$aCMDOptions
= array(
- "Create and setup nominatim search system",
+ 'Create and setup nominatim search system',
array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
- array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"),
+ array('no-partitions', '', 0, 1, 0, 0, 'bool', 'Do not partition search indices (speeds up import of single country extracts)'),
array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
}
-// This is a pretty hard core default - the number of processors in the box - 1
-$iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-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()) {
$iInstances = getProcessorCount();
- echo "WARNING: resetting threads to $iInstances\n";
+ warn("resetting threads to $iInstances");
}
// Assume we can steal all the cache memory in the box (unless told otherwise)
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
if ($aCMDResult['create-db'] || $aCMDResult['all']) {
- echo "Create DB\n";
+ info('Create DB');
$bDidSomething = true;
$oDB = DB::connect(CONST_Database_DSN, false);
if (!PEAR::isError($oDB)) {
}
if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
- echo "Setup DB\n";
+ 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.");
+ fail('Minimum supported version of Postgresql is 9.1.');
}
pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore');
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);
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 $oResult->userinfo."\n\n";
+ exit(1);
+ }
+
if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) {
- echo "Error: you need to download the country_osm_grid first:";
- echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz http://www.nominatim.org/data/country_grid.sql.gz\n";
+ echo 'Error: you need to download the country_osm_grid first:';
+ echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz https://www.nominatim.org/data/country_grid.sql.gz\n";
exit(1);
}
if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
} else {
- echo "WARNING: external UK postcode table not found.\n";
+ warn('external UK postcode table not found.');
}
if (CONST_Use_Extra_US_Postcodes) {
pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
}
if ($aCMDResult['import-data'] || $aCMDResult['all']) {
- echo "Import\n";
+ info('Import data');
$bDidSomething = true;
$osm2pgsql = CONST_Osm2pgsql_Binary;
fail("osm2pgsql not found in '$osm2pgsql'");
}
- if (!is_null(CONST_Osm2pgsql_Flatnode_File)) {
+ if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
$osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
}
if (CONST_Tablespace_Osm2pgsql_Data)
}
if ($aCMDResult['create-functions'] || $aCMDResult['all']) {
- echo "Functions\n";
+ 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']) {
+ 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);
$sTemplate = replace_tablespace(
pgsqlRunScript($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";
+ info('Create Partition Tables');
$bDidSomething = true;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) {
- echo "Partition Functions\n";
+ info('Create Partition Functions');
$bDidSomething = true;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
$sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin';
$sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin';
if (file_exists($sWikiArticlesFile)) {
- echo "Importing wikipedia articles...";
+ info('Importing wikipedia articles');
pgsqlRunDropAndRestore($sWikiArticlesFile);
- echo "...done\n";
} else {
- echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+ warn('wikipedia article dump file not found - places will have default importance');
}
if (file_exists($sWikiRedirectsFile)) {
- echo "Importing wikipedia redirects...";
+ info('Importing wikipedia redirects');
pgsqlRunDropAndRestore($sWikiRedirectsFile);
- echo "...done\n";
} else {
- echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+ warn('wikipedia redirect dump file not found - some place importance values may be missing');
}
}
if ($aCMDResult['load-data'] || $aCMDResult['all']) {
- echo "Drop old Data\n";
+ info('Drop old Data');
$bDidSomething = true;
$oDB =& getDB();
// pre-create the word list
if (!$aCMDResult['disable-token-precalc']) {
- echo "Loading word list\n";
+ 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();
$sSQL = "INSERT INTO placex ($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)";
+ $sSQL .= ' and ST_IsValid(geometry)';
if ($aCMDResult['verbose']) echo "$sSQL\n";
if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) {
fail(pg_last_error($aDBInstances[$i]->connection));
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) {
- echo "WARNING: could not determine database date.\n";
+ warn('could not determine database date.');
} else {
$sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')";
pg_query($oDB->connection, $sSQL);
}
if ($aCMDResult['import-tiger-data']) {
+ info('Import Tiger data');
$bDidSomething = true;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
foreach (glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) {
echo $sFile.': ';
- $hFile = fopen($sFile, "r");
+ $hFile = fopen($sFile, 'r');
$sSQL = fgets($hFile, 100000);
$iLines = 0;
if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
