]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
Merge remote-tracking branch 'origin/master' into osmline
[nominatim.git] / utils / setup.php
index 8cbb29781729d46bdee0d9f5407389ef845453d3..caaf0f9126141dba717782a8804cdfa03becc712 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('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
                array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
                array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
-               array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
+               array('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'),
                array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
-               array('create-partitions', '', 0, 1, 0, 0, 'bool', 'Create required partition tables and triggers'),
+               array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
+               array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
+               array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"),
                array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
                array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
                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', 'Calculate postcode centroids'),
                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";
        }
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
@@ -95,7 +99,7 @@
                {
                        fail('database already exists ('.CONST_Database_DSN.')');
                }
-               passthru('createdb -E UTF-8 '.$aDSNInfo['database']);
+               passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
        }
 
        if ($aCMDResult['setup-db'] || $aCMDResult['all'])
                // TODO: path detection, detection memory, etc.
 
                $oDB =& getDB();
-               passthru('createlang plpgsql '.$aDSNInfo['database']);
-               $pgver = (float) CONST_Postgresql_Version;
-               if ($pgver < 9.1) {
-                       pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
-                       pgsqlRunScriptFile(CONST_BasePath.'/sql/hstore_compatability_9_0.sql');
-               } else {
-                       pgsqlRunScript('CREATE EXTENSION hstore');
-               }
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
+
+               $fPostgresVersion = getPostgresVersion($oDB);
+               echo 'Postgres version found: '.$fPostgresVersion."\n";
+
+               if ($fPostgresVersion < 9.1)
+               {
+                       fail("Minimum supported version of Postgresql is 9.1.");
+               }
+
+               pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore');
+               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 = $oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'");
+               if (PEAR::isError($iNumFunc))
+               {
+                       fail("Cannot query stored procedures.", $iNumFunc);
+               }
+               if ($iNumFunc == 0)
+               {
+                       pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable");
+                       echo "WARNING: Postgresql is too old. extratags and namedetails API not available.";
+               }
+
+               $fPostgisVersion = getPostgisVersion($oDB);
+               echo 'Postgis version found: '.$fPostgisVersion."\n";
+
+               if ($fPostgisVersion < 2.1)
+               {
+                       // Function was renamed in 2.1 and throws an annoying deprecation warning
+                       pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
+               }
+
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/worldboundaries.sql');
+               pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
+               if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz'))
+               {
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
+               }
+               else
+               {
+                       echo "WARNING: external UK postcode table not found.\n";
+               }
+               if (CONST_Use_Extra_US_Postcodes)
+               {
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
+               }
+
+               if ($aCMDResult['no-partitions'])
+               {
+                       pgsqlRunScript('update country_name set partition = 0');
+               }
+
+               // the following will be needed by create_functions later but
+               // is only defined in the subsequently called create_tables.
+               // Create dummies here that will be overwritten by the proper
+               // versions in create-tables.
+               pgsqlRunScript('CREATE TABLE place_boundingbox ()');
+               pgsqlRunScript('create type wikipedia_article_match as ()');
        }
 
        if ($aCMDResult['import-data'] || $aCMDResult['all'])
                        echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
                        fail("osm2pgsql not found in '$osm2pgsql'");
                }
-               $osm2pgsql .= ' -lsc -O gazetteer --hstore';
+
+               if (!is_null(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';
                $osm2pgsql .= ' -C '.$iCacheMemory;
+               $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
 
        {
                echo "Functions\n";
                $bDidSomething = true;
-               if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
-               if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
-               if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
-               pgsqlRunScript($sTemplate);
+               if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
+               create_sql_functions($aCMDResult);
        }
 
-       if ($aCMDResult['create-minimal-tables'])
+       if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
-               echo "Minimal Tables\n";
                $bDidSomething = true;
-               pgsqlRunScriptFile(CONST_BasePath.'/sql/tables-minimal.sql');
 
