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(
97 '-c jit=off -c max_parallel_workers_per_gather=0'
102 $oIndexCmd = (new \Nominatim\Shell(CONST_BasePath.'/nominatim/nominatim.py'))
103 ->addParams('--database', $aDSNInfo['database'])
104 ->addParams('--port', $aDSNInfo['port'])
105 ->addParams('--threads', $aResult['index-instances']);
107 if ($aResult['verbose']) {
108 $oIndexCmd->addParams('--verbose');
110 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
111 $oIndexCmd->addParams('--host', $aDSNInfo['hostspec']);
113 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
114 $oIndexCmd->addParams('--username', $aDSNInfo['username']);
116 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
117 $oIndexCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
121 if ($aResult['init-updates']) {
122 // sanity check that the replication URL is correct
123 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
124 if ($sBaseState === false) {
125 echo "\nCannot find state.txt file at the configured replication URL.\n";
126 echo "Does the URL point to a directory containing OSM update data?\n\n";
127 fail('replication URL not reachable.');
129 // sanity check for pyosmium-get-changes
130 if (!CONST_Pyosmium_Binary) {
131 echo "\nCONST_Pyosmium_Binary not configured.\n";
132 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
133 echo "in your local settings file.\n\n";
134 fail('CONST_Pyosmium_Binary not configured');
138 $oCMD = new \Nominatim\Shell(CONST_Pyosmium_Binary, '--help');
139 exec($oCMD->escapedCmd(), $aOutput, $iRet);
142 echo "Cannot execute pyosmium-get-changes.\n";
143 echo "Make sure you have pyosmium installed correctly\n";
144 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
145 fail('pyosmium-get-changes not found or not usable');
148 if (!$aResult['no-update-functions']) {
149 // instantiate setupClass to use the function therein
150 $cSetup = new SetupFunctions(array(
151 'enable-diff-updates' => true,
152 'verbose' => $aResult['verbose']
154 $cSetup->createFunctions();
157 $sDatabaseDate = getDatabaseDate($oDB);
158 if (!$sDatabaseDate) {
159 fail('Cannot determine date of database.');
161 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
163 // get the appropriate state id
165 $oCMD = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
166 ->addParams('--start-date', $sWindBack)
167 ->addParams('--server', CONST_Replication_Url);
169 exec($oCMD->escapedCmd(), $aOutput, $iRet);
170 if ($iRet != 0 || $aOutput[0] == 'None') {
171 fail('Error running pyosmium tools');
174 $oDB->exec('TRUNCATE import_status');
175 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
176 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
180 } catch (\Nominatim\DatabaseError $e) {
181 fail('Could not enter sequence into database.');
184 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
187 if ($aResult['check-for-updates']) {
188 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
190 if (!$aLastState['sequence_id']) {
191 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
194 $oCmd = (new \Nominatim\Shell(CONST_BasePath.'/utils/check_server_for_updates.py'))
195 ->addParams(CONST_Replication_Url)
196 ->addParams($aLastState['sequence_id']);
197 $iRet = $oCmd->run();
202 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
203 // import diffs and files directly (e.g. from osmosis --rri)
204 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
206 if (!file_exists($sNextFile)) {
207 fail("Cannot open $sNextFile\n");
211 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sNextFile);
212 echo $oCMD->escapedCmd()."\n";
213 $iRet = $oCMD->run();
216 fail("Error from osm2pgsql, $iRet\n");
219 // Don't update the import status - we don't know what this file contains
222 if ($aResult['calculate-postcodes']) {
223 info('Update postcodes centroids');
224 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
225 runSQLScript($sTemplate, true, true);
228 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
230 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
232 if (isset($aResult['import-node']) && $aResult['import-node']) {
234 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
236 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
240 if (isset($aResult['import-way']) && $aResult['import-way']) {
242 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
244 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
248 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
250 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
252 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
257 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
262 // import generated change file
264 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sTemporaryFile);
265 echo $oCMD->escapedCmd()."\n";
267 $iRet = $oCMD->run();
269 fail("osm2pgsql exited with error level $iRet\n");
273 if ($aResult['deduplicate']) {
274 $oDB = new Nominatim\DB();
277 if ($oDB->getPostgresVersion() < 9.3) {
278 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
281 $sSQL = 'select partition from country_name order by country_code';
282 $aPartitions = $oDB->getCol($sSQL);
285 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
286 foreach ($aPartitions as $i => $sPartition) {
287 $sSQL = 'select count(*) from search_name_'.$sPartition;
288 $nEntries = $oDB->getOne($sSQL);
289 if ($nEntries == 0) {
290 unset($aPartitions[$i]);
294 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
295 $sSQL .= ' and class is null and type is null and country_code is null';
296 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
297 $aDuplicateTokens = $oDB->getAll($sSQL);
298 foreach ($aDuplicateTokens as $aToken) {
299 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
300 echo 'Deduping '.$aToken['word_token']."\n";
301 $sSQL = 'select word_id,';
302 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
303 $sSQL .= " from word where word_token = '".$aToken['word_token'];
