public AppleStoreAppModel ParseAppPage(string htmlResponse)
{
// Creating HTML Map based on the html response
HtmlDocument map = new HtmlDocument();
map.LoadHtml(htmlResponse);
// Instantiating Empty Parsed App
AppleStoreAppModel parsedApp = new AppleStoreAppModel();
// Reaching nodes of interest
parsedApp.name = GetNodeValue(map, Consts.XPATH_TITLE).Trim();
parsedApp.developerName = GetAppDeveloperName(map);
parsedApp.developerUrl = GetDeveloperUrl(map);
parsedApp.price = GetAppPrice(map);
parsedApp.isFree = parsedApp.price == 0.0 ? true : false;
parsedApp.category = GetAppCategory(map);
parsedApp.updateDate = GetAppUpdateDate(map);
//parsedApp.description = GetAppDescription (map); //TODO: Figure out how to get description on a separated request
parsedApp.version = GetAppVersion(map);
parsedApp.size = GetAppSize(map);
parsedApp.thumbnailUrl = GetThumbnailUrl(map);
parsedApp.languages = GetLanguages(map);
parsedApp.compatibility = GetCompatibility(map);
parsedApp.minimumAge = GetMinimumAge(map);
parsedApp.ageRatingReasons = GetRatingReasons(map);
parsedApp.rating = GetRatings(map);
parsedApp.topInAppPurchases = GetInAppPurchases(map);
return(parsedApp);
}
static void Main(string[] args)
{
// Creating Needed Instances
RequestsHandler httpClient = new RequestsHandler();
AppStoreParser parser = new AppStoreParser();
// Loading Configuration
LogSetup.InitializeLog("Apple_Store_Urls_Worker.log", "info");
_logger = LogManager.GetCurrentClassLogger();
// Loading Config
_logger.Info("Loading Configurations from App.config");
LoadConfiguration();
// Control Variable (Bool - Should the process use proxies? )
bool shouldUseProxies = false;
// Checking for the need to use proxies
if (args != null && args.Length == 1)
{
// Setting flag to true
shouldUseProxies = true;
// Loading proxies from .txt received as argument
String fPath = args[0];
// Sanity Check
if (!File.Exists(fPath))
{
_logger.Fatal("Couldnt find proxies on path : " + fPath);
System.Environment.Exit(-100);
}
// Reading Proxies from File
string[] fLines = File.ReadAllLines(fPath, Encoding.GetEncoding("UTF-8"));
try
{
// Actual Load of Proxies
ProxiesLoader.Load(fLines.ToList());
}
catch (Exception ex)
{
_logger.Fatal(ex);
System.Environment.Exit(-101);
}
}
// AWS Queue Handler
_logger.Info("Initializing Queues");
AWSSQSHelper appsUrlQueue = new AWSSQSHelper(_appUrlsQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
AWSSQSHelper appsDataQueue = new AWSSQSHelper(_appsDataQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
// Setting Error Flag to No Error ( 0 )
System.Environment.ExitCode = 0;
// Initialiazing Control Variables
int fallbackWaitTime = 1;
_logger.Info("Started Processing Individual Apps Urls");
do
{
try
{
// Dequeueing messages from the Queue
if (!appsUrlQueue.DeQueueMessages())
{
Thread.Sleep(_hiccupTime); // Hiccup
continue;
}
// Checking for no message received, and false positives situations
if (!appsUrlQueue.AnyMessageReceived())
{
// If no message was found, increases the wait time
int waitTime;
if (fallbackWaitTime <= 12)
{
// Exponential increase on the wait time, truncated after 12 retries
waitTime = Convert.ToInt32(Math.Pow(2, fallbackWaitTime) * 1000);
}
else // Reseting Wait after 12 fallbacks
{
waitTime = 2000;
fallbackWaitTime = 0;
}
fallbackWaitTime++;
// Sleeping before next try
Console.WriteLine("Fallback (seconds) => " + waitTime);
Thread.Sleep(waitTime);
continue;
}
// Reseting fallback time
fallbackWaitTime = 1;
// Iterating over dequeued Messages
foreach (var appUrl in appsUrlQueue.GetDequeuedMessages())
{
if (appUrl.Body.IndexOf("https://itunes.apple.com/us/app") < 0 && appUrl.Body.IndexOf("https://itunes.apple.com/app/") < 0)
{
_logger.Info("Invalid Message. Deleting [{0}]", appUrl.Body);
appsUrlQueue.DeleteMessage(appUrl);
