X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/0273e128f432c0a683d9c68cfd7d4bd29793c925..d4fa528d5c9d8435126b1d9286e84d5208bdfd9d:/lib/setup/SetupClass.php diff --git a/lib/setup/SetupClass.php b/lib/setup/SetupClass.php index 3b4f683d..013668fa 100755 --- a/lib/setup/SetupClass.php +++ b/lib/setup/SetupClass.php @@ -2,28 +2,28 @@ namespace Nominatim\Setup; +require_once(CONST_BasePath.'/lib/setup/AddressLevelParser.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 $oDB = null; // set in setupDB (earliest) or later in loadData, importData, drop, createSqlFunctions, importTigerData - // pgsqlRunPartitionScript, calculatePostcodes, ..if no already set - - public function __construct($callingFunction, array $aCMDResult = array()) + protected $iCacheMemory; + protected $iInstances; + protected $sModulePath; + protected $aDSNInfo; + protected $sVerbose; + protected $sIgnoreErrors; + protected $bEnableDiffUpdates; + protected $bEnableDebugStatements; + protected $bNoPartitions; + protected $oDB = null; + + public function __construct(array $aCMDResult) { // by default, use all but one processor, but never more than 15. $this->iInstances = isset($aCMDResult['threads']) ? $aCMDResult['threads'] : (min(16, getProcessorCount()) - 1); - if ($this->iInstances < 1) { $this->iInstances = 1; warn('resetting threads to '.$this->iInstances); @@ -38,11 +38,13 @@ class SetupFunctions $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 + $this->aDSNInfo = array_filter(\DB::parseDSN(CONST_Database_DSN)); + if (!isset($this->aDSNInfo['port'])) { + $this->aDSNInfo['port'] = 5432; + } + // setting member variables based on command line options stored in $aCMDResult $this->sVerbose = $aCMDResult['verbose']; @@ -62,13 +64,10 @@ class SetupFunctions } else { $this->bNoPartitions = false; } - - // if class is instantiated by update.php, we have to set EnableDiffUpdates to true - // otherwise set to value provided comand line to setup.php - if ($callingFunction == 'update') { - $this->bEnableDiffUpdates = true; - } elseif ($callingFunction == 'setup') { + if (isset($aCMDResult['enable-diff-updates'])) { $this->bEnableDiffUpdates = $aCMDResult['enable-diff-updates']; + } else { + $this->bEnableDiffUpdates = false; } } @@ -81,32 +80,31 @@ class SetupFunctions } $sCreateDBCmd = 'createdb -E UTF-8 -p '.$this->aDSNInfo['port'].' '.$this->aDSNInfo['database']; - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { + if (isset($this->aDSNInfo['username'])) { $sCreateDBCmd .= ' -U '.$this->aDSNInfo['username']; } - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { + if (isset($this->aDSNInfo['hostspec'])) { $sCreateDBCmd .= ' -h '.$this->aDSNInfo['hostspec']; } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); - } - - $result = runWithEnv($sCreateDBCmd, $aProcEnv); + $result = $this->runWithPgEnv($sCreateDBCmd); if ($result != 0) fail('Error executing external command: '.$sCreateDBCmd); } + public function connect() + { + $this->oDB =& getDB(); + } + public function setupDB() { info('Setup DB'); - $this->oDB =& getDB(); $fPostgresVersion = getPostgresVersion($this->oDB); echo 'Postgres version found: '.$fPostgresVersion."\n"; - if ($fPostgresVersion < 9.1) { + if ($fPostgresVersion < 9.01) { fail('Minimum supported version of Postgresql is 9.1.'); } @@ -154,15 +152,14 @@ 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_BasePath.'/data/gb_postcode_table.sql'); - - if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) { - $this->pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz'); + $sPostcodeFilename = CONST_BasePath.'/data/gb_postcode_data.sql.gz'; + if (file_exists($sPostcodeFilename)) { + $this->pgsqlRunScriptFile($sPostcodeFilename); } else { - warn('external UK postcode table not found.'); + warn('optional external UK postcode table file ('.$sPostcodeFilename.') not found. Skipping.'); } if (CONST_Use_Extra_US_Postcodes) { @@ -192,8 +189,6 @@ class SetupFunctions fail("osm2pgsql not found in '$osm2pgsql'"); } - - if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; } @@ -209,19 +204,16 @@ class SetupFunctions $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1'; $osm2pgsql .