X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/effd8e12af440586594a18a04beb33abed260d15..35df2bb75eae78f38b347f26862fa4cb8284d574:/utils/update.php

diff --git a/utils/update.php b/utils/update.php
index 84fb24e5..efdf6525 100755
--- a/utils/update.php
+++ b/utils/update.php
@@ -5,34 +5,35 @@ require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
 require_once(CONST_BasePath.'/lib/init-cmd.php');
 ini_set('memory_limit', '800M');
 
-$aCMDOptions = array(
-    "Import / update / index osm data",
-    array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
-    array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
-    array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
-
-    array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
-    array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
-    array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolate)'),
-    array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
-
-    array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
-
-    array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
-    array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
-    array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
-
-    array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
-    array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
-    array('import-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'),
-    array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'),
-
-    array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
-    array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
-    array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
-
-    array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
-);
+$aCMDOptions
+= array(
+   "Import / update / index osm data",
+   array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
+   array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
+   array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
+
+   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', '(obsolate)'),
+   array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
+
+   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-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('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
+  );
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 
 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
@@ -47,24 +48,20 @@ if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
 // 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())
-{
+if ($iCacheMemory + 500 > getTotalMemoryMB()) {
     $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))
-{
+if (!is_null(CONST_Osm2pgsql_Flatnode_File)) {
     $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
 }
 
 
-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))
-    {
+    if (!file_exists($sNextFile)) {
         fail("Cannot open $sNextFile\n");
     }
 
@@ -73,8 +70,7 @@ if (isset($aResult['import-diff']))
     echo $sCMD."\n";
     exec($sCMD, $sJunk, $iErrorLevel);
 
-    if ($iErrorLevel)
-    {
+    if ($iErrorLevel) {
         fail("Error from osm2pgsql, $iErrorLevel\n");
     }
 
@@ -83,68 +79,55 @@ if (isset($aResult['import-diff']))
 
 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 $bHaveDiff = false;
-if (isset($aResult['import-file']) && $aResult['import-file'])
-{
+if (isset($aResult['import-file']) && $aResult['import-file']) {
     $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)
-    {
+    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']) && $aResult['import-node'])
-{
-    if ($bUseOSMApi)
-    {
+if (isset($aResult['import-node']) && $aResult['import-node']) {
+    if ($bUseOSMApi) {
         $sContentURL = 'http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
-    }
-    else
-    {
+    } else {
         $sContentURL = 'http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
     }
 }
-if (isset($aResult['import-way']) && $aResult['import-way'])
-{
-    if ($bUseOSMApi)
-    {
+
+if (isset($aResult['import-way']) && $aResult['import-way']) {
+    if ($bUseOSMApi) {
         $sContentURL = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
-    }
-    else
-    {
+    } else {
         $sContentURL = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
     }
 }
-if (isset($aResult['import-relation']) && $aResult['import-relation'])
-{
-    if ($bUseOSMApi)
-    {
+
+if (isset($aResult['import-relation']) && $aResult['import-relation']) {
+    if ($bUseOSMApi) {
         $sContentURLsModifyXMLstr = 'http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
-    }
-    else
-    {
+    } else {
         $sContentURL = 'http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
     }
 }
-if ($sContentURL)
-{
+
+if ($sContentURL) {
     $sModifyXMLstr = file_get_contents($sContentURL);
     $bHaveDiff = true;
 
     $aSpec = array(
-        0 => array("pipe", "r"),  // stdin
-        1 => array("pipe", "w"),  // stdout
-        2 => array("pipe", "w") // stderr
-    );
+              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))
-    {
+    if (!is_resource($hProc)) {
         fail("Error converting osm to osc, osmosis failed\n");
     }
     fwrite($aPipes[0], $sModifyXMLstr);
@@ -155,31 +138,26 @@ if ($sContentURL)
     $sErrors = stream_get_contents($aPipes[2]);
     if ($aResult['verbose']) echo $sErrors;
     fclose($aPipes[2]);
-    if ($iError = proc_close($hProc))
-    {
+    if ($iError = proc_close($hProc)) {
         echo $sOut;
         echo $sErrors;
         fail("Error converting osm to osc, osmosis returned: $iError\n");
     }
 }
 
-if ($bHaveDiff)
-{
+if ($bHaveDiff) {
     // import generated change file
     $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
     echo $sCMD."\n";
     exec($sCMD, $sJunk, $iErrorLevel);
-    if ($iErrorLevel)
-    {
+    if ($iErrorLevel) {
         fail("osm2pgsql exited with error level $iErrorLevel\n");
     }
 }
 
-if ($aResult['deduplicate'])
-{
-
-    if (getPostgresVersion() < 9.3)
-    {
+if ($aResult['deduplicate']) {
+    //
+    if (getPostgresVersion() < 9.3) {
         fail("ERROR: deduplicate is only currently supported in postgresql 9.3");
     }
 
@@ -190,8 +168,7 @@ if ($aResult['deduplicate'])
 
