X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/388c7f706d43be1e025ac9250b5a068dff3c0a01..14dba3915725e0ff875656f5077945fdb1839017:/lib/setup/SetupClass.php?ds=sidebyside diff --git a/lib/setup/SetupClass.php b/lib/setup/SetupClass.php index 5e5b16d6..385eff70 100755 --- a/lib/setup/SetupClass.php +++ b/lib/setup/SetupClass.php @@ -2,19 +2,23 @@ namespace Nominatim\Setup; +require_once(CONST_BasePath.'/lib/setup/AddressLevelParser.php'); +require_once(CONST_BasePath.'/lib/Shell.php'); + class SetupFunctions { - protected $iCacheMemory; // set in constructor - protected $iInstances; // set in constructor - protected $sModulePath; // set in constructor - protected $aDSNInfo; // set in constructor = DB::parseDSN(CONST_Database_DSN); - protected $sVerbose; // set in constructor - protected $sIgnoreErrors; // set in constructor - protected $bEnableDiffUpdates; // set in constructor - protected $bEnableDebugStatements; // set in constructor - protected $bNoPartitions; // set in constructor - protected $oDB = null; // set in setupDB (earliest) or later in loadData, importData, drop, createSqlFunctions, importTigerData - // pgsqlRunPartitionScript, calculatePostcodes, ..if no already set + protected $iCacheMemory; + 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 $aCMDResult) { @@ -28,22 +32,28 @@ class SetupFunctions warn('resetting threads to '.$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 (!is_null(CONST_Osm2pgsql_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; info('module path: ' . $this->sModulePath); - // prepares DB for import or update, sets the Data Source Name - $this->aDSNInfo = \DB::parseDSN(CONST_Database_DSN); - if (!isset($this->aDSNInfo['port']) || !$this->aDSNInfo['port']) $this->aDSNInfo['port'] = 5432; + // parse database string + $this->aDSNInfo = \Nominatim\DB::parseDSN(CONST_Database_DSN); + if (!isset($this->aDSNInfo['port'])) { + $this->aDSNInfo['port'] = 5432; + } // setting member variables based on command line options stored in $aCMDResult - $this->sVerbose = $aCMDResult['verbose']; + $this->bQuiet = isset($aCMDResult['quiet']) && $aCMDResult['quiet']; + $this->bVerbose = $aCMDResult['verbose']; //setting default values which are not set by the update.php array if (isset($aCMDResult['ignore-errors'])) { @@ -66,73 +76,67 @@ class SetupFunctions } else { $this->bEnableDiffUpdates = false; } + + $this->bDrop = isset($aCMDResult['drop']) && $aCMDResult['drop']; } public function createDB() { info('Create DB'); - $sDB = \DB::connect(CONST_Database_DSN, false); - if (!\PEAR::isError($sDB)) { + $oDB = new \Nominatim\DB; + + if ($oDB->checkConnection()) { fail('database already exists ('.CONST_Database_DSN.')'); } - $sCreateDBCmd = 'createdb -E UTF-8 -p '.$this->aDSNInfo['port'].' '.$this->aDSNInfo['database']; - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { - $sCreateDBCmd .= ' -U '.$this->aDSNInfo['username']; - } + $oCmd = (new \Nominatim\Shell('createdb')) + ->addParams('-E', 'UTF-8') + ->addParams('-p', $this->aDSNInfo['port']); - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { - $sCreateDBCmd .= ' -h '.$this->aDSNInfo['hostspec']; + if (isset($this->aDSNInfo['username'])) { + $oCmd->addParams('-U', $this->aDSNInfo['username']); } - - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + if (isset($this->aDSNInfo['password'])) { + $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); + } + if (isset($this->aDSNInfo['hostspec'])) { + $oCmd->addParams('-h', $this->aDSNInfo['hostspec']); } + $oCmd->addParams($this->aDSNInfo['database']); + + $result = $oCmd->run(); + if ($result != 0) fail('Error executing external command: '.$oCmd->escapedCmd()); + } - $result = runWithEnv($sCreateDBCmd, $aProcEnv); - if ($result != 0) fail('Error executing external command: '.$sCreateDBCmd); + public function connect() + { + $this->oDB = new \Nominatim\DB(); + $this->oDB->connect(); } public function setupDB() { info('Setup DB'); - $this->oDB =& getDB(); - $fPostgresVersion = getPostgresVersion($this->oDB); + $fPostgresVersion = $this->oDB->getPostgresVersion(); echo 'Postgres version found: '.$fPostgresVersion."