$iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
if ($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);
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');
}
+ 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";
+ 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')) {
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;
$osm2pgsql = CONST_Osm2pgsql_Binary;
}
if ($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';
$aDBInstances = array();
for ($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' and ST_GeometryType(geometry) = 'ST_LineString')";
+ $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));
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']) {
+ info('Import Tiger data');
$bDidSomething = true;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
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, '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 .= "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 ";
- $sSQL .= "where ST_Point(x,y) is not null";
- if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+ 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) {
- $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";
+ // 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 (!$aCMDResult['all']) {
- fail("osmosis not found in '".CONST_Osmosis_Binary."'");
- }
- } else {
- if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) {
- echo "settings/configuration.txt already exists\n";
- } else {
- passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings');
- // update osmosis configuration.txt with our settings
- passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt');
- passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt');
+ if (!$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";
- // 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: <a href="000/">000/</a></td><td align="right">26-Feb-2013 11:53 </td>
- // planet.openstreetmap.org: <a href="273/">273/</a> 2013-03-11 07:41 -
- preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\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('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\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('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>\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 (!pg_query($oDB->connection, $sSQL)) {
+ fail(pg_last_error($oDB->connection));
}
}
+if ($aCMDResult['osmosis-init']) {
+ $bDidSomething = true;
+ echo "Command 'osmosis-init' no longer available, please use utils/update.php --init-updates.\n";
+}
+
if ($aCMDResult['index'] || $aCMDResult['all']) {
$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');
+
+ 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;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
}
if ($aCMDResult['create-country-names'] || $aCMDResult['all']) {
- echo 'Creating search index for default country names';
+ info('Create search index for default country names');
$bDidSomething = true;
pgsqlRunScript("select getorcreate_country(make_standard_name('uk'), 'gb')");
}
if ($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
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)) {
$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
$sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+ if (!$aCMDResult['verbose']) {
+ $sCMD .= ' -q';
+ }
if ($bfatal && !$aCMDResult['ignore-errors'])
$sCMD .= ' -v ON_ERROR_STOP=1';
$aDescriptors = array(
projects / nominatim.git / blobdiff