     $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";
     $aDuplicateTokens = chksql($oDB->getAll($sSQL));
-    foreach($aDuplicateTokens as $aToken)
-    {
+    foreach ($aDuplicateTokens as $aToken) {
         if (trim($aToken['word_token']) == '' || trim($aToken['word_token']) == '-') continue;
         echo "Deduping ".$aToken['word_token']."\n";
         $sSQL = "select word_id,(select count(*) from search_name where nameaddress_vector @> ARRAY[word_id]) as num from word where word_token = '".$aToken['word_token']."' and class is null and type is null and country_code is null order by num desc";
@@ -200,8 +177,7 @@ if ($aResult['deduplicate'])
         $aKeep = array_shift($aTokenSet);
         $iKeepID = $aKeep['word_id'];
 
-        foreach($aTokenSet as $aRemove)
-        {
+        foreach ($aTokenSet as $aRemove) {
             $sSQL = "update search_name set";
             $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),";
             $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")";
@@ -218,8 +194,7 @@ if ($aResult['deduplicate'])
             $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]";
             chksql($oDB->query($sSQL));
 
-            foreach ($aPartitions as $sPartition)
-            {
+            foreach ($aPartitions as $sPartition) {
                 $sSQL = "update search_name_".$sPartition." set";
                 $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")";
                 $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]";
@@ -237,14 +212,12 @@ if ($aResult['deduplicate'])
     }
 }
 
-if ($aResult['index'])
-{
+if ($aResult['index']) {
     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'] || $aResult['import-osmosis-all']) {
+    //
     if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) {
         fail("Error: Update interval too low for download.geofabrik.de.  Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n");
     }
@@ -256,33 +229,25 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
     $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
     $sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
 
-    while(true)
-    {
+    while (true) {
         $fStartTime = time();
         $iFileSize = 1001;
 
-        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 )
-            {
-
+            if (CONST_Replication_Update_Interval > 60) {
                 unset($aReplicationLag);
-                exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
-                while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
-                {
-                    if ($iErrorLevel)
-                    {
+                exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
+                while ($iErrorLevel > 0 || $aReplicationLag[0] < 1) {
+                    if ($iErrorLevel) {
                         echo "Error: $iErrorLevel. ";
                         echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
-                    }
-                    else
-                    {
+                    } else {
                         echo ".";
                     }
                     sleep(CONST_Replication_Recheck_Interval);
                     unset($aReplicationLag);
-                    exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
+                    exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
                 }
                 // There are new replication files - use osmosis to download the file
                 echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
@@ -291,8 +256,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
             $fCMDStartTime = time();
             echo $sCMDDownload."\n";
             exec($sCMDDownload, $sJunk, $iErrorLevel);
-            while ($iErrorLevel > 0)
-            {
+            while ($iErrorLevel > 0) {
                 echo "Error: $iErrorLevel\n";
                 sleep(60);
                 echo 'Re-trying: '.$sCMDDownload."\n";
@@ -300,10 +264,10 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
             }
             $iFileSize = filesize($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')";
+            $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s', $fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
             var_Dump($sSQL);
             $oDB->query($sSQL);
-            echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
+            echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
         }
 
         $iFileSize = filesize($sImportFile);
@@ -313,15 +277,14 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
         $fCMDStartTime = time();
         echo $sCMDImport."\n";
         exec($sCMDImport, $sJunk, $iErrorLevel);
-        if ($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')";
+        $sSQL = "INSERT INTO import_osmosis_log values ('$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";
+        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($sImportFile);
@@ -330,47 +293,52 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all'])
 
         // Index file
         $sThisIndexCmd = $sCMDIndex;
+        if (!isset($aResult['index-instances'])) {
+            if (getLoadAverage() < 24)
+                $iIndexInstances = 2;
+            else
+                $iIndexInstances = 1;
+        } else
+            $iIndexInstances = $aResult['index-instances'];
+
+        $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances;
         $fCMDStartTime = time();
 
-        if (!$aResult['no-index'])
-        {
+        if (!$aResult['no-index']) {
             echo "$sThisIndexCmd\n";
             exec($sThisIndexCmd, $sJunk, $iErrorLevel);
-            if ($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')."','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);
-        echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
+        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;
-        echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\n";
+        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 ( CONST_Replication_Update_Interval > 60 )
-        {
-            $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
-        }
-        else
-        {
-            $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
+        if (CONST_Replication_Update_Interval > 60) {
+            $iSleep = max(0, (strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time()));
+        } else {
+            $iSleep = max(0, CONST_Replication_Update_Interval-$fDuration);
         }
         echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
         sleep($iSleep);
     }
 }
 
+
 function getosmosistimestamp($sOsmosisConfigDirectory)
 {
     $sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');
     preg_match('#timestamp=(.+)#', $sStateFile, $aResult);
-    return str_replace('\:',':',$aResult[1]);
+    return str_replace('\:', ':', $aResult[1]);
 }