\n"; - if ($fPostgresVersion < 9.1) { - fail('Minimum supported version of Postgresql is 9.1.'); + if ($fPostgresVersion < 9.03) { + fail('Minimum supported version of Postgresql is 9.3.'); } $this->pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); $this->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. - $iNumFunc = chksql($this->oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'")); - - if ($iNumFunc == 0) { - $this->pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable"); - warn('Postgresql is too old. extratags and namedetails API not available.'); - } - - - $fPostgisVersion = getPostgisVersion($this->oDB); + $fPostgisVersion = $this->oDB->getPostgisVersion(); echo 'Postgis version found: '.$fPostgisVersion."\n"; - if ($fPostgisVersion < 2.1) { - // Functions were renamed in 2.1 and throw an annoying deprecation warning - $this->pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint'); - $this->pgsqlRunScript('ALTER FUNCTION ST_Line_Locate_Point(geometry, geometry) RENAME TO ST_LineLocatePoint'); - } if ($fPostgisVersion < 2.2) { - $this->pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid'); + echo "Minimum required Postgis version 2.2\n"; + exit(1); } - $i = chksql($this->oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'")); + $i = $this->oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'"); if ($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"; @@ -140,9 +144,7 @@ class SetupFunctions } // Try accessing the C module, so we know early if something is wrong - if (!checkModulePresence()) { - fail('error loading nominatim.so module'); - } + checkModulePresence(); // raises exception on failure if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) { echo 'Error: you need to download the country_osm_grid first:'; @@ -150,169 +152,130 @@ class SetupFunctions exit(1); } $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql'); - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz'); + $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'); $sPostcodeFilename = CONST_BasePath.'/data/gb_postcode_data.sql.gz'; if (file_exists($sPostcodeFilename)) { $this->pgsqlRunScriptFile($sPostcodeFilename); } else { - warn('optional external UK postcode table file ('.$sPostcodeFilename.') not found. Skipping.'); + warn('optional external GB postcode table file ('.$sPostcodeFilename.') not found. Skipping.'); } - if (CONST_Use_Extra_US_Postcodes) { - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); + $sPostcodeFilename = CONST_BasePath.'/data/us_postcode_data.sql.gz'; + if (file_exists($sPostcodeFilename)) { + $this->pgsqlRunScriptFile($sPostcodeFilename); + } else { + warn('optional external US postcode table file ('.$sPostcodeFilename.') not found. Skipping.'); } if ($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)) { + if (!file_exists(CONST_Osm2pgsql_Binary)) { echo "Check CONST_Osm2pgsql_Binary in your local settings file.\n"; echo "Normally you should not need to set this manually.\n"; - fail("osm2pgsql not found in '$osm2pgsql'"); + fail("osm2pgsql not found in '".CONST_Osm2pgsql_Binary."'"); } + $oCmd = new \Nominatim\Shell(CONST_Osm2pgsql_Binary); + $oCmd->addParams('--style', CONST_Import_Style); + if (!is_null(CONST_Osm2pgsql_Flatnode_File) && 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'; - $osm2pgsql .= ' -C '.$this->iCacheMemory; - $osm2pgsql .= ' -P '.$this->aDSNInfo['port']; - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { - $osm2pgsql .= ' -U '.$this->aDSNInfo['username']; - } - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { - $osm2pgsql .= ' -H '.$this->aDSNInfo['hostspec']; - } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); - } - $osm2pgsql .= ' -d '.$this->aDSNInfo['database'].' '.$sOSMFile; - runWithEnv($osm2pgsql, $aProcEnv); - if ($this->oDB == null) $this->oDB =& getDB(); - if (!