4 require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
5 require_once(CONST_BasePath.'/lib/init-cmd.php');
6 ini_set('memory_limit', '800M');
10 'Import / update / index osm data',
11 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
12 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
13 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
15 array('init-updates', '', 0, 1, 0, 0, 'bool', 'Set up database for updating'),
16 array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import updates once'),
17 array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import updates forever'),
18 array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
20 array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Update postcode centroid table'),
22 array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
23 array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
24 array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
26 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
27 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
28 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
29 array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
31 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
32 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
33 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
35 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
36 array('recompute-word-counts', '', 0, 1, 0, 0, 'bool', 'Compute frequency of full-word search terms'),
37 array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolete)'),
39 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
41 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
43 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
45 date_default_timezone_set('Etc/UTC');
49 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
50 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
52 // cache memory to be used by osm2pgsql, should not be more than the available memory
53 $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
54 if ($iCacheMemory + 500 > getTotalMemoryMB()) {
55 $iCacheMemory = getCacheMemoryMB();
56 echo "WARNING: resetting cache memory to $iCacheMemory\n";
58 $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
59 if (!is_null(CONST_Osm2pgsql_Flatnode_File)) {
60 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
63 if ($aResult['init-updates']) {
64 // sanity check that the replication URL is correct
65 $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
66 if ($sBaseState === false) {
67 echo "\nCannot find state.txt file at the configured replication URL.\n";
68 echo "Does the URL point to a directory containing OSM update data?\n\n";
69 fail('replication URL not reachable.');
71 $sSetup = CONST_InstallPath.'/utils/setup.php';
73 passthru($sSetup.' --create-functions --enable-diff-updates', $iRet);
75 fail('Error running setup script');
78 $sDatabaseDate = getDatabaseDate($oDB);
79 if ($sDatabaseDate === false) {
80 fail('Cannot determine date of database.');
82 $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ', strtotime($sDatabaseDate) - (3*60*60));
84 // get the appropriate state id
86 $sCmd = CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url;
87 exec($sCmd, $aOutput, $iRet);
88 if ($iRet != 0 || $aOutput[0] == 'None') {
89 fail('Error running pyosmium tools');
92 pg_query($oDB->connection, 'TRUNCATE import_status');
93 $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
94 $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
95 if (!pg_query($oDB->connection, $sSQL)) {
96 fail('Could not enter sequence into database.');
99 echo "Done. Database updates will start at sequence $aOutput[0] ($sWindBack)\n";
102 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
103 // import diffs and files directly (e.g. from osmosis --rri)
104 $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
106 if (!file_exists($sNextFile)) {
107 fail("Cannot open $sNextFile\n");
111 $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
113 exec($sCMD, $sJunk, $iErrorLevel);
116 fail("Error from osm2pgsql, $iErrorLevel\n");
119 // Don't update the import status - we don't know what this file contains
122 if ($aResult['calculate-postcodes']) {
123 info('Update postcodes centroids');
124 $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
125 runSQLScript($sTemplate, true, true);
128 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
130 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
132 if (isset($aResult['import-node']) && $aResult['import-node']) {
134 $sContentURL = 'https://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
136 $sContentURL = 'https://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
140 if (isset($aResult['import-way']) && $aResult['import-way']) {
142 $sContentURL = 'https://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
144 $sContentURL = 'https://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
148 if (isset($aResult['import-relation']) && $aResult['import-relation']) {
150 $sContentURLsModifyXMLstr = 'https://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
152 $sContentURL = 'https://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
157 file_put_contents($sTemporaryFile, file_get_contents($sContentURL));
162 // import generated change file
163 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
165 exec($sCMD, $sJunk, $iErrorLevel);
167 fail("osm2pgsql exited with error level $iErrorLevel\n");
171 if ($aResult['deduplicate']) {
174 if (getPostgresVersion($oDB) < 9.3) {
175 fail('ERROR: deduplicate is only currently supported in postgresql 9.3');
178 $sSQL = 'select partition from country_name order by country_code';
179 $aPartitions = chksql($oDB->getCol($sSQL));
182 // we don't care about empty search_name_* partitions, they can't contain mentions of duplicates
183 foreach ($aPartitions as $i => $sPartition) {
184 $sSQL = 'select count(*) from search_name_'.$sPartition;
185 $nEntries = chksql($oDB->getOne($sSQL));
186 if ($nEntries == 0) {
187 unset($aPartitions[$i]);
191 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' '";
