]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/update.php
projects / nominatim.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
update comments and formatting
[nominatim.git] / utils / update.php
diff --git a/utils/update.php b/utils/update.php
index 21ac524d3d64cd8a4a5e0ce95c71eae91e5e9be0..34d30a4ff20e3f1961609530a68b80d8e1fe1f09 100755 (executable)
--- a/utils/update.php
+++ b/utils/update.php
@@ -1,8 +1,9 @@
 #!/usr/bin/php -Cq
 <?php
 
 #!/usr/bin/php -Cq
 <?php
 
-        require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
-        ini_set('memory_limit', '800M');
+       require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
+       require_once(CONST_BasePath.'/lib/init-cmd.php');
+       ini_set('memory_limit', '800M');
 
        $aCMDOptions = array(
                "Import / update / index osm data",
 
        $aCMDOptions = array(
                "Import / update / index osm data",
@@ -10,215 +11,186 @@
                array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
                array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
 
                array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
                array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
 
-               array('max-load', '', 0, 1, 1, 1, 'float', 'Maximum load average - indexing is paused if this is exceeded'),
-               array('max-blocking', '', 0, 1, 1, 1, 'int', 'Maximum blocking processes - indexing is aborted / paused if this is exceeded'),
-
                array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
                array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
                array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
                array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
-               array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'),
+               array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolate)'),
                array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
 
                array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
 
-               array('import-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'),
-
-               array('import-hourly', '', 0, 1, 0, 0, 'bool', 'Import hourly diffs'),
-               array('import-daily', '', 0, 1, 0, 0, 'bool', 'Import daily diffs'),
                array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
 
                array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
                array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
                array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
 
                array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
                array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
+               array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
 
                array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
                array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
                array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
 
                array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
                array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
                array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
+               array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
 
                array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
                array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
                array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
 
                array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
                array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
                array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
-               array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
 
                array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 
 
                array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 
-       if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
-       {
-               showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
-       }
-
        if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
        if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
        if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
        if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
-/*
-       // Lock to prevent multiple copies running
-       if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
-       {
-               echo "Copy already running\n";
-               exit;
-       }
-       if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
-       if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
-       if (getBlockingProcesses() > $aResult['max-blocking'])
-       {
-               echo "Too many blocking processes for import\n";
-               exit;
-       }
-*/
 
 
-       // Assume osm2pgsql is in the folder above
-       $sBasePath = dirname(dirname(__FILE__));
+       date_default_timezone_set('Etc/UTC');
 
        $oDB =& getDB();
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 
        $oDB =& getDB();
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+       if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
 
-       $bFirst = true;
-       $bContinue = $aResult['import-all'];
-       while ($bContinue || $bFirst)
+       // cache memory to be used by osm2pgsql, should not be more than the available memory
+       $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
+       if ($iCacheMemory + 500 > getTotalMemoryMB())
        {
        {
-               $bFirst = false;
+               $iCacheMemory = getCacheMemoryMB();
+               echo "WARNING: resetting cache memory to $iCacheMemory\n";
+       }
+       $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
+       if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+       {
+               $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+       }
 
 
-               if ($aResult['import-hourly'])
-               {
-                       // Mirror the hourly diffs
-                       exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
-                       $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
-                       $sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
-                       $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
-               }
 
 
-               if ($aResult['import-daily'])
-               {
-                       // Mirror the daily diffs
-                       exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
-                       $sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
-                       $sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
-                       $sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
-               }
-               
-               if (isset($aResult['import-diff']))
+       if (isset($aResult['import-diff']))
+       {
+               // import diff directly (e.g. from osmosis --rri)
+               $sNextFile = $aResult['import-diff'];
+               if (!file_exists($sNextFile))
                {
                {
-                       // import diff directly (e.g. from osmosis --rri)
-                       $sNextFile = $aResult['import-diff'];
-                       if (!file_exists($sNextFile))
-                       {
-                               echo "Cannot open $nextFile\n";
-                               exit;
-                       }
-                       // Don't update the import status - we don't know what this file contains
-                       $sUpdateSQL = 'update import_status set lastimportdate = now() where false';
+                       fail("Cannot open $sNextFile\n");
                }
 