304 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
305 $aTokenSet = $oDB->getAll($sSQL);
307 $aKeep = array_shift($aTokenSet);
308 $iKeepID = $aKeep['word_id'];
310 foreach ($aTokenSet as $aRemove) {
311 $sSQL = 'update search_name set';
312 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
313 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
314 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
317 $sSQL = 'update search_name set';
318 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
319 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
322 $sSQL = 'update location_area_country set';
323 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
324 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
327 foreach ($aPartitions as $sPartition) {
328 $sSQL = 'update search_name_'.$sPartition.' set';
329 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
330 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
333 $sSQL = 'update location_area_country set';
334 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
335 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
339 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
345 if ($aResult['recompute-word-counts']) {
346 info('Recompute frequency of full-word search terms');
347 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
348 runSQLScript($sTemplate, true, true);
351 if ($aResult['index']) {
352 $oCmd = (clone $oIndexCmd)
353 ->addParams('--minrank', $aResult['index-rank']);
355 // echo $oCmd->escapedCmd()."\n";
358 $oDB->exec('update import_status set indexed = true');
361 if ($aResult['update-address-levels']) {
362 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
363 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
364 $oAlParser->createTable($oDB, 'address_levels');
367 if ($aResult['recompute-importance']) {
368 echo "Updating importance values for database.\n";
369 $oDB = new Nominatim\DB();
372 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
373 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
374 $sSQL .= ' (SELECT wikipedia, importance';
375 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
376 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
377 $sSQL .= ' FROM placex d';
378 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
379 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
380 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
384 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
386 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
387 fail('Error: Update interval too low for download.geofabrik.de. ' .
388 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
391 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
393 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
394 ->addParams('--server', CONST_Replication_Url)
395 ->addParams('--outfile', $sImportFile)
396 ->addParams('--size', CONST_Replication_Max_Diff_size);
398 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
401 $fStartTime = time();
402 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
404 if (!$aLastState['sequence_id']) {
405 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
409 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
411 $sBatchEnd = $aLastState['lastimportdate'];
412 $iEndSequence = $aLastState['sequence_id'];
414 if ($aLastState['indexed']) {
415 // Sleep if the update interval has not yet been reached.
416 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
417 if ($fNextUpdate > $fStartTime) {
418 $iSleepTime = $fNextUpdate - $fStartTime;
419 echo "Waiting for next update for $iSleepTime sec.";
423 // Download the next batch of changes.
425 $fCMDStartTime = time();
426 $iNextSeq = (int) $aLastState['sequence_id'];
429 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
430 echo $oCMD->escapedCmd()."\n";
431 if (file_exists($sImportFile)) {
432 unlink($sImportFile);
434 exec($oCMD->escapedCmd(), $aOutput, $iResult);
437 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
438 sleep(CONST_Replication_Recheck_Interval);
439 } elseif ($iResult != 0) {
440 echo 'ERROR: updates failed.';
443 $iEndSequence = (int)$aOutput[0];
447 // get the newest object from the diff file
450 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
451 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
453 echo "Diff file is empty. skipping import.\n";
454 if (!$aResult['import-osmosis-all']) {
461 fail('Error getting date from diff file.');
463 $sBatchEnd = $sBatchEnd[0];
466 $fCMDStartTime = time();
469 echo $oCMDImport->escapedCmd()."\n";
471 $iErrorLevel = $oCMDImport->run();
473 echo "Error executing osm2pgsql: $iErrorLevel\n";
477 // write the update logs
478 $iFileSize = filesize($sImportFile);
479 $sSQL = 'INSERT INTO import_osmosis_log';
480 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
481 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
482 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
483 $sSQL .= date('Y-m-d H:i:s')."','import')";
488 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
491 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
495 if (!$aResult['no-index']) {
496 $oThisIndexCmd = clone($oIndexCmd);
497 $fCMDStartTime = time();
499 echo $oThisIndexCmd->escapedCmd()."\n";
500 $iErrorLevel = $oThisIndexCmd->run();
502 echo "Error: $iErrorLevel\n";
506 $sSQL = 'INSERT INTO import_osmosis_log';
507 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
508 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
509 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
510 $sSQL .= date('Y-m-d H:i:s')."','index')";
513 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
515 $sSQL = 'update import_status set indexed = true';
518 if ($aResult['import-osmosis-all']) {
519 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
524 $fDuration = time() - $fStartTime;
525 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
526 if (!$aResult['import-osmosis-all']) exit(0);