static void Main(string[] args)
{
var arguments = new ConsoleSwitches(args);
var start = DateTime.UtcNow;
Console.WriteLine("Started At UTC: {0}", start);
Console.WriteLine("Reading files from {0}", arguments.DocumentDir.FullName);
//ExecuteResultImport();
ExecuteWordExtraction(arguments);
Console.WriteLine("Completed. DURATION: {0}", DateTime.UtcNow.Subtract(start));
Console.WriteLine("Results saved on: {0}", arguments.OutputFile);
Console.WriteLine("Press any key to end");
Console.ReadKey();
}
private static void ExecuteWordExtraction(ConsoleSwitches arguments)
{
DoExecute(() =>
{
var files = arguments.DocumentDir.GetFiles("*.doc*", SearchOption.AllDirectories).Where(i => !i.Name.StartsWith("~")).ToList();
var stats = new BlockingCollection <DocumentData>();
var tasks = new List <Task>(arguments.ThreadCount);
var tcount = 0;
for (int f = 0; f < files.Count; f++)
{
var file = files[f];
Console.WriteLine("File: {0} {1} of {2}", file.Name.PadRight(100), (files.IndexOf(file) + 1).ToString().PadLeft(4), files.Count.ToString().PadLeft(4));
if (tcount < arguments.ThreadCount)
{
var t = new Task(() =>
{
var doc = new WordDocument(file.FullName);
DoExecute(() =>
{
stats.Add(doc.GetStats());
});
});
tcount++;
tasks.Add(t);
}
if (tcount >= arguments.ThreadCount || f == (files.Count - 1)) /*is the last file*/
{
tasks.ForEach(t => t.Start());
tcount = 0;
Task.WaitAll(tasks.ToArray());
tasks.Clear();
}
}
Console.WriteLine();
Console.WriteLine();
File.WriteAllText(arguments.OutputFile, JsonConvert.SerializeObject(stats));
});
}
