X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/effd8e12af440586594a18a04beb33abed260d15..203b5f7de12ca09d275103eb2eb70fe5a4c20466:/utils/setup.php?ds=sidebyside diff --git a/utils/setup.php b/utils/setup.php index 9dc499be..e2810737 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -5,116 +5,104 @@ require_once(dirname(dirname(__FILE__)).'/settings/settings.php'); require_once(CONST_BasePath.'/lib/init-cmd.php'); ini_set('memory_limit', '800M'); -$aCMDOptions = array( - "Create and setup nominatim search system", - array('help', 'h', 0, 1, 0, 0, false, 'Show Help'), - array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'), - array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'), - - 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('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'), - array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'), - 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-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'), - array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"), - 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('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'), - 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('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('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'), - array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'), -); +$aCMDOptions += array( + 'Create and setup nominatim search system', + array('help', 'h', 0, 1, 0, 0, false, 'Show Help'), + array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'), + array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'), + + 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('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'), + array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'), + 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-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'), + array('no-partitions', '', 0, 1, 0, 0, 'bool', 'Do not partition search indices (speeds up import of single country extracts)'), + 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('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'), + 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('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('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'), + array('create-country-names', '', 0, 1, 0, 0, 'bool', 'Create default list of searchable country names'), + array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'), + ); getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true); $bDidSomething = false; // Check if osm-file is set and points to a valid file if --all or --import-data is given -if ($aCMDResult['import-data'] || $aCMDResult['all']) -{ - if (!isset($aCMDResult['osm-file'])) - { +if ($aCMDResult['import-data'] || $aCMDResult['all']) { + if (!isset($aCMDResult['osm-file'])) { fail('missing --osm-file for data import'); } - if (!file_exists($aCMDResult['osm-file'])) - { + if (!file_exists($aCMDResult['osm-file'])) { fail('the path supplied to --osm-file does not exist'); } - if (!is_readable($aCMDResult['osm-file'])) - { + if (!is_readable($aCMDResult['osm-file'])) { fail('osm-file "'.$aCMDResult['osm-file'].'" not readable'); } } -// This is a pretty hard core default - the number of processors in the box - 1 -$iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1); -if ($iInstances < 1) -{ +// by default, use all but one processor, but never more than 15. +$iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(min(16,getProcessorCount())-1); +if ($iInstances < 1) { $iInstances = 1; - echo "WARNING: resetting threads to $iInstances\n"; + warn("resetting threads to $iInstances"); } -if ($iInstances > getProcessorCount()) -{ +if ($iInstances > getProcessorCount()) { $iInstances = getProcessorCount(); - echo "WARNING: resetting threads to $iInstances\n"; + warn("resetting threads to $iInstances"); } // Assume we can steal all the cache memory in the box (unless told otherwise) -if (isset($aCMDResult['osm2pgsql-cache'])) -{ +if (isset($aCMDResult['osm2pgsql-cache'])) { $iCacheMemory = $aCMDResult['osm2pgsql-cache']; -} -else -{ +} else { $iCacheMemory = getCacheMemoryMB(); } $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; -if ($aCMDResult['create-db'] || $aCMDResult['all']) -{ - echo "Create DB\n"; +if ($aCMDResult['create-db'] || $aCMDResult['all']) { + info('Create DB'); $bDidSomething = true; $oDB = DB::connect(CONST_Database_DSN, false); - if (!PEAR::isError($oDB)) - { + if (!PEAR::isError($oDB)) { fail('database already exists ('.CONST_Database_DSN.')'); } passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']); } -if ($aCMDResult['setup-db'] || $aCMDResult['all']) -{ - echo "Setup DB\n"; +if ($aCMDResult['setup-db'] || $aCMDResult['all']) { + info('Setup DB'); $bDidSomething = true; - // TODO: path detection, detection memory, etc. $oDB =& getDB(); $fPostgresVersion = getPostgresVersion($oDB); echo 'Postgres version found: '.$fPostgresVersion."\n"; - if ($fPostgresVersion < 9.1) - { - fail("Minimum supported version of Postgresql is 9.1."); + if ($fPostgresVersion < 9.1) { + fail('Minimum supported version of Postgresql is 9.1.'); } pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore'); @@ -125,40 +113,63 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) // versions add a dummy function that returns nothing. $iNumFunc = chksql($oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'")); - if ($iNumFunc == 0) - { + if ($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."; + warn('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 + if ($fPostgisVersion < 2.1) { + // Functions were renamed in 2.1 and throw an annoying deprecation warning pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint'); + pgsqlRunScript('ALTER FUNCTION ST_Line_Locate_Point(geometry, geometry) RENAME TO ST_LineLocatePoint'); + } + if ($fPostgisVersion < 2.2) { + pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid'); + } + + $i = chksql($oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'")); + if ($i == 0) { + echo "\nERROR: Web user '".CONST_Database_Web_User."' does not exist. Create it with:\n"; + echo "\n createuser ".CONST_Database_Web_User."