$iLines++;
if ($iLines == 1000) {
- echo ".";
+ echo '.';
$iLines = 0;
}
}
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);
$sTemplate = replace_tablespace(
}
if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) {
+ info('Calculate Postcodes');
$bDidSomething = true;
$oDB =& getDB();
if (!pg_query($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 = '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 .= ' 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";
+ $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 = '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 .= ' 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));
}
+
+ // 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 (!$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));
+ }
+ }
+ $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM ';
+ $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p';
+
+ if (!pg_query($oDB->connection, $sSQL)) {
+ fail(pg_last_error($oDB->connection));
+ }
}
if ($aCMDResult['osmosis-init']) {
$bDidSomething = true;
$sOutputFile = '';
$sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
+ info('Index ranks 0 - 4');
passthruCheckReturn($sBaseCmd.' -R 4');
if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+ info('Index ranks 5 - 25');
passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+ info('Index ranks 26 - 30');
passthruCheckReturn($sBaseCmd.' -r 26');
- echo "Indexing postcodes....\n";
+ info('Index postcodes');
+ $oDB =& getDB();
$sSQL = 'UPDATE location_postcode SET indexed_status = 0';
if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
}
if ($aCMDResult['create-search-indices'] || $aCMDResult['all']) {
- echo "Search indices\n";
+ info('Create Search indices');
$bDidSomething = true;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
+ $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
$sTemplate = replace_tablespace(
'{ts:address-index}',
CONST_Tablespace_Address_Index,
}
if ($aCMDResult['create-country-names'] || $aCMDResult['all']) {
- echo 'Creating search index for default country names';
+ info('Create search index for default country names');
$bDidSomething = true;
pgsqlRunScript("select getorcreate_country(make_standard_name('uk'), 'gb')");
pgsqlRunScript("select getorcreate_country(make_standard_name('united states'), 'us')");
- pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(country_code), country_code) from country_name where country_code is not null) as x");
+ pgsqlRunScript('select count(*) from (select getorcreate_country(make_standard_name(country_code), country_code) from country_name where country_code is not null) as x');
pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(name->'name'), country_code) from country_name where name ? 'name') as x");
$sSQL = 'select count(*) from (select getorcreate_country(make_standard_name(v), country_code) from (select country_code, skeys(name) as k, svals(name) as v from country_name) x where k ';
}
if ($aCMDResult['drop']) {
+ info('Drop tables only required for updates');
// 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
// 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_*"
+ '*columns',
+ 'import_polygon_*',
+ 'import_status',
+ 'place_addressline',
+ 'location_postcode',
+ 'location_property*',
+ 'placex',
+ 'search_name',
+ 'seq_*',
+ 'word',
+ 'query_log',
+ 'new_query_log',
+ 'spatial_ref_sys',
+ 'country_name',
+ 'place_classtype_*'
);
$oDB =& getDB();
// been deleted already by CASCADE
}
- if (!is_null(CONST_Osm2pgsql_Flatnode_File)) {
- if ($aCMDResult['verbose']) echo "deleting ".CONST_Osm2pgsql_Flatnode_File."\n";
+ if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
+ if ($aCMDResult['verbose']) echo 'deleting '.CONST_Osm2pgsql_Flatnode_File."\n";
unlink(CONST_Osm2pgsql_Flatnode_File);
}
}
if (!$bDidSomething) {
showUsage($aCMDOptions, true);
} else {
- echo "Setup finished.\n";
+ echo "Summary of warnings:\n\n";
+ repeatWarnings();
+ echo "\n";
+ info('Setup finished.');
}
function pgsqlRunScriptFile($sFilename)
{
+ global $aCMDResult;
if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
// 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'].' -d '.$aDSNInfo['database'];
+ if (!$aCMDResult['verbose']) {
+ $sCMD .= ' -q';
+ }
$ahGzipPipes = null;
if (preg_match('/\\.gz$/', $sFilename)) {
function pgsqlRunScript($sScript, $bfatal = true)
{
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'].' -d '.$aDSNInfo['database'];
- if ($bfatal && !$aCMDResult['ignore-errors'])
- $sCMD .= ' -v ON_ERROR_STOP=1';
- $aDescriptors = array(
- 0 => array('pipe', 'r'),
- 1 => STDOUT,
- 2 => STDERR
- );
- $ahPipes = null;
- $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes);
- if (!is_resource($hProcess)) fail('unable to start pgsql');
-
- while (strlen($sScript)) {
- $written = fwrite($ahPipes[0], $sScript);
- if ($written <= 0) break;
- $sScript = substr($sScript, $written);
- }
- fclose($ahPipes[0]);
- $iReturn = proc_close($hProcess);
- if ($bfatal && $iReturn > 0) {
- fail("pgsql returned with error code ($iReturn)");
- }
+ runSQLScript(
+ $sScript,
+ $bfatal,
+ $aCMDResult['verbose'],
+ $aCMDResult['ignore-errors']
+ );
}
function pgsqlRunPartitionScript($sTemplate)
projects / nominatim.git / blobdiff