /** @@ -912,6 +963,22 @@ class SetupFunctions */ private function dbReverseOnly() { - return !($this->oDB->tableExists('search_name')); + return !($this->db()->tableExists('search_name')); + } + + /** + * Try accessing the C module, so we know early if something is wrong. + * + * Raises Nominatim\DatabaseError on failure + */ + private function checkModulePresence() + { + $sSQL = "CREATE FUNCTION nominatim_test_import_func(text) RETURNS text AS '"; + $sSQL .= $this->sModulePath . "/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT"; + $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);'; + + $oDB = new \Nominatim\DB(); + $oDB->connect(); + $oDB->exec($sSQL, null, 'Database server failed to load '.$this->sModulePath.'/nominatim.so module'); } }