private static string MapAthenaField(this FieldMapping fieldMapping)
{
switch (fieldMapping.MappedType)
{
case AthenaTypeEnum.athena_string:
return("string");
case AthenaTypeEnum.athena_integer:
return("int");
case AthenaTypeEnum.athena_boolean:
return("BOOLEAN");
case AthenaTypeEnum.athena_bigint:
return("bigint");
case AthenaTypeEnum.athena_smallint:
return("smallint");
case AthenaTypeEnum.athena_tinyint:
return("tinyint");
case AthenaTypeEnum.athena_double:
return("double");
case AthenaTypeEnum.athena_float:
return("float");
case AthenaTypeEnum.athena_timestamp:
return("timestamp");
case AthenaTypeEnum.athena_date:
return("date");
}
throw new Exception($"Unexpected Athena Type '{fieldMapping.MappedType}' in the field mapping '{fieldMapping.SourceFieldName}'");
}
public static void ReadFromCSVFile(this EtlSettings etlSettings, Stream stream, int lines = 20)
{
var newMapptings = new List <FieldMapping>();
etlSettings.Sample = new DataSample()
{
Rows = new List <DataRow>()
};
var config = new CsvConfiguration(CultureInfo.InvariantCulture)
{
Delimiter = etlSettings.CsvSourceOptoins.Delimiter
};
var csvStream = stream;
if (etlSettings.CsvSourceOptoins.GZip)
{
csvStream = new GZipStream(stream, CompressionMode.Decompress);
}
using (var streamReader = new StreamReader(csvStream))
{
using (var csvReader = new CsvReader(streamReader, config))
{
int numberOfColoumns = 0;
if (etlSettings.HasHeader)
{
if (csvReader.Read())
{
var value = "";
int i = 0;
while (csvReader.TryGetField(i, out value))
{
newMapptings.Add(new FieldMapping()
{
SourceFieldName = value,
MappedName = value.ToMappedName()
});
i++;
}
numberOfColoumns = i;
}
}
int rowCount = 0;
while (csvReader.Read() && rowCount < lines)
{
var value = "";
int i = 0;
var row = new List <string>();
while (csvReader.TryGetField(i, out value))
{
row.Add(value);
i++;
}
numberOfColoumns = numberOfColoumns <= i ? numberOfColoumns : i;
etlSettings.Sample.Rows.Add(new DataRow()
{
Items = row
});
rowCount++;
}
if (!etlSettings.HasHeader)
{
for (int i = 0; i < numberOfColoumns; i++)
{
var field = new FieldMapping()
{
SourceFieldName = $"Col{i}",
MappedName = $"Col{i}"
};
newMapptings.Add(field);
}
}
for (int i = 0; i < newMapptings.Count; i++)
{
newMapptings[i].MappedType = etlSettings.Sample.Rows
.Select(row => row.Items.Count > i ? row.Items[i].ToString() : "")
.DetectTypeString()
.DetectedTypeToAthenaType();
}
// update the mappings
if (etlSettings.Mappings != null && etlSettings.Mappings.Count > 0)
{
var oldMappings = etlSettings.Mappings;
etlSettings.Mappings = new List <FieldMapping>();
for (int i = 0; i < newMapptings.Count; i++)
{
if (oldMappings.Count > i && oldMappings[i].SourceFieldName == newMapptings[i].SourceFieldName)
{
etlSettings.Mappings.Add(oldMappings[i]);
}
else
{
etlSettings.Mappings.Add(newMapptings[i]);
}
}
}
else
{
etlSettings.Mappings = newMapptings;
}
}
}
}