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();
59 $aDSNInfo = Nominatim\DB::parseDSN(CONST_Database_DSN);
60 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
62 // cache memory to be used by osm2pgsql, should not be more than the available memory
63 $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
64 if ($iCacheMemory + 500 > getTotalMemoryMB()) {
65 $iCacheMemory = getCacheMemoryMB();
66 echo "WARNING: resetting cache memory to $iCacheMemory\n";
68 $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -S '.CONST_Import_Style.' -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
69 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
70 $sOsm2pgsqlCmd .= ' -U ' . $aDSNInfo['username'];
72 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
73 $sOsm2pgsqlCmd .= ' -H ' . $aDSNInfo['hostspec'];
76 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
77 $aProcEnv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
80 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
81 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
84 $sIndexCmd = CONST_BasePath.'/nominatim/nominatim.py';
85 if (!$aResult['quiet']) {
88 if ($aResult['verbose']) {
92 if ($aResult['init-updates']) {
93 // sanity check that the replication URL is correct
94 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
95 if ($sBaseState === false) {
96 echo "\nCannot find state.txt file at the configured replication URL.\n";
97 echo "Does the URL point to a directory containing OSM update data?\n\n";
98 fail('replication URL not reachable.');
100 // sanity check for pyosmium-get-changes
101 if (!CONST_Pyosmium_Binary) {
102 echo "\nCONST_Pyosmium_Binary not configured.\n";
103 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
104 echo "in your local settings file.\n\n";
105 fail('CONST_Pyosmium_Binary not configured');
108 $sCmd = CONST_Pyosmium_Binary.' --help';
109 exec($sCmd, $aOutput, $iRet);
111 echo "Cannot execute pyosmium-get-changes.\n";
112 echo "Make sure you have pyosmium installed correctly\n";
113 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
114 fail('pyosmium-get-changes not found or not usable');
117 if (!$aResult['no-update-functions']) {
118 // instantiate setupClass to use the function therein
119 $cSetup = new SetupFunctions(array(
120 'enable-diff-updates' => true,
121 'verbose' => $aResult['verbose']
124 $cSetup->createFunctions();
127 $sDatabaseDate = getDatabaseDate($oDB);
128 if (!$sDatabaseDate) {
129 fail('Cannot determine date of database.');
131 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
133 // get the appropriate state id
135 $sCmd = CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url;
136 exec($sCmd, $aOutput, $iRet);
137 if ($iRet != 0 || $aOutput[0] == 'None') {
138 fail('Error running pyosmium tools');
141 $oDB->exec('TRUNCATE import_status');
142 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
143 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
147 } catch (\Nominatim\DatabaseError $e) {
148 fail('Could not enter sequence into database.');
151 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
154 if ($aResult['check-for-updates']) {
155 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
157 if (!$aLastState['sequence_id']) {
158 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
161 system(CONST_BasePath.'/utils/check_server_for_updates.py '.CONST_Replication_Url.' '.$aLastState['sequence_id'], $iRet);
165 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
166 // import diffs and files directly (e.g. from osmosis --rri)
167 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
169 if (!file_exists($sNextFile)) {
170 fail("Cannot open $sNextFile\n");
174 $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
176 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
179 fail("Error from osm2pgsql, $iErrorLevel\n");
182 // Don't update the import status - we don't know what this file contains
185 if ($aResult['calculate-postcodes']) {
186 info('Update postcodes centroids');
187 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
188 runSQLScript($sTemplate, true, true);
191 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
193 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
195 if (isset($aResult['import-node']) && $aResult['import-node']) {
197 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
199 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
203 if (isset($aResult['import-way']) && $aResult['import-way']) {
205 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
207 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
211 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
213 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
215 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
220 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
225 // import generated change file
226 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
228 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
230 fail("osm2pgsql exited with error level $iErrorLevel\n");
234 if ($aResult['deduplicate']) {
235 $oDB = new Nominatim\DB();
238 if ($oDB->getPostgresVersion() < 9.3) {
239 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
242 $sSQL = 'select partition from country_name order by country_code';
243 $aPartitions = $oDB->getCol($sSQL);
246 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
247 foreach ($aPartitions as $i => $sPartition) {
248 $sSQL = 'select count(*) from search_name_'.$sPartition;
249 $nEntries = $oDB->getOne($sSQL);
250 if ($nEntries == 0) {
251 unset($aPartitions[$i]);
255 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
256 $sSQL .= ' and class is null and type is null and country_code is null';
257 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
258 $aDuplicateTokens = $oDB->getAll($sSQL);
259 foreach ($aDuplicateTokens as $aToken) {
260 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
261 echo 'Deduping '.$aToken['word_token']."\n";
262 $sSQL = 'select word_id,';
263 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
264 $sSQL .= " from word where word_token = '".$aToken['word_token'];
265 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
266 $aTokenSet = $oDB->getAll($sSQL);
268 $aKeep = array_shift($aTokenSet);
