Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / update.php

diff --git a/utils/update.php b/utils/update.php
index f7f20aa3f76fc14541e87dd0d5feaccfcd19f449..334ad123c41445eb0e9c4dce6435128125dcb952 100755 (executable)
--- a/utils/update.php
+++ b/utils/update.php
@@ -15,10 +15,9 @@ $aCMDOptions
    array('init-updates', '', 0, 1, 0, 0, 'bool', 'Set up database for updating'),
    array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import updates once'),
    array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import updates forever'),
    array('init-updates', '', 0, 1, 0, 0, 'bool', 'Set up database for updating'),
    array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import updates once'),
    array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import updates 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('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('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Update postcode centroid table'),

 
    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-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'),


@@ -34,6 +33,7 @@ $aCMDOptions
    array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
 
    array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
    array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
 
    array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),

+   array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolete)'),

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


@@ -60,6 +60,13 @@ if (!is_null(CONST_Osm2pgsql_Flatnode_File)) {
 }
 
 if ($aResult['init-updates']) {
 }
 
 if ($aResult['init-updates']) {

+    // sanity check that the replication URL is correct
+    $sBaseState = file_get_contents(CONST_Replication_Url.'/state.txt');
+    if ($sBaseState === false) {
+        echo "\nCannot find state.txt file at the configured replication URL.\n";
+        echo "Does the URL point to a directory containing OSM update data?\n\n";
+        fail("replication URL not reachable.");
+    }

     $sSetup = CONST_InstallPath.'/utils/setup.php';
     $iRet = -1;
     passthru($sSetup.' --create-functions --enable-diff-updates', $iRet);
     $sSetup = CONST_InstallPath.'/utils/setup.php';
     $iRet = -1;
     passthru($sSetup.' --create-functions --enable-diff-updates', $iRet);


@@ -68,14 +75,17 @@ if ($aResult['init-updates']) {
     }
 
     $sDatabaseDate = getDatabaseDate($oDB);
     }
 
     $sDatabaseDate = getDatabaseDate($oDB);

+    if ($sDatabaseDate === false) {
+        fail("Cannot determine date of database.");
+    }

     $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ',
                           strtotime($sDatabaseDate) - (3*60*60));
 
     // get the appropriate state id
     $aOutput = 0;
     $sWindBack = strftime('%Y-%m-%dT%H:%M:%SZ',
                           strtotime($sDatabaseDate) - (3*60*60));
 
     // get the appropriate state id
     $aOutput = 0;

-    exec(CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url,
-        $aOutput, $iRet);
-    if ($iRet != 0) {
+    $sCmd = CONST_Pyosmium_Binary.' -D '.$sWindBack.' --server '.CONST_Replication_Url;
+    exec($sCmd, $aOutput, $iRet);
+    if ($iRet != 0 || $aOutput[0] == 'None') {

         fail('Error running pyosmium tools');
     }
 
         fail('Error running pyosmium tools');
     }
 


@@ -92,6 +102,7 @@ if ($aResult['init-updates']) {
 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
     // import diffs and files directly (e.g. from osmosis --rri)
     $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];
 if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
     // import diffs and files directly (e.g. from osmosis --rri)
     $sNextFile = isset($aResult['import-diff']) ? $aResult['import-diff'] : $aResult['import-file'];

+

     if (!file_exists($sNextFile)) {
         fail("Cannot open $sNextFile\n");
     }
     if (!file_exists($sNextFile)) {
         fail("Cannot open $sNextFile\n");
     }


@@ -108,6 +119,12 @@ if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
     // Don't update the import status - we don't know what this file contains
 }
 
     // Don't update the import status - we don't know what this file contains
 }
 

+if ($aResult['calculate-postcodes']) {
+    info("Update postcodes centroids");
+    $sTemplate = file_get_contents(CONST_BasePath.'/sql/update-postcodes.sql');
+    runSQLScript($sTemplate, true, true);
+}
+

 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 $bHaveDiff = false;
 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
 $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
 $bHaveDiff = false;
 $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];


@@ -239,9 +256,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
 
     while (true) {
         $fStartTime = time();
 
     while (true) {
         $fStartTime = time();

-        $iFileSize = 1001;
-
-        $aLastState = chksql($oDB->getRow('SELECT * FROM import_status'));
+        $aLastState = chksql($oDB->getRow('SELECT *, EXTRACT (EPOCH FROM lastimportdate) as unix_ts FROM import_status'));

 
         if (!$aLastState['sequence_id']) {
             echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";
 
         if (!$aLastState['sequence_id']) {
             echo "Updates not set up. Please run ./utils/update.php --init-updates.\n";


@@ -255,7 +270,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
 
         if ($aLastState['indexed'] == 't') {
             // Sleep if the update interval has not yet been reached.
 
         if ($aLastState['indexed'] == 't') {
             // Sleep if the update interval has not yet been reached.

-            $fNextUpdate = $aLastState['lastimportdate'] + CONST_Replication_Update_Interval;
+            $fNextUpdate = $aLastState['unix_ts'] + CONST_Replication_Update_Interval;

             if ($fNextUpdate > $fStartTime) {
                 $iSleepTime = $fNextUpdate - $fStartTime;
                 echo "Waiting for next update for $iSleepTime sec.";
             if ($fNextUpdate > $fStartTime) {
                 $iSleepTime = $fNextUpdate - $fStartTime;
                 echo "Waiting for next update for $iSleepTime sec.";


@@ -282,6 +297,23 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
                 }
             } while ($iResult);
 
                 }
             } while ($iResult);
 

+            // get the newest object from the diff file
+            $sBatchEnd = 0;
+            $iRet = 0;
+            exec(CONST_BasePath.'/utils/osm_file_date.py '.$sImportFile, $sBatchEnd, $iRet);
+            if ($iRet == 5) {
+                echo "Diff file is empty. skipping import.\n";
+                if (!$aResult['import-osmosis-all']) {
+                    exit(0);
+                } else {
+                    continue;
+                }
+            }
+            if ($iRet != 0) {
+                fail('Error getting date from diff file.');
+            }
+            $sBatchEnd = $sBatchEnd[0];
+

             // Import the file
             $fCMDStartTime = time();
             echo $sCMDImport."\n";
             // Import the file
             $fCMDStartTime = time();
             echo $sCMDImport."\n";


@@ -294,7 +326,6 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
 
             // write the update logs
             $iFileSize = filesize($sImportFile);
 
             // write the update logs
             $iFileSize = filesize($sImportFile);

-            $sBatchEnd = getDatabaseDate($oDB);

             $sSQL = "INSERT INTO import_osmosis_log (batchend, batchseq, batchsize, starttime, endtime, event) values ('$sBatchEnd',$iEndSequence,$iFileSize,'".date('Y-m-d H:i:s', $fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
             var_Dump($sSQL);
             chksql($oDB->query($sSQL));
             $sSQL = "INSERT INTO import_osmosis_log (batchend, batchseq, batchsize, starttime, endtime, event) values ('$sBatchEnd',$iEndSequence,$iFileSize,'".date('Y-m-d H:i:s', $fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
             var_Dump($sSQL);
             chksql($oDB->query($sSQL));