]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
penalty for special searches
[nominatim.git] / utils / setup.php
index 8e3d1cb5fc108c49cd59f2363c49a3bc5c2a23ef..90474fcf2767f224b0c98dcc1d628b857dd48211 100755 (executable)
                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('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('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('create-roads', '', 0, 1, 0, 0, 'bool', ''),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
-               array('osmosis-init-date', '', 0, 1, 1, 1, 'string', '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'),
                echo "WARNING: resetting cache memory to $iCacheMemory\n";
        }
 
-       if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
-       {
-               $sBaseFile = basename($aCMDResult['osm-file']);
-               if (preg_match('#^planet-([0-9]{2})([0-9]{2})([0-9]{2})[.]#', $sBaseFile, $aMatch))
-               {
-                       $iTime = mktime(0, 0, 0, $aMatch[2], $aMatch[3], '20'.$aMatch[1]);
-                       $iTime -= (60*60*24);
-                       $aCMDResult['osmosis-init-date'] = date('Y-m-d', $iTime).'T22:00:00Z';
-               }
-       }
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
        if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
                {
                        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']);
+
+               $sVersionString = $oDB->getOne('select version()');
+               preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[.]([0-9]+) #', $sVersionString, $aMatches);
+               if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2])
+               {
+                       echo "ERROR: PostgreSQL version is not correct.  Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
+                       exit;
+               }
+
+               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');
                }
+
                pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+               $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])
+               {
+                       echo "ERROR: PostGIS version is not correct.  Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
+                       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/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['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 16000';
+               $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
 
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['create-partitions'] || $aCMDResult['all'])
+       if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
        {
-               echo "Partitions\n";
+               echo "Partition Tables\n";
                $bDidSomething = true;
                $oDB =& getDB();
                $sSQL = 'select partition from country_name order by country_code';
                }
                $aPartitions[] = 0;
 
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql');
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.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, $sTemplate);
+               }
+
+               pgsqlRunScript($sTemplate);
+       }
+
+
+       if ($aCMDResult['create-partition-functions'] || $aCMDResult['all'])
+       {
+               echo "Partition Functions\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/partition-functions.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
 
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
-               echo "Load Data\n";
+               echo "Drop old Data\n";
                $bDidSomething = true;
 
                $oDB =& getDB();
                        echo '.';
                }
 
+               // used by getorcreate_word_id to ignore frequent partial words
+               if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection));
+               echo ".\n";
+
                // pre-create the word list
                if (!$aCMDResult['disable-token-precalc'])
                {
-                       if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct svals(name) as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection));
-                       echo '.';
-                       if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct postcode as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection));
-                       echo '.';
-                       if (!pg_query($oDB->connection, 'select count(getorcreate_housenumber_id(v)) from (select distinct housenumber as v from place where housenumber is not null) as w;')) fail(pg_last_error($oDB->connection));
-                       echo '.';
+                       echo "Loading word list\n";
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
                }
 
+               echo "Load Data\n";
                $aDBInstances = array();
                for($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 .= '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));
                        echo '.';
                }
                echo "\n";
-               pgsqlRunScript('ALTER TABLE place SET TABLESPACE "data"');
                echo "Reanalysing database...\n";
                pgsqlRunScript('ANALYSE');
        }
                {
                        $aDBInstances[$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, country_code, geometry) from ';
+                       $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));
                $bDidSomething = true;
                $oDB =& getDB();
                if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
-               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
-               $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,country_code,";
-               $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,postcode,";
+               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
+               $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 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,country_code,geometry) ";
+               $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'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
+       if ($aCMDResult['osmosis-init'] || $aCMDResult['all'])
        {
                $bDidSomething = true;
                $oDB =& getDB();
 
-               if (!file_exists(CONST_Osmosis_Binary)) fail("please download osmosis");
-               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');
-
-               $sDate = $aCMDResult['osmosis-init-date'];
-               $aDate = date_parse_from_format("Y-m-d\TH-i", $sDate);
-               $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?';
-               $sURL .= 'Y='.$aDate['year'].'&m='.$aDate['month'].'&d='.$aDate['day'];
-               $sURL .= '&H='.$aDate['hour'].'&i='.$aDate['minute'].'&s=0';
-               $sURL .= '&stream=minute';
-               echo "Getting state file: $sURL\n";
-               $sStateFile = file_get_contents($sURL);
-               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');
-               $sSQL = "INSERT INTO import_status VALUES('".$sDate."')";
-               pg_query($oDB->connection, $sSQL);
+               if (!file_exists(CONST_Osmosis_Binary))
+               {
+                       echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+                       if (!$aCMDResult['all'])
+                       {
+                               fail("osmosis not found in '".CONST_Osmosis_Binary."'");
+                       }
+               }
+               else
+               {
+                       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');
+                       }
+
+                       // 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}.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 ($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;
+               $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
                passthruCheckReturn($sBaseCmd.' -R 4');
                if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
                passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
                @symlink(CONST_BasePath.'/website/reverse.php', $sTargetDir.'/reverse.php');
                @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/search.php');
                @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/index.php');
+               @symlink(CONST_BasePath.'/website/deletable.php', $sTargetDir.'/deletable.php');
+               @symlink(CONST_BasePath.'/website/polygons.php', $sTargetDir.'/polygons.php');
+               @symlink(CONST_BasePath.'/website/status.php', $sTargetDir.'/status.php');
                @symlink(CONST_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 (!$bDidSomething)