192 $sSQL .= ' and class is null and type is null and country_code is null';
193 $sSQL .= ' group by word_token having count(*) > 1 order by word_token';
194 $aDuplicateTokens = chksql($oDB->getAll($sSQL));
195 foreach ($aDuplicateTokens as $aToken) {
196 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
197 echo 'Deduping '.$aToken['word_token']."\n";
198 $sSQL = 'select word_id,';
199 $sSQL .= ' (select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num';
200 $sSQL .= " from word where word_token = '".$aToken['word_token'];
201 $sSQL .= "' and class is null and type is null and country_code is null order by num desc";
202 $aTokenSet = chksql($oDB->getAll($sSQL));
204 $aKeep = array_shift($aTokenSet);
205 $iKeepID = $aKeep['word_id'];
207 foreach ($aTokenSet as $aRemove) {
208 $sSQL = 'update search_name set';
209 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.'),';
210 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
211 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
212 chksql($oDB->query($sSQL));
214 $sSQL = 'update search_name set';
215 $sSQL .= ' nameaddress_vector = array_replace(nameaddress_vector,'.$aRemove['word_id'].','.$iKeepID.')';
216 $sSQL .= ' where nameaddress_vector @> ARRAY['.$aRemove['word_id'].']';
217 chksql($oDB->query($sSQL));
219 $sSQL = 'update location_area_country set';
220 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
221 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
222 chksql($oDB->query($sSQL));
224 foreach ($aPartitions as $sPartition) {
225 $sSQL = 'update search_name_'.$sPartition.' set';
226 $sSQL .= ' name_vector = array_replace(name_vector,'.$aRemove['word_id'].','.$iKeepID.')';
227 $sSQL .= ' where name_vector @> ARRAY['.$aRemove['word_id'].']';
228 chksql($oDB->query($sSQL));
230 $sSQL = 'update location_area_country set';
231 $sSQL .= ' keywords = array_replace(keywords,'.$aRemove['word_id'].','.$iKeepID.')';
232 $sSQL .= ' where keywords @> ARRAY['.$aRemove['word_id'].']';
233 chksql($oDB->query($sSQL));
236 $sSQL = 'delete from word where word_id = '.$aRemove['word_id'];
237 chksql($oDB->query($sSQL));
242 if ($aResult['recompute-word-counts']) {
243 info('Recompute frequency of full-word search terms');
244 $sTemplate = file_get_contents(CONST_BasePath.'/sql/words_from_search_name.sql');
245 runSQLScript($sTemplate, true, true);
248 if ($aResult['index']) {
249 passthru(CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
252 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
254 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
255 fail("Error: Update interval too low for download.geofabrik.de. Please check install documentation (http://nominatim.org/release-docs/latest/Import-and-Update#setting-up-the-update-process)\n");
258 $sImportFile = CONST_InstallPath.'/osmosischange.osc';
259 $sCMDDownload = CONST_Pyosmium_Binary.' --server '.CONST_Replication_Url.' -o '.$sImportFile.' -s '.CONST_Replication_Max_Diff_size;
260 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
261 $sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
264 $fStartTime = time();
265 $aLastState = chksql($oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status'));
267 if (!$aLastState['sequence_id']) {
268 echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
272 echo 'Currently at sequence '.$aLastState['sequence_id'].' ('.$aLastState['lastimportdate'].') - '.$aLastState['indexed']." indexed\n";
274 $sBatchEnd = $aLastState['lastimportdate'];
275 $iEndSequence = $aLastState['sequence_id'];
277 if ($aLastState['indexed'] == 't') {
278 // Sleep if the update interval has not yet been reached.
279 $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;
280 if ($fNextUpdate > $fStartTime) {
281 $iSleepTime = $fNextUpdate - $fStartTime;
282 echo "Waiting for next update for $iSleepTime sec.";
286 // Download the next batch of changes.
288 $fCMDStartTime = time();
289 $iNextSeq = (int) $aLastState['sequence_id'];
291 echo "$sCMDDownload -I $iNextSeq\n";
292 unlink($sImportFile);
293 exec($sCMDDownload.' -I '.$iNextSeq, $aOutput, $iResult);
296 echo 'No new updates. Sleeping for '.CONST_Replication_Recheck_Interval." sec.\n";
297 sleep(CONST_Replication_Recheck_Interval);
298 } elseif ($iResult != 0) {
299 echo 'ERROR: updates failed.';
302 $iEndSequence = (int)$aOutput[0];
306 // get the newest object from the diff file
309 exec(CONST_BasePath.'/utils/osm_file_date.py '.$sImportFile, $sBatchEnd, $iRet);
311 echo "Diff file is empty. skipping import.\n";
312 if (!$aResult['import-osmosis-all']) {
319 fail('Error getting date from diff file.');
321 $sBatchEnd = $sBatchEnd[0];
324 $fCMDStartTime = time();
325 echo $sCMDImport."\n";
327 exec($sCMDImport, $sJunk, $iErrorLevel);
329 echo "Error executing osm2pgsql: $iErrorLevel\n";
333 // write the update logs
334 $iFileSize = filesize($sImportFile);
335 $sSQL = 'INSERT INTO import_osmosis_log';
336 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
337 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
338 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
339 $sSQL .= date('Y-m-d H:i:s')."','import')";
341 chksql($oDB->query($sSQL));
344 $sSQL = "UPDATE import_status SET lastimportdate = '$sBatchEnd', indexed=false, sequence_id = $iEndSequence";
346 chksql($oDB->query($sSQL));
347 echo date('Y-m-d H:i:s')." Completed download step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
351 if (!$aResult['no-index']) {
352 $sThisIndexCmd = $sCMDIndex;
353 $fCMDStartTime = time();
355 echo "$sThisIndexCmd\n";
356 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
358 echo "Error: $iErrorLevel\n";
362 $sSQL = 'INSERT INTO import_osmosis_log';
363 $sSQL .= '(batchend, batchseq, batchsize, starttime, endtime, event)';
364 $sSQL .= " values ('$sBatchEnd',$iEndSequence,$iFileSize,'";
365 $sSQL .= date('Y-m-d H:i:s', $fCMDStartTime)."','";
366 $sSQL .= date('Y-m-d H:i:s')."','index')";
369 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
371 $sSQL = 'update import_status set indexed = true';
375 $fDuration = time() - $fStartTime;
376 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60, 2)." minutes\n";
377 if (!$aResult['import-osmosis-all']) exit(0);