\n\n"; + exit(1); + } + + // Try accessing the C module, so we know early if something is wrong + // and can simply error out. + $sSQL = "CREATE FUNCTION nominatim_test_import_func(text) RETURNS text AS '"; + $sSQL .= CONST_InstallPath."/module/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT"; + $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);'; + $oResult = $oDB->query($sSQL); + + if (PEAR::isError($oResult)) { + echo "\nERROR: Failed to load nominatim module. Reason:\n"; + echo $oResult->userinfo."\n\n"; + exit(1); + } + + if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) { + echo 'Error: you need to download the country_osm_grid first:'; + echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz https://www.nominatim.org/data/country_grid.sql.gz\n"; + exit(1); } pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql'); - pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz'); pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql'); - if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) - { + if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) { pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz'); + } else { + warn('external UK postcode table not found.'); } - else - { - echo "WARNING: external UK postcode table not found.\n"; - } - if (CONST_Use_Extra_US_Postcodes) - { + if (CONST_Use_Extra_US_Postcodes) { pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); } - if ($aCMDResult['no-partitions']) - { + if ($aCMDResult['no-partitions']) { pgsqlRunScript('update country_name set partition = 0'); } @@ -170,20 +181,18 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) pgsqlRunScript('create type wikipedia_article_match as ()'); } -if ($aCMDResult['import-data'] || $aCMDResult['all']) -{ - echo "Import\n"; +if ($aCMDResult['import-data'] || $aCMDResult['all']) { + info('Import data'); $bDidSomething = true; $osm2pgsql = CONST_Osm2pgsql_Binary; - if (!file_exists($osm2pgsql)) - { - echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + if (!file_exists($osm2pgsql)) { + echo "Check CONST_Osm2pgsql_Binary in your local settings file.\n"; + echo "Normally you should not need to set this manually.\n"; fail("osm2pgsql not found in '$osm2pgsql'"); } - if (!is_null(CONST_Osm2pgsql_Flatnode_File)) - { + if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) { $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; } if (CONST_Tablespace_Osm2pgsql_Data) @@ -201,72 +210,105 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) passthruCheckReturn($osm2pgsql); $oDB =& getDB(); - if (!chksql($oDB->getRow('select * from place limit 1'))) - { + if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) { fail('No Data'); } } -if ($aCMDResult['create-functions'] || $aCMDResult['all']) -{ - echo "Functions\n"; +if ($aCMDResult['create-functions'] || $aCMDResult['all']) { + info('Create Functions'); $bDidSomething = true; - if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built"); + if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) { + fail('nominatim module not built'); + } create_sql_functions($aCMDResult); } -if ($aCMDResult['create-tables'] || $aCMDResult['all']) -{ +if ($aCMDResult['create-tables'] || $aCMDResult['all']) { + info('Create Tables'); $bDidSomething = true; - 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); + $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); // re-run the functions - echo "Functions\n"; + info('Recreate Functions'); create_sql_functions($aCMDResult); } -if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) -{ - echo "Partition Tables\n"; +if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { + info('Create Partition Tables'); $bDidSomething = true; $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); + $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-partition-functions'] || $aCMDResult['all']) -{ - echo "Partition Functions\n"; +if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { + info('Create Partition Functions'); $bDidSomething = true; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); @@ -274,37 +316,27 @@ if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) pgsqlRunPartitionScript($sTemplate); } -if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) -{ +if ($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..."; + $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin'; + $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin'; + if (file_exists($sWikiArticlesFile)) { + info('Importing wikipedia articles'); pgsqlRunDropAndRestore($sWikiArticlesFile); - echo "...done\n"; + } else { + warn('wikipedia article dump file not found - places will have default importance'); } - else - { - echo "WARNING: wikipedia article dump file not found - places will have default importance\n"; - } - if (file_exists($sWikiRedirectsFile)) - { - echo "Importing wikipedia redirects..."; + if (file_exists($sWikiRedirectsFile)) { + info('Importing wikipedia redirects'); pgsqlRunDropAndRestore($sWikiRedirectsFile); - echo "...done\n"; - } - else - { - echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n"; + } else { + warn('wikipedia redirect dump file not found - some place importance values may be missing'); } } -if ($aCMDResult['load-data'] || $aCMDResult['all']) -{ - echo "Drop old Data\n"; +if ($aCMDResult['load-data'] || $aCMDResult['all']) { + info('Drop old Data'); $bDidSomething = true; $oDB =& getDB(); @@ -332,99 +364,115 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) $sSQL = 'select distinct partition from country_name'; $aPartitions = chksql($oDB->getCol($sSQL)); if (!$aCMDResult['no-partitions']) $aPartitions[] = 0; - foreach($aPartitions as $sPartition) - { + foreach ($aPartitions as $sPartition) { if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection)); echo '.'; } // 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)); + $sSQL = 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS '; + $sSQL .