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'),
30 array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
31 array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
32 array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
33 array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
35 array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
36 array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
37 array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
38 array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
40 array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
42 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
44 if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
46 showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
49 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
50 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
52 // Lock to prevent multiple copies running
53 if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
55 echo "Copy already running\n";
58 if (!isset($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 echo "Too many blocking processes for import\n";
67 // Assume osm2pgsql is in the folder above
68 $sBasePath = dirname(dirname(__FILE__));
70 date_default_timezone_set('Etc/UTC');
74 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
77 $bContinue = $aResult['import-all'];
78 while ($bContinue || $bFirst)
82 if ($aResult['import-hourly'])
84 // Mirror the hourly diffs
85 exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
86 $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
87 $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
88 $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
91 if ($aResult['import-daily'])
93 // Mirror the daily diffs
94 exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
95 $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
96 $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
97 $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
100 if (isset($aResult['import-diff']))
102 // import diff directly (e.g. from osmosis --rri)
103 $sNextFile = $aResult['import-diff'];
104 if (!file_exists($sNextFile))
106 echo "Cannot open $nextFile\n";
109 // Don't update the import status - we don't know what this file contains
110 $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
113 // Missing file is not an error - it might not be created yet
114 if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
117 $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;
119 exec($sCMD, $sJunk, $iErrorLevel);
123 echo "Error from osm2pgsql, $iErrorLevel\n";
127 // Move the date onwards
128 $oDB->query($sUpdateSQL);
138 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
139 if (isset($aResult['import-file']) && $aResult['import-file'])
143 if (isset($aResult['import-node']) && $aResult['import-node'])
148 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
152 $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
155 if (isset($aResult['import-way']) && $aResult['import-way'])
160 $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
164 $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
166 $sModifyXMLstr = file_get_contents($sCmd);
168 if (isset($aResult['import-relation']) && $aResult['import-relation'])
173 $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
177 $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
182 // derive change from normal osm file with osmosis
183 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
184 if (isset($aResult['import-file']) && $aResult['import-file'])
186 $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
188 exec($sCMD, $sJunk, $iErrorLevel);
191 echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n";
198 0 => array("pipe", "r"), // stdin
199 1 => array("pipe", "w"), // stdout
200 2 => array("pipe", "w") // stderr
202 $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
204 $hProc = proc_open($sCMD, $aSpec, $aPipes);
205 if (!is_resource($hProc))
207 echo "Error converting osm to osc, osmosis failed\n";
210 fwrite($aPipes[0], $sModifyXMLstr);
212 $sOut = stream_get_contents($aPipes[1]);
213 if ($aResult['verbose']) echo $sOut;
215 $sErrors = stream_get_contents($aPipes[2]);
216 if ($aResult['verbose']) echo $sErrors;
218 if ($iError = proc_close($hProc))
220 echo "Error converting osm to osc, osmosis returned: $iError\n";
227 // import generated change file
228 $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
230 exec($sCMD, $sJunk, $iErrorLevel);
233 echo "osm2pgsql exited with error level $iErrorLevel\n";
238 if ($aResult['deduplicate'])
241 $sSQL = 'select partition from country_name order by country_code';
242 $aPartitions = $oDB->getCol($sSQL);
243 if (PEAR::isError($aPartitions))
245 fail($aPartitions->getMessage());
249 $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";
250 $aDuplicateTokens = $oDB->getAll($sSQL);
251 foreach($aDuplicateTokens as $aToken)
253 if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
254 echo "Deduping ".$aToken['word_token']."\n";
255 $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
256 $aTokenSet = $oDB->getAll($sSQL);
257 if (PEAR::isError($aTokenSet))
259 var_dump($aTokenSet, $sSQL);
263 $aKeep = array_shift($aTokenSet);
264 $iKeepID = $aKeep['word_id'];
266 foreach($aTokenSet as $aRemove)
268 $sSQL = "update search_name set";
269 $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
270 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
271 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
272 $x = $oDB->query($sSQL);
273 if (PEAR::isError($x))
279 $sSQL = "update search_name set";
280 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
281 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
282 $x = $oDB->query($sSQL);
283 if (PEAR::isError($x))
289 $sSQL = "update location_area_country set";
290 $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
291 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
292 $x = $oDB->query($sSQL);
293 if (PEAR::isError($x))
299 foreach ($aPartitions as $sPartition)
301 $sSQL = "update search_name_".$sPartition." set";
302 $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
303 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
304 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
305 $x = $oDB->query($sSQL);
306 if (PEAR::isError($x))
312 $sSQL = "update search_name_".$sPartition." set";