-               $sScript = '';
-
-               // Backstop the import process - easliest possible import id
-               $sScript .= "insert into import_npi_log values (18022);\n";
-
-               $hFile = @fopen(CONST_BasePath.'/settings/partitionedtags.def', "r");
-               if (!$hFile) fail('unable to open list of partitions: '.CONST_BasePath.'/settings/partitionedtags.def');
-
-               while (($sLine = fgets($hFile, 4096)) !== false && $sLine && substr($sLine,0,1) !='#')
-               {
-                       list($sClass, $sType) = explode(' ', trim($sLine));
-                       $sScript .= "create table place_classtype_".$sClass."_".$sType." as ";
-                       $sScript .= "select place_id as place_id,geometry as centroid from placex limit 0;\n";
-
-                       $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_centroid ";
-                       $sScript .= "ON place_classtype_".$sClass."_".$sType." USING GIST (centroid);\n";
+               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);
 
-                       $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id ";
-                       $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n";
-               }
-               fclose($hFile);
-               pgsqlRunScript($sScript);
+               // re-run the functions
+               echo "Functions\n";
+               create_sql_functions($aCMDResult);
        }
 
-       if ($aCMDResult['create-tables'] || $aCMDResult['all'])
+       if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
        {
-               echo "Tables\n";
+               echo "Partition Tables\n";
                $bDidSomething = true;
-               pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
+               $oDB =& getDB();
+               $sSQL = 'select distinct partition from country_name';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               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)
+               {
+                       $sResult = '';
+                       foreach($aPartitions as $sPartitionName)
+                       {
+                               $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
+                       }
+                       $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
+               }
 
-               // re-run the functions
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['create-partitions'] || $aCMDResult['all'])
+
+       if ($aCMDResult['create-partition-functions'] || $aCMDResult['all'])
        {
-               echo "Partitions\n";
+               echo "Partition Functions\n";
                $bDidSomething = true;
                $oDB =& getDB();
-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql');
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection));
                echo '.';
+               if (!pg_query($oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($oDB->connection));
+               echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection));
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection));
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 
-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
                foreach($aPartitions as $sPartition)
                {
                        if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
 
                echo "Load Data\n";
                $aDBInstances = array();
+        
+        $aQueriesPlacex = array();
+        $aQueriesOsmline = array();
+        // the query is divided into parcels, so that the work between the processes, i.e. the DBInstances, will be evenly distributed
+        $iNumberOfParcels = 100;
+        for($i = 0; $i < $iNumberOfParcels; $i++)
+        {
+                       $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
+                       $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
+                       $sSQL .= 'geometry) select * from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and not ';
+                       $sSQL .= '(class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\');';
+                       array_push($aQueriesPlacex, $sSQL);
+                       $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, ';
+                       $sSQL .= 'geometry) from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and ';
+                       $sSQL .= 'class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\'';
+                       array_push($aQueriesOsmline, $sSQL);
+               }
+        
                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
-                       $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
-                       $sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, ';
-                       $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
-                       if ($aCMDResult['verbose']) echo "$sSQL\n";
-                       if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
                }
-               $bAnyBusy = true;
-               while($bAnyBusy)
+               // now execute the query blocks, in the first round for placex, then for osmline, 
+               // because insert_osmline depends on the placex table
+               echo 'Inserting from place to placex.';
+               $aQueries = $aQueriesPlacex;
+               for($j = 0; $j < 2; $j++)
                {
-                       $bAnyBusy = false;
-                       for($i = 0; $i < $iInstances; $i++)
+                       $bAnyBusy = true;
+                       while($bAnyBusy)
                        {
-                               if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
-                       }
-                       sleep(1);
-                       echo '.';
-               }
-               echo "\n";
-               echo "Reanalysing database...\n";
-               pgsqlRunScript('ANALYSE');
-       }
-
-       if ($aCMDResult['create-roads'])
-       {
-               $bDidSomething = true;
+                               $bAnyBusy = false;
 
