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']
155 $cSetup->createFunctions();
158 $sDatabaseDate = getDatabaseDate($oDB);
159 if (!$sDatabaseDate) {
160 fail('Cannot determine date of database.');
162 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
164 // get the appropriate state id
166 $oCMD = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
167 ->addParams('--start-date', $sWindBack)
168 ->addParams('--server', CONST_Replication_Url);
170 exec($oCMD->escapedCmd(), $aOutput, $iRet);
171 if ($iRet != 0 || $aOutput[0] == 'None') {
172 fail('Error running pyosmium tools');
175 $oDB->exec('TRUNCATE import_status');
176 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
177 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
181 } catch (\Nominatim\DatabaseError $e) {
182 fail('Could not enter sequence into database.');
185 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
188 if ($aResult['check-for-updates']) {
189 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
191 if (!$aLastState['sequence_id']) {
192 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
195 $oCmd = (new \Nominatim\Shell(CONST_BasePath.'/utils/check_server_for_updates.py'))
196 ->addParams(CONST_Replication_Url)
197 ->addParams($aLastState['sequence_id']);
198 $iRet = $oCmd->run();
203 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
204 // import diffs and files directly (e.g. from osmosis --rri)
205 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
207 if (!file_exists($sNextFile)) {
208 fail("Cannot open $sNextFile\n");
212 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sNextFile);
213 echo $oCMD->escapedCmd()."\n";
214 $iRet = $oCMD->run();
217 fail("Error from osm2pgsql, $iRet\n");
220 // Don't update the import status - we don't know what this file contains
223 if ($aResult['calculate-postcodes']) {
224 info('Update postcodes centroids');
225 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
226 runSQLScript($sTemplate, true, true);
229 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
231 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
233 if (isset($aResult['import-node']) && $aResult['import-node']) {
235 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
237 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
241 if (isset($aResult['import-way']) && $aResult['import-way']) {
243 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
245 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
249 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
251 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
253 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
258 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
263 // import generated change file
265 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sTemporaryFile);
266 echo $oCMD->escapedCmd()."\n";
268 $iRet = $oCMD->run();
270 fail("osm2pgsql exited with error level $iRet\n");
274 if ($aResult['deduplicate']) {
275 $oDB = new Nominatim\DB();
278 if ($oDB->getPostgresVersion() < 9.3) {
279 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
282 $sSQL = 'select partition from country_name order by country_code';
283 $aPartitions = $oDB->getCol($sSQL);
286 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
287 foreach ($aPartitions as $i => $sPartition) {
288 $sSQL = 'select count(*) from search_name_'.$sPartition;
289 $nEntries = $oDB->getOne($sSQL);
290 if ($nEntries == 0) {
291 unset($aPartitions[$i]);
295 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
296 $sSQL .= ' and class is null and type is null and country_code is null';
297 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
298 $aDuplicateTokens = $oDB->getAll($sSQL);
299 foreach ($aDuplicateTokens as $aToken) {
300 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
301 echo 'Deduping '.$aToken['word_token']."\n";
302 $sSQL = 'select word_id,';
303 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
304 $sSQL .= " from word where word_token = '".$aToken['word_token'];
305 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
306 $aTokenSet = $oDB->getAll($sSQL);
308 $aKeep = array_shift($aTokenSet);
309 $iKeepID = $aKeep['word_id'];
311 foreach ($aTokenSet as $aRemove) {
312 $sSQL = 'update search_name set';
313 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
314 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
315 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
318 $sSQL = 'update search_name set';
319 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
320 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
323 $sSQL = 'update location_area_country set';
324 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
325 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
328 foreach ($aPartitions as $sPartition) {
329 $sSQL = 'update search_name_'.$sPartition.' set';
330 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
331 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
334 $sSQL = 'update location_area_country set';
335 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
336 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
340 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
346 if ($aResult['recompute-word-counts']) {
347 info('Recompute frequency of full-word search terms');
348 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
349 runSQLScript($sTemplate, true, true);
352 if ($aResult['index']) {
353 $oCmd = (clone $oIndexCmd)
354 ->addParams('--minrank', $aResult['index-rank']);
356 // echo $oCmd->escapedCmd()."\n";
359 $oDB->exec('update import_status set indexed = true');
362 if ($aResult['update-address-levels']) {
363 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
364 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
365 $oAlParser->createTable($oDB, 'address_levels');
368 if ($aResult['recompute-importance']) {
369 echo "Updating importance values for database.\n";
370 $oDB = new Nominatim\DB();
373 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
374 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
375 $sSQL .= ' (SELECT wikipedia, importance';
376 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
377 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
378 $sSQL .= ' FROM placex d';
379 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
380 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
381 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
385 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
387 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
388 fail('Error: Update interval too low for download.geofabrik.de. ' .
389 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
392 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
394 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
395 ->addParams('--server', CONST_Replication_Url)
396 ->addParams('--outfile', $sImportFile)
397 ->addParams('--size', CONST_Replication_Max_Diff_size);
399 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
402 $fStartTime = time();
403 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
405 if (!$aLastState['sequence_id']) {
406 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
410 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
412 $sBatchEnd = $aLastState['lastimportdate'];
413 $iEndSequence = $aLastState['sequence_id'];
415 if ($aLastState['indexed']) {
416 // Sleep if the update interval has not yet been reached.
417 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
418 if ($fNextUpdate > $fStartTime) {
419 $iSleepTime = $fNextUpdate - $fStartTime;
420 echo "Waiting for next update for $iSleepTime sec.";
424 // Download the next batch of changes.
426 $fCMDStartTime = time();
427 $iNextSeq = (int) $aLastState['sequence_id'];
430 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
431 echo $oCMD->escapedCmd()."\n";
432 if (file_exists($sImportFile)) {
433 unlink($sImportFile);
435 exec($oCMD->escapedCmd(), $aOutput, $iResult);
438 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
439 sleep(CONST_Replication_Recheck_Interval);
440 } elseif ($iResult != 0) {
441 echo 'ERROR: updates failed.';
444 $iEndSequence = (int)$aOutput[0];
448 // get the newest object from the diff file
451 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
452 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
454 echo "Diff file is empty. skipping import.\n";
455 if (!$aResult['import-osmosis-all']) {
462 fail('Error getting date from diff file.');
464 $sBatchEnd = $sBatchEnd[0];
467 $fCMDStartTime = time();
470 echo $oCMDImport->escapedCmd()."\n";
472 $iErrorLevel = $oCMDImport->run();
474 echo "Error executing osm2pgsql: $iErrorLevel\n";
478 // write the update logs
479 $iFileSize = filesize($sImportFile);
480 $sSQL = 'INSERT INTO import_osmosis_log';
481 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
482 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
483 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
484 $sSQL .= date('Y-m-d H:i:s')."','import')";
489 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
492 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
496 if (!$aResult['no-index']) {
497 $oThisIndexCmd = clone($oIndexCmd);
498 $fCMDStartTime = time();
500 echo $oThisIndexCmd->escapedCmd()."\n";
501 $iErrorLevel = $oThisIndexCmd->run();
503 echo "Error: $iErrorLevel\n";
507 $sSQL = 'INSERT INTO import_osmosis_log';
508 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
509 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
510 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
511 $sSQL .= date('Y-m-d H:i:s')."','index')";
514 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
516 $sSQL = 'update import_status set indexed = true';
519 if ($aResult['import-osmosis-all']) {
520 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
525 $fDuration = time() - $fStartTime;
526 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
527 if (!$aResult['import-osmosis-all']) exit(0);