                }
 
-               // Missing file is not an error - it might not be created yet
-               if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
-               {
-                       // Import the file
-                       $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;
-                       echo $sCMD."\n";
-                       exec($sCMD, $sJunk, $iErrorLevel);
+               // Import the file
+               $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
+               echo $sCMD."\n";
+               exec($sCMD, $sJunk, $iErrorLevel);
 
 
-                       if ($iErrorLevel)
-                       {
-                               echo "Error from osm2pgsql, $iErrorLevel\n";
-                               exit;
-                       }
-       
-                       // Move the date onwards
-                       $oDB->query($sUpdateSQL);
-               }
-               else
+               if ($iErrorLevel)
                {
                {
-                       $bContinue = false;
+                       fail("Error from osm2pgsql, $iErrorLevel\n");
                }
                }
+
+               // Don't update the import status - we don't know what this file contains
        }
 
        }
 
-       $bModifyXML = false;
-       $sModifyXMLstr = '';
+       $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
+       $bHaveDiff = false;
        if (isset($aResult['import-file']) && $aResult['import-file'])
        {
        if (isset($aResult['import-file']) && $aResult['import-file'])
        {
-               $bModifyXML = true;
+               $bHaveDiff = true;
+               $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
+               echo $sCMD."\n";
+               exec($sCMD, $sJunk, $iErrorLevel);
+               if ($iErrorLevel)
+               {
+                       fail("Error converting osm to osc, osmosis returned: $iErrorLevel\n");
+               }
        }
        }
+
+       $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
+       $sContentURL = '';
        if (isset($aResult['import-node']) && $aResult['import-node'])
        {
        if (isset($aResult['import-node']) && $aResult['import-node'])
        {
-               $bModifyXML = true;
-               $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
+               if ($bUseOSMApi)
+               {
+                       $sContentURL = 'http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
+               }
+               else
+               {
+                       $sContentURL = 'http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
+               }
        }
        if (isset($aResult['import-way']) && $aResult['import-way'])
        {
        }
        if (isset($aResult['import-way']) && $aResult['import-way'])
        {
-               $bModifyXML = true;
-               $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full');
+               if ($bUseOSMApi)
+               {
+                       $sContentURL = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
+               }
+               else
+               {
+                       $sContentURL = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
+               }
        }
        if (isset($aResult['import-relation']) && $aResult['import-relation'])
        {
        }
        if (isset($aResult['import-relation']) && $aResult['import-relation'])
        {
-               $bModifyXML = true;
-               $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
+               if ($bUseOSMApi)
+               {
+                       $sContentURLsModifyXMLstr = 'http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
+               }
+               else
+               {
+                       $sContentURL = 'http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
+               }
        }
        }
-       if ($bModifyXML)
+       if ($sContentURL)
        {
        {
-               // derive change from normal osm file with osmosis
-               $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
-               if (isset($aResult['import-file']) && $aResult['import-file'])
+               $sModifyXMLstr = file_get_contents($sContentURL);
+               $bHaveDiff = true;
+
+               $aSpec = array(
+                       0 => array("pipe", "r"),  // stdin
+                       1 => array("pipe", "w"),  // stdout
+                       2 => array("pipe", "w") // stderr
+               );
+               $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
+               echo $sCMD."\n";
+               $hProc = proc_open($sCMD, $aSpec, $aPipes);
+               if (!is_resource($hProc))
                {
                {
-                       $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
-                       echo $sCMD."\n";
-                       exec($sCMD, $sJunk, $iErrorLevel);
-                       if ($iErrorLevel)
-                       {
-                               echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n";
-                               exit;
-                       }
+                       fail("Error converting osm to osc, osmosis failed\n");
                }
                }
-               else
+               fwrite($aPipes[0], $sModifyXMLstr);
+               fclose($aPipes[0]);
+               $sOut = stream_get_contents($aPipes[1]);
+               if ($aResult['verbose']) echo $sOut;
+               fclose($aPipes[1]);
+               $sErrors = stream_get_contents($aPipes[2]);
+               if ($aResult['verbose']) echo $sErrors;
+               fclose($aPipes[2]);
+               if ($iError = proc_close($hProc))
                {
                {
-                       $aSpec = array(
-                               0 => array("pipe", "r"),  // stdin
-                               1 => array("pipe", "w"),  // stdout
-                               2 => array("pipe", "w") // stderr
-                       );
-                       $sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
-                       echo $sCMD."\n";
-                       $hProc = proc_open($sCMD, $aSpec, $aPipes);
-                       if (!is_resource($hProc))
-                       {
-                               echo "Error converting osm to osc, osmosis failed\n";
-                               exit;
-                       }
-                       fwrite($aPipes[0], $sModifyXMLstr);
-                       fclose($aPipes[0]);
-                       $sOut = stream_get_contents($aPipes[1]);
-                       if ($aResult['verbose']) echo $sOut;
-                       fclose($aPipes[1]);
-                       $sErrors = stream_get_contents($aPipes[2]);
-                       if ($aResult['verbose']) echo $sErrors;
-                       fclose($aPipes[2]);
-                       if ($iError = proc_close($hProc))
-                       {
-                               echo "Error converting osm to osc, osmosis returned: $iError\n";
-                               echo $sOut;
-                               echo $sErrors;
-                               exit;
-                       }
+                       echo $sOut;
+                       echo $sErrors;
+                       fail("Error converting osm to osc, osmosis returned: $iError\n");
                }
                }
+       }
 
