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";
69 $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -S '.CONST_Import_Style.' -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
70 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
71 $sOsm2pgsqlCmd .= ' -U ' . $aDSNInfo['username'];
73 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
74 $sOsm2pgsqlCmd .= ' -H ' . $aDSNInfo['hostspec'];
77 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
78 $aProcEnv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
81 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
82 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
85 $sIndexCmd = CONST_BasePath.'/nominatim/nominatim.py';
86 if (!$aResult['quiet']) {
89 if ($aResult['verbose']) {
93 if ($aResult['init-updates']) {
94 // sanity check that the replication URL is correct
95 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
96 if ($sBaseState === false) {
97 echo "\nCannot find state.txt file at the configured replication URL.\n";
98 echo "Does the URL point to a directory containing OSM update data?\n\n";
99 fail('replication URL not reachable.');
101 // sanity check for pyosmium-get-changes
102 if (!CONST_Pyosmium_Binary) {
103 echo "\nCONST_Pyosmium_Binary not configured.\n";
104 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
105 echo "in your local settings file.\n\n";
106 fail('CONST_Pyosmium_Binary not configured');
109 $sCmd = CONST_Pyosmium_Binary.' --help';
110 exec($sCmd, $aOutput, $iRet);
112 echo "Cannot execute pyosmium-get-changes.\n";
113 echo "Make sure you have pyosmium installed correctly\n";
114 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
115 fail('pyosmium-get-changes not found or not usable');
118 if (!$aResult['no-update-functions']) {
119 // instantiate setupClass to use the function therein
120 $cSetup = new SetupFunctions(array(
121 'enable-diff-updates' => true,
122 'verbose' => $aResult['verbose']
125 $cSetup->createFunctions();
128 $sDatabaseDate = getDatabaseDate($oDB);
129 if (!$sDatabaseDate) {
130 fail('Cannot determine date of database.');
132 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
134 // get the appropriate state id
136 $sCmd = CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url;
137 exec($sCmd, $aOutput, $iRet);
138 if ($iRet != 0 || $aOutput[0] == 'None') {
139 fail('Error running pyosmium tools');
142 $oDB->exec('TRUNCATE import_status');
143 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
144 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
148 } catch (\Nominatim\DatabaseError $e) {
149 fail('Could not enter sequence into database.');
152 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
155 if ($aResult['check-for-updates']) {
156 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
158 if (!$aLastState['sequence_id']) {
159 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
162 system(CONST_BasePath.'/utils/check_server_for_updates.py '.CONST_Replication_Url.' '.$aLastState['sequence_id'], $iRet);
166 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
167 // import diffs and files directly (e.g. from osmosis --rri)
168 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
170 if (!file_exists($sNextFile)) {
171 fail("Cannot open $sNextFile\n");
175 $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
177 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
180 fail("Error from osm2pgsql, $iErrorLevel\n");
183 // Don't update the import status - we don't know what this file contains
186 if ($aResult['calculate-postcodes']) {
187 info('Update postcodes centroids');
188 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
189 runSQLScript($sTemplate, true, true);
192 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
194 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
196 if (isset($aResult['import-node']) && $aResult['import-node']) {
198 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
200 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
204 if (isset($aResult['import-way']) && $aResult['import-way']) {
206 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
208 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
212 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
214 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
216 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
221 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
226 // import generated change file
227 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
229 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
231 fail("osm2pgsql exited with error level $iErrorLevel\n");
235 if ($aResult['deduplicate']) {
236 $oDB = new Nominatim\DB();
239 if ($oDB->getPostgresVersion() < 9.3) {
240 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
243 $sSQL = 'select partition from country_name order by country_code';
244 $aPartitions = $oDB->getCol($sSQL);
247 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
248 foreach ($aPartitions as $i => $sPartition) {
249 $sSQL = 'select count(*) from search_name_'.$sPartition;
250 $nEntries = $oDB->getOne($sSQL);
251 if ($nEntries == 0) {
252 unset($aPartitions[$i]);
256 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
257 $sSQL .= ' and class is null and type is null and country_code is null';
258 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
259 $aDuplicateTokens = $oDB->getAll($sSQL);
260 foreach ($aDuplicateTokens as $aToken) {
261 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
262 echo 'Deduping '.$aToken['word_token']."\n";
263 $sSQL = 'select word_id,';
264 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
265 $sSQL .= " from word where word_token = '".$aToken['word_token'];
266 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
267 $aTokenSet = $oDB->getAll($sSQL);
269 $aKeep = array_shift($aTokenSet);
270 $iKeepID = $aKeep['word_id'];
272 foreach ($aTokenSet as $aRemove) {
273 $sSQL = 'update search_name set';
274 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
275 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
