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'])) $aResult['index-instances'] = 1;
53 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
55 date_default_timezone_set('Etc/UTC');
57 $oDB = new Nominatim\DB();
59 $fPostgresVersion = $oDB->getPostgresVersion();
61 $aDSNInfo = Nominatim\DB::parseDSN(CONST_Database_DSN);
62 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
64 // cache memory to be used by osm2pgsql, should not be more than the available memory
65 $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
66 if ($iCacheMemory + 500 > getTotalMemoryMB()) {
67 $iCacheMemory = getCacheMemoryMB();
68 echo "WARNING: resetting cache memory to $iCacheMemory\n";
71 $oOsm2pgsqlCmd = (new \Nominatim\Shell(CONST_Osm2pgsql_Binary))
72 ->addParams('--hstore')
73 ->addParams('--latlong')
74 ->addParams('--append')
76 ->addParams('--number-processes', 1)
77 ->addParams('--cache', $iCacheMemory)
78 ->addParams('--output', 'gazetteer')
79 ->addParams('--style', CONST_Import_Style)
80 ->addParams('--database', $aDSNInfo['database'])
81 ->addParams('--port', $aDSNInfo['port']);
83 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
84 $oOsm2pgsqlCmd->addParams('--host', $aDSNInfo['hostspec']);
86 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
87 $oOsm2pgsqlCmd->addParams('--user', $aDSNInfo['username']);
89 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
90 $oOsm2pgsqlCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
92 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
93 $oOsm2pgsqlCmd->addParams('--flat-nodes', CONST_Osm2pgsql_Flatnode_File);
95 if ($fPostgresVersion >= 11.0) {
96 $oOsm2pgsqlCmd->addEnvPair(
98 '-c jit=off -c max_parallel_workers_per_gather=0'
103 $oIndexCmd = (new \Nominatim\Shell(CONST_BasePath.'/nominatim/nominatim.py'))
104 ->addParams('--database', $aDSNInfo['database'])
105 ->addParams('--port', $aDSNInfo['port'])
106 ->addParams('--threads', $aResult['index-instances']);
107 if (!$aResult['quiet']) {
108 $oIndexCmd->addParams('--verbose');
110 if ($aResult['verbose']) {
111 $oIndexCmd->addParams('--verbose');
113 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
114 $oIndexCmd->addParams('--host', $aDSNInfo['hostspec']);
116 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
117 $oIndexCmd->addParams('--username', $aDSNInfo['username']);
119 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
120 $oIndexCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
124 if ($aResult['init-updates']) {
125 // sanity check that the replication URL is correct
126 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
127 if ($sBaseState === false) {
128 echo "\nCannot find state.txt file at the configured replication URL.\n";
129 echo "Does the URL point to a directory containing OSM update data?\n\n";
130 fail('replication URL not reachable.');
132 // sanity check for pyosmium-get-changes
133 if (!CONST_Pyosmium_Binary) {
134 echo "\nCONST_Pyosmium_Binary not configured.\n";
135 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
136 echo "in your local settings file.\n\n";
137 fail('CONST_Pyosmium_Binary not configured');
141 $oCMD = new \Nominatim\Shell(CONST_Pyosmium_Binary, '--help');
142 exec($oCMD->escapedCmd(), $aOutput, $iRet);
145 echo "Cannot execute pyosmium-get-changes.\n";
146 echo "Make sure you have pyosmium installed correctly\n";
147 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
148 fail('pyosmium-get-changes not found or not usable');
151 if (!$aResult['no-update-functions']) {
152 // instantiate setupClass to use the function therein
153 $cSetup = new SetupFunctions(array(
154 'enable-diff-updates' => true,
155 'verbose' => $aResult['verbose']
157 $cSetup->createFunctions();
160 $sDatabaseDate = getDatabaseDate($oDB);
161 if (!$sDatabaseDate) {
162 fail('Cannot determine date of database.');
164 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
166 // get the appropriate state id
168 $oCMD = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
169 ->addParams('--start-date', $sWindBack)
170 ->addParams('--server', CONST_Replication_Url);
172 exec($oCMD->escapedCmd(), $aOutput, $iRet);
173 if ($iRet != 0 || $aOutput[0] == 'None') {
174 fail('Error running pyosmium tools');
177 $oDB->exec('TRUNCATE import_status');
178 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
179 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
183 } catch (\Nominatim\DatabaseError $e) {
184 fail('Could not enter sequence into database.');
187 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
190 if ($aResult['check-for-updates']) {
191 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
193 if (!$aLastState['sequence_id']) {
194 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
197 $oCmd = (new \Nominatim\Shell(CONST_BasePath.'/utils/check_server_for_updates.py'))
198 ->addParams(CONST_Replication_Url)
199 ->addParams($aLastState['sequence_id']);
200 $iRet = $oCmd->run();
205 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
206 // import diffs and files directly (e.g. from osmosis --rri)
207 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
209 if (!file_exists($sNextFile)) {
210 fail("Cannot open $sNextFile\n");
214 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sNextFile);
215 echo $oCMD->escapedCmd()."\n";
216 $iRet = $oCMD->run();
219 fail("Error from osm2pgsql, $iRet\n");
222 // Don't update the import status - we don't know what this file contains
225 if ($aResult['calculate-postcodes']) {
226 info('Update postcodes centroids');
227 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
228 runSQLScript($sTemplate, true, true);
231 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
233 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
235 if (isset($aResult['import-node']) && $aResult['import-node']) {
237 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
239 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
243 if (isset($aResult['import-way']) && $aResult['import-way']) {
245 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
247 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
251 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
253 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
255 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
260 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
265 // import generated change file
267 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sTemporaryFile);
268 echo $oCMD->escapedCmd()."\n";
270 $iRet = $oCMD->run();
272 fail("osm2pgsql exited with error level $iRet\n");
276 if ($aResult['deduplicate']) {
277 $oDB = new Nominatim\DB();
280 if ($oDB->getPostgresVersion() < 9.3) {
281 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
284 $sSQL = 'select partition from country_name order by country_code';
