X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/7ed4c284052d1ac9259e561601db8f0e7bbb337d..be268f95e621ec767f1c4f26ca5dfe29a586d6b3:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index af0e1a5e..0026a813 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -1,7 +1,8 @@ #!/usr/bin/php -Cq getOne("select count(*) from pg_proc where proname = 'hstore_to_json'"); + if (PEAR::isError($iNumFunc)) + { + fail("Cannot query stored procedures.", $iNumFunc); + } + if ($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."; + } + + $fPostgisVersion = getPostgisVersion($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'); + } + 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'); + pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql'); + if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) + { + pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz'); + } + else + { + echo "WARNING: external UK postcode table not found.\n"; + } + if (CONST_Use_Extra_US_Postcodes) + { + pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); + } + + if ($aCMDResult['no-partitions']) + { + pgsqlRunScript('update country_name set partition = 0'); + } + + // the following will be needed by create_functions later but + // is only defined in the subsequently called create_tables. + // Create dummies here that will be overwritten by the proper + // versions in create-tables. + pgsqlRunScript('CREATE TABLE place_boundingbox ()'); + pgsqlRunScript('create type wikipedia_article_match as ()'); } if ($aCMDResult['import-data'] || $aCMDResult['all']) @@ -106,80 +180,135 @@ $bDidSomething = true; $osm2pgsql = CONST_Osm2pgsql_Binary; - if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql"); - passthru($osm2pgsql.' -lsc -O gazetteer -C 12000 --hstore -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']); + if (!file_exists($osm2pgsql)) + { + 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; + } + if (CONST_Tablespace_Osm2pgsql_Data) + $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data; + if (CONST_Tablespace_Osm2pgsql_Index) + $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index; + if (CONST_Tablespace_Place_Data) + $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data; + if (CONST_Tablespace_Place_Index) + $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index; + $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1'; + $osm2pgsql .= ' -C '.$iCacheMemory; + $osm2pgsql .= ' -P '.$aDSNInfo['port']; + $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']; + passthruCheckReturn($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']) { echo "Functions\n"; $bDidSomething = true; - if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built"); + 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_BasePath.'/module', $sTemplate); + $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 ($aCMDResult['create-minimal-tables']) + if ($aCMDResult['create-tables'] || $aCMDResult['all']) { - 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"; + echo "Tables\n"; + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-data}', + CONST_Tablespace_Address_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-data}', + CONST_Tablespace_Search_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); - $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id "; - $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n"; - } - fclose($hFile); - pgsqlRunScript($sScript); + // 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); } - if ($aCMDResult['create-tables'] || $aCMDResult['all']) + if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { - echo "Tables\n"; + echo "Partition Tables\n"; $bDidSomething = true; - pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql'); + $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}', + CONST_Tablespace_Address_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-data}', + CONST_Tablespace_Search_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + 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); + } - // 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'] || $aCMDResult['all']) + + if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { - echo "Partitions\n"; + echo "Partition Functions\n"; $bDidSomething = true; $oDB =& getDB(); - $sSQL = 'select partition from country_name order by country_code'; + $sSQL = 'select distinct partition from country_name'; $aPartitions = $oDB->getCol($sSQL); if (PEAR::isError($aPartitions)) { fail($aPartitions->getMessage()); } - $aPartitions[] = 0; + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; - $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql'); + $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 $aMatch) { @@ -194,9 +323,37 @@ pgsqlRunScript($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"; + echo "Drop old Data\n"; $bDidSomething = true; $oDB =& getDB(); @@ -219,44 +376,38 @@ if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection)); echo '.'; - $aDBInstances = array(); - for($i = 0; $i < $iInstances; $i++) + $sSQL = 'select distinct partition from country_name'; + $aPartitions = $oDB->getCol($sSQL); + if (PEAR::isError($aPartitions)) { - $aDBInstances[$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)); + fail($aPartitions->getMessage()); } - $bAnyBusy = true; - while($bAnyBusy) + if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; + foreach($aPartitions as $sPartition) { - $bAnyBusy = false; - for($i = 0; $i < $iInstances; $i++) - { - if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; - } - sleep(1); + if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection)); echo '.'; } - echo "\n"; - echo "Reanalysing database...\n"; - pgsqlRunScript('ANALYSE'); - } - if ($aCMDResult['create-roads']) - { - $bDidSomething = true; + // used by getorcreate_word_id to ignore frequent partial words + if (!pg_query($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"; - $oDB =& getDB(); + // pre-create the word list + if (!$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); - 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 '; + $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; + $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, '; + $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i; if ($aCMDResult['verbose']) echo "$sSQL\n"; if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); } @@ -272,19 +423,29 @@ echo '.'; } echo "\n"; + echo "Reanalysing database...\n"; + pgsqlRunScript('ANALYSE'); } if ($aCMDResult['import-tiger-data']) { $bDidSomething = true; + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); + $aDBInstances = array(); for($i = 0; $i < $iInstances; $i++) { $aDBInstances[$i] =& getDB(true); } - foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile) + foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) { echo $sFile.': '; $hFile = fopen($sFile, "r"); @@ -313,7 +474,7 @@ } fclose($hFile); - + $bAnyBusy = true; while($bAnyBusy) { @@ -326,6 +487,15 @@ } echo "\n"; } + + echo "Creating indexes\n"; + $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql'); + $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-data}', + CONST_Tablespace_Aux_Data, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); + pgsqlRunScript($sTemplate, false); } if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) @@ -333,46 +503,126 @@ $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"; + $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); - $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) "; - $sSQL .= "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'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date'])) + if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables { $bDidSomething = true; + $oDB =& getDB(); - if (!file_exists(CONST_Osmosis_Binary)) fail("please download osmosis"); - if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n"; - else passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings'); + if (!file_exists(CONST_Osmosis_Binary)) + { + echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + if (!$aCMDResult['all']) + { + fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + } + } + else + { + if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) + { + echo "settings/configuration.txt already exists\n"; + } + else + { + passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings'); + // update osmosis configuration.txt with our settings + passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt'); + passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt'); + } - $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); + // Find the last node in the DB + $iLastOSMID = $oDB->getOne("select max(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."/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;F=1"); + // download.geofabrik.de: 000/