- if ($aCMDResult['import-data'] || $aCMDResult['all']) {
- if (!isset($aCMDResult['osm-file'])) {
- fail('missing --osm-file for data import');
- }
-
- if (!file_exists($aCMDResult['osm-file'])) {
- fail('the path supplied to --osm-file does not exist');
- }
-
- if (!is_readable($aCMDResult['osm-file'])) {
- fail('osm-file "'.$aCMDResult['osm-file'].'" not readable');
- }
- }
-}
-
-function prepSystem($aCMDResult)
-{
- // 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;
- warn("resetting threads to $iInstances");
- }
-
- // Assume we can steal all the cache memory in the box (unless told otherwise)
- if (isset($aCMDResult['osm2pgsql-cache'])) {
- $iCacheMemory = $aCMDResult['osm2pgsql-cache'];
- } else {
- $iCacheMemory = getCacheMemoryMB();
- }
-
- $sModulePath = CONST_Database_Module_Path;
- info('module path: ' . $sModulePath);
-
- return array($iCacheMemory,$iInstances);
-}
-
-function prepDB($aCMDResult)
-{
- $sModulePath = CONST_Database_Module_Path;
- $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
- if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
-
- if ($aCMDResult['create-db'] || $aCMDResult['all']) {
- info('Create DB');
- $bDidSomething = true;
- $oDB = DB::connect(CONST_Database_DSN, false);
- if (!PEAR::isError($oDB)) {
- fail('database already exists ('.CONST_Database_DSN.')');
- }
-
- $sCreateDBCmd = 'createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database'];
- if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
- $sCreateDBCmd .= ' -U ' . $aDSNInfo['username'];
- }
- if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
- $sCreateDBCmd .= ' -h ' . $aDSNInfo['hostspec'];
- }
-
- $aProcEnv = null;
- if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
- $aProcEnv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
- }
-
- $result = runWithEnv($sCreateDBCmd, $aProcEnv);
- if ($result != 0) fail('Error executing external command: '.$sCreateDBCmd);
- }
-
- if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
- info('Setup DB');
- $bDidSomething = true;
-
- $oDB =& getDB();
-
- $fPostgresVersion = getPostgresVersion($oDB);
- echo 'Postgres version found: '.$fPostgresVersion."\n";
-
- if ($fPostgresVersion < 9.1) {
- fail('Minimum supported version of Postgresql is 9.1.');
- }
-
- pgsqlRunScript($aCMDResult, 'CREATE EXTENSION IF NOT EXISTS hstore');
- pgsqlRunScript($aCMDResult, '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 ($iNumFunc == 0) {
- pgsqlRunScript($aCMDResult, "create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable");
- warn('Postgresql is too old. extratags and namedetails API not available.');
- }
-
- $fPostgisVersion = getPostgisVersion($oDB);
- echo 'Postgis version found: '.$fPostgisVersion."\n";
-
- if ($fPostgisVersion < 2.1) {
- // Functions were renamed in 2.1 and throw an annoying deprecation warning
- pgsqlRunScript($aCMDResult, 'ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
- pgsqlRunScript($aCMDResult, 'ALTER FUNCTION ST_Line_Locate_Point(geometry, geometry) RENAME TO ST_LineLocatePoint');
- }
- if ($fPostgisVersion < 2.2) {
- pgsqlRunScript($aCMDResult, '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 ($i == 0) {
- echo "\nERROR: Web user '".CONST_Database_Web_User."' does not exist. Create it with:\n";
- echo "\n createuser ".CONST_Database_Web_User."\n\n";
- exit(1);
- }
-
- if (!checkModulePresence()) {
- fail('error loading nominatim.so module');
- }
-
- if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) {
- echo 'Error: you need to download the country_osm_grid first:';
- echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz https://www.nominatim.org/data/country_grid.sql.gz\n";
- exit(1);
- }
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/country_name.sql');
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/country_naturalearthdata.sql');
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/country_osm_grid.sql.gz');
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/gb_postcode_table.sql');
-
-
- if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/gb_postcode_data.sql.gz');
- } else {
- warn('external UK postcode table not found.');
- }
-
- if (CONST_Use_Extra_US_Postcodes) {
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/us_postcode.sql');
- }
-
- if ($aCMDResult['no-partitions']) {
- pgsqlRunScript($aCMDResult, 'update country_name set partition = 0');
- }
-
- // the following will be needed by create_functions later but
- // is only defined in the subsequently called T
- // Create dummies here that will be overwritten by the proper
- // versions in create-tables.