$this->sIgnoreErrors && !chksql($this->oDB->getRow('select * from place limit 1'))) { + $oCmd->addParams('--flat-nodes', CONST_Osm2pgsql_Flatnode_File); + } + if (CONST_Tablespace_Osm2pgsql_Data) { + $oCmd->addParams('--tablespace-slim-data', CONST_Tablespace_Osm2pgsql_Data); + } + if (CONST_Tablespace_Osm2pgsql_Index) { + $oCmd->addParams('--tablespace-slim-index', CONST_Tablespace_Osm2pgsql_Index); + } + if (CONST_Tablespace_Place_Data) { + $oCmd->addParams('--tablespace-main-data', CONST_Tablespace_Place_Data); + } + if (CONST_Tablespace_Place_Index) { + $oCmd->addParams('--tablespace-main-index', CONST_Tablespace_Place_Index); + } + $oCmd->addParams('--latlong', '--slim', '--create'); + $oCmd->addParams('--output', 'gazetteer'); + $oCmd->addParams('--hstore'); + $oCmd->addParams('--number-processes', 1); + $oCmd->addParams('--cache', $this->iCacheMemory); + $oCmd->addParams('--port', $this->aDSNInfo['port']); + + if (isset($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'])) { + $oCmd->addParams('--host', $this->aDSNInfo['hostspec']); + } + $oCmd->addParams('--database', $this->aDSNInfo['database']); + $oCmd->addParams($sOSMFile); + $oCmd->run(); + + if (!$this->sIgnoreErrors && !$this->oDB->getRow('select * from place limit 1')) { fail('No Data'); } + + if ($this->bDrop) { + $this->dropTable('planet_osm_nodes'); + $this->removeFlatnodeFile(); + } } public function createFunctions() { info('Create Functions'); - // Try accessing the C module, so we know eif something is wrong - // update.php calls this function - if (!checkModulePresence()) { - fail('error loading nominatim.so module'); - } + // Try accessing the C module, so we know early if something is wrong + checkModulePresence(); // raises exception on failure + $this->createSqlFunctions(); } - public function createTables() + public function createTables($bReverseOnly = false) { 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 = $this->replaceSqlPatterns($sTemplate); $this->pgsqlRunScript($sTemplate, false); + + if ($bReverseOnly) { + $this->dropTable('search_name'); + } + + $oAlParser = new AddressLevelParser(CONST_Address_Level_Config); + $oAlParser->createTable($this->oDB, '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 - ); + info('Create Tables'); - $sTemplate = $this->replaceTablespace( - '{ts:address-index}', - CONST_Tablespace_Address_Index, - $sTemplate - ); + $sTemplate = file_get_contents(CONST_BasePath.'/sql/table-triggers.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); - $sTemplate = $this->replaceTablespace( - '{ts:search-data}', - CONST_Tablespace_Search_Data, - $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_BasePath.'/sql/partition-tables.src.sql'); + $sTemplate = $this->replaceSqlPatterns($sTemplate); $this->pgsqlRunPartitionScript($sTemplate); } @@ -327,19 +290,14 @@ class SetupFunctions public function importWikipediaArticles() { - $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin'; - $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin'; + $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikimedia-importance.sql.gz'; if (file_exists($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'); } } @@ -347,43 +305,40 @@ class SetupFunctions { info('Drop old Data'); - if ($this->oDB == null) $this->oDB =& getDB(); - - if (!pg_query($this->oDB->connection, 'TRUNCATE word')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE word'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE placex'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE location_property_osmline'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE place_addressline'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE location_area'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($this->oDB->connection)); - echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($this->oDB->connection)); - echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($this->oDB->connection)); + if (!$this->dbReverseOnly()) { + $this->oDB->exec('TRUNCATE search_name'); + echo '.'; + } + $this->oDB->exec('TRUNCATE search_name_blank'); echo '.'; - if (!pg_query($this->oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('DROP SEQUENCE seq_place'); echo '.'; - if (!pg_query($this->oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('CREATE SEQUENCE seq_place start 100000'); echo '.'; $sSQL = 'select distinct partition from country_name'; - $aPartitions = chksql($this->oDB->getCol($sSQL)); + $aPartitions = $this->oDB->getCol($sSQL); + if (!