 
+       if ($bHaveDiff)
+       {
                // import generated change file
                // import generated change file
-               $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
+               $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
                echo $sCMD."\n";
                exec($sCMD, $sJunk, $iErrorLevel);
                if ($iErrorLevel)
                {
                echo $sCMD."\n";
                exec($sCMD, $sJunk, $iErrorLevel);
                if ($iErrorLevel)
                {
-                       echo "osm2pgsql exited with error level $iErrorLevel\n";
-                       exit;
+                       fail("osm2pgsql exited with error level $iErrorLevel\n");
                }
        }
 
        if ($aResult['deduplicate'])
        {
                }
        }
 
        if ($aResult['deduplicate'])
        {
-                $oDB =& getDB();
-                $sSQL = 'select partition from country_name order by country_code';
-                $aPartitions = $oDB->getCol($sSQL);
-                if (PEAR::isError($aPartitions))
-                {
-                        fail($aPartitions->getMessage());
-                }
-                $aPartitions[] = 0;
+
+               if (getPostgresVersion() < 9.3)
+               {
+                       fail("ERROR: deduplicate is only currently supported in postgresql 9.3");
+               }
+
+               $oDB =& getDB();
+               $sSQL = 'select partition from country_name order by country_code';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               $aPartitions[] = 0;
 
                $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";
                $aDuplicateTokens = $oDB->getAll($sSQL);
 
                $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";
                $aDuplicateTokens = $oDB->getAll($sSQL);
@@ -231,76 +203,65 @@
                        if (PEAR::isError($aTokenSet))
                        {
                                var_dump($aTokenSet, $sSQL);
                        if (PEAR::isError($aTokenSet))
                        {
                                var_dump($aTokenSet, $sSQL);
-                               exit;
+                               exit(1);
                        }
                        }
-                       
+
                        $aKeep = array_shift($aTokenSet);
                        $iKeepID = $aKeep['word_id'];
 