continue;
}
bool processingWorked = true;
try
{
// Retries Counter
int retries = 0;
string htmlResponse;
// Retrying if necessary
do
{
// Executing Http Request for the Category Url
htmlResponse = httpClient.Get(appUrl.Body, shouldUseProxies);
if (String.IsNullOrEmpty(htmlResponse))
{
// Extending Fallback time
retries++;
int sleepTime = retries * _hiccupTime <= 30000 ? retries * _hiccupTime : 30000;
_logger.Info("Retrying Request for App Page [ " + sleepTime / 1000 + " ]");
Thread.Sleep(sleepTime);
}
} while (String.IsNullOrWhiteSpace(htmlResponse) && retries <= _maxRetries);
// Checking if retries failed
if (String.IsNullOrWhiteSpace(htmlResponse))
{
continue;
}
// Feedback
_logger.Info("Current page " + appUrl.Body, "Parsing App Data");
// Parsing Data out of the Html Page
AppleStoreAppModel parsedApp = parser.ParseAppPage(htmlResponse);
parsedApp.url = appUrl.Body;
// Enqueueing App Data
appsDataQueue.EnqueueMessage(parsedApp.ToJson());
// Little Hiccup
Thread.Sleep(_hiccupTime);
}
catch (Exception ex)
{
_logger.Error(ex);
// Setting Flag to "False"
processingWorked = false;
}
finally
{
//Deleting the message - Only if the processing worked
if (processingWorked)
{
appsUrlQueue.DeleteMessage(appUrl);
}
}
}
}
catch (Exception ex)
{
_logger.Error(ex);
}
} while (true);
}
static void Main(string[] args)
{
// Loading Configuration
LogSetup.InitializeLog("Apple_Store_Recorder.log", "info");
_logger = LogManager.GetCurrentClassLogger();
// Loading Config
_logger.Info("Loading Configurations from App.config");
LoadConfiguration();
// Initializing Queue
_logger.Info("Initializing Queue");
AWSSQSHelper appsDataQueue = new AWSSQSHelper(_appsDataQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
AWSSQSHelper backup = new AWSSQSHelper("DeadLetter", _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
// Creating MongoDB Instance
_logger.Info("Loading MongoDB / Creating Instances");
MongoDBWrapper mongoDB = new MongoDBWrapper();
string serverAddr = String.Join(":", Consts.MONGO_SERVER, Consts.MONGO_PORT);
mongoDB.ConfigureDatabase(Consts.MONGO_USER, Consts.MONGO_PASS, Consts.MONGO_AUTH_DB, serverAddr, 10000, Consts.MONGO_DATABASE, Consts.MONGO_COLLECTION);
// Setting Error Flag to No Error ( 0 )
System.Environment.ExitCode = 0;
// Initialiazing Control Variables
int fallbackWaitTime = 1;
// Buffer of Messages to be recorder
List <AppleStoreAppModel> recordsBuffer = new List <AppleStoreAppModel> ();
List <Message> messagesBuffer = new List <Message> ();
// Insert Batch Size
int batchSize = 1000;
_logger.Info("Started Recording App Data");
do
{
try
{
// Dequeueing messages from the Queue
if (!appsDataQueue.DeQueueMessages())
{
Thread.Sleep(_hiccupTime); // Hiccup
continue;
}
// Checking for no message received, and false positives situations
if (!appsDataQueue.AnyMessageReceived())
{
// If no message was found, increases the wait time
int waitTime;
if (fallbackWaitTime <= 12)
{
// Exponential increase on the wait time, truncated after 12 retries
waitTime = Convert.ToInt32(Math.Pow(2, fallbackWaitTime) * 1000);
}
else // Reseting Wait after 12 fallbacks
{
waitTime = 2000;
fallbackWaitTime = 0;
}
fallbackWaitTime++;
// Sleeping before next try
Console.WriteLine("Fallback (seconds) => " + waitTime);
Thread.Sleep(waitTime);
continue;
}
// Reseting fallback time
fallbackWaitTime = 1;
// Iterating over dequeued Messages
foreach (var appDataMessage in appsDataQueue.GetDequeuedMessages())
{
try
{
// Deserializing message
var appData = AppleStoreAppModel.FromJson(appDataMessage.Body);
// Dumping "Url" to "_id"
appData._id = appData.url;
// Adding it to the buffer of records to be recorded
recordsBuffer.Add(appData);
// Adding message to the buffer of messages to be deleted
messagesBuffer.Add(appDataMessage);
// Is it time to batch insert ?
