4 require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
5 ini_set('memory_limit', '800M');
8 "Import / update / index osm data",
9 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
10 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
11 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
13 array('max-load', '', 0, 1, 1, 1, 'float', 'Maximum load average - indexing is paused if this is exceeded'),
14 array('max-blocking', '', 0, 1, 1, 1, 'int', 'Maximum blocking processes - indexing is aborted / paused if this is exceeded'),
16 array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
17 array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
18 array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'),
19 array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
21 array('import-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'),
23 array('import-hourly', '', 0, 1, 0, 0, 'bool', 'Import hourly diffs'),
24 array('import-daily', '', 0, 1, 0, 0, 'bool', 'Import daily diffs'),
25 array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
27 array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
28 array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
29 array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
31 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
32 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
33 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
34 array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
36 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
37 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
38 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
39 array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
41 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
43 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
45 if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
47 showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
50 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
53 // Lock to prevent multiple copies running
54 if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
56 fail("Copy already running\n");
58 if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
59 if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
60 if (getBlockingProcesses() > $aResult['max-blocking'])
62 fail("Too many blocking processes for import\n");
66 // Assume osm2pgsql is in the folder above
67 $sBasePath = dirname(dirname(__FILE__));
69 date_default_timezone_set('Etc/UTC');
73 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
74 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
76 // cache memory to be used by osm2pgsql, should not be more than the available memory
77 $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
78 if ($iCacheMemory + 500 > getTotalMemoryMB())
80 $iCacheMemory = getCacheMemoryMB();
81 echo "WARNING: resetting cache memory to $iCacheMemory\n";
83 $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
84 if (!is_null(CONST_Osm2pgsql_Flatnode_File))
86 $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
91 $bContinue = $aResult['import-all'];
92 while ($bContinue || $bFirst)
96 if ($aResult['import-hourly'])
98 // Mirror the hourly diffs
99 exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
100 $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
101 $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
102 $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
105 if ($aResult['import-daily'])
107 // Mirror the daily diffs
108 exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
109 $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
110 $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
111 $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
114 if (isset($aResult['import-diff']))
116 // import diff directly (e.g. from osmosis --rri)
117 $sNextFile = $aResult['import-diff'];
118 if (!file_exists($sNextFile))
120 fail("Cannot open $sNextFile\n");
122 // Don't update the import status - we don't know what this file contains
123 $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
126 // Missing file is not an error - it might not be created yet
127 if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
130 $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
132 exec($sCMD, $sJunk, $iErrorLevel);
136 fail("Error from osm2pgsql, $iErrorLevel\n");
139 // Move the date onwards
140 $oDB->query($sUpdateSQL);
150 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
151 if (isset($aResult['import-file']) && $aResult['import-file'])
155 if (isset($aResult['import-node']) && $aResult['import-node'])
160 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
164 $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
167 if (isset($aResult['import-way']) && $aResult['import-way'])
172 $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
176 $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
178 $sModifyXMLstr = file_get_contents($sCmd);
180 if (isset($aResult['import-relation']) && $aResult['import-relation'])
185 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
189 $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
194 // derive change from normal osm file with osmosis
195 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
196 if (isset($aResult['import-file']) && $aResult['import-file'])
198 $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
200 exec($sCMD, $sJunk, $iErrorLevel);
203 fail("Error converting osm to osc, osmosis returned: $iErrorLevel\n");
209 0 => array("pipe", "r"), // stdin
210 1 => array("pipe", "w"), // stdout
211 2 => array("pipe", "w") // stderr
213 $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
215 $hProc = proc_open($sCMD, $aSpec, $aPipes);
216 if (!is_resource($hProc))
218 fail("Error converting osm to osc, osmosis failed\n");
220 fwrite($aPipes[0], $sModifyXMLstr);
222 $sOut = stream_get_contents($aPipes[1]);
223 if ($aResult['verbose']) echo $sOut;
225 $sErrors = stream_get_contents($aPipes[2]);
226 if ($aResult['verbose']) echo $sErrors;
228 if ($iError = proc_close($hProc))
230 echo "Error converting osm to osc, osmosis returned: $iError\n";
237 // import generated change file
238 $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
240 exec($sCMD, $sJunk, $iErrorLevel);
243 fail("osm2pgsql exited with error level $iErrorLevel\n");
247 if ($aResult['deduplicate'])
250 $pgver = (float) CONST_Postgresql_Version;
252 fail("ERROR: deduplicate is only currently supported in postgresql 9.3");
256 $sSQL = 'select partition from country_name order by country_code';
257 $aPartitions = $oDB->getCol($sSQL);
258 if (PEAR::isError($aPartitions))
260 fail($aPartitions->getMessage());
264 $sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' ' and class is null and type is null and country_code is null group by word_token having count(*) > 1 order by word_token";
265 $aDuplicateTokens = $oDB->getAll($sSQL);
266 foreach($aDuplicateTokens as $aToken)
268 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
269 echo "Deduping ".$aToken['word_token']."\n";
270 $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
271 $aTokenSet = $oDB->getAll($sSQL);
272 if (PEAR::isError($aTokenSet))
274 var_dump($aTokenSet, $sSQL);
278 $aKeep = array_shift($aTokenSet);
279 $iKeepID = $aKeep['word_id'];
281 foreach($aTokenSet as $aRemove)
283 $sSQL = "update search_name set";
284 $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),";
285 $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
