private void BuildMappingTable()
{
//Get a new mapping table in memory
_mappingTable = new Dictionary <object, List <object> >();
//connect to server and run distinct query
var server = MappingFromColumn.TableInfo.Discover(DataAccessContext.DataLoad).Database.Server;
var fromColumnName = MappingFromColumn.GetRuntimeName();
var toColumnName = MappingToColumn.GetRuntimeName();
//pull back all the data
using (var con = server.GetConnection())
{
con.Open();
var sql = GetMappingTableSql();
var cmd = server.GetCommand(sql, con);
cmd.CommandTimeout = Timeout;
var r = cmd.ExecuteReader();
while (r.Read())
{
if (!_mappingTable.ContainsKey(r[fromColumnName]))
{
_mappingTable.Add(r[fromColumnName], new List <object>());
}
_mappingTable[r[fromColumnName]].Add(r[toColumnName]);
}
}
}
public DataTable ProcessPipelineData(DataTable toProcess, IDataLoadEventListener listener, GracefulCancellationToken cancellationToken)
{
var fromColumnName = MappingFromColumn.GetRuntimeName();
var toColumnName = MappingToColumn.GetRuntimeName();
listener.OnNotify(this, new NotifyEventArgs(ProgressEventType.Information, "About to build mapping table"));
if (!toProcess.Columns.Contains(fromColumnName))
{
throw new Exception("DataTable did not contain a field called '" + fromColumnName + "'");
}
if (toProcess.Columns.Contains(toColumnName))
{
throw new Exception("DataTable already contained a field '" + toColumnName + "'");
}
if (_mappingTable == null)
{
BuildMappingTable();
}
listener.OnNotify(this, new NotifyEventArgs(ProgressEventType.Information, "Mapping table resulted in " + _mappingTable.Count + " unique possible input values"));
listener.OnNotify(this, new NotifyEventArgs(ProgressEventType.Information, "Mapping table resulted in " + _mappingTable.Sum(kvp => kvp.Value.Count) + " unique possible output values"));
//add the new column (the output column)
toProcess.Columns.Add(toColumnName);
int idxFrom = toProcess.Columns.IndexOf(fromColumnName);
int idxTo = toProcess.Columns.IndexOf(toColumnName);
int numberOfElementsPerRow = toProcess.Columns.Count;
List <object[]> newRows = new List <object[]>();
List <DataRow> toDrop = new List <DataRow>();
foreach (DataRow row in toProcess.Rows)
{
var fromValue = row[idxFrom];
//if we don't have the key value
if (!_mappingTable.ContainsKey(fromValue))
{
if (CrashIfNoMappingsFound)
{
throw new KeyNotFoundException("Could not find mapping for " + fromValue);
}
else
{
listener.OnNotify(this, new NotifyEventArgs(ProgressEventType.Warning, "No mapping for '" + fromValue + "' dropping row"));
toDrop.Add(row);
continue;
}
}
//we do have the key value!
var results = _mappingTable[fromValue];
//yes 1
if (results.Count == 1)
{
row[idxTo] = results.Single();
}
else
{
//great we have multiple mappings, bob=>Frank and bob=>Jesus. What does the user want to do about that
switch (AliasResolutionStrategy)
{
case AliasResolutionStrategy.CrashIfAliasesFound:
throw new AliasException("The value '" + fromValue + "' maps to mulitple ouptut values:" + string.Join(",", results.Select(v => "'" + v.ToString() + "'")));
case AliasResolutionStrategy.MultiplyInputDataRowsByAliases:
//substitute for the first alias (bob=>Frank)
row[idxTo] = results.First();
//then clone the row and do a row with bob=>Jesus
foreach (object next in results.Skip(1))
{
//Create a copy of the input row
object[] newRow = new object[numberOfElementsPerRow];
row.ItemArray.CopyTo(newRow, 0);
//Set the aliasable element to the alias
newRow[idxTo] = next;
//Add it to our new rows collection
newRows.Add(newRow);
}
break;
default:
throw new ArgumentOutOfRangeException();
}
}
}
//add any alias multiplication rows
foreach (object[] newRow in newRows)
{
toProcess.Rows.Add(newRow);
}
//drop rows with missing identifiers
foreach (DataRow dropRow in toDrop)
{
toProcess.Rows.Remove(dropRow);
}
if (!KeepInputColumnToo)
{
toProcess.Columns.Remove(fromColumnName);
}
return(toProcess);
}