if ((recordsBuffer.Count % batchSize) == 0)
{
// Batch Insertion
mongoDB.BatchInsert <AppleStoreAppModel> (recordsBuffer);
// Logging Feedback
_logger.Info("\tApps Recorded : " + recordsBuffer.Count);
// Deleting Messages
messagesBuffer.ForEach((msg) => appsDataQueue.DeleteMessage(msg));
_logger.Info("\tMessages Deleted: " + messagesBuffer.Count);
// Clearing Buffers
recordsBuffer.Clear();
messagesBuffer.Clear();
}
}
catch (Exception ex)
{
_logger.Error(ex);
}
finally
{
// Deleting the message
appsDataQueue.DeleteMessage(appDataMessage);
backup.EnqueueMessage(appDataMessage.Body);
}
}
}
catch (Exception ex)
{
_logger.Error(ex);
}
} while (true);
}
static void Main(string[] args)
{
// Creating Needed Instances
RequestsHandler httpClient = new RequestsHandler();
AppStoreParser parser = new AppStoreParser();
_logger = new LogWrapper();
// Loading Configuration
_logger.LogMessage("Loading Configurations from App.config");
LoadConfiguration();
// AWS Queue Handler
_logger.LogMessage("Initializing Queues");
AWSSQSHelper appsUrlQueue = new AWSSQSHelper(_appUrlsQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
AWSSQSHelper appsDataQueue = new AWSSQSHelper(_appsDataQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
// Setting Error Flag to No Error ( 0 )
System.Environment.ExitCode = 0;
// Initialiazing Control Variables
int fallbackWaitTime = 1;
_logger.LogMessage("Started Processing Individual Apps Urls");
do
{
try
{
// Dequeueing messages from the Queue
if (!appsUrlQueue.DeQueueMessages())
{
Thread.Sleep(_hiccupTime); // Hiccup
continue;
}
// Checking for no message received, and false positives situations
if (!appsUrlQueue.AnyMessageReceived())
{
// If no message was found, increases the wait time
int waitTime;
if (fallbackWaitTime <= 12)
{
// Exponential increase on the wait time, truncated after 12 retries
waitTime = Convert.ToInt32(Math.Pow(2, fallbackWaitTime) * 1000);
}
else // Reseting Wait after 12 fallbacks
{
waitTime = 2000;
fallbackWaitTime = 0;
}
fallbackWaitTime++;
// Sleeping before next try
Console.WriteLine("Fallback (seconds) => " + waitTime);
Thread.Sleep(waitTime);
continue;
}
// Reseting fallback time
fallbackWaitTime = 1;
// Iterating over dequeued Messages
foreach (var appUrl in appsUrlQueue.GetDequeuedMessages())
{
try
{
// Retries Counter
int retries = 0;
string htmlResponse;
// Retrying if necessary
do
{
// Executing Http Request for the Category Url
//appUrl.Body = "https://itunes.apple.com/us/app/action-run-3d/id632371832?mt=8";
//appUrl.Body = "https://itunes.apple.com/us/app/emoji-2-free-new-emoticons/id521863802?mt=8";
//appUrl.Body = "https://itunes.apple.com/us/app/candy-crush-saga/id553834731?mt=8";
//appUrl.Body = "https://itunes.apple.com/us/app/dba-den-bla-avis/id448605988?mt=8";
htmlResponse = httpClient.Get(appUrl.Body);
if (String.IsNullOrEmpty(htmlResponse))
{
// Extending Fallback time
retries++;
int sleepTime = retries * _hiccupTime <= 30000 ? retries * _hiccupTime : 30000;
_logger.LogMessage("Retrying Request for App Page [ " + sleepTime / 1000 + " ]", "Request Error", BDC.BDCCommons.TLogEventLevel.Error);
Thread.Sleep(sleepTime);
}
