From 9afb6bced6e66683efe2ba570fccecbe754bfde7 Mon Sep 17 00:00:00 2001
From: Jacek Kowalski <Jacek@jacekk.info>
Date: Mon, 10 Jun 2019 20:44:20 +0000
Subject: [PATCH] Check if data timestamps are recent before doing any DB changes

---
 parse.php |   61 +++++++++++++++++-------------
 1 files changed, 35 insertions(+), 26 deletions(-)

diff --git a/parse.php b/parse.php
index 70f92ed..b3014b8 100644
--- a/parse.php
+++ b/parse.php
@@ -6,9 +6,9 @@
 $logger = new Monolog\Logger('Parse changes');
 
 $sources = [
-	'buses' => [
-		'gtfs' => 'ftp://ztp.krakow.pl/VehiclePositions_A.pb',
-		'gtfs_file' => 'VehiclePositions_A.pb',
+	'bus' => [
+		'gtfsrt' => 'ftp://ztp.krakow.pl/VehiclePositions_A.pb',
+		'gtfsrt_file' => 'VehiclePositions_A.pb',
 		'ttss' => 'http://91.223.13.70/internetservice/geoserviceDispatcher/services/vehicleinfo/vehicles',
 		'ttss_file' => 'vehicles_A.json',
 		'database' => 'mapping_A.sqlite3',
@@ -19,25 +19,35 @@
 foreach($sources as $name => $source) {
 	$logger = new Monolog\Logger('fetch_'.$name);
 	try {
-		foreach(['gtfs_file', 'ttss_file', 'database', 'result'] as $field) {
+		foreach(['gtfsrt_file', 'ttss_file', 'database', 'result'] as $field) {
 			$source[$field] = __DIR__.'/data/'.$source[$field];
 		}
 		$source['result_temp'] = $source['result'].'.tmp';
 		
 		$logger->info('Fetching '.$name.' position data from FTP...');
-		$updated = ftp_fetch_if_newer($source['gtfs'], $source['gtfs_file']);
+		$updated = ftp_fetch_if_newer($source['gtfsrt'], $source['gtfsrt_file']);
 		if(!$updated) {
 			$logger->info('Nothing to do, remote file not newer than local one');
 			continue;
 		}
 		
-		$logger->info('Fetching '.$name.' positions from TTSS...');
-		fetch($source['ttss'],$source['ttss_file']);
+		$logger->info('Fetching '.$name.' position data from TTSS...');
+		fetch($source['ttss'], $source['ttss_file']);
 		
 		$logger->info('Loading data...');
 		$mapper = new Mapper();
+		
 		$mapper->loadTTSS($source['ttss_file']);
-		$mapper->loadGTFS($source['gtfs_file']);
+		$timeDifference = time() - $mapper->getTTSSDate();
+		if(abs($timeDifference) > 60) {
+			throw new Exception('TTSS timestamp difference ('.$timeDifference.'s) is too high, aborting!');
+		}
+		
+		$mapper->loadGTFSRT($source['gtfsrt_file']);
+		$timeDifference = time() - $mapper->getGTFSRTDate();
+		if(abs($timeDifference) > 60) {
+			throw new Exception('GTFSRT timestamp difference ('.$timeDifference.'s) is too high, aborting!');
+		}
 		
 		$db = new Database($source['database']);
 		
@@ -48,40 +58,39 @@
 		}
 		
 		$logger->info('Got offset '.$offset.', creating mapping...');
-		$mapping = $mapper->getMapping($offset);
+		$mapping = $mapper->mapUsingOffset($offset);
 		
 		$logger->info('Checking the data for correctness...');
 		$weight = count($mapping);
-		$replace = 0;
-		$ignore = 0;
+		
+		$correct = 0;
+		$incorrect = 0;
+		$old = 0;
+		$maxWeight = 0;
 		foreach($mapping as $id => $vehicle) {
 			$dbVehicle = $db->getById($id);
 			if($dbVehicle) {
-				if((int)substr($vehicle['num'], 2) != (int)$dbVehicle['num']) {
-					if($weight > $dbVehicle['weight']) {
-						$replace += 1;
-						$logger->warn($vehicle['num'].' voting to replace '.$dbVehicle['num'].' (same ID: '.$id.')');
-					} else {
-						$ignore += 1;
-						$logger->warn($vehicle['num'].' voting to ignore '.$dbVehicle['num'].' (same ID: '.$id.')');
-					}
+				$maxWeight = max($maxWeight, $dbVehicle['weight']);
+				if((int)substr($vehicle['num'], 2) == (int)$dbVehicle['num']) {
+					$correct += 1;
+				} else {
+					$incorrect += 1;
 				}
 				continue;
 			}
 			
 			$dbVehicle = $db->getByNum($vehicle['num']);
 			if($dbVehicle && $dbVehicle['id'] != $id) {
-				$replace += 1;
-				$logger->warn($vehicle['id'].' voting to replace '.$dbVehicle['id'].' (same num: '.$vehicle['num'].')');
+				$old += 1;
 			}
 		}
-		$logger->info('Weight: '.$weight.', ignore: '.$ignore.', replace: '.$replace);
+		$logger->info('Weight: '.$weight.', correct: '.$correct.', incorrect: '.$incorrect.', old: '.$old);
 		
 		$previousMapping = NULL;
-		if($ignore > 0 && $ignore >= $replace) {
+		if($incorrect > $correct && $maxWeight > $weight) {
 			throw new Exception('Ignoring result due to better data already present');
-		} elseif($replace > 0) {
-			$logger->warn('Replacing DB data with the mapping');
+		} elseif($old > $correct) {
+			$logger->warn('Replacing DB data with the new mapping');
 			$db->clear();
 		} else {
 			$previousMapping = @json_decode(@file_get_contents($source['result']), TRUE);
@@ -91,7 +100,7 @@
 		
 		if(is_array($previousMapping)) {
 			$logger->info('Merging previous data with current mapping');
-			$mapping = $previousMapping + $mapping;
+			$mapping = $mapping + $previousMapping;
 			ksort($mapping);
 		}
 		

--
Gitblit v1.9.1