private static bool ValidateColumns(string[] splitLine, string line, SharedStateObject sharedState, StreamWriter errorFile)
{
if (!splitLine.Count().Equals(sharedState.Arguments.Columns))
{
lock (errorFile)
{
errorFile.WriteLine("Invalid number of columns expecting '" + sharedState.Arguments.Columns.ToString() + "' found '" + splitLine.Count().ToString() + "' in data \n\t" + line);
}
return(true);
}
return(false);
}
internal static void ProcessFiles(SharedStateObject sharedState, List <string> fileList, StreamWriter outputFile, StreamWriter errorFile)
{
foreach (var file in fileList)
{
Console.WriteLine("Processing File: " + file);
var stopwatch = new Stopwatch();
stopwatch.Start();
using (FileStream fs = File.Open(file, FileMode.Open, FileAccess.Read))
using (BufferedStream bs = new BufferedStream(fs))
using (StreamReader sr = new StreamReader(bs))
{
var errorOnLine = false;
string line;
while ((line = sr.ReadLine()) != null)
{
errorOnLine = false;
//split row
var splitLine = line.Split('|');
//validate the number of columns
errorOnLine = ValidateColumns(splitLine, line, sharedState, errorFile);
//grab the id assume it's first
var ID = splitLine[0];
//concatenate and hash the row
var hash = HashLine(string.Join("", splitLine));
if (sharedState.Arguments.HashIncluded)
{
//compare the hash generated to the one included assume it's last
errorOnLine = ValidateHash(splitLine, line, hash, errorFile);
}
//output the id and hash if there are no errors
if (!errorOnLine)
{
lock (outputFile)
{
outputFile.WriteLine(ID + "\t" + hash);
}
}
}
}
stopwatch.Stop();
Console.WriteLine("Done Processing File: " + file + " in {0} seconds.", ((TimeSpan)stopwatch.Elapsed).Seconds);
stopwatch.Reset();
}
}
static void Main(string[] args)
{
var stopwatch = new Stopwatch();
var parameters = ArgumentParser.ParseArguments(args);
const int numOfThreads = 4;
stopwatch.Start();
try
{
var fileList = Directory.GetFiles(parameters.DirectoryPath).ToList();
/*split the file list into four pieces
* on each small list start a thread
* thread runs the hashing algorithm
*/
var listToProcess = fileList.SplitList(numOfThreads);
var sharedStateObj = new SharedStateObject()
{
Arguments = parameters
};
var threadList = new List <Thread>();
var errorFile = new StreamWriter("error.txt", true);
var outputFile = new StreamWriter("output.txt", true);
foreach (var item in listToProcess)
{
threadList.Add(new Thread(() => FileProcessor.ProcessFiles(sharedStateObj, item, outputFile, errorFile)));
}
foreach (var thread in threadList)
{
thread.Start();
}
var stillRunning = false;
do
{
Thread.Sleep(5000);
stillRunning = false;
foreach (var thread in threadList)
{
if (thread.IsAlive)
{
stillRunning = true;
}
}
} while (stillRunning);
errorFile.Flush();
errorFile.Close();
outputFile.Flush();
outputFile.Close();
}
catch (Exception e)
{
throw;
}
stopwatch.Stop();
Console.WriteLine("Hashing finished in {0} seconds.", ((TimeSpan)stopwatch.Elapsed).Seconds);
Console.Read();
}