} while (String.IsNullOrWhiteSpace(htmlResponse) && retries <= _maxRetries);
// Checking if retries failed
if (String.IsNullOrWhiteSpace(htmlResponse))
{
// Deletes Message and moves on
//appsUrlQueue.DeleteMessage (appUrl);
continue;
}
// Feedback
_logger.LogMessage("Current page " + appUrl.Body, "Parsing App Data");
// Parsing Data out of the Html Page
AppleStoreAppModel parsedApp = parser.ParseAppPage(htmlResponse);
parsedApp.url = appUrl.Body;
// Enqueueing App Data
appsDataQueue.EnqueueMessage(parsedApp.ToJson());
// Little Hiccup
Thread.Sleep(_hiccupTime);
}
catch (Exception ex)
{
_logger.LogMessage(ex.Message, "App Url Processing", BDC.BDCCommons.TLogEventLevel.Error);
}
finally
{
//Deleting the message
appsUrlQueue.DeleteMessage(appUrl);
}
}
}
catch (Exception ex)
{
_logger.LogMessage(ex);
}
} while (true);
}
static void Main(string[] args)
{
// Creating needed Instances
_logger = new LogWrapper();
// Loading Configuration
_logger.LogMessage("Loading Configurations from App.config");
LoadConfiguration();
// Initializing Queue
_logger.LogMessage("Initializing Queue");
AWSSQSHelper appsDataQueue = new AWSSQSHelper(_appsDataQueueName, _maxMessagesPerDequeue, _awsKey, _awsKeySecret);
// Creating MongoDB Instance
_logger.LogMessage("Loading MongoDB / Creating Instances");
MongoDBWrapper mongoDB = new MongoDBWrapper();
string serverAddr = String.Join(":", Consts.MONGO_SERVER, Consts.MONGO_PORT);
mongoDB.ConfigureDatabase(Consts.MONGO_USER, Consts.MONGO_PASS, Consts.MONGO_AUTH_DB, serverAddr, 10000, Consts.MONGO_DATABASE, Consts.MONGO_COLLECTION);
// Setting Error Flag to No Error ( 0 )
System.Environment.ExitCode = 0;
// Initialiazing Control Variables
int fallbackWaitTime = 1;
_logger.LogMessage("Started Processing App Urls");
do
{
try
{
// Dequeueing messages from the Queue
if (!appsDataQueue.DeQueueMessages())
{
Thread.Sleep(_hiccupTime); // Hiccup
continue;
}
// Checking for no message received, and false positives situations
if (!appsDataQueue.AnyMessageReceived())
{
// If no message was found, increases the wait time
int waitTime;
if (fallbackWaitTime <= 12)
{
// Exponential increase on the wait time, truncated after 12 retries
waitTime = Convert.ToInt32(Math.Pow(2, fallbackWaitTime) * 1000);
}
else // Reseting Wait after 12 fallbacks
{
waitTime = 2000;
fallbackWaitTime = 0;
}
fallbackWaitTime++;
// Sleeping before next try
Console.WriteLine("Fallback (seconds) => " + waitTime);
Thread.Sleep(waitTime);
continue;
}
// Reseting fallback time
fallbackWaitTime = 1;
// Iterating over dequeued Messages
foreach (var appDataMessage in appsDataQueue.GetDequeuedMessages())
{
try
{
// Deserializing message
var appData = AppleStoreAppModel.FromJson(appDataMessage.Body);
// Checking for duplicates
if (!mongoDB.IsAppOnDatabase <AppleStoreAppModel> (appData.url))
{
// Recording App Data
mongoDB.Insert <AppleStoreAppModel> (appData);
}
}
catch (Exception ex)
{
_logger.LogMessage(ex.Message, "App Recording", BDC.BDCCommons.TLogEventLevel.Error);
}
finally
{
// Deleting the message
appsDataQueue.DeleteMessage(appDataMessage);
}
}
}
catch (Exception ex)
{
_logger.LogMessage(ex);
}
} while (true);
}