X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/b8db117bd0211cc974cb9e3468d431adb6da21c6..5c187e131a372d0cbdbba3372668862e3f1c47f9:/utils/setup.php?ds=inline diff --git a/utils/setup.php b/utils/setup.php index db15aa44..992034c7 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -1,7 +1,8 @@ #!/usr/bin/php -Cq getTotalMemoryMB()) + if (isset($aCMDResult['osm2pgsql-cache'])) { - $iCacheMemory = getCacheMemoryMB(); - echo "WARNING: resetting cache memory to $iCacheMemory\n"; + $iCacheMemory = $aCMDResult['osm2pgsql-cache']; } - - if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date'])) + else { - $sBaseFile = basename($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'; - } + $iCacheMemory = getCacheMemoryMB(); } + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; @@ -76,38 +93,81 @@ { 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.')'); } - passthru('createdb -E UTF-8 '.$aDSNInfo['database']); + passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']); } - if ($aCMDResult['create-db'] || $aCMDResult['all']) + if ($aCMDResult['setup-db'] || $aCMDResult['all']) { - echo "Create DB (2)\n"; + echo "Setup DB\n"; $bDidSomething = true; // TODO: path detection, detection memory, etc. $oDB =& getDB(); - passthru('createlang plpgsql '.$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'); + + $fPostgresVersion = getPostgresVersion($oDB); + echo 'Postgres version found: '.$fPostgresVersion."\n"; + + if ($fPostgresVersion < 9.1) + { + fail("Minimum supported version of Postgresql is 9.1."); + } + + pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); + 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."; + } + + $fPostgisVersion = getPostgisVersion($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']) @@ -116,103 +176,135 @@ $bDidSomething = true; $osm2pgsql = CONST_Osm2pgsql_Binary; - if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql"); - $osm2pgsql .= ' -lsc -O gazetteer --hstore'; + if (!file_exists($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'].' '.$aCMDResult['osm-file']; - passthru($osm2pgsql); + 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']) { 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 (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built"); + create_sql_functions($aCMDResult); } - if ($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'); + 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); - 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 "; - $sScript .= "select place_id as place_id,geometry as centroid from placex limit 0;\n"; + // re-run the functions + echo "Functions\n"; + create_sql_functions($aCMDResult); + } - $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_centroid "; - $sScript .= "ON place_classtype_".$sClass."_".$sType." USING GIST (centroid);\n"; + if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) + { + echo "Partition Tables\n"; + $bDidSomething = true; - $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id "; - $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n"; - } - fclose($hFile); - pgsqlRunScript($sScript); + $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, $sTemplate); + + pgsqlRunPartitionScript($sTemplate); } - if ($aCMDResult['create-tables'] || $aCMDResult['all']) + + if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { - echo "Tables\n"; + echo "Partition Functions\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); + $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); + + pgsqlRunPartitionScript($sTemplate); } - if ($aCMDResult['create-partitions'] || $aCMDResult['all']) + if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) { - 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)) + $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin'; + $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin'; + if (file_exists($sWikiArticlesFile)) { - fail($aPartitions->getMessage()); + echo "Importing wikipedia articles..."; + pgsqlRunDropAndRestore($sWikiArticlesFile); + echo "...done\n"; } - $aPartitions[] = 0; - - $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql'); - preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER); - foreach($aMatches as $aMatch) + else { - $sResult = ''; - foreach($aPartitions as $sPartitionName) - { - $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]); - } - $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate); + echo "WARNING: wikipedia article dump file not found - places will have default importance\n"; + } + if (file_exists($sWikiRedirectsFile)) + { + echo "Importing wikipedia redirects..."; + pgsqlRunDropAndRestore($sWikiRedirectsFile); + echo "...done\n"; + } + else + { + echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n"; } - - pgsqlRunScript($sTemplate); } + if ($aCMDResult['load-data'] || $aCMDResult['all']) { - echo "Load Data\n"; + echo "Drop old Data\n"; $bDidSomething = true; $oDB =& getDB(); @@ -220,6 +312,8 @@ echo '.'; if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection)); echo '.'; + if (!pg_query($oDB->connection, 'TRUNCATE location_property_osmline')) 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($oDB->connection)); @@ -235,52 +329,52 @@ if (!pg_query($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) + $sSQL = 'select distinct partition from country_name'; + $aPartitions = chksql($oDB->getCol($sSQL)); + if (!$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++) + $iLoadThreads = max(1, $iInstances - 1); + for($i = 0; $i < $iLoadThreads; $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, '; + $sSQL .= 'geometry) select * from place where osm_id % '.$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 < $iInstances; $i++) + for($i = 0; $i <= $iLoadThreads; $i++) { if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; } @@ -288,19 +382,29 @@ echo '.'; } echo "\n"; + echo "Reanalysing database...\n"; + pgsqlRunScript('ANALYSE'); } if ($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"); @@ -329,7 +433,7 @@ } fclose($hFile); - + $bAnyBusy = true; while($bAnyBusy) { @@ -342,6 +446,15 @@ } 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']) @@ -349,115 +462,243 @@ $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'); - - $sDate = $aCMDResult['osmosis-init-date']; - $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, $sSQL); + if (!file_exists(CONST_Osmosis_Binary)) + { + echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + if (!$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'); + } + // 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: 000/26-Feb-2013 11:53 + // planet.openstreetmap.org: 273/ 2013-03-11 07:41 - + preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER); + if ($aRepMatches) + { + $aPrevRepMatch = false; + foreach($aRepMatches as $aRepMatch) + { + if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break; + $aPrevRepMatch = $aRepMatch; + } + if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch; + + $sRepURL .= $aRepMatch[1]; + $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1"); + preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [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."?C=M;O=D;F=1"); + preg_match_all('#([0-9]{3}).state.txt\s*([-0-9a-zA-Z]+ [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_InstallPath.'/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, $sSQL); + } + else + { + if (!$aCMDResult['all']) + { + fail("Cannot read state file directory."); + } + } + } } 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; - passthru($sBaseCmd.' -R 4'); - pgsqlRunScript('ANALYSE'); - passthru($sBaseCmd.' -r 5 -R 25'); - pgsqlRunScript('ANALYSE'); - passthru($sBaseCmd.' -r 26'); + $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile; + passthruCheckReturn($sBaseCmd.' -R 4'); + if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE'); + passthruCheckReturn($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); - } + $sTemplate = replace_tablespace('{ts:address-index}', + CONST_Tablespace_Address_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:search-index}', + CONST_Tablespace_Search_Index, $sTemplate); + $sTemplate = replace_tablespace('{ts:aux-index}', + CONST_Tablespace_Aux_Index, $sTemplate); pgsqlRunScript($sTemplate); } - if (isset($aCMDResult['create-website'])) + if ($aCMDResult['drop']) { + // The implementation is potentially a bit dangerous because it uses + // a positive selection of tables to keep, and deletes everything else. + // Including any tables that the unsuspecting user might have manually + // created. USE AT YOUR OWN PERIL. $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"; + + // tables we want to keep. everything else goes. + $aKeepTables = array( + "*columns", + "import_polygon_*", + "import_status", + "place_addressline", + "location_property*", + "placex", + "search_name", + "seq_*", + "word", + "query_log", + "new_query_log", + "gb_postcode", + "spatial_ref_sys", + "country_name", + "place_classtype_*" + ); + + $oDB =& getDB(); + $aDropTables = array(); + $aHaveTables = chksql($oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'")); + + foreach($aHaveTables as $sTable) + { + $bFound = false; + foreach ($aKeepTables as $sKeep) + { + if (fnmatch($sKeep, $sTable)) + { + $bFound = true; + break; + } + } + if (!