$this->bNoPartitions) $aPartitions[] = 0; foreach ($aPartitions as $sPartition) { - if (!pg_query($this->oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec('TRUNCATE location_road_'.$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'; - if (!pg_query($this->oDB->connection, $sSQL)) { - fail(pg_last_error($this->oDB->connection)); - } + $this->oDB->exec($sSQL); echo ".\n"; // pre-create the word list @@ -394,34 +349,48 @@ class SetupFunctions info('Load Data'); $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry'; + $aDBInstances = array(); $iLoadThreads = max(1, $this->iInstances - 1); for ($i = 0; $i < $iLoadThreads; $i++) { - $aDBInstances[$i] =& getDB(true); + // https://secure.php.net/manual/en/function.pg-connect.php + $DSN = CONST_Database_DSN; + $DSN = preg_replace('/^pgsql:/', '', $DSN); + $DSN = preg_replace('/;/', ' ', $DSN); + $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW); + pg_ping($aDBInstances[$i]); + } + + for ($i = 0; $i < $iLoadThreads; $i++) { $sSQL = "INSERT INTO placex ($sColumns) SELECT $sColumns FROM place WHERE osm_id % $iLoadThreads = $i"; $sSQL .= " and not (class='place' and type='houses' and osm_type='W'"; $sSQL .= " and ST_GeometryType(geometry) = 'ST_LineString')"; $sSQL .= ' and ST_IsValid(geometry)'; - if ($this->sVerbose) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) { - fail(pg_last_error($aDBInstances[$i]->connection)); + if ($this->bVerbose) echo "$sSQL\n"; + if (!pg_send_query($aDBInstances[$i], $sSQL)) { + fail(pg_last_error($aDBInstances[$i])); } } // last thread for interpolation lines - $aDBInstances[$iLoadThreads] =& getDB(true); + // https://secure.php.net/manual/en/function.pg-connect.php + $DSN = CONST_Database_DSN; + $DSN = preg_replace('/^pgsql:/', '', $DSN); + $DSN = preg_replace('/;/', ' ', $DSN); + $aDBInstances[$iLoadThreads] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW); + pg_ping($aDBInstances[$iLoadThreads]); $sSQL = 'insert into location_property_osmline'; $sSQL .= ' (osm_id, address, linegeo)'; $sSQL .= ' SELECT osm_id, address, geometry from place where '; $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'"; - if ($this->sVerbose) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$iLoadThreads]->connection, $sSQL)) { - fail(pg_last_error($aDBInstances[$iLoadThreads]->connection)); + if ($this->bVerbose) echo "$sSQL\n"; + if (!pg_send_query($aDBInstances[$iLoadThreads], $sSQL)) { + fail(pg_last_error($aDBInstances[$iLoadThreads])); } $bFailed = false; for ($i = 0; $i <= $iLoadThreads; $i++) { - while (($hPGresult = pg_get_result($aDBInstances[$i]->connection)) !== false) { + while (($hPGresult = pg_get_result($aDBInstances[$i])) !== false) { $resultStatus = pg_result_status($hPGresult); // PGSQL_EMPTY_QUERY, PGSQL_COMMAND_OK, PGSQL_TUPLES_OK, // PGSQL_COPY_OUT, PGSQL_COPY_IN, PGSQL_BAD_RESPONSE, @@ -437,17 +406,22 @@ class SetupFunctions if ($bFailed) { fail('SQL errors loading placex and/or location_property_osmline tables'); } + + for ($i = 0; $i < $this->iInstances; $i++) { + pg_close($aDBInstances[$i]); + } + echo "\n"; info('Reanalysing database'); $this->pgsqlRunScript('ANALYSE'); $sDatabaseDate = getDatabaseDate($this->oDB); - pg_query($this->oDB->connection, 'TRUNCATE import_status'); - if ($sDatabaseDate === false) { + $this->oDB->exec('TRUNCATE import_status'); + if (!$sDatabaseDate) { warn('could not determine database date.'); } else { $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')"; - pg_query($this->oDB->connection, $sSQL); + $this->oDB->exec($sSQL); echo "Latest data imported from $sDatabaseDate.