]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
exclude Tiger stuff from required tests
[nominatim.git] / utils / setup.php
index 65c3d88aaf619df098a6025ec9467e8c4adcdffb..a363fab35142757fef707e491c25a7dfa0504909 100755 (executable)
@@ -1,7 +1,8 @@
 #!/usr/bin/php -Cq
 <?php
 
-       require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
+       require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
+       require_once(CONST_BasePath.'/lib/init-cmd.php');
        ini_set('memory_limit', '800M');
 
        $aCMDOptions = array(
                array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
                array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
                array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
-               array('create-roads', '', 0, 1, 0, 0, 'bool', ''),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
                array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
                array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
                array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
                array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
                array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
+               array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'),
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
        }
 
        // Assume we can steal all the cache memory in the box (unless told otherwise)
-       $iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB());
-       if ($iCacheMemory > getTotalMemoryMB())
+       if (isset($aCMDResult['osm2pgsql-cache']))
+       {
+               $iCacheMemory = $aCMDResult['osm2pgsql-cache'];
+       }
+       else
        {
                $iCacheMemory = getCacheMemoryMB();
-               echo "WARNING: resetting cache memory to $iCacheMemory\n";
        }
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                        pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
                        pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
                } else {
-                       pgsqlRunScript('CREATE EXTENSION postgis');
+                       pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis');
                }
                if ($fPostgisVersion < 2.1) {
                        // Function was renamed in 2.1 and throws an annoying deprecation warning
                {
                        $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
                }
-               $osm2pgsql .= ' -lsc -O gazetteer --hstore';
+               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';
                $osm2pgsql .= ' -C '.$iCacheMemory;
                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
        {
                echo "Functions\n";
                $bDidSomething = true;
-               if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
+               if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
+               $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/module', $sTemplate);
                if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
                if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
                if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
                pgsqlRunScript($sTemplate);
+
                if ($fPostgisVersion < 2.0) {
                        echo "Helper functions for postgis < 2.0\n";
                        $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql');
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
-               echo "Tables\n";
                $bDidSomething = true;
-               pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
+
+               echo "Tables\n";
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql');
+               $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:address-data}',
+                                               CONST_Tablespace_Address_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:address-index}',
+                                               CONST_Tablespace_Address_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:search-data}',
+                                               CONST_Tablespace_Search_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:search-index}',
+                                               CONST_Tablespace_Search_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-data}',
+                                               CONST_Tablespace_Aux_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-index}',
+                                               CONST_Tablespace_Aux_Index, $sTemplate);
+               pgsqlRunScript($sTemplate, false);
 
                // re-run the functions
+               echo "Functions\n";
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
+               $sTemplate = str_replace('{modulepath}',
+                                            CONST_InstallPath.'/module', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
                if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
+               $sTemplate = replace_tablespace('{ts:address-data}',
+                                               CONST_Tablespace_Address_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:address-index}',
+                                               CONST_Tablespace_Address_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:search-data}',
+                                               CONST_Tablespace_Search_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:search-index}',
+                                               CONST_Tablespace_Search_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-data}',
+                                               CONST_Tablespace_Aux_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-index}',
+                                               CONST_Tablespace_Aux_Index, $sTemplate);
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
                pgsqlRunScript('ANALYSE');
        }
 
-       if ($aCMDResult['create-roads'])
-       {
-               $bDidSomething = true;
-
-               $oDB =& getDB();
-               $aDBInstances = array();
-               for($i = 0; $i < $iInstances; $i++)
-               {
-                       $aDBInstances[$i] =& getDB(true);
-                       if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
-                       $sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, calculated_country_code, geometry) from ';
-                       $sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
-                       if ($aCMDResult['verbose']) echo "$sSQL\n";
-                       if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
-               }
-               $bAnyBusy = true;
-               while($bAnyBusy)
-               {
-                       $bAnyBusy = false;
-                       for($i = 0; $i < $iInstances; $i++)
-                       {
-                               if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
-                       }
-                       sleep(1);
-                       echo '.';
-               }
-               echo "\n";
-       }
-
        if ($aCMDResult['import-tiger-data'])
        {
                $bDidSomething = true;
 
