public ColumnSchema Clone()
{
var clone = new ColumnSchema
{
ColumnName = this.ColumnName,
DataType = this.DataType,
DataTypeName = this.DataTypeName,
ColumnSize = this.ColumnSize,
NumericPrecision = this.NumericPrecision,
NumericScale = this.NumericScale,
AllowNull = this.AllowNull
};
return(clone);
}
protected override void OnBeginReading()
{
// Use Parquet.NET to do the heavy reading.
var options = new Parquet.ParquetOptions();
_pqReader = new Parquet.ParquetReader(_reader.BaseStream);
_pqDataFields = _pqReader.Schema.GetDataFields();
// Extract the column names from the parquet file in a separate array for fast lookups.
for (int i = 0; i < _pqDataFields.Length; i++)
{
var columnSchema = new ColumnSchema()
{
ColumnName = _pqDataFields[i].Name,
DataTypeName = _pqDataFields[i].DataType.ToString(),
ColumnSize = -1,
NumericPrecision = -1,
NumericScale = -1,
AllowNull = _pqDataFields[i].HasNulls
};
_columnSchema.Add(columnSchema);
}
}
protected override void OnBeginReading()
{
// Even if no header is set, we still need to know how many columns there are.
_firstLine = _reader.ReadLine();
var matches = Regex.Matches(_firstLine, _regexParseExpression, RegexOptions.IgnoreCase | RegexOptions.Multiline);
// Foreach of the extract columns from the first row
for (var i = 0; i < matches.Count; i++)
{
// Even though we only support the string data type, we must generate a SchemaTable
// to adhere to the IDataReader standard (and required by importers).
var columnSchema = new ColumnSchema()
{
ColumnName = _header ? matches[i].Groups[2].Value : i.ToString(), // if no header, use the column ordinal
ColumnSize = -1,
DataType = typeof(System.String),
DataTypeName = "STRING",
AllowNull = true
};
_columnSchema.Add(columnSchema);
}
}