                        foreach($aTokenSet as $aRemove)
                        {
                                $sSQL = "update search_name set";
                        $aKeep = array_shift($aTokenSet);
                        $iKeepID = $aKeep['word_id'];
 
                        foreach($aTokenSet as $aRemove)
                        {
                                $sSQL = "update search_name set";
-                               $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
-                               $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
+                               $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),";
+                               $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
                                $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
                                $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
-                                       exit;
+                                       exit(1);
                                }
 
                                $sSQL = "update search_name set";
                                }
 
                                $sSQL = "update search_name set";
-                               $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
+                               $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
                                $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
                                $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
-                                       exit;
+                                       exit(1);
                                }
 
                                $sSQL = "update location_area_country set";
                                }
 
                                $sSQL = "update location_area_country set";
-                               $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
+                               $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
                                $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
                                $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
                                $x = $oDB->query($sSQL);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
-                                       exit;
+                                       exit(1);
                                }
 
                                foreach ($aPartitions as $sPartition)
                                {
                                        $sSQL = "update search_name_".$sPartition." set";
                                }
 
                                foreach ($aPartitions as $sPartition)
                                {
                                        $sSQL = "update search_name_".$sPartition." set";
-                                       $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.",";
-                                       $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
+                                       $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")";
                                        $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
                                        $x = $oDB->query($sSQL);
                                        if (PEAR::isError($x))
                                        {
                                                var_dump($x);
                                        $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
                                        $x = $oDB->query($sSQL);
                                        if (PEAR::isError($x))
                                        {
                                                var_dump($x);
-                                               exit;
-                                       }
-
-                                       $sSQL = "update search_name_".$sPartition." set";
-                                       $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID;
-                                       $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]";
-                                       $x = $oDB->query($sSQL);
-                                       if (PEAR::isError($x))
-                                       {
-                                               var_dump($x);
-                                               exit;
+                                               exit(1);
                                        }
 
                                        $sSQL = "update location_area_country set";
                                        }
 
                                        $sSQL = "update location_area_country set";
-                                       $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID;
+                                       $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")";
                                        $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
                                        $x = $oDB->query($sSQL);
                                        if (PEAR::isError($x))
                                        {
                                                var_dump($x);
                                        $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
                                        $x = $oDB->query($sSQL);
                                        if (PEAR::isError($x))
                                        {
                                                var_dump($x);
-                                               exit;
+                                               exit(1);
                                        }
                                }
 
                                        }
                                }
 
@@ -309,112 +270,106 @@
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
                                if (PEAR::isError($x))
                                {
                                        var_dump($x);
-                                       exit;
+                                       exit(1);
                                }
                        }
                                }
                        }
-
                }
        }
 
        if ($aResult['index'])
        {
                }
        }
 
        if ($aResult['index'])
        {
-               passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
+               passthru(CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
        }
 
        if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
        {
        }
 
        if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
        {
-               $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
-               $sOsmosisCMD = CONST_Osmosis_Binary;
-               $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
-               $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
-               $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile;
-               $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
-               if (!$aResult['no-npi']) {
-                       $sCMDIndex .= '-F ';
+
+               if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
+                       fail("Error: Update interval too low for download.geofabrik.de.  Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n");
                }
                }
+
+               $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
+               $sOsmosisConfigDirectory = CONST_InstallPath.'/settings';
+               $sCMDDownload = CONST_Osmosis_Binary.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
+               $sCMDCheckReplicationLag = CONST_Osmosis_Binary.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
+               $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
+               $sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
+
                while(true)
                {
                        $fStartTime = time();
                        $iFileSize = 1001;
 
                while(true)
                {
                        $fStartTime = time();
                        $iFileSize = 1001;
 