- pgsqlRunScript($aCMDResult, 'CREATE TABLE IF NOT EXISTS place_boundingbox ()');
- pgsqlRunScript($aCMDResult, 'CREATE TYPE wikipedia_article_match AS ()', false);
- }
- return $aDSNInfo;
-}
-
-function importData($aCMDResult, $iCacheMemory, $aDSNInfo)
-{
- info('Import data');
-
- $osm2pgsql = CONST_Osm2pgsql_Binary;
- if (!file_exists($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) && 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'];
- if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
- $osm2pgsql .= ' -U ' . $aDSNInfo['username'];
- }
- if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
- $osm2pgsql .= ' -H ' . $aDSNInfo['hostspec'];
- }
-
- $aProcEnv = null;
- if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
- $aProcEnv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
- }
-
- $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
- runWithEnv($osm2pgsql, $aProcEnv);
-
- $oDB =& getDB();
- if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) {
- fail('No Data');
- }
-}
-
-function createFunctions($aCMDResult)
-{
- info('Create Functions');
-
- if (!checkModulePresence()) {
- fail('error loading nominatim.so module');
- }
-
- createSqlFunctions($aCMDResult);
-}
-
-function createTables($aCMDResult)
-{
- info('Create Tables');
-
- $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql');
- $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
- $sTemplate = replaceTablespace(
- '{ts:address-data}',
- CONST_Tablespace_Address_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:address-index}',
- CONST_Tablespace_Address_Index,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:search-data}',
- CONST_Tablespace_Search_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:search-index}',
- CONST_Tablespace_Search_Index,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-data}',
- CONST_Tablespace_Aux_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-index}',
- CONST_Tablespace_Aux_Index,
- $sTemplate
- );
- pgsqlRunScript($aCMDResult, $sTemplate, false);
-
- // re-run the functions
- info('Recreate Functions');
- createSqlFunctions($aCMDResult);
-}
-
-function createPartitionTables($aCMDResult)
-{
- info('Create Partition Tables');
-
- $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
- $sTemplate = replaceTablespace(
- '{ts:address-data}',
- CONST_Tablespace_Address_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:address-index}',
- CONST_Tablespace_Address_Index,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:search-data}',
- CONST_Tablespace_Search_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:search-index}',
- CONST_Tablespace_Search_Index,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-data}',
- CONST_Tablespace_Aux_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-index}',
- CONST_Tablespace_Aux_Index,
- $sTemplate
- );
-
- pgsqlRunPartitionScript($aCMDResult, $sTemplate);
-}
-
-function createPartitionFunctions($aCMDResult)
-{
- info('Create Partition Functions');
-
- $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
-
- pgsqlRunPartitionScript($aCMDResult, $sTemplate);
-}
-
-function importWikipediaArticles()
-{
- $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($sWikiArticlesFile);
- } else {
- warn('wikipedia article dump file not found - places will have default importance');
- }
- if (file_exists($sWikiRedirectsFile)) {
- info('Importing wikipedia redirects');
- pgsqlRunDropAndRestore($sWikiRedirectsFile);
- } else {
- warn('wikipedia redirect dump file not found - some place importance values may be missing');
- }
-}
-
-function loadData($aCMDResult, $iInstances)
-{
- info('Drop old Data');
-
- $oDB =& getDB();
- if (!pg_query($oDB->connection, 'TRUNCATE word')) fail(pg_last_error($oDB->connection));
- echo '.';
- if (!pg_query($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($oDB->connection));
- echo '.';
