X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/4c2c499860d631b746e0db10a5b2ff9cd279c241..1028c2444cd5789860de283cc77eae914daae503:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index 26179d04..5a54ef77 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -10,37 +10,170 @@ array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'), array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'), - array('create-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'), - array('load-data', '', 0, 1, 0, 0, 'realpath', 'Import a osm file'), + array('osm-file', '', 0, 1, 1, 1, 'realpath', 'File to import'), + array('threads', '', 0, 1, 1, 1, 'int', 'Number of threads (where possible)'), + + array('all', '', 0, 1, 0, 0, 'bool', 'Do the complete process'), + + array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'), + array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'), + array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'), + array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'), + 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('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'), + 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('osmosis-init-date', '', 0, 1, 1, 1, 'string', 'Generate default osmosis configuration'), + array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'), + array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'), + array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'), ); getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true); - if ($aCMDResult['create-db']) + $bDidSomething = false; + + // This is a pretty hard core defult - the number of processors in the box - 1 + $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1); + if ($iInstances < 1) + { + $iInstances = 1; + echo "WARNING: resetting threads to $iInstances\n"; + } + if ($iInstances > getProcessorCount()) + { + $iInstances = getProcessorCount(); + echo "WARNING: resetting threads to $iInstances\n"; + } + if (isset($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']) $aDSNInfo['port'] = 5432; + + if ($aCMDResult['create-db'] || $aCMDResult['all']) { + echo "Create DB\n"; + $bDidSomething = true; + $oDB =& DB::connect(CONST_Database_DSN, false); + if (!PEAR::isError($oDB)) + { + fail('database already exists ('.CONST_Database_DSN.')'); + } + passthru('createdb '.$aDSNInfo['database']); + } + + if ($aCMDResult['create-db'] || $aCMDResult['all']) + { + echo "Create DB (2)\n"; + $bDidSomething = true; // TODO: path detection, detection memory, etc. - passthru('createdb nominatim'); - passthru('createlang plpgsql nominatim'); - passthru('psql -f /mqdata/mapquest/postgres-9.0.1-server/share/contrib/_int.sql nominatim'); - passthru('psql -f /mqdata/mapquest/postgres-9.0.1-server/share/contrib/hstore.sql nominatim'); - passthru('psql -f /mqdata/mapquest/postgres-9.0.1-server/share/contrib/postgis-1.5/postgis.sql nominatim'); - passthru('psql -f /mqdata/mapquest/postgres-9.0.1-server/share/contrib/postgis-1.5/spatial_ref_sys.sql nominatim'); + + $oDB =& getDB(); + passthru('createlang plpgsql '.$aDSNInfo['database']); + pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/_int.sql'); + pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql'); + pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); + 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/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 "Import\n"; + $bDidSomething = true; + + $osm2pgsql = CONST_BasePath.'/osm2pgsql/osm2pgsql'; + if (!file_exists($osm2pgsql)) $osm2pgsql = trim(`which osm2pgsql`); + if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql"); + passthru($osm2pgsql.' -lsc -O gazetteer -C 10000 --hstore -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']); + + $oDB =& getDB(); + $x = $oDB->getRow('select * from place limit 1'); + if (!$x || PEAR::isError($x)) fail('No Data'); + } + + if ($aCMDResult['create-functions'] || $aCMDResult['all']) + { + 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', $sTemplate); + pgsqlRunScript($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($sScript); } - if ($aCMDResult['load-data']) + if ($aCMDResult['create-tables'] || $aCMDResult['all']) { - passthru(CONST_BasePath.'/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['load-data']); + echo "Tables\n"; + $bDidSomething = true; + pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql'); + + // re-run the functions + $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); + $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate); + pgsqlRunScript($sTemplate); } - if ($aCMDResult['create-partitions']) + if ($aCMDResult['create-partitions'] || $aCMDResult['all']) { - $sSQL = 'select distinct country_code from country_name order by country_code'; + echo "Partitions\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[] = 'none'; + $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql'); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); @@ -53,8 +186,252 @@ } $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); } - echo $sTemplate; - exit; + + pgsqlRunScript($sTemplate); + } + + if ($aCMDResult['load-data'] || $aCMDResult['all']) + { + echo "Load 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 '.'; + + $aDBInstances = array(); + for($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)) $bAnyBusy = true; + } + sleep(1); + echo '.'; + } + echo "\n"; + } + + if ($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, 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; + + $aDBInstances = array(); + for($i = 0; $i < $iInstances; $i++) + { + $aDBInstances[$i] =& getDB(true); + } + + foreach(glob(CONST_BasePath.'/data/tiger2009/*.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)) $bAnyBusy = true; + } + usleep(10); + } + echo "\n"; + } + } + + if ($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,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 .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y "; + $sSQL .= "from placex where postcode is not null group by country_code,postcode) as x"; + if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + + $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) "; + $sSQL .= "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'])) + { + $bDidSomething = true; + + if (!file_exists(CONST_BasePath.'/osmosis-0.38/bin/osmosis')) fail("please download osmosis"); + if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n"; + else passthru(CONST_BasePath.'/osmosis-0.38/bin/osmosis --read-replication-interval-init '.CONST_BasePath.'/settings'); + + $sDate = $aCMDResult['osmosis-init-date']; + $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?'.$sDate; + 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', $sStateFile); + } + + if ($aCMDResult['index'] || $aCMDResult['all']) + { + $bDidSomething = true; + $sOutputFile = ''; + if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output']; + passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$iInstances.$sOutputFile); + } + + if (isset($aCMDResult['create-website'])) + { + $bDidSomething = true; + $sTargetDir = $aCMDResult['create-website']; + if (!is_dir($sTargetDir)) fail('please specify a directory to setup'); + @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/images', $sTargetDir.'/images'); + @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js'); + echo "Symlinks created\n"; + } + + if (!$bDidSomething) + { + showUsage($aCMDOptions, true); + } + + function pgsqlRunScriptFile($sFilename) + { + if (!file_exists($sFilename)) fail('unable to find '.$sFilename); + + // Convert database DSN to psql paramaters + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + $sCMD = 'psql -f '.$sFilename.' '.$aDSNInfo['database']; + + $aDescriptors = array( + 0 => array('pipe', 'r'), + 1 => array('pipe', 'w'), + 2 => array('file', '/dev/null', 'a') + ); + $ahPipes = null; + $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); + if (!is_resource($hProcess)) fail('unable to start pgsql'); + + fclose($ahPipes[0]); + + // TODO: error checking + while(!feof($ahPipes[1])) + { + echo fread($ahPipes[1], 4096); + } + fclose($ahPipes[1]); + + proc_close($hProcess); } - showUsage($aCMDOptions, true); + function pgsqlRunScript($sScript) + { + // Convert database DSN to psql paramaters + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $sCMD = 'psql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']; + $aDescriptors = array( + 0 => array('pipe', 'r'), + 1 => STDOUT, + 2 => STDERR + ); + $ahPipes = null; + $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); + if (!is_resource($hProcess)) fail('unable to start pgsql'); + + while(strlen($sScript)) + { + $written = fwrite($ahPipes[0], $sScript); + $sScript = substr($sScript, $written); + } + fclose($ahPipes[0]); + proc_close($hProcess); + }