-               $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; $i++)
+                               for($i = 0; $i < $iInstances; $i++)
+                               {
+                                       if (pg_connection_busy($aDBInstances[$i]->connection)) 
+                                       {
+                                               $bAnyBusy = true;
+                                       }
+                                       else if (count($aQueries) > 0)
+                                       {
+                                               $query = array_pop($aQueries);
+                                               if (!pg_send_query($aDBInstances[$i]->connection, $query))
+                                               {
+                                                       fail(pg_last_error($oDB->connection));
+                                               }
+                                               else
+                                               {
+                                                       pg_get_result($aDBInstances[$i]->connection);
+                                                       $bAnyBusy = true;
+                                               }
+                                       }
+                               }
+                               sleep(1);
+                               echo '.';
+                       }
+                       echo "\n";
+                       if ($j == 0)  //for the second round with osmline
                        {
-                               if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+                               echo 'Inserting from place to osmline.';
+                               $aQueries = $aQueriesOsmline;
                        }
-                       sleep(1);
-                       echo '.';
                }
-               echo "\n";
+               
+               echo "Reanalysing database...\n";
+               pgsqlRunScript('ANALYSE');
        }
 
        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'])
                $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
-               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
-               $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
-               $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
-               if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+               if (CONST_Use_Extra_US_Postcodes)
+               {
+                       $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
+                       $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
+                       $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
+                       if (!pg_query($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();
 
-               if (!file_exists(CONST_Osmosis_Binary)) fail("please download osmosis");
-               if (file_exists(CONST_BasePath.'/settings/configuration.txt'))
+               if (!file_exists(CONST_Osmosis_Binary))
                {
-                       echo "settings/configuration.txt already exists\n";
+                       echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+                       if (!$aCMDResult['all'])
+                       {
+                               fail("osmosis not found in '".CONST_Osmosis_Binary."'");
+                       }
                }
                else
                {
-                       passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
-                       // server layout changed afer license change, fix path to minutely diffs
-                       passthru("sed -i 's:minute-replicate:replication/minute:' ".CONST_BasePath.'/settings/configuration.txt');
-               }
-
-               // Find the last node in the DB
-               $iLastOSMID = $oDB->getOne("select max(osm_id) as 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;
-               $sLastNodeXML = file_get_contents($sLastNodeURL);
-               preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate);
-               $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60);
-
+                       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_InstallPath.'/settings');
+                               // update osmosis configuration.txt with our settings
+                               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');
+                       }
 
-               // Search for the correct state file - uses file timestamps
-               $sRepURL = 'http://planet.openstreetmap.org/replication/minute/';
-               $sRep = file_get_contents($sRepURL);
-               preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a> *(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
-               $aPrevRepMatch = false;
-               foreach($aRepMatches as $aRepMatch)
-               {
-                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                       $aPrevRepMatch = $aRepMatch;
-               }
-               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+                       // Find the last node in the DB
+                       $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";
+                       $sLastNodeXML = file_get_contents($sLastNodeURL);
+                       preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate);
+                       $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60);
+
+                       // Search for the correct state file - uses file timestamps so need to sort by date descending
+                       $sRepURL = CONST_Replication_Url."/";
+                       $sRep = file_get_contents($sRepURL."?C=M;O=D;F=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, $aRepMatches, PREG_SET_ORDER);
+                       if ($aRepMatches)
+                       {
+                               $aPrevRepMatch = false;
+                               foreach($aRepMatches as $aRepMatch)
+                               {
+                                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+                                       $aPrevRepMatch = $aRepMatch;
+                               }
+                               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
 
-               $sRepURL .= $aRepMatch[1];
-               $sRep = file_get_contents($sRepURL);
-               preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a> *(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
-               $aPrevRepMatch = false;
-               foreach($aRepMatches as $aRepMatch)
-               {
-                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                       $aPrevRepMatch = $aRepMatch;
-               }
-               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+                               $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);
+                               $aPrevRepMatch = false;
+                               foreach($aRepMatches as $aRepMatch)
+                               {
+                                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+                                       $aPrevRepMatch = $aRepMatch;
+                               }
+                               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
 
