]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/setup/SetupClass.php
move SQL functions for normalisation in separate file
[nominatim.git] / lib / setup / SetupClass.php
index b8070e4a2d326255497a979d6ee2b1e5ccac0e0d..92d6347b0ad770f684d03f41c24f2a31d9ef2ca3 100755 (executable)
@@ -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();
         }
 
@@ -139,7 +142,7 @@ class SetupFunctions
             exit(1);
         }
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.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');
 
@@ -160,13 +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)
@@ -227,42 +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);
@@ -274,41 +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);
     }
@@ -323,19 +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');
         }
     }
 
@@ -343,27 +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';
@@ -371,14 +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
@@ -470,18 +395,13 @@ 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)) 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);
 
         $aDBInstances = array();
@@ -494,7 +414,7 @@ class SetupFunctions
             pg_ping($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);
@@ -534,24 +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) ';
@@ -562,7 +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';
@@ -572,7 +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';
@@ -581,17 +492,17 @@ 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)
@@ -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);
     }
 
@@ -728,10 +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 $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) {
@@ -742,25 +636,6 @@ class SetupFunctions
         }
     }
 
-    private function pgsqlRunDropAndRestore($sDumpFile)
-    {
-        $sCMD = 'pg_restore'
-            .' -p '.escapeshellarg($this->aDSNInfo['port'])
-            .' -d '.escapeshellarg($this->aDSNInfo['database'])
-            .' --no-owner -Fc --clean '.escapeshellarg($sDumpFile);
-        if ($this->oDB->getPostgresVersion() >= 9.04) {
-            $sCMD .= ' --if-exists';
-        }
-        if (isset($this->aDSNInfo['hostspec'])) {
-            $sCMD .= ' -h '.escapeshellarg($this->aDSNInfo['hostspec']);
-        }
-        if (isset($this->aDSNInfo['username'])) {
-            $sCMD .= ' -U '.escapeshellarg($this->aDSNInfo['username']);
-        }
-
-        $this->runWithPgEnv($sCMD);
-    }
-
     private function pgsqlRunScript($sScript, $bfatal = true)
     {
         runSQLScript(
@@ -773,7 +648,9 @@ class SetupFunctions
 
     private function createSqlFunctions()
     {
+        $sBasePath = CONST_BasePath.'/sql/functions/';
         $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
+        $sTemplate .= file_get_contents($sBasePath.'normalization.sql');
         $sTemplate = str_replace('{modulepath}', $this->sModulePath, $sTemplate);
         if ($this->bEnableDiffUpdates) {
             $sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate);
@@ -873,13 +750,27 @@ 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;
     }
 
@@ -899,17 +790,18 @@ 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');
     }
 
     /**