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();
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);
96 $oIndexCmd = (new \Nominatim\Shell(CONST_BasePath.'/nominatim/nominatim.py'))
97 ->addParams('--database', $aDSNInfo['database'])
98 ->addParams('--port', $aDSNInfo['port'])
99 ->addParams('--threads', $aResult['index-instances']);
101 if ($aResult['verbose']) {
102 $oIndexCmd->addParams('--verbose');
104 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
105 $oIndexCmd->addParams('--host', $aDSNInfo['hostspec']);
107 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
108 $oIndexCmd->addParams('--username', $aDSNInfo['username']);
110 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
111 $oIndexCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
115 if ($aResult['init-updates']) {
116 // sanity check that the replication URL is correct
117 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
118 if ($sBaseState === false) {
119 echo "\nCannot find state.txt file at the configured replication URL.\n";
120 echo "Does the URL point to a directory containing OSM update data?\n\n";
121 fail('replication URL not reachable.');
123 // sanity check for pyosmium-get-changes
124 if (!CONST_Pyosmium_Binary) {
125 echo "\nCONST_Pyosmium_Binary not configured.\n";
126 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
127 echo "in your local settings file.\n\n";
128 fail('CONST_Pyosmium_Binary not configured');
132 $oCMD = new \Nominatim\Shell(CONST_Pyosmium_Binary, '--help');
133 exec($oCMD->escapedCmd(), $aOutput, $iRet);
136 echo "Cannot execute pyosmium-get-changes.\n";
137 echo "Make sure you have pyosmium installed correctly\n";
138 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
139 fail('pyosmium-get-changes not found or not usable');
142 if (!$aResult['no-update-functions']) {
143 // instantiate setupClass to use the function therein
144 $cSetup = new SetupFunctions(array(
145 'enable-diff-updates' => true,
146 'verbose' => $aResult['verbose']
149 $cSetup->createFunctions();
152 $sDatabaseDate = getDatabaseDate($oDB);
153 if (!$sDatabaseDate) {
154 fail('Cannot determine date of database.');
156 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
158 // get the appropriate state id
160 $oCMD = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
161 ->addParams('--start-date', $sWindBack)
162 ->addParams('--server', CONST_Replication_Url);
164 exec($oCMD->escapedCmd(), $aOutput, $iRet);
165 if ($iRet != 0 || $aOutput[0] == 'None') {
166 fail('Error running pyosmium tools');
169 $oDB->exec('TRUNCATE import_status');
170 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
171 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
175 } catch (\Nominatim\DatabaseError $e) {
176 fail('Could not enter sequence into database.');
179 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
182 if ($aResult['check-for-updates']) {
183 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
185 if (!$aLastState['sequence_id']) {
186 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
189 $oCmd = (new \Nominatim\Shell(CONST_BasePath.'/utils/check_server_for_updates.py'))
190 ->addParams(CONST_Replication_Url)
191 ->addParams($aLastState['sequence_id']);
192 $iRet = $oCmd->run();
197 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
198 // import diffs and files directly (e.g. from osmosis --rri)
199 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
201 if (!file_exists($sNextFile)) {
202 fail("Cannot open $sNextFile\n");
206 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sNextFile);
207 echo $oCMD->escapedCmd()."\n";
208 $iRet = $oCMD->run();
211 fail("Error from osm2pgsql, $iRet\n");
214 // Don't update the import status - we don't know what this file contains
217 if ($aResult['calculate-postcodes']) {
218 info('Update postcodes centroids');
219 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
220 runSQLScript($sTemplate, true, true);
223 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
225 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
227 if (isset($aResult['import-node']) && $aResult['import-node']) {
229 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
231 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
235 if (isset($aResult['import-way']) && $aResult['import-way']) {
237 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
239 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
243 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
245 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
247 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
252 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
257 // import generated change file
259 $oCMD = (clone $oOsm2pgsqlCmd)->addParams($sTemporaryFile);
260 echo $oCMD->escapedCmd()."\n";
262 $iRet = $oCMD->run();
264 fail("osm2pgsql exited with error level $iRet\n");
268 if ($aResult['deduplicate']) {
269 $oDB = new Nominatim\DB();
272 if ($oDB->getPostgresVersion() < 9.3) {
273 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
276 $sSQL = 'select partition from country_name order by country_code';
277 $aPartitions = $oDB->getCol($sSQL);
280 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
281 foreach ($aPartitions as $i => $sPartition) {
282 $sSQL = 'select count(*) from search_name_'.$sPartition;
283 $nEntries = $oDB->getOne($sSQL);
284 if ($nEntries == 0) {
285 unset($aPartitions[$i]);
289 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
290 $sSQL .= ' and class is null and type is null and country_code is null';
291 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
292 $aDuplicateTokens = $oDB->getAll($sSQL);
293 foreach ($aDuplicateTokens as $aToken) {
294 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
295 echo 'Deduping '.$aToken['word_token']."\n";
296 $sSQL = 'select word_id,';
297 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
298 $sSQL .= " from word where word_token = '".$aToken['word_token'];
