3 require_once(CONST_BasePath.'/lib/init-cmd.php');
4 require_once(CONST_BasePath.'/lib/setup_functions.php');
5 require_once(CONST_BasePath.'/lib/setup/SetupClass.php');
6 require_once(CONST_BasePath.'/lib/setup/AddressLevelParser.php');
8 ini_set('memory_limit', '800M');
10 use Nominatim\Setup\SetupFunctions as SetupFunctions;
12 // (long-opt, short-opt, min-occurs, max-occurs, num-arguments, num-arguments, type, help)
15 'Import / update / index osm data',
16 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
17 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
18 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
20 array('init-updates', '', 0, 1, 0, 0, 'bool', 'Set up database for updating'),
21 array('check-for-updates', '', 0, 1, 0, 0, 'bool', 'Check if new updates are available'),
22 array('no-update-functions', '', 0, 1, 0, 0, 'bool', 'Do not update trigger functions to support differential updates (assuming the diff update logic is already present)'),
23 array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import updates once'),
24 array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import updates forever'),
25 array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
27 array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Update postcode centroid table'),
29 array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
30 array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
31 array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
33 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
34 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
35 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
36 array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
38 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
39 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
40 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
42 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
43 array('recompute-word-counts', '', 0, 1, 0, 0, 'bool', 'Compute frequency of full-word search terms'),
44 array('update-address-levels', '', 0, 1, 0, 0, 'bool', 'Reimport address level configuration (EXPERT)'),
45 array('recompute-importance', '', 0, 1, 0, 0, 'bool', 'Recompute place importances'),
46 array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolete)'),
49 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
51 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
52 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
54 date_default_timezone_set('Etc/UTC');
56 $oDB = new Nominatim\DB();
58 $fPostgresVersion = $oDB->getPostgresVersion();
60 $aDSNInfo = Nominatim\DB::parseDSN(CONST_Database_DSN);
61 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
63 // cache memory to be used by osm2pgsql, should not be more than the available memory
64 $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
65 if ($iCacheMemory + 500 > getTotalMemoryMB()) {
66 $iCacheMemory = getCacheMemoryMB();
67 echo "WARNING: resetting cache memory to $iCacheMemory\n";
70 $oOsm2pgsqlCmd = (new \Nominatim\Shell(CONST_Osm2pgsql_Binary))
71 ->addParams('--hstore')
72 ->addParams('--latlong')
73 ->addParams('--append')
75 ->addParams('--number-processes', 1)
76 ->addParams('--cache', $iCacheMemory)
77 ->addParams('--output', 'gazetteer')
78 ->addParams('--style', CONST_Import_Style)
79 ->addParams('--database', $aDSNInfo['database'])
80 ->addParams('--port', $aDSNInfo['port']);
82 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
83 $oOsm2pgsqlCmd->addParams('--host', $aDSNInfo['hostspec']);
85 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
86 $oOsm2pgsqlCmd->addParams('--user', $aDSNInfo['username']);
88 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
89 $oOsm2pgsqlCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
91 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
92 $oOsm2pgsqlCmd->addParams('--flat-nodes', CONST_Osm2pgsql_Flatnode_File);
94 if ($fPostgresVersion >= 11.0) {
95 $oOsm2pgsqlCmd->addEnvPair('PGOPTIONS',
96 '-c jit=off -c max_parallel_workers_per_gather=0');
100 $oIndexCmd = (new \Nominatim\Shell(CONST_BasePath.'/nominatim/nominatim.py'))
101 ->addParams('--database', $aDSNInfo['database'])
102 ->addParams('--port', $aDSNInfo['port'])
103 ->addParams('--threads', $aResult['index-instances']);
105 if ($aResult['verbose']) {
106 $oIndexCmd->addParams('--verbose');
108 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
109 $oIndexCmd->addParams('--host', $aDSNInfo['hostspec']);
111 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
112 $oIndexCmd->addParams('--username', $aDSNInfo['username']);
114 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
115 $oIndexCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
119 if ($aResult['init-updates']) {
120 // sanity check that the replication URL is correct
121 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
122 if ($sBaseState === false) {
123 echo "\nCannot find state.txt file at the configured replication URL.\n";
124 echo "Does the URL point to a directory containing OSM update data?\n\n";
125 fail('replication URL not reachable.');
127 // sanity check for pyosmium-get-changes
128 if (!CONST_Pyosmium_Binary) {
129 echo "\nCONST_Pyosmium_Binary not configured.\n";
