upper bound on bulks

[nominatim.git] / utils / update.php

diff --git a/utils/update.php b/utils/update.php
index 21ac524d3d64cd8a4a5e0ce95c71eae91e5e9be0..e8972f1d26f4d3cefa962667459966d38f548050 100755 (executable)
--- a/utils/update.php
+++ b/utils/update.php
@@ -30,6 +30,7 @@
                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'),
@@ -45,7 +46,6 @@
                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;
 /*
        // Lock to prevent multiple copies running
@@ -66,6 +66,8 @@
        // 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);
@@ -132,6 +134,7 @@
 
        $bModifyXML = false;
        $sModifyXMLstr = '';
+       $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
        if (isset($aResult['import-file']) && $aResult['import-file'])
        {
                $bModifyXML = true;
@@ -139,17 +142,39 @@
        if (isset($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)
+               {
+                       $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
+               }
+               else
+               {
+                       $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
+               }
        }
        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)
+               {
+                       $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
+               }
+               else
+               {
+                       $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
+               }
+               $sModifyXMLstr = file_get_contents($sCmd);
        }
        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)
+               {
+                       $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
+               }
+               else
+               {
+                       $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
+               }
        }
        if ($bModifyXML)
        {
@@ -318,6 +343,7 @@
 
        if ($aResult['index'])
        {
+               if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
                passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
        }
 
@@ -327,8 +353,9 @@
                $sOsmosisCMD = CONST_Osmosis_Binary;
                $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
                $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
+               $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
                $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile;
-               $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
+               $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'];
                if (!$aResult['no-npi']) {
                        $sCMDIndex .= '-F ';
                }
@@ -346,7 +373,30 @@
 //                     {
                                if (!file_exists($sImportFile))
                                {
-                                       // Use osmosis to download the file
+                                       // First check if there are new updates published (except for minutelies - there's always new diffs to process)
+                                       if ( CONST_Replication_Update_Interval > 60 )
+                                       {
+
+                                               unset($aReplicationLag);
+                                               exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
+                                               while ($iErrorLevel == 1 || $aReplicationLag[0] < 1)
+                                               {
+                                                       if ($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, $iErrorLevel); 
+                                               }
+                                               // There are new replication files - use osmosis to download the file
+                                               echo "\nReplication Delay is ".$aReplicationLag[0]."\n";
+                                       }
                                        $fCMDStartTime = time();
                                        echo $sCMDDownload."\n";
                                        exec($sCMDDownload, $sJunk, $iErrorLevel);
@@ -388,7 +438,16 @@
                        $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
 
                        // Index file
-                       $sThisIndexCmd = $sCMDIndex;
+                       if (!isset($aResult['index-instances']))
+                       {
+                               if (getLoadAverage() < 15)
+                                       $iIndexInstances = 2;
+                               else
+                                       $iIndexInstances = 1;
+                       } else
+                               $iIndexInstances = $aResult['index-instances'];
+
+                       $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;
 
                        if (!$aResult['no-npi'])
                        {
@@ -457,8 +516,16 @@
                        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));
+                       if ( CONST_Replication_Update_Interval > 60 )
+                       {
+                               $iSleep = round(CONST_Replication_Update_Interval*0.8);
+                       }
+                       else
+                       {
+                               $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
+                       }
+                       echo "Sleeping $iSleep seconds\n";
+                       sleep($iSleep);
                }
 
        }