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 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
85 if ($aResult['init-updates']) {
86 // sanity check that the replication URL is correct
87 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
88 if ($sBaseState === false) {
89 echo "\nCannot find state.txt file at the configured replication URL.\n";
90 echo "Does the URL point to a directory containing OSM update data?\n\n";
91 fail('replication URL not reachable.');
93 // sanity check for pyosmium-get-changes
94 if (!CONST_Pyosmium_Binary) {
95 echo "\nCONST_Pyosmium_Binary not configured.\n";
96 echo "You need to install pyosmium and set up the path to pyosmium-get-changes\n";
97 echo "in your local settings file.\n\n";
98 fail('CONST_Pyosmium_Binary not configured');
101 $sCmd = CONST_Pyosmium_Binary.' --help';
102 exec($sCmd, $aOutput, $iRet);
104 echo "Cannot execute pyosmium-get-changes.\n";
105 echo "Make sure you have pyosmium installed correctly\n";
106 echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
107 fail('pyosmium-get-changes not found or not usable');
110 if (!$aResult['no-update-functions']) {
111 // instantiate setupClass to use the function therein
112 $cSetup = new SetupFunctions(array(
113 'enable-diff-updates' => true,
114 'verbose' => $aResult['verbose']
117 $cSetup->createFunctions();
120 $sDatabaseDate = getDatabaseDate($oDB);
121 if (!$sDatabaseDate) {
122 fail('Cannot determine date of database.');
124 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
126 // get the appropriate state id
128 $sCmd = CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url;
129 exec($sCmd, $aOutput, $iRet);
130 if ($iRet != 0 || $aOutput[0] == 'None') {
131 fail('Error running pyosmium tools');
134 $oDB->exec('TRUNCATE import_status');
135 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
136 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
140 } catch (\Nominatim\DatabaseError $e) {
141 fail('Could not enter sequence into database.');
144 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
147 if ($aResult['check-for-updates']) {
148 $aLastState = $oDB->getRow('SELECT sequence_id FROM import_status');
150 if (!$aLastState['sequence_id']) {
151 fail('Updates not set up. Please run ./utils/update.php --init-updates.');
154 system(CONST_BasePath.'/utils/check_server_for_updates.py '.CONST_Replication_Url.' '.$aLastState['sequence_id'], $iRet);
158 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
159 // import diffs and files directly (e.g. from osmosis --rri)
160 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
162 if (!file_exists($sNextFile)) {
163 fail("Cannot open $sNextFile\n");
167 $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
169 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
172 fail("Error from osm2pgsql, $iErrorLevel\n");
175 // Don't update the import status - we don't know what this file contains
178 if ($aResult['calculate-postcodes']) {
179 info('Update postcodes centroids');
180 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
181 runSQLScript($sTemplate, true, true);
184 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
186 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
188 if (isset($aResult['import-node']) && $aResult['import-node']) {
190 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
192 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
196 if (isset($aResult['import-way']) && $aResult['import-way']) {
198 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
200 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
204 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
206 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
208 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
213 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
218 // import generated change file
219 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
221 $iErrorLevel = runWithEnv($sCMD, $aProcEnv);
223 fail("osm2pgsql exited with error level $iErrorLevel\n");
227 if ($aResult['deduplicate']) {
228 $oDB = new Nominatim\DB();
231 if ($oDB->getPostgresVersion() < 9.3) {
232 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
235 $sSQL = 'select partition from country_name order by country_code';
236 $aPartitions = $oDB->getCol($sSQL);
239 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
240 foreach ($aPartitions as $i => $sPartition) {
241 $sSQL = 'select count(*) from search_name_'.$sPartition;
242 $nEntries = $oDB->getOne($sSQL);
243 if ($nEntries == 0) {
244 unset($aPartitions[$i]);
248 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
249 $sSQL .= ' and class is null and type is null and country_code is null';
250 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
251 $aDuplicateTokens = $oDB->getAll($sSQL);
252 foreach ($aDuplicateTokens as $aToken) {
253 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
254 echo 'Deduping '.$aToken['word_token']."\n";
255 $sSQL = 'select word_id,';
256 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
257 $sSQL .= " from word where word_token = '".$aToken['word_token'];
258 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
259 $aTokenSet = $oDB->getAll($sSQL);
261 $aKeep = array_shift($aTokenSet);
262 $iKeepID = $aKeep['word_id'];
264 foreach ($aTokenSet as $aRemove) {
265 $sSQL = 'update search_name set';
266 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
267 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
268 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
271 $sSQL = 'update search_name set';
272 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
273 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
276 $sSQL = 'update location_area_country set';
277 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
278 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