-                       // Logic behind this is that osm2pgsql locks the database quite a bit
-                       // So it is better to import lots of small files
-                       // But indexing works most efficiently on large amounts of data
-                       // So do lots of small imports and a BIG index
-
-//                     while($aResult['import-osmosis-all'] && $iFileSize > 1000)
-//                     {
-                               if (!file_exists($sImportFile))
+                       if (!file_exists($sImportFile))
+                       {
+                               // First check if there are new updates published (except for minutelies - there's always new diffs to process)
+                               if ( CONST_Replication_Update_Interval > 60 )
                                {
                                {
-                                       // Use osmosis to download the file
-                                       $fCMDStartTime = time();
-                                       echo $sCMDDownload."\n";
-                                       exec($sCMDDownload, $sJunk, $iErrorLevel);
-                                       while ($iErrorLevel == 1)
+
+                                       unset($aReplicationLag);
+                                       exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
+                                       while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
                                        {
                                        {
-                                               echo "Error: $iErrorLevel\n";
-                                               sleep(60);
-                                               echo 'Re-trying: '.$sCMDDownload."\n";
-                                               exec($sCMDDownload, $sJunk, $iErrorLevel);
+                                               if ($iErrorLevel)
+                                               {
+                                                       echo "Error: $iErrorLevel. ";
+                                                       echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
+                                               }
+                                               else
+                                               {
+                                                       echo ".";
+                                               }
+                                               sleep(CONST_Replication_Recheck_Interval);
+                                               unset($aReplicationLag);
+                                               exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
                                        }
                                        }
-                                       $iFileSize = filesize($sImportFile);
-                                       $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
-                                       echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
-                                       $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
-                                       $oDB->query($sSQL);
+                                       // There are new replication files - use osmosis to download the file
+                                       echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
                                }
                                }
-
-                               $iFileSize = filesize($sImportFile);
-                               $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
-               
-                               // Import the file
+                               $fStartTime = time();
                                $fCMDStartTime = time();
                                $fCMDStartTime = time();
-                               echo $sCMDImport."\n";
-                               exec($sCMDImport, $sJunk, $iErrorLevel);
-                               if ($iErrorLevel)
+                               echo $sCMDDownload."\n";
+                               exec($sCMDDownload, $sJunk, $iErrorLevel);
+                               while ($iErrorLevel > 0)
                                {
                                        echo "Error: $iErrorLevel\n";
                                {
                                        echo "Error: $iErrorLevel\n";
-                                       exit;
+                                       sleep(60);
+                                       echo 'Re-trying: '.$sCMDDownload."\n";
+                                       exec($sCMDDownload, $sJunk, $iErrorLevel);
                                }
                                }
-                               echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
-                               $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
+                               $iFileSize = filesize($sImportFile);
+                               $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
+                               $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
                                var_Dump($sSQL);
                                $oDB->query($sSQL);
                                var_Dump($sSQL);
                                $oDB->query($sSQL);
+                               echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
+                       }
+
+                       $iFileSize = filesize($sImportFile);
+                       $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
+       
+                       // Import the file
+                       $fCMDStartTime = time();
+                       echo $sCMDImport."\n";
+                       exec($sCMDImport, $sJunk, $iErrorLevel);
+                       if ($iErrorLevel)
+                       {
+                               echo "Error: $iErrorLevel\n";
+                               exit($iErrorLevel);
+                       }
+                       $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
+                       var_Dump($sSQL);
+                       $oDB->query($sSQL);
+                       echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 
 
-                               // Archive for debug?
-                               unlink($sImportFile);
-//                     }
+                       // Archive for debug?
+                       unlink($sImportFile);
 
                        $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 
                        // Index file
                        $sThisIndexCmd = $sCMDIndex;
 