- if (!pg_query($oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($oDB->connection));
- echo '.';
- if (!pg_query($oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($oDB->connection));
- echo '.';
- if (!pg_query($oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($oDB->connection));
- echo '.';
- if (!pg_query($oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($oDB->connection));
- echo '.';
- if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
- echo '.';
-
- $sSQL = 'select distinct partition from country_name';
- $aPartitions = chksql($oDB->getCol($sSQL));
- if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
- foreach ($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
- $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 (!$aCMDResult['disable-token-precalc']) {
- info('Loading word list');
- pgsqlRunScriptFile($aCMDResult, CONST_BasePath.'/data/words.sql');
- }
-
- 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++) {
- $aDBInstances[$i] =& getDB(true);
- $sSQL = "INSERT INTO placex ($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, $sSQL)) {
- fail(pg_last_error($aDBInstances[$i]->connection));
- }
- }
- // last thread for interpolation lines
- $aDBInstances[$iLoadThreads] =& getDB(true);
- $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[$iLoadThreads]->connection, $sSQL)) {
- fail(pg_last_error($aDBInstances[$iLoadThreads]->connection));
- }
-
- $bFailed = false;
- for ($i = 0; $i <= $iLoadThreads; $i++) {
- while (($hPGresult = pg_get_result($aDBInstances[$i]->connection)) !== false) {
- $resultStatus = pg_result_status($hPGresult);
- // PGSQL_EMPTY_QUERY, PGSQL_COMMAND_OK, PGSQL_TUPLES_OK,
- // PGSQL_COPY_OUT, PGSQL_COPY_IN, PGSQL_BAD_RESPONSE,
- // PGSQL_NONFATAL_ERROR and PGSQL_FATAL_ERROR
- echo 'Query result ' . $i . ' is: ' . $resultStatus . "\n";
- if ($resultStatus != PGSQL_COMMAND_OK && $resultStatus != PGSQL_TUPLES_OK) {
- $resultError = pg_result_error($hPGresult);
- echo '-- error text ' . $i . ': ' . $resultError . "\n";
- $bFailed = true;
- }
- }
- }
- if ($bFailed) {
- fail('SQL errors loading placex and/or location_property_osmline tables');
- }
- echo "\n";
- info('Reanalysing database');
- pgsqlRunScript($aCMDResult, '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 $sDatabaseDate.\n";
- }
-}
-
-function importTigerData($aCMDResult, $iInstances)
-{
- info('Import Tiger data');
-
- $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
- $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
- $sTemplate = replaceTablespace(
- '{ts:aux-data}',
- CONST_Tablespace_Aux_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-index}',
- CONST_Tablespace_Aux_Index,
- $sTemplate
- );
- pgsqlRunScript($aCMDResult, $sTemplate, false);
-
- $aDBInstances = array();
- for ($i = 0; $i < $iInstances; $i++) {
- $aDBInstances[$i] =& getDB(true);
- }
-
- foreach (glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) {
- echo $sFile.': ';
- $hFile = fopen($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));
- $sSQL = fgets($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 '.';
- $iLines = 0;
- }
- }
- }
- usleep(10);
- }
-
- fclose($hFile);
-
- $bAnyBusy = true;
- while ($bAnyBusy) {
- $bAnyBusy = false;
- for ($i = 0; $i < $iInstances; $i++) {
- if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
- }
- usleep(10);
- }
- echo "\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 = replaceTablespace(
- '{ts:aux-data}',
- CONST_Tablespace_Aux_Data,
- $sTemplate
- );
- $sTemplate = replaceTablespace(
- '{ts:aux-index}',
- CONST_Tablespace_Aux_Index,
- $sTemplate
- );
- pgsqlRunScript($aCMDResult, $sTemplate, false);
-}
-
-function calculatePostcodes($aCMDResult)
-{
- info('Calculate Postcodes');
- $oDB =& getDB();
- if (!pg_query($oDB->connection, 'TRUNCATE location_postcode')) {
- fail(pg_last_error($oDB->connection));
projects / nominatim.git / blobdiff