269 $iKeepID = $aKeep['word_id'];
271 foreach ($aTokenSet as $aRemove) {
272 $sSQL = 'update search_name set';
273 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
274 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
275 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
278 $sSQL = 'update search_name set';
279 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
280 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
283 $sSQL = 'update location_area_country set';
284 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
285 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
288 foreach ($aPartitions as $sPartition) {
289 $sSQL = 'update search_name_'.$sPartition.' set';
290 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
291 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
294 $sSQL = 'update location_area_country set';
295 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
296 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
300 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
306 if ($aResult['recompute-word-counts']) {
307 info('Recompute frequency of full-word search terms');
308 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
309 runSQLScript($sTemplate, true, true);
312 if ($aResult['index']) {
314 .' -d '.$aDSNInfo['database']
315 .' -P '.$aDSNInfo['port']
316 .' -t '.$aResult['index-instances']
317 .' -r '.$aResult['index-rank'];
318 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
319 $sCmd .= ' -H ' . $aDSNInfo['hostspec'];
321 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
322 $sCmd .= ' -U ' . $aDSNInfo['username'];
325 runWithEnv($sCmd, $aProcEnv);
327 $oDB->exec('update import_status set indexed = true');
330 if ($aResult['update-address-levels']) {
331 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
332 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
333 $oAlParser->createTable($oDB, 'address_levels');
336 if ($aResult['recompute-importance']) {
337 echo "Updating importance values for database.\n";
338 $oDB = new Nominatim\DB();
341 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
342 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
343 $sSQL .= ' (SELECT wikipedia, importance';
344 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
345 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
346 $sSQL .= ' FROM placex d';
347 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
348 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
349 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
353 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
355 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
356 fail('Error: Update interval too low for download.geofabrik.de. ' .
357 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
360 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
361 $sCMDDownload = CONST_Pyosmium_Binary.' --server '.CONST_Replication_Url.' -o '.$sImportFile.' -s '.CONST_Replication_Max_Diff_size;
362 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
363 $sCMDIndex = $sIndexCmd
364 .' -d '.$aDSNInfo['database']
365 .' -P '.$aDSNInfo['port']
366 .' -t '.$aResult['index-instances'];
367 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
368 $sCMDIndex .= ' -H ' . $aDSNInfo['hostspec'];
370 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
371 $sCMDIndex .= ' -U ' . $aDSNInfo['username'];
375 $fStartTime = time();
376 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
378 if (!$aLastState['sequence_id']) {
379 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
383 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
385 $sBatchEnd = $aLastState['lastimportdate'];
386 $iEndSequence = $aLastState['sequence_id'];
388 if ($aLastState['indexed']) {
389 // Sleep if the update interval has not yet been reached.
390 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
391 if ($fNextUpdate > $fStartTime) {
392 $iSleepTime = $fNextUpdate - $fStartTime;
393 echo "Waiting for next update for $iSleepTime sec.";
397 // Download the next batch of changes.
399 $fCMDStartTime = time();
400 $iNextSeq = (int) $aLastState['sequence_id'];
402 echo "$sCMDDownload -I $iNextSeq\n";
403 if (file_exists($sImportFile)) {
404 unlink($sImportFile);
406 exec($sCMDDownload.' -I '.$iNextSeq, $aOutput, $iResult);
409 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
410 sleep(CONST_Replication_Recheck_Interval);
411 } elseif ($iResult != 0) {
412 echo 'ERROR: updates failed.';
415 $iEndSequence = (int)$aOutput[0];
419 // get the newest object from the diff file
422 exec(CONST_BasePath.'/utils/osm_file_date.py '.$sImportFile, $sBatchEnd, $iRet);
424 echo "Diff file is empty. skipping import.\n";
425 if (!$aResult['import-osmosis-all']) {
432 fail('Error getting date from diff file.');
434 $sBatchEnd = $sBatchEnd[0];
437 $fCMDStartTime = time();
438 echo $sCMDImport."\n";
440 $iErrorLevel = runWithEnv($sCMDImport, $aProcEnv);
442 echo "Error executing osm2pgsql: $iErrorLevel\n";
446 // write the update logs
447 $iFileSize = filesize($sImportFile);
448 $sSQL = 'INSERT INTO import_osmosis_log';
449 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
450 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
451 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
452 $sSQL .= date('Y-m-d H:i:s')."','import')";
457 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
460 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
464 if (!$aResult['no-index']) {
465 $sThisIndexCmd = $sCMDIndex;
466 $fCMDStartTime = time();
468 echo "$sThisIndexCmd\n";
469 $iErrorLevel = runWithEnv($sThisIndexCmd, $aProcEnv);
471 echo "Error: $iErrorLevel\n";
475 $sSQL = 'INSERT INTO import_osmosis_log';
476 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
477 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
478 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
479 $sSQL .= date('Y-m-d H:i:s')."','index')";
482 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
484 $sSQL = 'update import_status set indexed = true';
487 if ($aResult['import-osmosis-all']) {
488 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
493 $fDuration = time() - $fStartTime;
494 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
495 if (!$aResult['import-osmosis-all']) exit(0);