= ' -C '.$this->iCacheMemory; $osm2pgsql .= ' -P '.$this->aDSNInfo['port']; - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { + if (isset($this->aDSNInfo['username'])) { $osm2pgsql .= ' -U '.$this->aDSNInfo['username']; } - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { + if (isset($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(); + + $this->runWithPgEnv($osm2pgsql); + if (!$this->sIgnoreErrors && !chksql($this->oDB->getRow('select * from place limit 1'))) { fail('No Data'); } @@ -239,7 +231,7 @@ class SetupFunctions $this->createSqlFunctions(); } - public function createTables() + public function createTables($bReverseOnly = false) { info('Create Tables'); @@ -277,6 +269,13 @@ class SetupFunctions ); $this->pgsqlRunScript($sTemplate, false); + + if ($bReverseOnly) { + $this->pgExec('DROP TABLE search_name'); + } + + $oAlParser = new AddressLevelParser(CONST_Address_Level_Config); + $oAlParser->createTable($this->oDB, 'address_levels'); } public function createPartitionTables() @@ -353,43 +352,41 @@ 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)); + $this->pgExec('TRUNCATE word'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('TRUNCATE placex'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('TRUNCATE location_property_osmline'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('TRUNCATE place_addressline'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('TRUNCATE place_boundingbox'); echo '.'; - if (!pg_query($this->oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('TRUNCATE location_area'); echo '.'; - if (!pg_query($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->pgExec('TRUNCATE search_name'); + echo '.'; + } + $this->pgExec('TRUNCATE search_name_blank'); echo '.'; - if (!pg_query($this->oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('DROP SEQUENCE seq_place'); echo '.'; - if (!pg_query($this->oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($this->oDB->connection)); + $this->pgExec('CREATE SEQUENCE seq_place start 100000'); echo '.'; $sSQL = 'select distinct partition from country_name'; $aPartitions = chksql($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)); + $this->pgExec('TRUNCATE location_road_'.$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)); - } + $this->pgExec($sSQL); echo ".\n"; // pre-create the word list @@ -534,11 +531,7 @@ class SetupFunctions 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->pgExec('TRUNCATE location_postcode'); $sSQL = 'INSERT INTO location_postcode'; $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; @@ -549,10 +542,7 @@ class SetupFunctions $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'"; $sSQL .= ' AND geometry IS NOT null'; $sSQL .= ' GROUP BY country_code, pc'; - - if (!pg_query($this->oDB->connection, $sSQL)) { - fail(pg_last_error($this->oDB->connection)); - } + $this->pgExec($sSQL); if (CONST_Use_Extra_US_Postcodes) { // only add postcodes that are not yet available in OSM @@ -563,7 +553,7 @@ class SetupFunctions $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)); + $this->pgExec($sSQL); } // add missing postcodes for GB (if available) @@ -573,21 +563,17 @@ 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)); + $this->pgExec($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)); - } + $this->pgExec($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)); - } + $this->pgExec($sSQL); } public function index($bIndexNoanalyse) @@ -595,38 +581,36 @@ class SetupFunctions $sOutputFile = ''; $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$this->aDSNInfo['database'].' -P ' .$this->aDSNInfo['port'].' -t '.$this->iInstances.$sOutputFile; - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { + if (isset($this->aDSNInfo['hostspec'])) { $sBaseCmd .= ' -H '.$this->aDSNInfo['hostspec']; } - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { + if (isset($this->aDSNInfo['username'])) { $sBaseCmd .= ' -U '.