-               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
+               $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-data}',
+                                               CONST_Tablespace_Aux_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-index}',
+                                               CONST_Tablespace_Aux_Index, $sTemplate);
+               pgsqlRunScript($sTemplate, false);
 
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                        $aDBInstances[$i] =& getDB(true);
                }
 
-               foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile)
+               foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile)
                {
                        echo $sFile.': ';
                        $hFile = fopen($sFile, "r");
                }
 
                echo "Creating indexes\n";
-               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql');
+               $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-data}',
+                                               CONST_Tablespace_Aux_Data, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-index}',
+                                               CONST_Tablespace_Aux_Index, $sTemplate);
+               pgsqlRunScript($sTemplate, false);
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
        }
 
-       if ($aCMDResult['osmosis-init'] || $aCMDResult['all'])
+       if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables
        {
                $bDidSomething = true;
                $oDB =& getDB();
                }
                else
                {
-                       if (file_exists(CONST_BasePath.'/settings/configuration.txt'))
+                       if (file_exists(CONST_InstallPath.'/settings/configuration.txt'))
                        {
                                echo "settings/configuration.txt already exists\n";
                        }
                        else
                        {
-                               passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
+                               passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings');
                                // update osmosis configuration.txt with our settings
-                               passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_BasePath.'/settings/configuration.txt');
-                               passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_BasePath.'/settings/configuration.txt');
+                               passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt');
+                               passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt');
                        }
 
                        // Find the last node in the DB
-                       $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes");
+                       $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'");
 
                        // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed)
                        $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1";
                        // download.geofabrik.de:    <a href="000/">000/</a></td><td align="right">26-Feb-2013 11:53  </td>
                        // planet.openstreetmap.org: <a href="273/">273/</a>                    2013-03-11 07:41    -
                        preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-                       var_dump($aRepMatches);
                        if ($aRepMatches)
                        {
                                $aPrevRepMatch = false;
                                $sRepURL .= $aRepMatch[1];
                                $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
                                preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-                               var_dump($aRepMatches);
                                $aPrevRepMatch = false;
                                foreach($aRepMatches as $aRepMatch)
                                {
                                $sRepURL .= $aRepMatch[1];
                                $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
                                preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-                               var_dump($aRepMatches);
                                $aPrevRepMatch = false;
                                foreach($aRepMatches as $aRepMatch)
                                {
                                echo "Getting state file: $sRepURL\n";
                                $sStateFile = file_get_contents($sRepURL);
                                if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
-                               file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
+                               file_put_contents(CONST_InstallPath.'/settings/state.txt', $sStateFile);
                                echo "Updating DB status\n";
                                pg_query($oDB->connection, 'TRUNCATE import_status');
                                $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
                $bDidSomething = true;
                $sOutputFile = '';
                if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
-               $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
+               $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
                passthruCheckReturn($sBaseCmd.' -R 4');
                if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
                passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
                if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
+               $sTemplate = replace_tablespace('{ts:address-index}',
+                                               CONST_Tablespace_Address_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:search-index}',
+                                               CONST_Tablespace_Search_Index, $sTemplate);
+               $sTemplate = replace_tablespace('{ts:aux-index}',
+                                               CONST_Tablespace_Aux_Index, $sTemplate);
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
                        fail("Target directory does not exist.");
                }
 