299 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
300 $aTokenSet = $oDB->getAll($sSQL);
302 $aKeep = array_shift($aTokenSet);
303 $iKeepID = $aKeep['word_id'];
305 foreach ($aTokenSet as $aRemove) {
306 $sSQL = 'update search_name set';
307 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
308 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
309 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
312 $sSQL = 'update search_name set';
313 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
314 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
317 $sSQL = 'update location_area_country set';
318 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
319 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
322 foreach ($aPartitions as $sPartition) {
323 $sSQL = 'update search_name_'.$sPartition.' set';
324 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
325 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
328 $sSQL = 'update location_area_country set';
329 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
330 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
334 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
340 if ($aResult['recompute-word-counts']) {
341 info('Recompute frequency of full-word search terms');
342 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
343 runSQLScript($sTemplate, true, true);
346 if ($aResult['index']) {
347 $oCmd = (clone $oIndexCmd)
348 ->addParams('--minrank', $aResult['index-rank']);
350 // echo $oCmd->escapedCmd()."\n";
353 $oDB->exec('update import_status set indexed = true');
356 if ($aResult['update-address-levels']) {
357 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
358 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
359 $oAlParser->createTable($oDB, 'address_levels');
362 if ($aResult['recompute-importance']) {
363 echo "Updating importance values for database.\n";
364 $oDB = new Nominatim\DB();
367 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
368 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
369 $sSQL .= ' (SELECT wikipedia, importance';
370 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
371 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
372 $sSQL .= ' FROM placex d';
373 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
374 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
375 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
379 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
381 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
382 fail('Error: Update interval too low for download.geofabrik.de. ' .
383 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
386 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
388 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
389 ->addParams('--server', CONST_Replication_Url)
390 ->addParams('--outfile', $sImportFile)
391 ->addParams('--size', CONST_Replication_Max_Diff_size);
393 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
396 $fStartTime = time();
397 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
399 if (!$aLastState['sequence_id']) {
400 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
404 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
406 $sBatchEnd = $aLastState['lastimportdate'];
407 $iEndSequence = $aLastState['sequence_id'];
409 if ($aLastState['indexed']) {
410 // Sleep if the update interval has not yet been reached.
411 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
412 if ($fNextUpdate > $fStartTime) {
413 $iSleepTime = $fNextUpdate - $fStartTime;
414 echo "Waiting for next update for $iSleepTime sec.";
418 // Download the next batch of changes.
420 $fCMDStartTime = time();
421 $iNextSeq = (int) $aLastState['sequence_id'];
424 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
425 echo $oCMD->escapedCmd()."\n";
426 if (file_exists($sImportFile)) {
427 unlink($sImportFile);
429 exec($oCMD->escapedCmd(), $aOutput, $iResult);
432 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
433 sleep(CONST_Replication_Recheck_Interval);
434 } elseif ($iResult != 0) {
435 echo 'ERROR: updates failed.';
438 $iEndSequence = (int)$aOutput[0];
442 // get the newest object from the diff file
445 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
446 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
448 echo "Diff file is empty. skipping import.\n";
449 if (!$aResult['import-osmosis-all']) {
456 fail('Error getting date from diff file.');
458 $sBatchEnd = $sBatchEnd[0];
461 $fCMDStartTime = time();
464 echo $oCMDImport->escapedCmd()."\n";
466 $iErrorLevel = $oCMDImport->run();
468 echo "Error executing osm2pgsql: $iErrorLevel\n";
472 // write the update logs
473 $iFileSize = filesize($sImportFile);
474 $sSQL = 'INSERT INTO import_osmosis_log';
475 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
476 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
477 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
478 $sSQL .= date('Y-m-d H:i:s')."','import')";
483 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
486 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
490 if (!$aResult['no-index']) {
491 $oThisIndexCmd = clone($oIndexCmd);
492 $fCMDStartTime = time();
494 echo $oThisIndexCmd->escapedCmd()."\n";
495 $iErrorLevel = $oThisIndexCmd->run();
497 echo "Error: $iErrorLevel\n";
501 $sSQL = 'INSERT INTO import_osmosis_log';
502 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
503 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
504 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
505 $sSQL .= date('Y-m-d H:i:s')."','index')";
508 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
510 $sSQL = 'update import_status set indexed = true';
513 if ($aResult['import-osmosis-all']) {
514 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
519 $fDuration = time() - $fStartTime;
520 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
521 if (!$aResult['import-osmosis-all']) exit(0);