\n"; } } @@ -456,37 +430,39 @@ class SetupFunctions { 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)) { + warn('Tiger data import selected but no files found in path '.CONST_Tiger_Data_Path); + return; + } $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 - ); + $sTemplate = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate, false); $aDBInstances = array(); for ($i = 0; $i < $this->iInstances; $i++) { - $aDBInstances[$i] =& getDB(true); + // https://secure.php.net/manual/en/function.pg-connect.php + $DSN = CONST_Database_DSN; + $DSN = preg_replace('/^pgsql:/', '', $DSN); + $DSN = preg_replace('/;/', ' ', $DSN); + $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW | PGSQL_CONNECT_ASYNC); + pg_ping($aDBInstances[$i]); } - foreach (glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) { + foreach ($aFilenames as $sFile) { echo $sFile.': '; $hFile = fopen($sFile, 'r'); $sSQL = fgets($hFile, 100000); $iLines = 0; while (true) { for ($i = 0; $i < $this->iInstances; $i++) { - if (!pg_connection_busy($aDBInstances[$i]->connection)) { - while (pg_get_result($aDBInstances[$i]->connection)); + if (!pg_connection_busy($aDBInstances[$i])) { + while (pg_get_result($aDBInstances[$i])); $sSQL = fgets($hFile, 100000); if (!$sSQL) break 2; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($this->oDB->connection)); + if (!pg_send_query($aDBInstances[$i], $sSQL)) fail(pg_last_error($aDBInstances[$i])); $iLines++; if ($iLines == 1000) { echo '.'; @@ -502,37 +478,28 @@ class SetupFunctions while ($bAnyBusy) { $bAnyBusy = false; for ($i = 0; $i < $this->iInstances; $i++) { - if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; + if (pg_connection_busy($aDBInstances[$i])) $bAnyBusy = true; } usleep(10); } echo "\n"; } + for ($i = 0; $i < $this->iInstances; $i++) { + pg_close($aDBInstances[$i]); + } + 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 = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate, false); } public function calculatePostcodes($bCMDResultAll) { info('Calculate Postcodes'); - if ($this->oDB == null) $this->oDB =& getDB(); - if (!pg_query($this->oDB->connection, 'TRUNCATE location_postcode')) { - fail(pg_last_error($this->oDB->connection)); - } - + $this->oDB->exec('TRUNCATE location_postcode'); $sSQL = 'INSERT INTO location_postcode'; $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; @@ -543,22 +510,17 @@ class SetupFunctions $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'"; $sSQL .= ' AND geometry IS NOT null'; $sSQL .= ' GROUP BY country_code, pc'; + $this->oDB->exec($sSQL); - if (!pg_query($this->oDB->connection, $sSQL)) { - fail(pg_last_error($this->oDB->connection)); - } - - if (CONST_Use_Extra_US_Postcodes) { - // only add postcodes that are not yet available in OSM - $sSQL = 'INSERT INTO location_postcode'; - $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; - $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,"; - $sSQL .= ' ST_SetSRID(ST_Point(x,y),4326)'; - $sSQL .= ' FROM us_postcode WHERE postcode NOT IN'; - $sSQL .= ' (SELECT postcode FROM location_postcode'; - $sSQL .= " WHERE country_code = 'us')"; - if (!pg_query($this->oDB->connection, $sSQL)) fail(pg_last_error($this->oDB->connection)); - } + // only add postcodes that are not yet available in OSM + $sSQL = 'INSERT INTO location_postcode'; + $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; + $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,"; + $sSQL .= ' ST_SetSRID(ST_Point(x,y),4326)'; + $sSQL .= ' FROM us_postcode WHERE postcode NOT IN'; + $sSQL .= ' (SELECT postcode FROM location_postcode'; + $sSQL .= " WHERE country_code = 'us')"; + $this->oDB->exec($sSQL); // add missing postcodes for GB (if available) $sSQL = 'INSERT INTO location_postcode'; @@ -567,83 +529,94 @@ class SetupFunctions $sSQL .= ' FROM gb_postcode WHERE postcode NOT IN'; $sSQL .= ' (SELECT postcode FROM location_postcode'; $sSQL .= " WHERE country_code = 'gb')"; - if (!pg_query($this->oDB->connection, $sSQL)) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec($sSQL); if (!$bCMDResultAll) { $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'"; $sSQL .= 'and word NOT IN (SELECT postcode FROM location_postcode)'; - if (!pg_query($this->oDB->connection, $sSQL)) { - fail(pg_last_error($this->oDB->connection)); - } + $this->oDB->exec($sSQL); } + $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM '; $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p'; - - if (!pg_query($this->oDB->connection, $sSQL)) { - fail(pg_last_error($this->oDB->connection)); - } + $this->oDB->exec($sSQL); } public function index($bIndexNoanalyse) { - $sOutputFile = ''; - $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$this->aDSNInfo['database'].' -P ' - .