-               @symlink(CONST_BasePath.'/website/details.php', $sTargetDir.'/details.php');
-               @symlink(CONST_BasePath.'/website/reverse.php', $sTargetDir.'/reverse.php');
-               @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/search.php');
-               @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/index.php');
-               @symlink(CONST_BasePath.'/website/deletable.php', $sTargetDir.'/deletable.php');
-               @symlink(CONST_BasePath.'/website/polygons.php', $sTargetDir.'/polygons.php');
-               @symlink(CONST_BasePath.'/website/status.php', $sTargetDir.'/status.php');
+               @symlink(CONST_InstallPath.'/website/details.php', $sTargetDir.'/details.php');
+               @symlink(CONST_InstallPath.'/website/reverse.php', $sTargetDir.'/reverse.php');
+               @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/search.php');
+               @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/index.php');
+               @symlink(CONST_InstallPath.'/website/lookup.php', $sTargetDir.'/lookup.php');
+               @symlink(CONST_InstallPath.'/website/deletable.php', $sTargetDir.'/deletable.php');
+               @symlink(CONST_InstallPath.'/website/polygons.php', $sTargetDir.'/polygons.php');
+               @symlink(CONST_InstallPath.'/website/status.php', $sTargetDir.'/status.php');
                @symlink(CONST_BasePath.'/website/images', $sTargetDir.'/images');
                @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js');
                @symlink(CONST_BasePath.'/website/css', $sTargetDir.'/css');
                }
        }
 
+       if ($aCMDResult['drop'])
+       {
+               // The implementation is potentially a bit dangerous because it uses
+               // a positive selection of tables to keep, and deletes everything else.
+               // Including any tables that the unsuspecting user might have manually
+               // created. USE AT YOUR OWN PERIL.
+               $bDidSomething = true;
+
+               // tables we want to keep. everything else goes.
+               $aKeepTables = array(
+                  "*columns",
+                  "import_polygon_*",
+                  "import_status",
+                  "place_addressline",
+                  "location_property*",
+                  "placex",
+                  "search_name",
+                  "seq_*",
+                  "word",
+                  "query_log",
+                  "new_query_log",
+                  "gb_postcode",
+                  "spatial_ref_sys",
+                  "country_name",
+                  "place_classtype_*"
+               );
+
+               $oDB =& getDB();
+               $aDropTables = array();
+               $aHaveTables = $oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'");
+               if (PEAR::isError($aHaveTables))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               foreach($aHaveTables as $sTable)
+               {
+                       $bFound = false;
+                       foreach ($aKeepTables as $sKeep)
+                       {
+                               if (fnmatch($sKeep, $sTable))
+                               {
+                                       $bFound = true;
+                                       break;
+                               }
+                       }
+                       if (!$bFound) array_push($aDropTables, $sTable);
+               }
+
+               foreach ($aDropTables as $sDrop)
+               {
+                       if ($aCMDResult['verbose']) echo "dropping table $sDrop\n";
+                       @pg_query($oDB->connection, "DROP TABLE $sDrop CASCADE");
+                       // ignore warnings/errors as they might be caused by a table having
+                       // been deleted already by CASCADE
+               }
+
+               if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+               {
+                       if ($aCMDResult['verbose']) echo "deleting ".CONST_Osm2pgsql_Flatnode_File."\n";
+                       unlink(CONST_Osm2pgsql_Flatnode_File);
+               }
+       }
+
        if (!$bDidSomething)
        {
                showUsage($aCMDOptions, true);
 
        }
 
-       function pgsqlRunScript($sScript)
+       function pgsqlRunScript($sScript, $bfatal = true)
        {
                global $aCMDResult;
                // Convert database DSN to psql parameters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
                $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
-               if (!$aCMDResult['ignore-errors'])
+               if ($bfatal && !$aCMDResult['ignore-errors'])
                        $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                }
                fclose($ahPipes[0]);
                $iReturn = proc_close($hProcess);
-               if ($iReturn > 0)
+               if ($bfatal && $iReturn > 0)
                {
                        fail("pgsql returned with error code ($iReturn)");
                }
                passthru($cmd, $result);
                if ($result != 0) fail('Error executing external command: '.$cmd);
        }
+
+       function replace_tablespace($sTemplate, $sTablespace, $sSql)
+       {
+               if ($sTablespace)
+                       $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"',
+                                           $sSql);
+               else
+                       $sSql = str_replace($sTemplate, '', $sSql);
+
+               return $sSql;
+       }
+