-               $sRepURL .= $aRepMatch[1];
-               $sRep = file_get_contents($sRepURL);
-               preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a> *(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
-               $aPrevRepMatch = false;
-               foreach($aRepMatches as $aRepMatch)
-               {
-                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                       $aPrevRepMatch = $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);
+                               $aPrevRepMatch = false;
+                               foreach($aRepMatches as $aRepMatch)
+                               {
+                                       if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+                                       $aPrevRepMatch = $aRepMatch;
+                               }
+                               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+
+                               $sRepURL .= $aRepMatch[1].'.state.txt';
+                               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_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]."')";
+                               pg_query($oDB->connection, $sSQL);
+                       }
+                       else
+                       {
+                               if (!$aCMDResult['all'])
+                               {
+                                       fail("Cannot read state file directory.");
+                               }
+                       }
                }
-               if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
-               $sRepURL .= $aRepMatch[1].'.state.txt';
-               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', $sStateFile);
-               echo "Updating DB status\n";
-               pg_query($oDB->connection, 'TRUNCATE import_status');
-               $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
-               pg_query($oDB->connection, $sSQL);
        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])
        {
                $bDidSomething = true;
                $sOutputFile = '';
-               if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
-               $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -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');
        {
                echo "Search indices\n";
                $bDidSomething = true;
-               $oDB =& getDB();
-               $sSQL = 'select partition from country_name order by country_code';
-               $aPartitions = $oDB->getCol($sSQL);
-               if (PEAR::isError($aPartitions))
-               {
-                       fail($aPartitions->getMessage());
-               }
-               $aPartitions[] = 0;
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
-               preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
-               foreach($aMatches as $aMatch)
-               {
-                       $sResult = '';
-                       foreach($aPartitions as $sPartitionName)
-                       {
-                               $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
-                       }
-                       $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
-               }
+               $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);
 
                pgsqlRunScript($sTemplate);
        }
                        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_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');
                echo "Symlinks created\n";
+
+               $sTestFile = @file_get_contents(CONST_Website_BaseURL.'js/tiles.js');
+               if (!$sTestFile)
+               {
+                       echo "\nWARNING: Unable to access the website at ".CONST_Website_BaseURL."\n";
+                       echo "You may want to update settings/local.php with @define('CONST_Website_BaseURL', 'http://[HOST]/[PATH]/');\n";
+               }
+       }
+
+       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);
        }
+       else
+       {
+               echo "Setup finished.\n";
+       }
 
        function pgsqlRunScriptFile($sFilename)
        {
                // 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 '.$aDSNInfo['database'].' -f '.$sFilename;
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+
+               $ahGzipPipes = null;
+               if (preg_match('/\\.gz$/', $sFilename))
+               {
+                       $aDescriptors = array(
+                               0 => array('pipe', 'r'),
+                               1 => array('pipe', 'w'),
+                               2 => array('file', '/dev/null', 'a')
+                       );
+                       $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes);
+                       if (!is_resource($hGzipProcess)) fail('unable to start zcat');
+                       $aReadPipe = $ahGzipPipes[1];
+                       fclose($ahGzipPipes[0]);
+               }
+               else
+               {
+                       $sCMD .= ' -f '.$sFilename;
+                       $aReadPipe = array('pipe', 'r');
+               }
 
                $aDescriptors = array(
-                       0 => array('pipe', 'r'),
+                       0 => $aReadPipe,
                        1 => array('pipe', 'w'),
                        2 => array('file', '/dev/null', 'a')
                );
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                if (!is_resource($hProcess)) fail('unable to start pgsql');
 
-               fclose($ahPipes[0]);
 
                // TODO: error checking
                while(!feof($ahPipes[1]))
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
+               if ($ahGzipPipes)
+               {
+                       fclose($ahGzipPipes[1]);
+                       proc_close($hGzipProcess);
+               }
+
        }
 
-       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 '.$aDSNInfo['database'];
+               if ($bfatal && !$aCMDResult['ignore-errors'])
+                       $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
                while(strlen($sScript))
                {
                        $written = fwrite($ahPipes[0], $sScript);
+                       if ($written <= 0) break;
                        $sScript = substr($sScript, $written);
                }
                fclose($ahPipes[0]);
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($bfatal && $iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        function pgsqlRunRestoreData($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
        }
 
        function pgsqlRunDropAndRestore($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
        }
 
        function passthruCheckReturn($cmd)
                passthru($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 $sSql;
+       }
+
+       function create_sql_functions($aCMDResult)
+       {
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
+               $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);
+               }
+               if (!CONST_Use_US_Tiger_Data)
+               {
+                       $sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate);
+               }
+               if (!CONST_Use_Aux_Location_data)
+               {
+                       $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
+               }
+               pgsqlRunScript($sTemplate);
+
+       }
+