281 foreach ($aPartitions as $sPartition) {
282 $sSQL = 'update search_name_'.$sPartition.' set';
283 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
284 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
287 $sSQL = 'update location_area_country set';
288 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
289 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
293 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
299 if ($aResult['recompute-word-counts']) {
300 info('Recompute frequency of full-word search terms');
301 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
302 runSQLScript($sTemplate, true, true);
305 if ($aResult['index']) {
306 $sCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank'];
307 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
308 $sCmd .= ' -H ' . $aDSNInfo['hostspec'];
310 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
311 $sCmd .= ' -U ' . $aDSNInfo['username'];
314 runWithEnv($sCmd, $aProcEnv);
316 $oDB->exec('update import_status set indexed = true');
319 if ($aResult['update-address-levels']) {
320 echo 'Updating address levels from '.CONST_Address_Level_Config.".\n";
321 $oAlParser = new \Nominatim\Setup\AddressLevelParser(CONST_Address_Level_Config);
322 $oAlParser->createTable($oDB, 'address_levels');
325 if ($aResult['recompute-importance']) {
326 echo "Updating importance values for database.\n";
327 $oDB = new Nominatim\DB();
330 $sSQL = 'ALTER TABLE placex DISABLE TRIGGER ALL;';
331 $sSQL .= 'UPDATE placex SET (wikipedia, importance) =';
332 $sSQL .= ' (SELECT wikipedia, importance';
333 $sSQL .= ' FROM compute_importance(extratags, country_code, osm_type, osm_id));';
334 $sSQL .= 'UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance';
335 $sSQL .= ' FROM placex d';
336 $sSQL .= ' WHERE s.place_id = d.linked_place_id and d.wikipedia is not null';
337 $sSQL .= ' and (s.wikipedia is null or s.importance < d.importance);';
338 $sSQL .= 'ALTER TABLE placex ENABLE TRIGGER ALL;';
342 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
344 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
345 fail('Error: Update interval too low for download.geofabrik.de. ' .
346 "Please check install documentation (http://nominatim.org/release-docs/latest/Import-and-Update#setting-up-the-update-process)\n");
349 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
350 $sCMDDownload = CONST_Pyosmium_Binary.' --server '.CONST_Replication_Url.' -o '.$sImportFile.' -s '.CONST_Replication_Max_Diff_size;
351 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
352 $sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
353 if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
354 $sCMDIndex .= ' -H ' . $aDSNInfo['hostspec'];
356 if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
357 $sCMDIndex .= ' -U ' . $aDSNInfo['username'];
361 $fStartTime = time();
362 $aLastState = $oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status');
364 if (!$aLastState['sequence_id']) {
365 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
369 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
371 $sBatchEnd = $aLastState['lastimportdate'];
372 $iEndSequence = $aLastState['sequence_id'];
374 if ($aLastState['indexed']) {
375 // Sleep if the update interval has not yet been reached.
376 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
377 if ($fNextUpdate > $fStartTime) {
378 $iSleepTime = $fNextUpdate - $fStartTime;
379 echo "Waiting for next update for $iSleepTime sec.";
383 // Download the next batch of changes.
385 $fCMDStartTime = time();
386 $iNextSeq = (int) $aLastState['sequence_id'];
388 echo "$sCMDDownload -I $iNextSeq\n";
389 if (file_exists($sImportFile)) {
390 unlink($sImportFile);
392 exec($sCMDDownload.' -I '.$iNextSeq, $aOutput, $iResult);
395 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
396 sleep(CONST_Replication_Recheck_Interval);
397 } elseif ($iResult != 0) {
398 echo 'ERROR: updates failed.';
401 $iEndSequence = (int)$aOutput[0];
405 // get the newest object from the diff file
408 exec(CONST_BasePath.'/utils/osm_file_date.py '.$sImportFile, $sBatchEnd, $iRet);
410 echo "Diff file is empty. skipping import.\n";
411 if (!$aResult['import-osmosis-all']) {
418 fail('Error getting date from diff file.');
420 $sBatchEnd = $sBatchEnd[0];
423 $fCMDStartTime = time();
424 echo $sCMDImport."\n";
426 $iErrorLevel = runWithEnv($sCMDImport, $aProcEnv);
428 echo "Error executing osm2pgsql: $iErrorLevel\n";
432 // write the update logs
433 $iFileSize = filesize($sImportFile);
434 $sSQL = 'INSERT INTO import_osmosis_log';
435 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
436 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
437 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
438 $sSQL .= date('Y-m-d H:i:s')."','import')";
443 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
446 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
450 if (!$aResult['no-index']) {
451 $sThisIndexCmd = $sCMDIndex;
452 $fCMDStartTime = time();
454 echo "$sThisIndexCmd\n";
455 $iErrorLevel = runWithEnv($sThisIndexCmd, $aProcEnv);
457 echo "Error: $iErrorLevel\n";
461 $sSQL = 'INSERT INTO import_osmosis_log';
462 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
463 $sSQL .= " values ('$sBatchEnd',$iEndSequence,NULL,'";
464 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
465 $sSQL .= date('Y-m-d H:i:s')."','index')";
468 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
470 $sSQL = 'update import_status set indexed = true';
473 if ($aResult['import-osmosis-all']) {
474 echo "Error: --no-index cannot be used with continuous imports (--import-osmosis-all).\n";
479 $fDuration = time() - $fStartTime;
480 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
481 if (!$aResult['import-osmosis-all']) exit(0);