restrict depth of word set calculation

[nominatim.git] / utils / update.php

diff --git a/utils/update.php b/utils/update.php
index f8bbfb15ee955ed9069d7fad8fdda835fbf1bbf1..76d6f8582c2e1e9646845890b1d451b5ff8c6527 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-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', '', 0, 1, 0, 0, 'bool', 'Index'),
                array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),


@@ -46,6 +47,7 @@
        }
 
        if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
        }
 
        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)
 /*
        // 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)


@@ -65,8 +67,12 @@
        // Assume osm2pgsql is in the folder above
        $sBasePath = dirname(dirname(__FILE__));
 
        // Assume osm2pgsql is in the folder above
        $sBasePath = dirname(dirname(__FILE__));
 

+       date_default_timezone_set('Etc/UTC');
+

        $oDB =& getDB();
 
        $oDB =& getDB();
 

+       $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+

        $bFirst = true;
        $bContinue = $aResult['import-all'];
        while ($bContinue || $bFirst)
        $bFirst = true;
        $bContinue = $aResult['import-all'];
        while ($bContinue || $bFirst)


@@ -95,7 +101,7 @@
                {
                        // import diff directly (e.g. from osmosis --rri)
                        $sNextFile = $aResult['import-diff'];
                {
                        // import diff directly (e.g. from osmosis --rri)
                        $sNextFile = $aResult['import-diff'];

-                       if (!file_exists($nextFile))
+                       if (!file_exists($sNextFile))

                        {
                                echo "Cannot open $nextFile\n";
                                exit;
                        {
                                echo "Cannot open $nextFile\n";
                                exit;


@@ -105,10 +111,10 @@
                }
 
                // Missing file is not an error - it might not be created yet
                }
 
                // Missing file is not an error - it might not be created yet

-               if (($aResult['import-hourly'] || $aResult['import-daily']) && file_exists($sNextFile))
+               if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))

                {
                        // Import the file
                {
                        // Import the file

-                       $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sNextFile;
+                       $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;

                        echo $sCMD."\n";
                        exec($sCMD, $sJunk, $iErrorLevel);
 
                        echo $sCMD."\n";
                        exec($sCMD, $sJunk, $iErrorLevel);
 


@@ -127,63 +133,104 @@
                }
        }
 
                }
        }
 

-       $sModifyXML = false;
+       $bModifyXML = false;
+       $sModifyXMLstr = '';
+       $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];

        if (isset($aResult['import-file']) && $aResult['import-file'])
        {
        if (isset($aResult['import-file']) && $aResult['import-file'])
        {

-               $sModifyXML = file_get_contents($aResult['import-file']);
+               $bModifyXML = true;

        }
        if (isset($aResult['import-node']) && $aResult['import-node'])
        {
        }
        if (isset($aResult['import-node']) && $aResult['import-node'])
        {

-               $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
+               $bModifyXML = true;
+               if ($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']) && $aResult['import-way'])
        {
        }
        if (isset($aResult['import-way']) && $aResult['import-way'])
        {

-               $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full');
+               $bModifyXML = true;
+               if ($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']) && $aResult['import-relation'])
        {
        }
        if (isset($aResult['import-relation']) && $aResult['import-relation'])
        {

-               $sModifyXML = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
+               $bModifyXML = true;
+               if ($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 ($sModifyXML)
+       if ($bModifyXML)

        {
        {

-               // Hack into a modify request
-               $sModifyXML = str_replace('<osm version="0.6" generator="OpenStreetMap server">',
-                       '<osmChange version="0.6" generator="OpenStreetMap server"><modify>', $sModifyXML);
-               $sModifyXML = str_replace('<osm version=\'0.6\' upload=\'true\' generator=\'JOSM\'>',
-                       '<osmChange version="0.6" generator="OpenStreetMap server"><modify>', $sModifyXML);
-               $sModifyXML = str_replace('</osm>', '</modify></osmChange>', $sModifyXML);
-
-               // Outputing this is too verbose
-               if ($aResult['verbose'] && false) var_dump($sModifyXML);
-
-               $sDatabaseName = 'nominatim';
-               $aSpec = array(
-                       0 => array("pipe", "r"),  // stdin
-                       1 => array("pipe", "w"),  // stdout
-                       2 => array("pipe", "w") // stderr
-               );
-               $aPipes = array();
-               $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' -';
-               echo $sCMD."\n";
-               $hProc = proc_open($sCMD, $aSpec, $aPipes);
-               if (!is_resource($hProc))
+               // derive change from normal osm file with osmosis
+               $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
+               if (isset($aResult['import-file']) && $aResult['import-file'])

                {
                {

-                       echo "$sBasePath/osm2pgsql failed\n";
-                       exit;   
+                       $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$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;
+                       }

                }
                }

-               fwrite($aPipes[0], $sModifyXML);
-               fclose($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))
+               else

                {
                {

-                       echo "osm2pgsql existed with error level $iError\n";
-                       echo $sOut;
-                       echo $sErrors;
+                       $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;
+                       }
+               }
+
+               // import generated change file
+               $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
+               echo $sCMD."\n";
+               exec($sCMD, $sJunk, $iErrorLevel);
+               if ($iErrorLevel)
+               {
+                       echo "osm2pgsql exited with error level $iErrorLevel\n";

                        exit;
                }
        }
                        exit;
                }
        }


@@ -297,7 +344,7 @@
 
        if ($aResult['index'])
        {
 
        if ($aResult['index'])
        {

-               index($aResult, $sDatabaseDSN);
+               passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);

        }
 
        if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
        }
 
        if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])


@@ -305,10 +352,10 @@
                $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
                $sOsmosisCMD = CONST_Osmosis_Binary;
                $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
                $sImportFile = CONST_BasePath.'/data/osmosischange.osc';
                $sOsmosisCMD = CONST_Osmosis_Binary;
                $sOsmosisConfigDirectory = CONST_BasePath.'/settings';

-               $sDatabaseName = 'nominatim';

                $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
                $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;

-               $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sImportFile;
-               $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -t '.$aResult['index-instances'];
+               $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'];

                if (!$aResult['no-npi']) {
                        $sCMDIndex .= '-F ';
                }
                if (!$aResult['no-npi']) {
                        $sCMDIndex .= '-F ';
                }


@@ -326,7 +373,30 @@
 //                     {
                                if (!file_exists($sImportFile))
                                {
 //                     {
                                if (!file_exists($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);
                                        $fCMDStartTime = time();
                                        echo $sCMDDownload."\n";
                                        exec($sCMDDownload, $sJunk, $iErrorLevel);


@@ -437,8 +507,16 @@
                        echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
                        if (!$aResult['import-osmosis-all']) exit;
 
                        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);

                }
 
        }
                }
 
        }