$this->aDSNInfo['port'].' -t '.$this->iInstances.$sOutputFile; - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { - $sBaseCmd .= ' -H '.$this->aDSNInfo['hostspec']; + $oBaseCmd = (new \Nominatim\Shell(CONST_BasePath.'/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['username']) && $this->aDSNInfo['username']) { - $sBaseCmd .= ' -U '.$this->aDSNInfo['username']; + if (isset($this->aDSNInfo['hostspec'])) { + $oBaseCmd->addParams('--host', $this->aDSNInfo['hostspec']); } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + if (isset($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'); - $iStatus = runWithEnv($sBaseCmd.' -R 4', $aProcEnv); + $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 ranks 5 - 25'); - $iStatus = runWithEnv($sBaseCmd.' -r 5 -R 25', $aProcEnv); + $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'); - $iStatus = runWithEnv($sBaseCmd.' -r 26', $aProcEnv); + $oCmd = (clone $oBaseCmd)->addParams('--minrank', 26); + $iStatus = $oCmd->run(); if ($iStatus != 0) { fail('error status ' . $iStatus . ' running nominatim!'); } + info('Index postcodes'); - if ($this->oDB == null) $this->oDB =& getDB(); $sSQL = 'UPDATE location_postcode SET indexed_status = 0'; - if (!pg_query($this->oDB->connection, $sSQL)) fail(pg_last_error($this->oDB->connection)); + $this->oDB->exec($sSQL); } public function createSearchIndices() { info('Create Search indices'); + $sSQL = 'SELECT relname FROM pg_class, pg_index '; + $sSQL .= 'WHERE pg_index.indisvalid = false AND pg_index.indexrelid = pg_class.oid'; + $aInvalidIndices = $this->oDB->getCol($sSQL); + + foreach ($aInvalidIndices as $sIndexName) { + info("Cleaning up invalid index $sIndexName"); + $this->oDB->exec("DROP INDEX $sIndexName;"); + } + $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.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, - $sTemplate - ); + if (!$this->bDrop) { + $sTemplate .= file_get_contents(CONST_BasePath.'/sql/indices_updates.src.sql'); + } + if (!$this->dbReverseOnly()) { + $sTemplate .= file_get_contents(CONST_BasePath.'/sql/indices_search.src.sql'); + } + $sTemplate = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate); } @@ -697,12 +670,12 @@ class SetupFunctions 'new_query_log', 'spatial_ref_sys', 'country_name', - 'place_classtype_*' + 'place_classtype_*', + 'country_osm_grid' ); - if ($this->oDB = null) $this->oDB =& getDB(); $aDropTables = array(); - $aHaveTables = chksql($this->oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'")); + $aHaveTables = $this->oDB->getListOfTables(); foreach ($aHaveTables as $sTable) { $bFound = false; @@ -715,33 +688,20 @@ class SetupFunctions if (!$bFound) array_push($aDropTables, $sTable); } foreach ($aDropTables as $sDrop) { - if ($this->sVerbose) echo "dropping table $sDrop\n"; - @pg_query($this->oDB->connection, "DROP TABLE $sDrop CASCADE"); - // ignore warnings/errors as they might be caused by a table having - // been deleted already by CASCADE + $this->dropTable($sDrop); } - if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { - if ($sVerbose) echo 'deleting '.CONST_Osm2pgsql_Flatnode_File."\n"; - unlink(CONST_Osm2pgsql_Flatnode_File); - } + $this->removeFlatnodeFile(); } - private function pgsqlRunDropAndRestore($sDumpFile) + private function removeFlatnodeFile() { - if (!isset($this->aDSNInfo['port']) || !$this->aDSNInfo['port']) $this->aDSNInfo['port'] = 5432; - $sCMD = 'pg_restore -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database'].' -Fc --clean '.$sDumpFile; - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { - $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; - } - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; - } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { + if (file_exists(CONST_Osm2pgsql_Flatnode_File)) { + if ($this->bVerbose) echo 'Deleting '.CONST_Osm2pgsql_Flatnode_File."