$this->aDSNInfo['username']; } - $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { - $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); - } info('Index ranks 0 - 4'); - $iStatus = runWithEnv($sBaseCmd.' -R 4', $aProcEnv); + $iStatus = $this->runWithPgEnv($sBaseCmd.' -R 4'); 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); + $iStatus = $this->runWithPgEnv($sBaseCmd.' -r 5 -R 25'); 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); + $iStatus = $this->runWithPgEnv($sBaseCmd.' -r 26'); 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)); + $this->pgExec($sSQL); } public function createSearchIndices() @@ -634,6 +618,9 @@ class SetupFunctions info('Create Search indices'); $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql'); + if (!$this->dbReverseOnly()) { + $sTemplate .= file_get_contents(CONST_BasePath.'/sql/indices_search.src.sql'); + } $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); $sTemplate = $this->replaceTablespace( '{ts:address-index}', @@ -706,7 +693,6 @@ class SetupFunctions 'place_classtype_*' ); - if ($this->oDB = null) $this->oDB =& getDB(); $aDropTables = array(); $aHaveTables = chksql($this->oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'")); @@ -721,35 +707,33 @@ class SetupFunctions if (!$bFound) array_push($aDropTables, $sTable); } foreach ($aDropTables as $sDrop) { - if ($this->sVerbose) echo "dropping table $sDrop\n"; + if ($this->sVerbose) echo "Dropping table $sDrop\n"; @pg_query($this->oDB->connection, "DROP TABLE $sDrop CASCADE"); // ignore warnings/errors as they might be caused by a table having // been deleted already by CASCADE } if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { - if ($sVerbose) echo 'deleting '.CONST_Osm2pgsql_Flatnode_File."\n"; - unlink(CONST_Osm2pgsql_Flatnode_File); + if (file_exists(CONST_Osm2pgsql_Flatnode_File)) { + if ($this->sVerbose) echo 'Deleting '.CONST_Osm2pgsql_Flatnode_File."\n"; + unlink(CONST_Osm2pgsql_Flatnode_File); + } } } private function pgsqlRunDropAndRestore($sDumpFile) { - 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']) && $this->aDSNInfo['hostspec']) { + if (isset($this->aDSNInfo['hostspec'])) { $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; } - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { + if (isset($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); - } - $iReturn = runWithEnv($sCMD, $aProcEnv); // /lib/cmd.php "function runWithEnv($sCmd, $aEnv)" + + $this->runWithPgEnv($sCMD); } - + private function pgsqlRunScript($sScript, $bfatal = true) { runSQLScript( @@ -779,13 +763,15 @@ 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($sSQL)); if (!$this->bNoPartitions) $aPartitions[] = 0; @@ -810,14 +796,14 @@ class SetupFunctions if (!$this->sVerbose) { $sCMD .= ' -q'; } - if (isset($this->aDSNInfo['hostspec']) && $this->aDSNInfo['hostspec']) { + if (isset($this->aDSNInfo['hostspec'])) { $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; } - if (isset($this->aDSNInfo['username']) && $this->aDSNInfo['username']) { + if (isset($this->aDSNInfo['username'])) { $sCMD .= ' -U '.$this->aDSNInfo['username']; } $aProcEnv = null; - if (isset($this->aDSNInfo['password']) && $this->aDSNInfo['password']) { + if (isset($this->aDSNInfo['password'])) { $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); } $ahGzipPipes = null; @@ -867,4 +853,42 @@ class SetupFunctions } return $sSql; } + + private function runWithPgEnv($sCmd) + { + $aProcEnv = null; + + if (isset($this->aDSNInfo['password'])) { + $aProcEnv = array_merge(array('PGPASSWORD' => $this->aDSNInfo['password']), $_ENV); + } + + return runWithEnv($sCmd, $aProcEnv); + } + + /** + * Execute the SQL command on the open database. + * + * @param string $sSQL SQL command to execute. + * + * @return null + * + * @pre connect() must have been called. + */ + private function pgExec($sSQL) + { + if (!pg_query($this->oDB->connection, $sSQL)) { + fail(pg_last_error($this->oDB->connection)); + } + } + + /** + * Check if the database is in reverse-only mode. + * + * @return True if there is no search_name table and infrastructure. + */ + private function dbReverseOnly() + { + $sSQL = "SELECT count(*) FROM pg_tables WHERE tablename = 'search_name'"; + return !(chksql($this->oDB->getOne($sSQL))); + } }