130 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
131 echo "in your local settings file.\n\n";
132 fail('CONST_Pyosmium_Binary not configured');
136 $oCMD = new \Nominatim\Shell(CONST_Pyosmium_Binary, '--help');
137 exec($oCMD->escapedCmd(), $aOutput, $iRet);
140 echo "Cannot execute pyosmium-get-changes.\n";
141 echo "Make sure you have pyosmium installed correctly\n";
142 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
143 fail('pyosmium-get-changes not found or not usable');
146 if (!$aResult['no-update-functions']) {
147 // instantiate setupClass to use the function therein
148 $cSetup = new SetupFunctions(array(
149 'enable-diff-updates' => true,
150 'verbose' => $aResult['verbose']
153 $cSetup->createFunctions();
156 $sDatabaseDate = getDatabaseDate($oDB);
157 if (!$sDatabaseDate) {
158 fail('Cannot determine date of database.');
160 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
162 // get the appropriate state id
164 $oCMD = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
165 ->addParams('--start-date', $sWindBack)
166 ->addParams('--server', CONST_Replication_Url);
168 exec($oCMD->escapedCmd(), $aOutput, $iRet);
169 if ($iRet != 0 || $aOutput[0] == 'None') {
170 fail('Error running pyosmium tools');
173 $oDB->exec('TRUNCATE import_status');
174 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
175 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
179 } catch (\Nominatim\DatabaseError $e) {
180 fail('Could not enter sequence into database.');
183 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
186 if ($aResult['check-for-updates']) {
187 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
189 if (!$aLastState['sequence_id']) {
190 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
193 $oCmd = (new \Nominatim\Shell(CONST_BasePath.'/utils/check_server_for_updates.py'))
194 ->addParams(CONST_Replication_Url)
195 ->addParams($aLastState['sequence_id']);
196 $iRet = $oCmd->run();
201 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
202 // import diffs and files directly (e.g. from osmosis --rri)
203 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
205 if (!file_exists($sNextFile)) {
206 fail("Cannot open $sNextFile\n");
210 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sNextFile);
211 echo $oCMD->escapedCmd()."\n";
212 $iRet = $oCMD->run();
215 fail("Error from osm2pgsql, $iRet\n");
218 // Don't update the import status - we don't know what this file contains
221 if ($aResult['calculate-postcodes']) {
222 info('Update postcodes centroids');
223 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
224 runSQLScript($sTemplate, true, true);
227 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
229 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
231 if (isset($aResult['import-node']) && $aResult['import-node']) {
233 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
235 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
239 if (isset($aResult['import-way']) && $aResult['import-way']) {
241 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
243 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
247 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
249 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
251 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
256 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
261 // import generated change file
263 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sTemporaryFile);
264 echo $oCMD->escapedCmd()."\n";
266 $iRet = $oCMD->run();
268 fail("osm2pgsql exited with error level $iRet\n");
272 if ($aResult['deduplicate']) {
273 $oDB = new Nominatim\DB();
276 if ($oDB->getPostgresVersion() < 9.3) {
277 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
280 $sSQL = 'select partition from country_name order by country_code';
281 $aPartitions = $oDB->getCol($sSQL);
284 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
285 foreach ($aPartitions as $i => $sPartition) {
286 $sSQL = 'select count(*) from search_name_'.$sPartition;
287 $nEntries = $oDB->getOne($sSQL);
288 if ($nEntries == 0) {
289 unset($aPartitions[$i]);
293 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
294 $sSQL .= ' and class is null and type is null and country_code is null';
295 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
296 $aDuplicateTokens = $oDB->getAll($sSQL);
297 foreach ($aDuplicateTokens as $aToken) {
298 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
299 echo 'Deduping '.$aToken['word_token']."\n";
300 $sSQL = 'select word_id,';
301 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
302 $sSQL .= " from word where word_token = '".$aToken['word_token'];
303 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
304 $aTokenSet = $oDB->getAll($sSQL);
306 $aKeep = array_shift($aTokenSet);
307 $iKeepID = $aKeep['word_id'];
309 foreach ($aTokenSet as $aRemove) {
310 $sSQL = 'update search_name set';
311 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
312 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
313 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
316 $sSQL = 'update search_name set';
317 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
318 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
321 $sSQL = 'update location_area_country set';
322 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
