add max age for access control headers

[nominatim.git] / utils / update.php

diff --git a/utils/update.php b/utils/update.php
index 5c5b64ead5d6f2a934698c718c6e958c43a945d3..f82696c215c6217d17237d05383158433c91bdd4 100755 (executable)
--- a/utils/update.php
+++ b/utils/update.php
@@ -15,6 +15,8 @@
 
                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-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-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'),
 


@@ -28,6 +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'),


@@ -35,7 +38,6 @@
                array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
 
                array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
                array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
 
                array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),

-               array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'),

        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 


@@ -44,7 +46,7 @@
                showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
        }
 
                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
        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)


@@ -64,8 +66,12 @@
        // 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)


@@ -94,7 +100,7 @@
                {
                        // 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;


@@ -104,10 +110,14 @@
                }
 
                // 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;
+                       if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+                       {
+                               $sCMD .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+                       }

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


@@ -126,61 +136,108 @@
                }
        }
 
                }
        }
 

-       $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>', '</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'])
+               {
+                       $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$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;
+                       }
+               }
+               else
+               {
+                       $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;
+               if (!is_null(CONST_Osm2pgsql_Flatnode_File))

                {
                {

-                       echo "$sBasePath/osm2pgsql failed\n";
-                       exit;   
+                       $sCMD .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;

                }
                }

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

                {
                {

-                       echo "osm2pgsql existed with error level $iError\n";
-                       echo $sOut;
-                       echo $sErrors;
+                       echo "osm2pgsql exited with error level $iErrorLevel\n";

                        exit;
                }
        }
                        exit;
                }
        }


@@ -294,7 +351,8 @@
 
        if ($aResult['index'])
        {
 
        if ($aResult['index'])
        {

-               index($aResult, $sDatabaseDSN);
+               if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
+               passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);

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


@@ -302,10 +360,14 @@
                $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;
+               if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+               {
+                       $sCMDImport .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+               }
+               $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'];

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


@@ -323,7 +385,30 @@
 //                     {
                                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);


@@ -360,10 +445,24 @@
 
                                // Archive for debug?
                                unlink($sImportFile);
 
                                // Archive for debug?
                                unlink($sImportFile);

+//                     }

 
 

-                               $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
+                       $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
+
+                       // Index file
+                       if (!isset($aResult['index-instances']))
+                       {
+                               if (getLoadAverage() < 15)
+                                       $iIndexInstances = 2;
+                               else
+                                       $iIndexInstances = 1;
+                       } else
+                               $iIndexInstances = $aResult['index-instances'];
+
+                       $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;

 
 

-                               // Index file
+                       if (!$aResult['no-npi'])
+                       {

                                $fCMDStartTime = time();
                                $iFileID = $oDB->getOne('select nextval(\'file\')');
                                if (PEAR::isError($iFileID))
                                $fCMDStartTime = time();
                                $iFileID = $oDB->getOne('select nextval(\'file\')');
                                if (PEAR::isError($iFileID))


@@ -376,20 +475,20 @@
                                $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 
                                if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
                                $sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
 
                                if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);

-                               $sThisIndexCmd = $sCMDIndex;
-                               if (!$aResult['no-npi']) {
-                                       $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);
-                               }
+                               $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);
+                       }
+
+                       if (!$aResult['no-index'])
+                       {

                                echo "$sThisIndexCmd\n";
                                exec($sThisIndexCmd, $sJunk, $iErrorLevel);
                                if ($iErrorLevel)
                                echo "$sThisIndexCmd\n";
                                exec($sThisIndexCmd, $sJunk, $iErrorLevel);
                                if ($iErrorLevel)


@@ -398,7 +497,8 @@
                                        exit;
                                }
 
                                        exit;
                                }
 

-                               if (!$aResult['no-npi']) {
+                               if (!$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);
                                        $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);


@@ -415,23 +515,31 @@
                                        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");
                                }
                                        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");
                                }

+                       }

 
 

-                               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')";
-                               $oDB->query($sSQL);
+                       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')";
+                       $oDB->query($sSQL);

 
 

-                               $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
-                               $oDB->query($sSQL);
+                       $sSQL = "update import_status set lastimportdate = '$sBatchEnd'";
+                       $oDB->query($sSQL);

 
 

+                       $fDuration = time() - $fStartTime;
+                       echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
+                       if (!$aResult['import-osmosis-all']) exit;

 
 

-                               $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));
+                       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);

                }
                }

-               
+

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