X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/7879ad44cd1e36ae1235c3dcdfd9e477491813e4..797c2d1e74a14d8d52d963f1efad9be53fca5f5f:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index 4ed343df..992034c7 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -24,7 +24,6 @@ array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'), array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'), array('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'), - 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-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'), array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'), @@ -37,9 +36,7 @@ array('osmosis-init', '', 0, 1, 0, 0, 'bool', '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'), 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'), array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'), ); getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true); @@ -92,13 +89,11 @@ $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; - $fPostgisVersion = (float) CONST_Postgis_Version; - if ($aCMDResult['create-db'] || $aCMDResult['all']) { echo "Create DB\n"; $bDidSomething = true; - $oDB =& DB::connect(CONST_Database_DSN, false); + $oDB = DB::connect(CONST_Database_DSN, false); if (!PEAR::isError($oDB)) { fail('database already exists ('.CONST_Database_DSN.')'); @@ -114,40 +109,36 @@ $oDB =& getDB(); - $sVersionString = $oDB->getOne('select version()'); - preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches); - if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2]) + $fPostgresVersion = getPostgresVersion($oDB); + echo 'Postgres version found: '.$fPostgresVersion."\n"; + + if ($fPostgresVersion < 9.1) { - echo "ERROR: PostgreSQL version is not correct. Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; - exit; + fail("Minimum supported version of Postgresql is 9.1."); } - 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'); - } + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis'); - if ($fPostgisVersion < 2.0) { - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql'); - pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql'); - } else { - pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis'); + // For extratags and namedetails the hstore_to_json converter is + // needed which is only available from Postgresql 9.3+. For older + // versions add a dummy function that returns nothing. + $iNumFunc = chksql($oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'")); + + if ($iNumFunc == 0) + { + pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable"); + echo "WARNING: Postgresql is too old. extratags and namedetails API not available."; } - if ($fPostgisVersion < 2.1) { + + $fPostgisVersion = getPostgisVersion($oDB); + echo 'Postgis version found: '.$fPostgisVersion."\n"; + + if ($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]) - { - echo "ERROR: PostGIS version is not correct. Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n"; - exit; - } pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql'); @@ -161,9 +152,10 @@ { 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'); + if (CONST_Use_Extra_US_Postcodes) + { + pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); + } if ($aCMDResult['no-partitions']) { @@ -209,11 +201,10 @@ passthruCheckReturn($osm2pgsql); $oDB =& getDB(); - $x = $oDB->getRow('select * from place limit 1'); - if (PEAR::isError($x)) { - fail($x->getMessage()); + if (!chksql($oDB->getRow('select * from place limit 1'))) + { + fail('No Data'); } - if (!$x) fail('No Data'); } if ($aCMDResult['create-functions'] || $aCMDResult['all']) @@ -221,51 +212,7 @@ echo "Functions\n"; $bDidSomething = true; if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built"); - $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); - $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/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); - - if ($fPostgisVersion < 2.0) { - echo "Helper functions for postgis < 2.0\n"; - $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql'); - } else { - echo "Helper functions for postgis >= 2.0\n"; - $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_20_aux.sql'); - } - pgsqlRunScript($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($sScript); + create_sql_functions($aCMDResult); } if ($aCMDResult['create-tables'] || $aCMDResult['all']) @@ -291,24 +238,13 @@ // re-run the functions echo "Functions\n"; - $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); - $sTemplate = str_replace('{modulepath}', - CONST_InstallPath.'/module', $sTemplate); - pgsqlRunScript($sTemplate); + create_sql_functions($aCMDResult); } if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { echo "Partition Tables\n"; $bDidSomething = true; - $oDB =& getDB(); - $sSQL = 'select distinct partition from country_name'; - $aPartitions = $oDB->getCol($sSQL); - if (PEAR::isError($aPartitions)) - { - fail($aPartitions->getMessage()); - } - if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); $sTemplate = replace_tablespace('{ts:address-data}', @@ -323,18 +259,8 @@ CONST_Tablespace_Aux_Data, $sTemplate); $sTemplate = replace_tablespace('{ts:aux-index}', CONST_Tablespace_Aux_Index, $sTemplate); - 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); + pgsqlRunPartitionScript($sTemplate); } @@ -342,28 +268,10 @@ { echo "Partition Functions\n"; $bDidSomething = true; - $oDB =& getDB(); - $sSQL = 'select distinct partition from country_name'; - $aPartitions = $oDB->getCol($sSQL); - if (PEAR::isError($aPartitions)) - { - fail($aPartitions->getMessage()); - } - if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); - preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); - foreach($aMatches as $aMatch) - { - $sResult = ''; - foreach($aPartitions as $sPartitionName) - { - $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]); - } - $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); - } - pgsqlRunScript($sTemplate); + pgsqlRunPartitionScript($sTemplate); } if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) @@ -422,11 +330,7 @@ echo '.'; $sSQL = 'select distinct partition from country_name'; - $aPartitions = $oDB->getCol($sSQL); - if (PEAR::isError($aPartitions)) - { - fail($aPartitions->getMessage()); - } + $aPartitions = chksql($oDB->getCol($sSQL)); if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; foreach($aPartitions as $sPartition) { @@ -447,70 +351,37 @@ echo "Load Data\n"; $aDBInstances = array(); - - $aQueriesPlacex = array(); - $aQueriesOsmline = array(); - // the query is divided into parcels, so that the work between the processes, i.e. the DBInstances, will be evenly distributed - $iNumberOfParcels = 100; - for($i = 0; $i < $iNumberOfParcels; $i++) - { - $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; - $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, '; - $sSQL .