\n"; + unlink(CONST_Osm2pgsql_Flatnode_File); + } } - $iReturn = runWithEnv($sCMD, $aProcEnv); // /lib/cmd.php "function runWithEnv($sCmd, $aEnv)" } private function pgsqlRunScript($sScript, $bfatal = true) @@ -749,14 +709,29 @@ class SetupFunctions runSQLScript( $sScript, $bfatal, - $this->sVerbose, + $this->bVerbose, $this->sIgnoreErrors ); } private function createSqlFunctions() { - $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); + $sBasePath = CONST_BasePath.'/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->oDB->tableExists('place')) { + $sTemplate .= file_get_contents($sBasePath.'place_triggers.sql'); + } + if ($this->oDB->tableExists('placex')) { + $sTemplate .= file_get_contents($sBasePath.'placex_triggers.sql'); + } + if ($this->oDB->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); @@ -773,15 +748,17 @@ class SetupFunctions if (!CONST_Use_Aux_Location_data) { $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate); } + + $sReverseOnly = $this->dbReverseOnly() ? 'true' : 'false'; + $sTemplate = str_replace('%REVERSE-ONLY%', $sReverseOnly, $sTemplate); + $this->pgsqlRunScript($sTemplate); } private function pgsqlRunPartitionScript($sTemplate) { - if ($this->oDB == null) $this->oDB =& getDB(); - $sSQL = 'select distinct partition from country_name'; - $aPartitions = chksql($this->oDB->getCol($sSQL)); + $aPartitions = $this->oDB->getCol($sSQL); if (!$this->bNoPartitions) $aPartitions[] = 0; preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -800,19 +777,21 @@ class SetupFunctions { if (!file_exists($sFilename)) fail('unable to find '.$sFilename); - $sCMD = 'psql -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database']; - if (!$this->sVerbose) { - $sCMD .= ' -q'; + $oCmd = (new \Nominatim\Shell('psql')) + ->addParams('--port', $this->aDSNInfo['port']) + ->addParams('--dbname', $this->aDSNInfo['database']); + + if (!$this->bVerbose) { + $oCmd->addParams('--quiet'); } - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { - $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; + if (isset($this->aDSNInfo['hostspec'])) { + $oCmd->addParams('--host', $this->aDSNInfo['hostspec']); } - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; + if (isset($this->aDSNInfo['username'])) { + $oCmd->addParams('--username', $this->aDSNInfo['username']); } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + if (isset($this->aDSNInfo['password'])) { + $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']); } $ahGzipPipes = null; if (preg_match('/\\.gz$/', $sFilename)) { @@ -821,12 +800,14 @@ 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( @@ -835,7 +816,8 @@ 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($ahPipes[1])) { @@ -852,13 +834,52 @@ class SetupFunctions } } - private function replaceTablespace($sTemplate, $sTablespace, $sSql) + private function replaceSqlPatterns($sSql) { - if ($sTablespace) { - $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"', $sSql); - } else { - $sSql = str_replace($sTemplate, '', $sSql); + $sSql = str_replace('{www-user}', CONST_Database_Web_User, $sSql); + + $aPatterns = array( + '{ts:address-data}' => CONST_Tablespace_Address_Data, + '{ts:address-index}' => CONST_Tablespace_Address_Index, + '{ts:search-data}' => CONST_Tablespace_Search_Data, + '{ts:search-index}' => CONST_Tablespace_Search_Index, + '{ts:aux-data}' => CONST_Tablespace_Aux_Data, + '{ts:aux-index}' => CONST_Tablespace_Aux_Index, + ); + + foreach ($aPatterns as $sPattern => $sTablespace) { + if ($sTablespace) { + $sSql = str_replace($sPattern, 'TABLESPACE "'.$sTablespace.'"', $sSql); + } else { + $sSql = str_replace($sPattern, '', $sSql); + } } + return $sSql; } + + /** + * Drop table with the given name if it exists. + * + * @param string $sName Name of table to remove. + * + * @return null + * + * @pre connect() must have been called. + */ + private function dropTable($sName) + { + if ($this->bVerbose) echo "Dropping table $sName\n"; + $this->oDB->deleteTable($sName); + } + + /** + * Check if the database is in reverse-only mode. + * + * @return True if there is no search_name table and infrastructure. + */ + private function dbReverseOnly() + { + return !($this->oDB->tableExists('search_name')); + } }