private void ReadTheSource(string path)
{
if (!File.Exists(path) || new FileInfo(path).Length == 0)
{
throw new FileNotFoundException("Source file isn't exists or it's empty!");
}
_documents.Clear();
WriteDebugInfo("Start reading of the file...");
try
{
using (CsvFileReader reader = new CsvFileReader(path, EmptyLineBehavior.Ignore))
{
bool isFirstRow = true;
bool hasComplianceScore = false;
List <string> row = new List <string>();
List <string> services = new List <string>();
while (reader.ReadRow(row))
{
if (isFirstRow)
{
isFirstRow = false;
if (row.Count <= 1)
{
throw new Exception("Source file isn't proper CSV file for SemantAPI.Human application");
}
if (!row[0].Equals("Document ID"))
{
throw new Exception("Source file doesn't have \"Document ID\" column or it's not the first one.");
}
if (row[row.Count - 1] != "Source text")
{
throw new Exception("Source file doesn't have \"Source text\" column or it's not the last one.");
}
for (int index = 1; index < row.Count - 1; index++)
{
string service = row[index].Split(' ').First();
if (!services.Contains(service))
{
services.Add(service);
}
}
hasComplianceScore = row.Any(item => item.Contains("Compliance"));
continue;
}
int col = 1;
string polarity = string.Empty;
ResultSet result = new ResultSet(row[row.Count - 1]);
foreach (string service in services)
{
result.AddOutput(service, double.Parse(row[col + 1]), row[col]);
if (hasComplianceScore)
{
polarity = row[col + 2];
}
col += (hasComplianceScore && service != "MechanicalTurk") ? 3 : 2;
}
if (hasComplianceScore)
{
result.AddReferencePolarity(polarity, "MechanicalTurk");
}
_documents.Add(row[0], result);
Application.DoEvents();
}
WriteDebugInfo("Source file has been loaded.");
}
}
catch (IOException ex)
{
throw new Exception("File is locked or another IO problem happened.", ex);
}
catch (Exception ex)
{
throw new Exception("File is locked or another IO problem happened.", ex);
}
return;
}