/// <summary>
/// Construct a new CSV reader off a streamed source
/// </summary>
/// <param name="source">The stream source. Note that when disposed, the CSV Reader will dispose the stream reader.</param>
/// <param name="settings">The CSV settings to use for this reader (Default: CSV)</param>
public CSVReader(Stream source, CSVSettings settings = null)
{
_settings = settings;
if (_settings == null)
{
_settings = CSVSettings.CSV;
}
_stream = new StreamReader(source, _settings.Encoding);
// Do we need to parse headers?
if (_settings.HeaderRowIncluded)
{
var line = _stream.ReadLine();
if (_settings.AllowSepLine)
{
var newDelimiter = CSV.ParseSepLine(line);
if (newDelimiter != null)
{
// We don't want to change the original settings, since they may be a singleton
_settings = _settings.CloneWithNewDelimiter(newDelimiter.Value);
line = _stream.ReadLine();
}
}
Headers = CSV.ParseLine(line, _settings);
}
else
{
Headers = _settings.AssumedHeaders;
}
}
/// <summary>
/// Read this file into a data table in memory
/// </summary>
/// <returns></returns>
public DataTable ReadAsDataTable()
{
var dt = new DataTable();
string[] firstLine = null;
// File contains column names - so name each column properly
if (Headers == null)
{
var rawLine = _stream.ReadLine();
firstLine = CSV.ParseLine(rawLine, _settings);
var list = new List <string>();
for (var i = 0; i < firstLine.Length; i++)
{
list.Add($"Column{i}");
}
this.Headers = list.ToArray();
}
// Add headers
var numColumns = Headers.Length;
foreach (var t in Headers)
{
dt.Columns.Add(new DataColumn(t, typeof(string)));
}
// If we had to read the first line to get dimensions, add it
var row_num = 1;
if (firstLine != null)
{
dt.Rows.Add(firstLine);
row_num++;
}
// Start reading through the file
foreach (var line in CSV.ParseStream(_stream, _settings))
{
// Does this line match the length of the first line?
if (line.Length != numColumns)
{
if (!_settings.IgnoreDimensionErrors)
{
throw new Exception($"Line #{row_num} contains {line.Length} columns; expected {numColumns}");
}
else
{
// Add as best we can - construct a new line and make it fit
var list = new List <string>();
list.AddRange(line);
while (list.Count < numColumns)
{
list.Add("");
}
dt.Rows.Add(list.GetRange(0, numColumns).ToArray());
}
}
else
{
dt.Rows.Add(line);
}
// Keep track of where we are in the file
row_num++;
}
// Here's your data table
return(dt);
}
