]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
Merge remote-tracking branch 'upstream/master'
[nominatim.git] / utils / setup.php
index d404f4df85057e422440d77402a4bd446cd20222..edf0b14d460f27156569b65f872f2bb8aa934e6d 100755 (executable)
@@ -26,6 +26,7 @@
                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'),
+               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)'),
                        pgsqlRunScript('CREATE EXTENSION hstore');
                }
 
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+               $fPostgisVersion = (float) CONST_Postgis_Version;
+               if ($fPostgisVersion < 2.0) {
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
+               } else {
+                       pgsqlRunScript('CREATE EXTENSION postgis');
+               }
+               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');
+               }
                $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])
                        exit;
                }
 
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
                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/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";
+               }
                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');
+
+               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'");
                }
+
+               if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+               {
+                       $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+               }
+               $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';
-               $osm2pgsql .= ' -C '.$iCacheMemory;
+               $osm2pgsql .= ' -C 18000';
                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
                $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);
        }
 
                echo "Partition Tables\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/partition-tables.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                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/partition-functions.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                if (!pg_query($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));
                {
                        $aDBInstances[$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 .= '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));
                $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,";
                $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,";
                $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-               $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
+               $sSQL .= "from placex where postcode is not null and calculated_country_code not in ('ie') group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
                $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
                $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');
-                       // 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');
-               }
-
-               // 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_BasePath.'/settings/configuration.txt'))
+                       {
+                               echo "settings/configuration.txt already exists\n";
+                       }
+                       else
+                       {
+                               passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/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');
+                       }
 
-               // 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");
-               // 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>                    22-Mar-2013 07:41    -
-               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(id) from planet_osm_nodes");
+
+                       // 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");
+                       // 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>                    22-Mar-2013 07:41    -
+                       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");
-               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");
+                       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");
-               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");
+                       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 = $aRepMatch;
+                       }
+                       if ($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, $sSQL);
                }
-               if ($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'])
                echo "Search indices\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/indices.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                @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 (!$bDidSomething)
                // 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)
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
                $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+               $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
                        $sScript = substr($sScript, $written);
                }
                fclose($ahPipes[0]);
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        function pgsqlRunRestoreData($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        function pgsqlRunDropAndRestore($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        function passthruCheckReturn($cmd)