                        $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 
                        // Index file
                        $sThisIndexCmd = $sCMDIndex;
-
-                       if (!$aResult['no-npi'])
-                       {
-                               $fCMDStartTime = time();
-                               $iFileID = $oDB->getOne('select nextval(\'file\')');
-                               if (PEAR::isError($iFileID))
-                               {
-                                       echo $iFileID->getMessage()."\n";
-                                       exit;
-                               } 
-                               $sFileDir = CONST_BasePath.'/export/diff/';
-                               $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
-                               $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
-
-                               if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
-                               $sThisIndexCmd .= $sFileDir;
-                               $sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
-                               $sThisIndexCmd .= ".npi.out";
-
-                               preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
-                               $sFileDir = CONST_BasePath.'/export/index/';
-                               $sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
-
-                               if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
-                               file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
-                       }
+                       $fCMDStartTime = time();
 
                        if (!$aResult['no-index'])
                        {
 
                        if (!$aResult['no-index'])
                        {
@@ -423,93 +378,33 @@
                                if ($iErrorLevel)
                                {
                                        echo "Error: $iErrorLevel\n";
                                if ($iErrorLevel)
                                {
                                        echo "Error: $iErrorLevel\n";
-                                       exit;
-                               }
-
-                               if (!$aResult['no-npi'])
-                               {
-                                       $sFileDir = CONST_BasePath.'/export/diff/';
-                                       $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
-                                       $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
-
-                                       $sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
-                                       echo "$sThisIndexCmd\n";
-                                       exec($sThisIndexCmd, $sJunk, $iErrorLevel);
-                                       if ($iErrorLevel)
-                                       {
-                                               echo "Error: $iErrorLevel\n";
-                                               exit;
-                                       }
-
-                                       rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
-                                               $sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
+                                       exit($iErrorLevel);
                                }
                        }
 
                                }
                        }
 
-                       echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
                        $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
                        $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
+                       var_Dump($sSQL);
                        $oDB->query($sSQL);
                        $oDB->query($sSQL);
+                       echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
 
                        $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
                        $oDB->query($sSQL);
 
                        $fDuration = time() - $fStartTime;
 
                        $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
                        $oDB->query($sSQL);
 
                        $fDuration = time() - $fStartTime;
-                       echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
-                       if (!$aResult['import-osmosis-all']) exit;
-
-                       echo "Sleeping ".max(0,60-$fDuration)." seconds\n";
-                       sleep(max(0,60-$fDuration));
-               }
-
-       }
+                       echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n";
+                       if (!$aResult['import-osmosis-all']) exit(0);
 
 
-       if ($aResult['import-npi-all'])
-       {
-               $iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
-               if (PEAR::isError($iNPIID))
-               {
-                       var_dump($iNPIID);
-                       exit;
-               }
-               $sConfigDirectory = CONST_BasePath.'/settings';
-               $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F ';
-               while(true)
-               {
-                       $fStartTime = time();
-
-                       $iNPIID++;
-
-                       $sImportFile = CONST_BasePath.'/export/diff/';
-                       $sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
-                       $sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
-                       $sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
-                       $sImportFile .= ".npi";
-                       while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
+                       if ( CONST_Replication_Update_Interval > 60 )
                        {
                        {
-                               echo "sleep (waiting for $sImportFile)\n";
-                               sleep(10);
+                               $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
                        }
                        }
-                       if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
-
-                       $iFileSize = filesize($sImportFile);
-               
-                       // Import the file
-                       $fCMDStartTime = time();
-                       $sCMDImport = $sCMDImportTemplate . $sImportFile;
-                       echo $sCMDImport."\n";
-                       exec($sCMDImport, $sJunk, $iErrorLevel);
-                       if ($iErrorLevel)
+                       else
                        {
                        {
-                               echo "Error: $iErrorLevel\n";
-                               exit;
+                               $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
                        }
                        }
-                       $sBatchEnd = $iNPIID;
-                       echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
-                       $sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
-                       var_Dump($sSQL);
-                       $oDB->query($sSQL);
+                       echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
+                       sleep($iSleep);
                }
                }
-               
        }
 
        function getosmosistimestamp($sOsmosisConfigDirectory)
        }
 
        function getosmosistimestamp($sOsmosisConfigDirectory)
Open Source search based on OpenStreetMap data