/// <summary>
/// Searches and redacts text from all PDF-files found in input directory. Writes result to output directory.
/// </summary>
/// <param name="inDir">Directory to read PDFs from</param>
/// <param name="outDir">Directory to write redacted PDFs to</param>
/// <param name="textToSearchAfter">Text to be redacted</param>
/// <param name="regexStr">Regular expression for searching specific filenames</param>
public void SearchAndRedact(string inDir, string outDir, string textToSearchAfter, string regexStr)
{
List <RectAndText> RATs = null;
string newFilePath = "";
System.IO.DirectoryInfo directoryInfo = null;
int counterSearchedFiles = 0;
int counterRedactedFiles = 0;
Regex regex = new Regex(regexStr);
// Turns off AGPL licence note when looping many times (thousands).
CounterFactory.getInstance().SetCounter(new NoOpCounter());
List <string> files = Directory.GetFiles(inDir, "*.pdf", System.IO.SearchOption.AllDirectories)
.Where(path => regex.IsMatch(System.IO.Path.GetFileName(path)))
.ToList();
foreach (string filePath in files)
{
logger.Debug("Checking if following file needs redact: " + filePath);
counterSearchedFiles++;
if (counterSearchedFiles % 100 == 0)
{
logger.Info("Number of searched files: " + counterSearchedFiles);
Console.WriteLine("Number of searched files: " + counterSearchedFiles);
}
RATs = GetTextAndCoords(filePath, textToSearchAfter);
if (RATs.Count != 0)
{
logger.Info("Executing redact on: " + filePath);
directoryInfo = System.IO.Directory.GetParent(filePath);
newFilePath = outDir + directoryInfo.Name + @"\" + System.IO.Path.GetFileName(filePath);
new FileInfo(newFilePath).Directory.Create();
Redact(filePath, newFilePath, RATs);
logger.Info("Done... saved to: " + newFilePath);
counterRedactedFiles++;
if (counterRedactedFiles % 100 == 0)
{
logger.Info("Number of redacted files: " + counterRedactedFiles);
Console.WriteLine("Number of redacted files: " + counterRedactedFiles);
}
}
}
logger.Info("Total number of searched files: " + counterSearchedFiles);
Console.WriteLine("Total number of searched files: " + counterSearchedFiles);
logger.Info("Total number of redacted files: " + counterRedactedFiles);
Console.WriteLine("Total number of redacted files: " + counterRedactedFiles);
}
static void Main(string[] args)
{
CounterFactory.getInstance().SetCounter(new AGPLWarningRemoverCounter());
var uri = "http://localhost:8888";
Console.WriteLine("Starting Nancy on " + uri);
// initialize an instance of NancyHost
var host = new NancyHost(new HostConfiguration()
{
UrlReservations = new UrlReservations()
{
CreateAutomatically = true
},
RewriteLocalhost = true
}, new Uri(uri));
host.Start(); // start hosting
Console.ReadLine();
Console.WriteLine("Stopping Nancy");
host.Stop(); // stop hosting
}