313 $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
314 $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
315 $x = $oDB->query($sSQL);
316 if (PEAR::isError($x))
322 $sSQL = "update location_area_country set";
323 $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
324 $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
325 $x = $oDB->query($sSQL);
326 if (PEAR::isError($x))
333 $sSQL = "delete from word where word_id = ".$aRemove['word_id'];
334 $x = $oDB->query($sSQL);
335 if (PEAR::isError($x))
345 if ($aResult['index'])
347 passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
350 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
352 $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
353 $sOsmosisCMD = CONST_Osmosis_Binary;
354 $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
355 $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
356 $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
357 $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile;
358 $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
359 if (!$aResult['no-npi']) {
364 $fStartTime = time();
367 // Logic behind this is that osm2pgsql locks the database quite a bit
368 // So it is better to import lots of small files
369 // But indexing works most efficiently on large amounts of data
370 // So do lots of small imports and a BIG index
372 // while($aResult['import-osmosis-all'] && $iFileSize > 1000)
374 if (!file_exists($sImportFile))
376 // First check if there are new updates published (except for minutelies - there's always new diffs to process)
377 if ( CONST_Replication_Update_Interval > 60 )
380 unset($aReplicationLag);
381 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
382 while ($iErrorLevel == 1 || $aReplicationLag[0] < 1)
386 echo "Error: $iErrorLevel. ";
387 echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
393 sleep(CONST_Replication_Recheck_Interval);
394 unset($aReplicationLag);
395 exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
397 // There are new replication files - use osmosis to download the file
398 echo "\nReplication Delay is ".$aReplicationLag[0]."\n";
400 $fCMDStartTime = time();
401 echo $sCMDDownload."\n";
402 exec($sCMDDownload, $sJunk, $iErrorLevel);
403 while ($iErrorLevel == 1)
405 echo "Error: $iErrorLevel\n";
407 echo 'Re-trying: '.$sCMDDownload."\n";
408 exec($sCMDDownload, $sJunk, $iErrorLevel);
410 $iFileSize = filesize($sImportFile);
411 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
412 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
413 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
417 $iFileSize = filesize($sImportFile);
418 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
421 $fCMDStartTime = time();
422 echo $sCMDImport."\n";
423 exec($sCMDImport, $sJunk, $iErrorLevel);
426 echo "Error: $iErrorLevel\n";
429 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
430 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
434 // Archive for debug?
435 unlink($sImportFile);
438 $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
441 $sThisIndexCmd = $sCMDIndex;
443 if (!$aResult['no-npi'])
445 $fCMDStartTime = time();
446 $iFileID = $oDB->getOne('select nextval(\'file\')');
447 if (PEAR::isError($iFileID))
449 echo $iFileID->getMessage()."\n";
452 $sFileDir = CONST_BasePath.'/export/diff/';
453 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
454 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
456 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
457 $sThisIndexCmd .= $sFileDir;
458 $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
459 $sThisIndexCmd .= ".npi.out";
461 preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
462 $sFileDir = CONST_BasePath.'/export/index/';
463 $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
465 if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
466 file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
469 if (!$aResult['no-index'])
471 echo "$sThisIndexCmd\n";
472 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
475 echo "Error: $iErrorLevel\n";
479 if (!$aResult['no-npi'])
481 $sFileDir = CONST_BasePath.'/export/diff/';
482 $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
483 $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
485 $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
486 echo "$sThisIndexCmd\n";
487 exec($sThisIndexCmd, $sJunk, $iErrorLevel);
490 echo "Error: $iErrorLevel\n";
494 rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
495 $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
499 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
500 $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
503 $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
506 $fDuration = time() - $fStartTime;
507 echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
508 if (!$aResult['import-osmosis-all']) exit;
510 if ( CONST_Replication_Update_Interval > 60 )
512 $iSleep = round(CONST_Replication_Update_Interval*0.8);
516 $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
518 echo "Sleeping $iSleep seconds\n";
524 if ($aResult['import-npi-all'])
526 $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
527 if (PEAR::isError($iNPIID))
532 $sConfigDirectory = CONST_BasePath.'/settings';
533 $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
536 $fStartTime = time();
540 $sImportFile = CONST_BasePath.'/export/diff/';
541 $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
542 $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
543 $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
544 $sImportFile .= ".npi";
545 while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
547 echo "sleep (waiting for $sImportFile)\n";
550 if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
552 $iFileSize = filesize($sImportFile);
555 $fCMDStartTime = time();
556 $sCMDImport = $sCMDImportTemplate . $sImportFile;
557 echo $sCMDImport."\n";
558 exec($sCMDImport, $sJunk, $iErrorLevel);
561 echo "Error: $iErrorLevel\n";
564 $sBatchEnd = $iNPIID;
565 echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
566 $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
573 function getosmosistimestamp($sOsmosisConfigDirectory)
575 $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
576 preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
577 return str_replace('\:',':',$aResult[1]);