X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/2be00f37c175630e23e1ee6a4aa6897fda41eb33..f17c5345ec0ec0b85347b6ccb1b50900d645b4b2:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index bab94e76..8254b6cc 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -18,9 +18,12 @@ 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('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'), 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('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('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'), @@ -29,6 +32,8 @@ 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-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'), ); getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true); @@ -46,6 +51,15 @@ $iInstances = getProcessorCount(); echo "WARNING: resetting threads to $iInstances\n"; } + + // Assume we can steal all the cache memory in the box (unless told otherwise) + $iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB()); + if ($iCacheMemory > getTotalMemoryMB()) + { + $iCacheMemory = getCacheMemoryMB(); + echo "WARNING: resetting cache memory to $iCacheMemory\n"; + } + if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date'])) { $sBaseFile = basename($aCMDResult['osm-file']); @@ -56,6 +70,8 @@ $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']) { @@ -64,9 +80,9 @@ $oDB =& DB::connect(CONST_Database_DSN, false); if (!PEAR::isError($oDB)) { - fail('database already exists'); + fail('database already exists ('.CONST_Database_DSN.')'); } - passthru('createdb nominatim'); + passthru('createdb -E UTF-8 '.$aDSNInfo['database']); } if ($aCMDResult['create-db'] || $aCMDResult['all']) @@ -76,9 +92,13 @@ // TODO: path detection, detection memory, etc. $oDB =& getDB(); - passthru('createlang plpgsql nominatim'); - pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/_int.sql'); - pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql'); + passthru('createlang plpgsql '.$aDSNInfo['database']); + $pgver = (float) CONST_Postgresql_Version; + if ($pgver < 9.1) { + pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql'); + } else { + pgsqlRunScript('CREATE EXTENSION hstore'); + } pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); @@ -96,12 +116,19 @@ echo "Import\n"; $bDidSomething = true; - if (!file_exists(CONST_BasePath.'/osm2pgsql/osm2pgsql')) fail("please download and build osm2pgsql"); - passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['osm-file']); + $osm2pgsql = CONST_Osm2pgsql_Binary; + if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql"); + $osm2pgsql .= ' -lsc -O gazetteer --hstore'; + $osm2pgsql .= ' -C '.$iCacheMemory; + $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; + passthru($osm2pgsql); $oDB =& getDB(); $x = $oDB->getRow('select * from place limit 1'); - if (!$x || PEAR::isError($x)) fail('No Data'); + if (PEAR::isError($x)) { + fail($x->getMessage()); + } + if (!$x) fail('No Data'); } if ($aCMDResult['create-functions'] || $aCMDResult['all']) @@ -114,6 +141,36 @@ pgsqlRunScript($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['create-tables'] || $aCMDResult['all']) { echo "Tables\n"; @@ -130,13 +187,9 @@ { echo "Partitions\n"; $bDidSomething = true; -echo "here"; $oDB =& getDB(); -echo "there"; $sSQL = 'select partition from country_name order by country_code'; -var_dump($sSQL); $aPartitions = $oDB->getCol($sSQL); -var_dump($aPartitions); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); @@ -147,11 +200,9 @@ var_dump($aPartitions); preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); foreach($aMatches as $aMatch) { -var_dump($aMatch); $sResult = ''; foreach($aPartitions as $sPartitionName) { -var_dump($sPartitionName); $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]); } $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); @@ -160,6 +211,34 @@ var_dump($sPartitionName); 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'] || $aCMDResult['all']) { echo "Load Data\n"; @@ -185,6 +264,14 @@ var_dump($sPartitionName); if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection)); echo '.'; + // pre-create the word list + 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 '.'; + $aDBInstances = array(); for($i = 0; $i < $iInstances; $i++) { @@ -207,6 +294,8 @@ var_dump($sPartitionName); echo '.'; } echo "\n"; + echo "Reanalysing database...\n"; + pgsqlRunScript('ANALYSE'); } if ($aCMDResult['create-roads']) @@ -248,7 +337,7 @@ var_dump($sPartitionName); $aDBInstances[$i] =& getDB(true); } - foreach(glob(CONST_BasePath.'/data/tiger2009/*.sql') as $sFile) + foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile) { echo $sFile.': '; $hFile = fopen($sFile, "r"); @@ -277,7 +366,7 @@ var_dump($sPartitionName); } fclose($hFile); - + $bAnyBusy = true; while($bAnyBusy) { @@ -294,6 +383,7 @@ var_dump($sPartitionName); if ($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)); $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) "; @@ -312,15 +402,27 @@ var_dump($sPartitionName); if (($aCMDResult['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date'])) { $bDidSomething = true; + $oDB =& getDB(); - if (!file_exists(CONST_BasePath.'/osmosis-0.38/bin/osmosis')) fail("please download osmosis"); + 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_BasePath.'/osmosis-0.38/bin/osmosis --read-replication-interval-init '.CONST_BasePath.'/settings'); + else passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings'); $sDate = $aCMDResult['osmosis-init-date']; - $sStateFile = file_get_contents('http://toolserver.org/~mazder/replicate-sequences/?'.$sDate); + $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, $sSQL); + } if ($aCMDResult['index'] || $aCMDResult['all']) @@ -328,7 +430,50 @@ var_dump($sPartitionName); $bDidSomething = true; $sOutputFile = ''; if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output']; - passthru(CONST_BasePath.'/nominatim/nominatim -i -d nominatim -t '.$iInstances.$sOutputFile); + $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$iInstances.$sOutputFile; + passthru($sBaseCmd); + } + + if ($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, $sTemplate); + } + + pgsqlRunScript($sTemplate); + } + + 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) @@ -342,7 +487,8 @@ var_dump($sPartitionName); // Convert database DSN to psql paramaters $aDSNInfo = DB::parseDSN(CONST_Database_DSN); - $sCMD = 'psql -f '.$sFilename.' '.$aDSNInfo['database']; + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -f '.$sFilename; $aDescriptors = array( 0 => array('pipe', 'r'), @@ -369,7 +515,32 @@ var_dump($sPartitionName); { // Convert database DSN to psql paramaters $aDSNInfo = DB::parseDSN(CONST_Database_DSN); - $sCMD = 'psql '.$aDSNInfo['database']; + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$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($hProcess); + } + + function pgsqlRunRestoreData($sDumpFile) + { + // Convert database DSN to psql paramaters + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc -a '.$sDumpFile; $aDescriptors = array( 0 => array('pipe', 'r'), @@ -378,24 +549,44 @@ var_dump($sPartitionName); ); $ahPipes = null; $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); - if (!is_resource($hProcess)) fail('unable to start pgsql'); + if (!is_resource($hProcess)) fail('unable to start pg_restore'); + + fclose($ahPipes[0]); + + // TODO: error checking + while(!feof($ahPipes[1])) + { + echo fread($ahPipes[1], 4096); + } + fclose($ahPipes[1]); + + proc_close($hProcess); + } + + function pgsqlRunDropAndRestore($sDumpFile) + { + // Convert database DSN to psql paramaters + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile; + + $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 pg_restore'); -echo "write"; - fwrite($ahPipes[0], $sScript); -echo "close"; fclose($ahPipes[0]); -echo "done"; // TODO: error checking while(!feof($ahPipes[1])) { -echo "read"; echo fread($ahPipes[1], 4096); } -echo "done"; fclose($ahPipes[1]); -echo "done1"; proc_close($hProcess); -echo "done2"; }