public RegexColumnHandler(string regex, MapColumnHandler inner)
{
// Pre-Compile the Regex
this.Regex = new Regex(regex, RegexOptions.Compiled);
this.Inner = inner;
this.Block = new String8Block();
}
public void LoadSpec()
{
using (ITabularReader r = new CsvReader(this.SpecFilePath, false))
{
while (r.NextRow())
{
string command = r.Current(0).ToString();
string columnName = r.Current(1).ToString();
switch (command.ToLowerInvariant())
{
case "sample":
// Sample,ColumnName,Probability
this.SampleColumnName = columnName;
this.SampleProbability = double.Parse(r.Current(2).ToString());
if (this.SampleProbability < 0.0 || this.SampleProbability > 1.0)
{
throw new UsageException($"SanitizeSpec sample probability ({r.Current(2)}) is out of range. It must be between zero and one.");
}
break;
case "echo":
// Echo,Value1,Value2,...
for (int i = 1; i < r.CurrentRowColumns; ++i)
{
this.EchoValues.Add(r.Current(i).ToString8());
}
break;
case "drop":
// Drop,ColumnName
this.DropColumns.Add(columnName);
break;
case "map":
// Map,ColumnName,MapperName
this.HandlersByColumn.Add(columnName, new EchoColumnHandler(this.EchoValues, new MapColumnHandler(this.HashKeyHash, this.Provider.Mapper(r.Current(2).ToString()))));
break;
case "keep":
// Keep is the default behavior, so it is in the sanispec only as a comment
break;
case "regex":
// Regex,ColumnName,Expression,MapperName
MapColumnHandler handler = new MapColumnHandler(this.HashKeyHash, this.Provider.Mapper(r.Current(3).ToString()));
this.HandlersByColumn.Add(columnName, new EchoColumnHandler(this.EchoValues, new RegexColumnHandler(r.Current(2).ToString(), handler)));
break;
default:
throw new UsageException($"SanitizeSpec mode '{command}' is unknown. Supported modes: sample, echo, drop, map, keep, regex.");
}
}
}
}
