X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/260dbe302a753841d7a6cd0f4cc6d4d56e3496d5..a437f5d18a48d3f982e965ef81c93f71a73e53e0:/lib/setup/SetupClass.php?ds=sidebyside diff --git a/lib/setup/SetupClass.php b/lib/setup/SetupClass.php index a26b7dae..b96b186b 100755 --- a/lib/setup/SetupClass.php +++ b/lib/setup/SetupClass.php @@ -29,10 +29,13 @@ 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(); } @@ -80,13 +83,15 @@ class SetupFunctions fail('database already exists ('.CONST_Database_DSN.')'); } - $sCreateDBCmd = 'createdb -E UTF-8 -p '.$this->aDSNInfo['port'].' '.$this->aDSNInfo['database']; + $sCreateDBCmd = 'createdb -E UTF-8' + .' -p '.escapeshellarg($this->aDSNInfo['port']) + .' '.escapeshellarg($this->aDSNInfo['database']); if (isset($this->aDSNInfo['username'])) { - $sCreateDBCmd .= ' -U '.$this->aDSNInfo['username']; + $sCreateDBCmd .= ' -U '.escapeshellarg($this->aDSNInfo['username']); } if (isset($this->aDSNInfo['hostspec'])) { - $sCreateDBCmd .= ' -h '.$this->aDSNInfo['hostspec']; + $sCreateDBCmd .= ' -h '.escapeshellarg($this->aDSNInfo['hostspec']); } $result = $this->runWithPgEnv($sCreateDBCmd); @@ -158,13 +163,6 @@ 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) @@ -178,30 +176,30 @@ class SetupFunctions fail("osm2pgsql not found in '$osm2pgsql'"); } - $osm2pgsql .= ' -S '.CONST_Import_Style; + $osm2pgsql .= ' -S '.escapeshellarg(CONST_Import_Style); if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { - $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; + $osm2pgsql .= ' --flat-nodes '.escapeshellarg(CONST_Osm2pgsql_Flatnode_File); } if (CONST_Tablespace_Osm2pgsql_Data) - $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data; + $osm2pgsql .= ' --tablespace-slim-data '.escapeshellarg(CONST_Tablespace_Osm2pgsql_Data); if (CONST_Tablespace_Osm2pgsql_Index) - $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index; + $osm2pgsql .= ' --tablespace-slim-index '.escapeshellarg(CONST_Tablespace_Osm2pgsql_Index); if (CONST_Tablespace_Place_Data) - $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data; + $osm2pgsql .= ' --tablespace-main-data '.escapeshellarg(CONST_Tablespace_Place_Data); if (CONST_Tablespace_Place_Index) - $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index; + $osm2pgsql .= ' --tablespace-main-index '.escapeshellarg(CONST_Tablespace_Place_Index); $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1'; - $osm2pgsql .= ' -C '.$this->iCacheMemory; - $osm2pgsql .= ' -P '.$this->aDSNInfo['port']; + $osm2pgsql .= ' -C '.escapeshellarg($this->iCacheMemory); + $osm2pgsql .= ' -P '.escapeshellarg($this->aDSNInfo['port']); if (isset($this->aDSNInfo['username'])) { - $osm2pgsql .= ' -U '.$this->aDSNInfo['username']; + $osm2pgsql .= ' -U '.escapeshellarg($this->aDSNInfo['username']); } if (isset($this->aDSNInfo['hostspec'])) { - $osm2pgsql .= ' -H '.$this->aDSNInfo['hostspec']; + $osm2pgsql .= ' -H '.escapeshellarg($this->aDSNInfo['hostspec']); } - $osm2pgsql .= ' -d '.$this->aDSNInfo['database'].' '.$sOSMFile; + $osm2pgsql .= ' -d '.escapeshellarg($this->aDSNInfo['database']).' '.escapeshellarg($sOSMFile); $this->runWithPgEnv($osm2pgsql); @@ -225,42 +223,12 @@ 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 = $this->replaceSqlPatterns($sTemplate); $this->pgsqlRunScript($sTemplate, false); if ($bReverseOnly) { - $this->pgExec('DROP TABLE search_name'); + $this->dropTable('search_name'); } $oAlParser = new AddressLevelParser(CONST_Address_Level_Config); @@ -272,41 +240,7 @@ class SetupFunctions 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 - ); - - $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); $this->pgsqlRunPartitionScript($sTemplate); } @@ -321,19 +255,14 @@ 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); + info('Importing wikipedia articles and redirects'); + $this->dropTable('wikipedia_article'); + $this->dropTable('wikipedia_redirect'); + $this->pgsqlRunScriptFile($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); - } 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'); } } @@ -341,27 +270,25 @@ class SetupFunctions { info('Drop old Data'); - $this->pgExec('TRUNCATE word'); - echo '.'; - $this->pgExec('TRUNCATE placex'); + $this->oDB->exec('TRUNCATE word'); echo '.'; - $this->pgExec('TRUNCATE location_property_osmline'); + $this->oDB->exec('TRUNCATE placex'); echo '.'; - $this->pgExec('TRUNCATE place_addressline'); + $this->oDB->exec('TRUNCATE location_property_osmline'); echo '.'; - $this->pgExec('TRUNCATE place_boundingbox'); + $this->oDB->exec('TRUNCATE place_addressline'); echo '.'; - $this->pgExec('TRUNCATE location_area'); + $this->oDB->exec('TRUNCATE location_area'); echo '.'; if (!$this->dbReverseOnly()) { - $this->pgExec('TRUNCATE search_name'); + $this->oDB->exec('TRUNCATE search_name'); echo '.'; } - $this->pgExec('TRUNCATE search_name_blank'); + $this->oDB->exec('TRUNCATE search_name_blank'); echo '.'; - $this->pgExec('DROP SEQUENCE seq_place'); + $this->oDB->exec('DROP SEQUENCE seq_place'); echo '.'; - $this->pgExec('CREATE SEQUENCE seq_place start 100000'); + $this->oDB->exec('CREATE SEQUENCE seq_place start 100000'); echo '.'; $sSQL = 'select distinct partition from country_name'; @@ -369,14 +296,14 @@ class SetupFunctions if (!$this->bNoPartitions) $aPartitions[] = 0; foreach ($aPartitions as $sPartition) { - $this->pgExec('TRUNCATE location_road_'.$sPartition); + $this->oDB->exec('TRUNCATE location_road_'.$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'; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); echo ".\n"; // pre-create the word list @@ -473,17 +400,8 @@ class SetupFunctions if (empty($aFilenames)) 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(); @@ -536,24 +454,15 @@ 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 = $this->replaceSqlPatterns($sTemplate); + $this->pgsqlRunScript($sTemplate, false); } public function calculatePostcodes($bCMDResultAll) { info('Calculate Postcodes'); - $this->pgExec('TRUNCATE location_postcode'); + $this->oDB->exec('TRUNCATE location_postcode'); $sSQL = 'INSERT INTO location_postcode'; $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; @@ -564,7 +473,7 @@ 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); + $this->oDB->exec($sSQL); // only add postcodes that are not yet available in OSM $sSQL = 'INSERT INTO location_postcode'; @@ -574,7 +483,7 @@ class SetupFunctions $sSQL .= ' FROM us_postcode WHERE postcode NOT IN'; $sSQL .= ' (SELECT postcode FROM location_postcode'; $sSQL .= " WHERE country_code = 'us')"; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); // add missing postcodes for GB (if available) $sSQL = 'INSERT INTO location_postcode'; @@ -583,29 +492,31 @@ class SetupFunctions $sSQL .= ' FROM gb_postcode WHERE postcode NOT IN'; $sSQL .= ' (SELECT postcode FROM location_postcode'; $sSQL .= " WHERE country_code = 'gb')"; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); if (!$bCMDResultAll) { $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'"; $sSQL .= 'and word NOT IN (SELECT postcode FROM location_postcode)'; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); } $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM '; $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p'; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); } public function index($bIndexNoanalyse) { $sOutputFile = ''; - $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$this->aDSNInfo['database'].' -P ' - .