]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
Merge remote-tracking branch 'origin/master' into osmline
[nominatim.git] / utils / setup.php
index 90374e74785175a146a9f567537c23aca5dfdb64..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(
@@ -23,7 +24,6 @@
                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('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'),
-               array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
                array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
                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'),
@@ -36,7 +36,6 @@
                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)'),
@@ -91,8 +90,6 @@
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
        if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
-       $fPostgisVersion = (float) CONST_Postgis_Version;
-
        if ($aCMDResult['create-db'] || $aCMDResult['all'])
        {
                echo "Create DB\n";
 
                $oDB =& getDB();
 
-               $sVersionString = $oDB->getOne('select version()');
-               preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches);
-               if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2])
+               $fPostgresVersion = getPostgresVersion($oDB);
+               echo 'Postgres version found: '.$fPostgresVersion."\n";
+
+               if ($fPostgresVersion < 9.1)
                {
-                       echo "ERROR: PostgreSQL version is not correct.  Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
-                       exit;
+                       fail("Minimum supported version of Postgresql is 9.1.");
                }
 
-               passthru('createlang plpgsql -p '.$aDSNInfo['port'].' '.$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');
-               }
+               pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore');
+               pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis');
 
-               if ($fPostgisVersion < 2.0) {
-                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
-                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
-               } else {
-                       pgsqlRunScript('CREATE EXTENSION 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.", $iNumFunc);
                }
-               if ($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');
+               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.";
                }
-               $sVersionString = $oDB->getOne('select postgis_full_version()');
-               preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
-               if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2])
+
+               $fPostgisVersion = getPostgisVersion($oDB);
+               echo 'Postgis version found: '.$fPostgisVersion."\n";
+
+               if ($fPostgisVersion < 2.1)
                {
-                       echo "ERROR: PostGIS version is not correct.  Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
-                       exit;
+                       // 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');
                {
                        echo "WARNING: external UK postcode table not found.\n";
                }
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
+               if (CONST_Use_Extra_US_Postcodes)
+               {
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
+               }
 
                if ($aCMDResult['no-partitions'])
                {
        {
                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);
-               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');
-               } else {
-                       echo "Helper functions for postgis >= 2.0\n";
-                       $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_20_aux.sql');
-               }
-               pgsqlRunScript($sTemplate);
-       }
-
-       if ($aCMDResult['create-minimal-tables'])
-       {
-               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";
-
-                       $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id ";
-                       $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n";
-               }
-               fclose($hFile);
-               pgsqlRunScript($sScript);
+               if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
+               create_sql_functions($aCMDResult);
        }
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
 
                // re-run the functions
                echo "Functions\n";
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}',
-                                            CONST_BasePath.'/module', $sTemplate);
-               pgsqlRunScript($sTemplate);
+               create_sql_functions($aCMDResult);
        }
 
        if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
                echo '.';
                if (!pg_query($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));
 
                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, addr_place, 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)
+                       {
+                               $bAnyBusy = false;
+
+                               for($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');
        }
                $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
                if (!pg_query($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'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables
                }
                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
                                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');
        {
                echo "Search indices\n";
                $bDidSomething = true;
-               $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/indices.src.sql');
                $sTemplate = replace_tablespace('{ts:address-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)
-               {
-                       $sResult = '';
-                       foreach($aPartitions as $sPartitionName)
-                       {
-                               $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
-                       }
-                       $sTemplate = str_replace($aMatch[0], $sResult, $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_BasePath.'/website/lookup.php', $sTargetDir.'/lookup.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');
                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);
+
+       }
+