286 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
287 $x = $oDB->query($sSQL);
288 if (PEAR::isError($x))
294 $sSQL = "update search_name set";
295 $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
296 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
297 $x = $oDB->query($sSQL);
298 if (PEAR::isError($x))
304 $sSQL = "update location_area_country set";
305 $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
306 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
307 $x = $oDB->query($sSQL);
308 if (PEAR::isError($x))
314 foreach ($aPartitions as $sPartition)
316 $sSQL = "update search_name_".$sPartition." set";
317 $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")";
318 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
319 $x = $oDB->query($sSQL);
320 if (PEAR::isError($x))
326 $sSQL = "update location_area_country set";
327 $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
328 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
329 $x = $oDB->query($sSQL);
330 if (PEAR::isError($x))
337 $sSQL = "delete from word where word_id = ".$aRemove['word_id'];
338 $x = $oDB->query($sSQL);
339 if (PEAR::isError($x))
349 if ($aResult['index'])
351 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
352 passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
355 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
358 if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
359 fail("Error: Update interval too low for download.geofabrik.de. Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n");
362 $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
363 $sOsmosisCMD = CONST_Osmosis_Binary;
364 $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
365 $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
366 $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
367 $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
368 $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
369 if (!$aResult['no-npi']) {
374 $fStartTime = time();
377 // Logic behind this is that osm2pgsql locks the database quite a bit
378 // So it is better to import lots of small files
379 // But indexing works most efficiently on large amounts of data
380 // So do lots of small imports and a BIG index
382 // while($aResult['import-osmosis-all'] && $iFileSize > 1000)
384 if (!file_exists($sImportFile))
386 // First check if there are new updates published (except for minutelies - there's always new diffs to process)
387 if ( CONST_Replication_Update_Interval > 60 )
390 unset($aReplicationLag);
391 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
392 while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
396 echo "Error: $iErrorLevel. ";
397 echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
403 sleep(CONST_Replication_Recheck_Interval);
404 unset($aReplicationLag);
405 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
407 // There are new replication files - use osmosis to download the file
408 echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
410 $fStartTime = time();
411 $fCMDStartTime = time();
412 echo $sCMDDownload."\n";
413 exec($sCMDDownload, $sJunk, $iErrorLevel);
414 while ($iErrorLevel > 0)
416 echo "Error: $iErrorLevel\n";
418 echo 'Re-trying: '.$sCMDDownload."\n";
419 exec($sCMDDownload, $sJunk, $iErrorLevel);
421 $iFileSize = filesize($sImportFile);
422 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
423 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
426 echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
429 $iFileSize = filesize($sImportFile);
430 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
433 $fCMDStartTime = time();
434 echo $sCMDImport."\n";
435 exec($sCMDImport, $sJunk, $iErrorLevel);
438 echo "Error: $iErrorLevel\n";
441 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
444 echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
446 // Archive for debug?
447 unlink($sImportFile);
450 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
453 if (!isset($aResult['index-instances']))
455 if (getLoadAverage() < 24)
456 $iIndexInstances = 2;
458 $iIndexInstances = 1;
460 $iIndexInstances = $aResult['index-instances'];
462 $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;
463 $fCMDStartTime = time();
465 if (!$aResult['no-npi'])
467 $iFileID = $oDB->getOne('select nextval(\'file\')');
468 if (PEAR::isError($iFileID))
470 echo $iFileID->getMessage()."\n";
473 $sFileDir = CONST_BasePath.'/export/diff/';
474 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
475 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
477 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
478 $sThisIndexCmd .= $sFileDir;
479 $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
480 $sThisIndexCmd .= ".npi.out";
482 preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
483 $sFileDir = CONST_BasePath.'/export/index/';
484 $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
486 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
487 file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
490 if (!$aResult['no-index'])
492 echo "$sThisIndexCmd\n";
493 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
496 echo "Error: $iErrorLevel\n";
500 if (!$aResult['no-npi'])
502 $sFileDir = CONST_BasePath.'/export/diff/';
503 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
504 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
506 $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
507 echo "$sThisIndexCmd\n";
508 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
511 echo "Error: $iErrorLevel\n";
515 rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
516 $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
520 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
523 echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
525 $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
528 $fDuration = time() - $fStartTime;
529 echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n";
530 if (!$aResult['import-osmosis-all']) exit(0);
532 if ( CONST_Replication_Update_Interval > 60 )
534 $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
538 $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
540 echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
546 if ($aResult['import-npi-all'])
548 $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
549 if (PEAR::isError($iNPIID))
554 $sConfigDirectory = CONST_BasePath.'/settings';
555 $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
558 $fStartTime = time();
562 $sImportFile = CONST_BasePath.'/export/diff/';
563 $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
564 $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
565 $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
566 $sImportFile .= ".npi";
567 while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
569 echo "sleep (waiting for $sImportFile)\n";
572 if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
574 $iFileSize = filesize($sImportFile);
577 $fCMDStartTime = time();
578 $sCMDImport = $sCMDImportTemplate . $sImportFile;
579 echo $sCMDImport."\n";
580 exec($sCMDImport, $sJunk, $iErrorLevel);
583 fail("Error: $iErrorLevel\n");
585 $sBatchEnd = $iNPIID;
586 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
587 $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
594 function getosmosistimestamp($sOsmosisConfigDirectory)
596 $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
597 preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
598 return str_replace('\:',':',$aResult[1]);