$bFound) array_push($aDropTables, $sTable); + } + + foreach ($aDropTables as $sDrop) + { + if ($aCMDResult['verbose']) echo "dropping table $sDrop\n"; + @pg_query($oDB->connection, "DROP TABLE $sDrop CASCADE"); + // ignore warnings/errors as they might be caused by a table having + // been deleted already by CASCADE + } + + if (!is_null(CONST_Osm2pgsql_Flatnode_File)) + { + if ($aCMDResult['verbose']) echo "deleting ".CONST_Osm2pgsql_Flatnode_File."\n"; + unlink(CONST_Osm2pgsql_Flatnode_File); + } } if (!$bDidSomething) { showUsage($aCMDOptions, true); } + else + { + echo "Setup finished.\n"; + } function pgsqlRunScriptFile($sFilename) { if (!file_exists($sFilename)) fail('unable to find '.$sFilename); - // Convert database DSN to psql paramaters + // Convert database DSN to psql parameters $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']; + + $ahGzipPipes = null; + if (preg_match('/\\.gz$/', $sFilename)) + { + $aDescriptors = array( + 0 => array('pipe', 'r'), + 1 => array('pipe', 'w'), + 2 => array('file', '/dev/null', 'a') + ); + $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes); + if (!is_resource($hGzipProcess)) fail('unable to start zcat'); + $aReadPipe = $ahGzipPipes[1]; + fclose($ahGzipPipes[0]); + } + else + { + $sCMD .= ' -f '.$sFilename; + $aReadPipe = array('pipe', 'r'); + } $aDescriptors = array( - 0 => array('pipe', 'r'), + 0 => $aReadPipe, 1 => array('pipe', 'w'), 2 => array('file', '/dev/null', 'a') ); @@ -465,7 +706,6 @@ $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes); if (!is_resource($hProcess)) fail('unable to start pgsql'); - fclose($ahPipes[0]); // TODO: error checking while(!feof($ahPipes[1])) @@ -474,15 +714,28 @@ } fclose($ahPipes[1]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } + if ($ahGzipPipes) + { + fclose($ahGzipPipes[1]); + proc_close($hGzipProcess); + } + } - function pgsqlRunScript($sScript) + function pgsqlRunScript($sScript, $bfatal = true) { - // Convert database DSN to psql paramaters + global $aCMDResult; + // Convert database DSN to psql parameters $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; - $sCMD = 'psql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']; + $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database']; + if ($bfatal && !$aCMDResult['ignore-errors']) + $sCMD .= ' -v ON_ERROR_STOP=1'; $aDescriptors = array( 0 => array('pipe', 'r'), 1 => STDOUT, @@ -495,8 +748,139 @@ while(strlen($sScript)) { $written = fwrite($ahPipes[0], $sScript); + if ($written <= 0) break; $sScript = substr($sScript, $written); } fclose($ahPipes[0]); - proc_close($hProcess); + $iReturn = proc_close($hProcess); + if ($bfatal && $iReturn > 0) + { + fail("pgsql returned with error code ($iReturn)"); + } } + + 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 + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($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'), + 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'); + + fclose($ahPipes[0]); + + // TODO: error checking + while(!feof($ahPipes[1])) + { + echo fread($ahPipes[1], 4096); + } + fclose($ahPipes[1]); + + $iReturn = proc_close($hProcess); + } + + function pgsqlRunDropAndRestore($sDumpFile) + { + // Convert database DSN to psql parameters + $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($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'); + + fclose($ahPipes[0]); + + // TODO: error checking + while(!feof($ahPipes[1])) + { + echo fread($ahPipes[1], 4096); + } + fclose($ahPipes[1]); + + $iReturn = proc_close($hProcess); + } + + function passthruCheckReturn($cmd) + { + $result = -1; + passthru($cmd, $result); + if ($result != 0) fail('Error executing external command: '.$cmd); + } + + function replace_tablespace($sTemplate, $sTablespace, $sSql) + { + if ($sTablespace) + $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"', + $sSql); + else + $sSql = str_replace($sTemplate, '', $sSql); + + 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); + + } +