276 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
279 $sSQL = 'update search_name set';
280 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
281 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
284 $sSQL = 'update location_area_country set';
285 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
286 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
289 foreach ($aPartitions as $sPartition) {
290 $sSQL = 'update search_name_'.$sPartition.' set';
291 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
292 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
295 $sSQL = 'update location_area_country set';
296 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
297 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
301 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
307 if ($aResult['recompute-word-counts']) {
308 info('Recompute frequency of full-word search terms');
309 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
310 runSQLScript($sTemplate, true, true);
313 if ($aResult['index']) {
315 .' -d '.$aDSNInfo['database']
316 .' -P '.$aDSNInfo['port']
317 .' -t '.$aResult['index-instances']
318 .' -r '.$aResult['index-rank'];
319 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
320 $sCmd .= ' -H ' . $aDSNInfo['hostspec'];
322 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
323 $sCmd .= ' -U ' . $aDSNInfo['username'];
326 runWithEnv($sCmd, $aProcEnv);
328 $oDB->exec('update import_status set indexed = true');
331 if ($aResult['update-address-levels']) {
332 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
333 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
334 $oAlParser->createTable($oDB, 'address_levels');
337 if ($aResult['recompute-importance']) {
338 echo "Updating importance values for database.\n";
339 $oDB = new Nominatim\DB();
342 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
343 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
344 $sSQL .= ' (SELECT wikipedia, importance';
345 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
346 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
347 $sSQL .= ' FROM placex d';
348 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
349 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
350 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
354 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
356 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
357 fail('Error: Update interval too low for download.geofabrik.de. ' .
358 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
361 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
362 $sCMDDownload = CONST_Pyosmium_Binary.' --server '.CONST_Replication_Url.' -o '.$sImportFile.' -s '.CONST_Replication_Max_Diff_size;
363 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
364 $sCMDIndex = $sIndexCmd
365 .' -d '.$aDSNInfo['database']
366 .' -P '.$aDSNInfo['port']
367 .' -t '.$aResult['index-instances'];
368 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
369 $sCMDIndex .= ' -H ' . $aDSNInfo['hostspec'];
371 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
372 $sCMDIndex .= ' -U ' . $aDSNInfo['username'];
376 $fStartTime = time();
377 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
379 if (!$aLastState['sequence_id']) {
380 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
384 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
386 $sBatchEnd = $aLastState['lastimportdate'];
387 $iEndSequence = $aLastState['sequence_id'];
389 if ($aLastState['indexed']) {
390 // Sleep if the update interval has not yet been reached.
391 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
392 if ($fNextUpdate > $fStartTime) {
393 $iSleepTime = $fNextUpdate - $fStartTime;
394 echo "Waiting for next update for $iSleepTime sec.";
398 // Download the next batch of changes.
400 $fCMDStartTime = time();
401 $iNextSeq = (int) $aLastState['sequence_id'];
403 echo "$sCMDDownload -I $iNextSeq\n";
404 if (file_exists($sImportFile)) {
405 unlink($sImportFile);
407 exec($sCMDDownload.' -I '.$iNextSeq, $aOutput, $iResult);
410 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
411 sleep(CONST_Replication_Recheck_Interval);
412 } elseif ($iResult != 0) {
413 echo 'ERROR: updates failed.';
416 $iEndSequence = (int)$aOutput[0];
420 // get the newest object from the diff file
423 exec(CONST_BasePath.'/utils/osm_file_date.py '.$sImportFile, $sBatchEnd, $iRet);
425 echo "Diff file is empty. skipping import.\n";
426 if (!$aResult['import-osmosis-all']) {
433 fail('Error getting date from diff file.');
435 $sBatchEnd = $sBatchEnd[0];
438 $fCMDStartTime = time();
439 echo $sCMDImport."\n";
441 $iErrorLevel = runWithEnv($sCMDImport, $aProcEnv);
443 echo "Error executing osm2pgsql: $iErrorLevel\n";
447 // write the update logs
448 $iFileSize = filesize($sImportFile);
449 $sSQL = 'INSERT INTO import_osmosis_log';
450 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
451 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
452 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
453 $sSQL .= date('Y-m-d H:i:s')."','import')";
458 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
461 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
465 if (!$aResult['no-index']) {
466 $sThisIndexCmd = $sCMDIndex;
467 $fCMDStartTime = time();
469 echo "$sThisIndexCmd\n";
470 $iErrorLevel = runWithEnv($sThisIndexCmd, $aProcEnv);
472 echo "Error: $iErrorLevel\n";
476 $sSQL = 'INSERT INTO import_osmosis_log';
477 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
478 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
479 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
480 $sSQL .= date('Y-m-d H:i:s')."','index')";
483 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
485 $sSQL = 'update import_status set indexed = true';
488 if ($aResult['import-osmosis-all']) {
489 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
494 $fDuration = time() - $fStartTime;
495 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
496 if (!$aResult['import-osmosis-all']) exit(0);