= '$$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE'; + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); + } echo ".\n"; // pre-create the word list - if (!$aCMDResult['disable-token-precalc']) - { - echo "Loading word list\n"; + if (!$aCMDResult['disable-token-precalc']) { + info('Loading word list'); pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql'); } - echo "Load Data\n"; + info('Load Data'); + $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry'; + $aDBInstances = array(); $iLoadThreads = max(1, $iInstances - 1); - for($i = 0; $i < $iLoadThreads; $i++) - { + for ($i = 0; $i < $iLoadThreads; $i++) { $aDBInstances[$i] =& getDB(true); - $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')"; + $sSQL = "INSERT INTO placex ($sColumns) SELECT $sColumns FROM place WHERE osm_id % $iLoadThreads = $i"; + $sSQL .= " and not (class='place' and type='houses' and osm_type='W'"; + $sSQL .= " and ST_GeometryType(geometry) = 'ST_LineString')"; + $sSQL .= ' and ST_IsValid(geometry)'; if ($aCMDResult['verbose']) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) { + fail(pg_last_error($aDBInstances[$i]->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 = 'insert into location_property_osmline'; + $sSQL .= ' (osm_id, address, linegeo)'; + $sSQL .= ' SELECT osm_id, address, 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)); + if (!pg_send_query($aDBInstances[$iLoadThreads]->connection, $sSQL)) { + fail(pg_last_error($aDBInstances[$iLoadThreads]->connection)); + } $bAnyBusy = true; - while($bAnyBusy) - { + while ($bAnyBusy) { $bAnyBusy = false; - for($i = 0; $i <= $iLoadThreads; $i++) - { + for ($i = 0; $i <= $iLoadThreads; $i++) { if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; } sleep(1); echo '.'; } echo "\n"; - echo "Reanalysing database...\n"; + info('Reanalysing database'); pgsqlRunScript('ANALYSE'); + + $sDatabaseDate = getDatabaseDate($oDB); + pg_query($oDB->connection, 'TRUNCATE import_status'); + if ($sDatabaseDate === false) { + warn('could not determine database date.'); + } else { + $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')"; + pg_query($oDB->connection, $sSQL); + echo "Latest data imported from $sDatabaseDate.\n"; + } } -if ($aCMDResult['import-tiger-data']) -{ +if ($aCMDResult['import-tiger-data']) { + info('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); + $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; $i++) - { + for ($i = 0; $i < $iInstances; $i++) { $aDBInstances[$i] =& getDB(true); } - foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) - { + foreach (glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) { echo $sFile.': '; - $hFile = fopen($sFile, "r"); + $hFile = fopen($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[$i]->connection)); + while (true) { + for ($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 "."; + if ($iLines == 1000) { + echo '.'; $iLines = 0; } } @@ -435,11 +483,9 @@ if ($aCMDResult['import-tiger-data']) fclose($hFile); $bAnyBusy = true; - while($bAnyBusy) - { + while ($bAnyBusy) { $bAnyBusy = false; - for($i = 0; $i < $iInstances; $i++) - { + for ($i = 0; $i < $iInstances; $i++) { if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true; } usleep(10); @@ -447,161 +493,157 @@ if ($aCMDResult['import-tiger-data']) echo "\n"; } - echo "Creating indexes\n"; + info('Creating indexes on Tiger data'); $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); + $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'] || $aCMDResult['all']) -{ +if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) { + info('Calculate Postcodes'); $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,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 "; - $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; - if (!pg_query($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, 'TRUNCATE location_postcode')) { + fail(pg_last_error($oDB->connection)); + } + + $sSQL = 'INSERT INTO location_postcode'; + $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; + $sSQL .= "SELECT nextval('seq_place'), 1, country_code,"; + $sSQL .= " upper(trim (both ' ' from address->'postcode')) as pc,"; + $sSQL .= ' ST_Centroid(ST_Collect(ST_Centroid(geometry)))'; + $sSQL .= ' FROM placex'; + $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'"; + $sSQL .= ' AND geometry IS NOT null'; + $sSQL .= ' GROUP BY country_code, pc'; + + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); + } + + if (CONST_Use_Extra_US_Postcodes) { + // only add postcodes that are not yet available in OSM + $sSQL = 'INSERT INTO location_postcode'; + $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; + $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,"; + $sSQL .= ' ST_SetSRID(ST_Point(x,y),4326)'; + $sSQL .= ' FROM us_postcode WHERE postcode NOT IN'; + $sSQL .= ' (SELECT postcode FROM location_postcode'; + $sSQL .= " WHERE country_code = 'us')"; if (!pg_query($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 -{ - $bDidSomething = true; - $oDB =& getDB(); + // add missing postcodes for GB (if available) + $sSQL = 'INSERT INTO location_postcode'; + $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) '; + $sSQL .= "SELECT nextval('seq_place'), 1, 'gb', postcode, geometry"; + $sSQL .= ' FROM gb_postcode WHERE postcode NOT IN'; + $sSQL .= ' (SELECT postcode FROM location_postcode'; + $sSQL .= " WHERE country_code = 'gb')"; + if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); - 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 (!$aCMDResult['all']) - { - fail("osmosis not found in '".CONST_Osmosis_Binary."'"); + if (!$aCMDResult['all']) { + $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'"; + $sSQL .= 'and word NOT IN (SELECT postcode FROM location_postcode)'; + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); } } - 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'); - } + $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM '; + $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p'; - // 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/