+
+ pgsqlRunScript($sTemplate);
+ }
+
+ if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all'])
+ {
+ $bDidSomething = true;
+ $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin';
+ $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin';
+ if (file_exists($sWikiArticlesFile))
+ {
+ echo "Importing wikipedia articles...";
+ pgsqlRunDropAndRestore($sWikiArticlesFile);
+ echo "...done\n";
+ }
+ else
+ {
+ echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+ }
+ if (file_exists($sWikiRedirectsFile))
+ {
+ echo "Importing wikipedia redirects...";
+ pgsqlRunDropAndRestore($sWikiRedirectsFile);
+ echo "...done\n";
+ }
+ else
+ {
+ echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+ }
+ }
+
+
+ if ($aCMDResult['load-data'] || $aCMDResult['all'])
+ {
+ echo "Drop old Data\n";
+ $bDidSomething = true;
+
+ $oDB =& getDB();
+ if (!pg_query($oDB->connection, 'TRUNCATE word')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE placex')) 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($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($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';
+ $aPartitions = $oDB->getCol($sSQL);
+ if (PEAR::isError($aPartitions))
+ {
+ fail($aPartitions->getMessage());
+ }
+ $aPartitions[] = 0;
+ foreach($aPartitions as $sPartition)
+ {
+ if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
+ 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'])
+ {
+ 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, ';
+ $sSQL .= 'geometry) select * from place where osm_id % '.$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)
+ {
+ $bAnyBusy = false;
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+ }
+ sleep(1);
+ echo '.';
+ }
+ echo "\n";
+ echo "Reanalysing database...\n";
+ pgsqlRunScript('ANALYSE');
+ }
+
+ if ($aCMDResult['create-roads'])
+ {
+ $bDidSomething = true;
+
+ $oDB =& getDB();
+ $aDBInstances = array();
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ $aDBInstances[$i] =& getDB(true);
+ if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
+ $sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, calculated_country_code, geometry) from ';
+ $sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
+ if ($aCMDResult['verbose']) echo "$sSQL\n";
+ if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+ }
+ $bAnyBusy = true;
+ while($bAnyBusy)
+ {
+ $bAnyBusy = false;
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+ }
+ sleep(1);
+ echo '.';
+ }
+ echo "\n";
+ }
+
+ if ($aCMDResult['import-tiger-data'])
+ {
+ $bDidSomething = true;
+
+ pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+
+ $aDBInstances = array();
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ $aDBInstances[$i] =& getDB(true);
+ }
+
+ foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile)
+ {
+ echo $sFile.': ';
+ $hFile = fopen($sFile, "r");
+ $sSQL = fgets($hFile, 100000);
+ $iLines = 0;
+
+ while(true)
+ {
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (!pg_connection_busy($aDBInstances[$i]->connection))
+ {
+ while(pg_get_result($aDBInstances[$i]->connection));
+ $sSQL = fgets($hFile, 100000);
+ if (!$sSQL) break 2;
+ if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+ $iLines++;
+ if ($iLines == 1000)
+ {
+ echo ".";
+ $iLines = 0;
+ }
+ }
+ }
+ usleep(10);
+ }
+
+ fclose($hFile);
+
+ $bAnyBusy = true;
+ while($bAnyBusy)
+ {
+ $bAnyBusy = false;
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+ }
+ usleep(10);
+ }
+ echo "\n";
+ }
+
+ echo "Creating indexes\n";
+ pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
+ }
+
+ if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
+ {
+ $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,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 ";
+ $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
+ if (!pg_query($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 ($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');
+ // 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);
+
+
+ // 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;
+
+ $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);
+ 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;
+ passthruCheckReturn($sBaseCmd.' -R 4');
+ if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+ passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
+ if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+ passthruCheckReturn($sBaseCmd.' -r 26');
+ }
+
+ if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
+ {
+ 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);
+ }
+
projects / nominatim.git / blobdiff