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('recompute-word-counts', '', 0, 1, 0, 0, 'bool', 'Compute frequency of full-word search terms'),
43 array('update-address-levels', '', 0, 1, 0, 0, 'bool', 'Reimport address level configuration (EXPERT)'),
44 array('recompute-importance', '', 0, 1, 0, 0, 'bool', 'Recompute place importances')
47 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
49 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
51 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
53 date_default_timezone_set('Etc/UTC');
55 $oDB = new Nominatim\DB();
57 $fPostgresVersion = $oDB->getPostgresVersion();
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";
69 $oOsm2pgsqlCmd = (new \Nominatim\Shell(CONST_Osm2pgsql_Binary))
70 ->addParams('--hstore')
71 ->addParams('--latlong')
72 ->addParams('--append')
74 ->addParams('--number-processes', 1)
75 ->addParams('--cache', $iCacheMemory)
76 ->addParams('--output', 'gazetteer')
77 ->addParams('--style', CONST_Import_Style)
78 ->addParams('--database', $aDSNInfo['database'])
79 ->addParams('--port', $aDSNInfo['port']);
81 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
82 $oOsm2pgsqlCmd->addParams('--host', $aDSNInfo['hostspec']);
84 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
85 $oOsm2pgsqlCmd->addParams('--user', $aDSNInfo['username']);
87 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
88 $oOsm2pgsqlCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
90 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
91 $oOsm2pgsqlCmd->addParams('--flat-nodes', CONST_Osm2pgsql_Flatnode_File);
93 if ($fPostgresVersion >= 11.0) {
94 $oOsm2pgsqlCmd->addEnvPair(
96 '-c jit=off -c max_parallel_workers_per_gather=0'
101 $oIndexCmd = (new \Nominatim\Shell(CONST_BasePath.'/nominatim/nominatim.py'))
102 ->addParams('--database', $aDSNInfo['database'])
103 ->addParams('--port', $aDSNInfo['port'])
104 ->addParams('--threads', $aResult['index-instances']);
105 if (!$aResult['quiet']) {
106 $oIndexCmd->addParams('--verbose');
108 if ($aResult['verbose']) {
109 $oIndexCmd->addParams('--verbose');
111 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
112 $oIndexCmd->addParams('--host', $aDSNInfo['hostspec']);
114 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
115 $oIndexCmd->addParams('--username', $aDSNInfo['username']);
117 if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
118 $oIndexCmd->addEnvPair('PGPASSWORD', $aDSNInfo['password']);
122 if ($aResult['init-updates']) {
123 // sanity check that the replication URL is correct
124 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
125 if ($sBaseState === false) {
126 echo "\nCannot find state.txt file at the configured replication URL.\n";
127 echo "Does the URL point to a directory containing OSM update data?\n\n";
128 fail('replication URL not reachable.');
130 // sanity check for pyosmium-get-changes
131 if (!CONST_Pyosmium_Binary) {
132 echo "\nCONST_Pyosmium_Binary not configured.\n";
133 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
134 echo "in your local settings file.\n\n";
135 fail('CONST_Pyosmium_Binary not configured');
139 $oCMD = new \Nominatim\Shell(CONST_Pyosmium_Binary, '--help');
140 exec($oCMD->escapedCmd(), $aOutput, $iRet);
143 echo "Cannot execute pyosmium-get-changes.\n";
144 echo "Make sure you have pyosmium installed correctly\n";
145 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
146 fail('pyosmium-get-changes not found or not usable');
149 if (!$aResult['no-update-functions']) {
150 // instantiate setupClass to use the function therein
151 $cSetup = new SetupFunctions(array(
152 'enable-diff-updates' => true,
153 '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['recompute-word-counts']) {
275 info('Recompute frequency of full-word search terms');
276 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
277 runSQLScript($sTemplate, true, true);
280 if ($aResult['index']) {
281 $oCmd = (clone $oIndexCmd)
282 ->addParams('--minrank', $aResult['index-rank']);
284 // echo $oCmd->escapedCmd()."\n";
287 $oDB->exec('update import_status set indexed = true');
290 if ($aResult['update-address-levels']) {
291 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
292 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
293 $oAlParser->createTable($oDB, 'address_levels');
296 if ($aResult['recompute-importance']) {
297 echo "Updating importance values for database.\n";
298 $oDB = new Nominatim\DB();
301 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
302 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
303 $sSQL .= ' (SELECT wikipedia, importance';
304 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
305 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
306 $sSQL .= ' FROM placex d';
307 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
308 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
309 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
313 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
315 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
316 fail('Error: Update interval too low for download.geofabrik.de. ' .