= 'geometry) select * from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and not '; - $sSQL .= '(class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\');'; - array_push($aQueriesPlacex, $sSQL); - $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, '; - $sSQL .= 'geometry) from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and '; - $sSQL .= 'class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\''; - array_push($aQueriesOsmline, $sSQL); - } - - for($i = 0; $i < $iInstances; $i++) + $iLoadThreads = max(1, $iInstances - 1); + for($i = 0; $i < $iLoadThreads; $i++) { $aDBInstances[$i] =& getDB(true); - } - // now execute the query blocks, in the first round for placex, then for osmline, - // because insert_osmline depends on the placex table - echo 'Inserting from place to placex.'; - $aQueries = $aQueriesPlacex; - for($j = 0; $j < 2; $j++) - { - $bAnyBusy = true; - while($bAnyBusy) - { - $bAnyBusy = false; - - for($i = 0; $i < $iInstances; $i++) - { - if (pg_connection_busy($aDBInstances[$i]->connection)) - { - $bAnyBusy = true; - } - else if (count($aQueries) > 0) - { - $query = array_pop($aQueries); - if (!pg_send_query($aDBInstances[$i]->connection, $query)) - { - fail(pg_last_error($oDB->connection)); - } - else - { - pg_get_result($aDBInstances[$i]->connection); - $bAnyBusy = true; - } - } - } - sleep(1); - echo '.'; - } - echo "\n"; - if ($j == 0) //for the second round with osmline + $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; + $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, '; + $sSQL .= 'geometry) select * from place where osm_id % '.$iLoadThreads.' = '.$i; + $sSQL .= " and not (class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString')"; + if ($aCMDResult['verbose']) echo "$sSQL\n"; + if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + } + // last thread for interpolation lines + $aDBInstances[$iLoadThreads] =& getDB(true); + $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, '; + $sSQL .= 'geometry) from place where '; + $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'"; + if ($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 <= $iLoadThreads; $i++) { - echo 'Inserting from place to osmline.'; - $aQueries = $aQueriesOsmline; + if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; } + sleep(1); + echo '.'; } - + echo "\n"; echo "Reanalysing database...\n"; pgsqlRunScript('ANALYSE'); } @@ -597,10 +468,14 @@ $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 (CONST_Use_Extra_US_Postcodes) + { + $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'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables @@ -701,7 +576,6 @@ { $bDidSomething = true; $sOutputFile = ''; - if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output']; $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile; passthruCheckReturn($sBaseCmd.' -R 4'); if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE'); @@ -714,14 +588,6 @@ { echo "Search indices\n"; $bDidSomething = true; - $oDB =& getDB(); - $sSQL = 'select distinct partition from country_name'; - $aPartitions = $oDB->getCol($sSQL); - if (PEAR::isError($aPartitions)) - { - fail($aPartitions->getMessage()); - } - if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql'); $sTemplate = replace_tablespace('{ts:address-index}', @@ -730,51 +596,10 @@ CONST_Tablespace_Search_Index, $sTemplate); $sTemplate = replace_tablespace('{ts:aux-index}', CONST_Tablespace_Aux_Index, $sTemplate); - 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)) - { - echo "You must create the website directory before calling this function.\n"; - fail("Target directory does not exist."); - } - - @symlink(CONST_InstallPath.'/website/details.php', $sTargetDir.'/details.php'); - @symlink(CONST_InstallPath.'/website/reverse.php', $sTargetDir.'/reverse.php'); - @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/search.php'); - @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/index.php'); - @symlink(CONST_InstallPath.'/website/lookup.php', $sTargetDir.'/lookup.php'); - @symlink(CONST_InstallPath.'/website/deletable.php', $sTargetDir.'/deletable.php'); - @symlink(CONST_InstallPath.'/website/polygons.php', $sTargetDir.'/polygons.php'); - @symlink(CONST_InstallPath.'/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 ($aCMDResult['drop']) { // The implementation is potentially a bit dangerous because it uses @@ -804,11 +629,8 @@ $oDB =& getDB(); $aDropTables = array(); - $aHaveTables = $oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'"); - if (PEAR::isError($aHaveTables)) - { - fail($aPartitions->getMessage()); - } + $aHaveTables = chksql($oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'")); + foreach($aHaveTables as $sTable) { $bFound = false; @@ -937,6 +759,29 @@ } } + function pgsqlRunPartitionScript($sTemplate) + { + global $aCMDResult; + $oDB =& getDB(); + + $sSQL = 'select distinct partition from country_name'; + $aPartitions = chksql($oDB->getCol($sSQL)); + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; + + 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); + } + function pgsqlRunRestoreData($sDumpFile) { // Convert database DSN to psql parameters @@ -1011,3 +856,31 @@ return $sSql; } + function create_sql_functions($aCMDResult) + { + $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql'); + $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/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); + } + if (!CONST_Use_US_Tiger_Data) + { + $sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate); + } + if (!CONST_Use_Aux_Location_data) + { + $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate); + } + pgsqlRunScript($sTemplate); + + } +