$this->aDSNInfo['port'].' -t '.$this->iInstances.$sOutputFile; + $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i' + .' -d '.escapeshellarg($this->aDSNInfo['database']) + .' -P '.escapeshellarg($this->aDSNInfo['port']) + .' -t '.escapeshellarg($this->iInstances.$sOutputFile); if (isset($this->aDSNInfo['hostspec'])) { - $sBaseCmd .= ' -H '.$this->aDSNInfo['hostspec']; + $sBaseCmd .= ' -H '.escapeshellarg($this->aDSNInfo['hostspec']); } if (isset($this->aDSNInfo['username'])) { - $sBaseCmd .= ' -U '.$this->aDSNInfo['username']; + $sBaseCmd .= ' -U '.escapeshellarg($this->aDSNInfo['username']); } info('Index ranks 0 - 4'); @@ -630,7 +541,7 @@ class SetupFunctions info('Index postcodes'); $sSQL = 'UPDATE location_postcode SET indexed_status = 0'; - $this->pgExec($sSQL); + $this->oDB->exec($sSQL); } public function createSearchIndices() @@ -641,22 +552,8 @@ class SetupFunctions 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}', - 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); } @@ -728,8 +625,7 @@ 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) { @@ -740,22 +636,6 @@ class SetupFunctions } } - private function pgsqlRunDropAndRestore($sDumpFile) - { - $sCMD = 'pg_restore -p '.$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 (isset($this->aDSNInfo['username'])) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; - } - - $this->runWithPgEnv($sCMD); - } - private function pgsqlRunScript($sScript, $bfatal = true) { runSQLScript( @@ -768,7 +648,15 @@ class SetupFunctions 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.'importance.sql'); + $sTemplate .= file_get_contents($sBasePath.'address_lookup.sql'); + $sTemplate .= file_get_contents($sBasePath.'interpolation.sql'); + $sTemplate .= file_get_contents($sBasePath.'place_triggers.sql'); + $sTemplate .= file_get_contents($sBasePath.'placex_triggers.sql'); + $sTemplate .= file_get_contents($sBasePath.'postcode_triggers.sql'); $sTemplate = str_replace('{modulepath}', $this->sModulePath, $sTemplate); if ($this->bEnableDiffUpdates) { $sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate); @@ -814,15 +702,17 @@ class SetupFunctions { if (!file_exists($sFilename)) fail('unable to find '.$sFilename); - $sCMD = 'psql -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database']; + $sCMD = 'psql' + .' -p '.escapeshellarg($this->aDSNInfo['port']) + .' -d '.escapeshellarg($this->aDSNInfo['database']); if (!$this->bVerbose) { $sCMD .= ' -q'; } if (isset($this->aDSNInfo['hostspec'])) { - $sCMD .= ' -h '.$this->aDSNInfo['hostspec']; + $sCMD .= ' -h '.escapeshellarg($this->aDSNInfo['hostspec']); } if (isset($this->aDSNInfo['username'])) { - $sCMD .= ' -U '.$this->aDSNInfo['username']; + $sCMD .= ' -U '.escapeshellarg($this->aDSNInfo['username']); } $aProcEnv = null; if (isset($this->aDSNInfo['password'])) { @@ -835,12 +725,12 @@ class SetupFunctions 1 => array('pipe', 'w'), 2 => array('file', '/dev/null', 'a') ); - $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes); + $hGzipProcess = proc_open('zcat '.escapeshellarg($sFilename), $aDescriptors, $ahGzipPipes); if (!is_resource($hGzipProcess)) fail('unable to start zcat'); $aReadPipe = $ahGzipPipes[1]; fclose($ahGzipPipes[0]); } else { - $sCMD .= ' -f '.$sFilename; + $sCMD .= ' -f '.escapeshellarg($sFilename); $aReadPipe = array('pipe', 'r'); } $aDescriptors = array( @@ -866,13 +756,27 @@ class SetupFunctions } } - private function replaceTablespace($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; } @@ -892,17 +796,18 @@ class SetupFunctions } /** - * 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"; + $this->oDB->exec('DROP TABLE IF EXISTS '.$sName.' CASCADE'); } /**