317 "Please check install documentation (https://nominatim.org/release-docs/latest/admin/Import-and-Update#setting-up-the-update-process)\n");
320 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
322 $oCMDDownload = (new \Nominatim\Shell(CONST_Pyosmium_Binary))
323 ->addParams('--server', CONST_Replication_Url)
324 ->addParams('--outfile', $sImportFile)
325 ->addParams('--size', CONST_Replication_Max_Diff_size);
327 $oCMDImport = (clone $oOsm2pgsqlCmd)->addParams($sImportFile);
330 $fStartTime = time();
331 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
333 if (!$aLastState['sequence_id']) {
334 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
338 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
340 $sBatchEnd = $aLastState['lastimportdate'];
341 $iEndSequence = $aLastState['sequence_id'];
343 if ($aLastState['indexed']) {
344 // Sleep if the update interval has not yet been reached.
345 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
346 if ($fNextUpdate > $fStartTime) {
347 $iSleepTime = $fNextUpdate - $fStartTime;
348 echo "Waiting for next update for $iSleepTime sec.";
352 // Download the next batch of changes.
354 $fCMDStartTime = time();
355 $iNextSeq = (int) $aLastState['sequence_id'];
358 $oCMD = (clone $oCMDDownload)->addParams('--start-id', $iNextSeq);
359 echo $oCMD->escapedCmd()."\n";
360 if (file_exists($sImportFile)) {
361 unlink($sImportFile);
363 exec($oCMD->escapedCmd(), $aOutput, $iResult);
366 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
367 sleep(CONST_Replication_Recheck_Interval);
368 } elseif ($iResult != 0) {
369 echo 'ERROR: updates failed.';
372 $iEndSequence = (int)$aOutput[0];
376 // get the newest object from the diff file
379 $oCMD = new \Nominatim\Shell(CONST_BasePath.'/utils/osm_file_date.py', $sImportFile);
380 exec($oCMD->escapedCmd(), $sBatchEnd, $iRet);
382 echo "Diff file is empty. skipping import.\n";
383 if (!$aResult['import-osmosis-all']) {
390 fail('Error getting date from diff file.');
392 $sBatchEnd = $sBatchEnd[0];
395 $fCMDStartTime = time();
398 echo $oCMDImport->escapedCmd()."\n";
400 $iErrorLevel = $oCMDImport->run();
402 echo "Error executing osm2pgsql: $iErrorLevel\n";
406 // write the update logs
407 $iFileSize = filesize($sImportFile);
408 $sSQL = 'INSERT INTO import_osmosis_log';
409 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
410 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
411 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
412 $sSQL .= date('Y-m-d H:i:s')."','import')";
417 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
420 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
424 if (!$aResult['no-index']) {
425 $oThisIndexCmd = clone($oIndexCmd);
426 $fCMDStartTime = time();
428 echo $oThisIndexCmd->escapedCmd()."\n";
429 $iErrorLevel = $oThisIndexCmd->run();
431 echo "Error: $iErrorLevel\n";
435 $sSQL = 'INSERT INTO import_osmosis_log';
436 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
437 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
438 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
439 $sSQL .= date('Y-m-d H:i:s')."','index')";
442 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
444 $sSQL = 'update import_status set indexed = true';
447 if ($aResult['import-osmosis-all']) {
448 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
453 $fDuration = time() - $fStartTime;
454 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
455 if (!$aResult['import-osmosis-all']) exit(0);