323 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
326 foreach ($aPartitions as $sPartition) {
327 $sSQL = 'update search_name_'.$sPartition.' set';
328 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
329 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
332 $sSQL = 'update location_area_country set';
333 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
334 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
338 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
344 if ($aResult['recompute-word-counts']) {
345 info('Recompute frequency of full-word search terms');
346 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
347 runSQLScript($sTemplate, true, true);
350 if ($aResult['index']) {
351 $oCmd = (clone $oIndexCmd)
352 ->addParams('--minrank', $aResult['index-rank']);
354 // echo $oCmd->escapedCmd()."\n";
357 $oDB->exec('update import_status set indexed = true');
360 if ($aResult['update-address-levels']) {
361 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
362 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
363 $oAlParser->createTable($oDB, 'address_levels');
366 if ($aResult['recompute-importance']) {
367 echo "Updating importance values for database.\n";
368 $oDB = new Nominatim\DB();
371 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
372 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
373 $sSQL .= ' (SELECT wikipedia, importance';
374 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
375 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
376 $sSQL .= ' FROM placex d';
377 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
378 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
379 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
383 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
385 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
386 fail('Error: Update interval too low for download.geofabrik.de. ' .
387 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
390 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
392 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
393 ->addParams('--server', CONST_Replication_Url)
394 ->addParams('--outfile', $sImportFile)
395 ->addParams('--size', CONST_Replication_Max_Diff_size);
397 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
400 $fStartTime = time();
401 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
403 if (!$aLastState['sequence_id']) {
404 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
408 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
410 $sBatchEnd = $aLastState['lastimportdate'];
411 $iEndSequence = $aLastState['sequence_id'];
413 if ($aLastState['indexed']) {
414 // Sleep if the update interval has not yet been reached.
415 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
416 if ($fNextUpdate > $fStartTime) {
417 $iSleepTime = $fNextUpdate - $fStartTime;
418 echo "Waiting for next update for $iSleepTime sec.";
422 // Download the next batch of changes.
424 $fCMDStartTime = time();
425 $iNextSeq = (int) $aLastState['sequence_id'];
428 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
429 echo $oCMD->escapedCmd()."\n";
430 if (file_exists($sImportFile)) {
431 unlink($sImportFile);
433 exec($oCMD->escapedCmd(), $aOutput, $iResult);
436 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
437 sleep(CONST_Replication_Recheck_Interval);
438 } elseif ($iResult != 0) {
439 echo 'ERROR: updates failed.';
442 $iEndSequence = (int)$aOutput[0];
446 // get the newest object from the diff file
449 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
450 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
452 echo "Diff file is empty. skipping import.\n";
453 if (!$aResult['import-osmosis-all']) {
460 fail('Error getting date from diff file.');
462 $sBatchEnd = $sBatchEnd[0];
465 $fCMDStartTime = time();
468 echo $oCMDImport->escapedCmd()."\n";
470 $iErrorLevel = $oCMDImport->run();
472 echo "Error executing osm2pgsql: $iErrorLevel\n";
476 // write the update logs
477 $iFileSize = filesize($sImportFile);
478 $sSQL = 'INSERT INTO import_osmosis_log';
479 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
480 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
481 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
482 $sSQL .= date('Y-m-d H:i:s')."','import')";
487 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
490 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
494 if (!$aResult['no-index']) {
495 $oThisIndexCmd = clone($oIndexCmd);
496 $fCMDStartTime = time();
498 echo $oThisIndexCmd->escapedCmd()."\n";
499 $iErrorLevel = $oThisIndexCmd->run();
501 echo "Error: $iErrorLevel\n";
505 $sSQL = 'INSERT INTO import_osmosis_log';
506 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
507 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
508 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
509 $sSQL .= date('Y-m-d H:i:s')."','index')";
512 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
514 $sSQL = 'update import_status set indexed = true';
517 if ($aResult['import-osmosis-all']) {
518 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
523 $fDuration = time() - $fStartTime;
524 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
525 if (!$aResult['import-osmosis-all']) exit(0);