285 $aPartitions = $oDB->getCol($sSQL);
288 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
289 foreach ($aPartitions as $i => $sPartition) {
290 $sSQL = 'select count(*) from search_name_'.$sPartition;
291 $nEntries = $oDB->getOne($sSQL);
292 if ($nEntries == 0) {
293 unset($aPartitions[$i]);
297 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
298 $sSQL .= ' and class is null and type is null and country_code is null';
299 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
300 $aDuplicateTokens = $oDB->getAll($sSQL);
301 foreach ($aDuplicateTokens as $aToken) {
302 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
303 echo 'Deduping '.$aToken['word_token']."\n";
304 $sSQL = 'select word_id,';
305 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
306 $sSQL .= " from word where word_token = '".$aToken['word_token'];
307 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
308 $aTokenSet = $oDB->getAll($sSQL);
310 $aKeep = array_shift($aTokenSet);
311 $iKeepID = $aKeep['word_id'];
313 foreach ($aTokenSet as $aRemove) {
314 $sSQL = 'update search_name set';
315 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
316 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
317 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
320 $sSQL = 'update search_name set';
321 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
322 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
325 $sSQL = 'update location_area_country set';
326 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
327 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
330 foreach ($aPartitions as $sPartition) {
331 $sSQL = 'update search_name_'.$sPartition.' set';
332 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
333 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
336 $sSQL = 'update location_area_country set';
337 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
338 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
342 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
348 if ($aResult['recompute-word-counts']) {
349 info('Recompute frequency of full-word search terms');
350 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
351 runSQLScript($sTemplate, true, true);
354 if ($aResult['index']) {
355 $oCmd = (clone $oIndexCmd)
356 ->addParams('--minrank', $aResult['index-rank']);
358 // echo $oCmd->escapedCmd()."\n";
361 $oDB->exec('update import_status set indexed = true');
364 if ($aResult['update-address-levels']) {
365 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
366 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
367 $oAlParser->createTable($oDB, 'address_levels');
370 if ($aResult['recompute-importance']) {
371 echo "Updating importance values for database.\n";
372 $oDB = new Nominatim\DB();
375 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
376 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
377 $sSQL .= ' (SELECT wikipedia, importance';
378 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
379 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
380 $sSQL .= ' FROM placex d';
381 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
382 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
383 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
387 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
389 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
390 fail('Error: Update interval too low for download.geofabrik.de. ' .
391 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
394 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
396 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
397 ->addParams('--server', CONST_Replication_Url)
398 ->addParams('--outfile', $sImportFile)
399 ->addParams('--size', CONST_Replication_Max_Diff_size);
401 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
404 $fStartTime = time();
405 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
407 if (!$aLastState['sequence_id']) {
408 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
412 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
414 $sBatchEnd = $aLastState['lastimportdate'];
415 $iEndSequence = $aLastState['sequence_id'];
417 if ($aLastState['indexed']) {
418 // Sleep if the update interval has not yet been reached.
419 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
420 if ($fNextUpdate > $fStartTime) {
421 $iSleepTime = $fNextUpdate - $fStartTime;
422 echo "Waiting for next update for $iSleepTime sec.";
426 // Download the next batch of changes.
428 $fCMDStartTime = time();
429 $iNextSeq = (int) $aLastState['sequence_id'];
432 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
433 echo $oCMD->escapedCmd()."\n";
434 if (file_exists($sImportFile)) {
435 unlink($sImportFile);
437 exec($oCMD->escapedCmd(), $aOutput, $iResult);
440 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
441 sleep(CONST_Replication_Recheck_Interval);
442 } elseif ($iResult != 0) {
443 echo 'ERROR: updates failed.';
446 $iEndSequence = (int)$aOutput[0];
450 // get the newest object from the diff file
453 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
454 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
456 echo "Diff file is empty. skipping import.\n";
457 if (!$aResult['import-osmosis-all']) {
464 fail('Error getting date from diff file.');
466 $sBatchEnd = $sBatchEnd[0];
469 $fCMDStartTime = time();
472 echo $oCMDImport->escapedCmd()."\n";
474 $iErrorLevel = $oCMDImport->run();
476 echo "Error executing osm2pgsql: $iErrorLevel\n";
480 // write the update logs
481 $iFileSize = filesize($sImportFile);
482 $sSQL = 'INSERT INTO import_osmosis_log';
483 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
484 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
485 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
486 $sSQL .= date('Y-m-d H:i:s')."','import')";
491 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
494 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
498 if (!$aResult['no-index']) {
499 $oThisIndexCmd = clone($oIndexCmd);
500 $fCMDStartTime = time();
502 echo $oThisIndexCmd->escapedCmd()."\n";
503 $iErrorLevel = $oThisIndexCmd->run();
505 echo "Error: $iErrorLevel\n";
509 $sSQL = 'INSERT INTO import_osmosis_log';
510 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
511 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
512 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
513 $sSQL .= date('Y-m-d H:i:s')."','index')";
516 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
518 $sSQL = 'update import_status set indexed = true';
521 if ($aResult['import-osmosis-all']) {
522 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
527 $fDuration = time() - $fStartTime;
528 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
529 if (!$aResult['import-osmosis-all']) exit(0);