private void BuildRegexSchemaIntoDataSet()
{
if (ContentExpression != null || ContentExpressionHasChanged)
{
RemoveDataTable();
foreach (string sGroup in ContentExpression.GetGroupNames())
{
if (sGroup != DefaultGroup)
{
DataColumn newDc = new DataColumn
{
DataType = typeof(string)
};
if (_regexColumns != null)
{
foreach (RegexColumn r in _regexColumns)
{
if (r.ColumnName == sGroup)
{
newDc.DataType = r.ColumnTypeAsType; break;
}
}
}
newDc.ColumnName = sGroup;
DataTable.Columns.Add(newDc);
}
}
}
}
/// <summary>
/// Reads every line in the text file and tries to match
/// it with the given regular expression.
/// Every match will be placed as a new row in the
/// datatable
/// </summary>
/// <returns></returns>
public void Fill()
{
BuildRegexSchemaIntoDataSet();
if (TextFile == null)
{
throw new ApplicationException("No stream available to convert to a DataSet");
}
TextFile.Seek(0, SeekOrigin.Begin);
StreamReader sr = new StreamReader(this.TextFile);
string Line = sr.ReadLine();
bool IsFirstLine = true;
while (Line != null)
{
if (IsFirstLine && UseFirstRowNamesAsColumnNames && !SkipFirstRow)
{
if (FirstRowExpression == null)
{
throw new TextFileDataSetException("FirstRowExpression is not set, but UseFirstRowNamesAsColumnNames is set to true");
}
if (!FirstRowExpression.IsMatch(Line))
{
throw new TextFileDataSetException("The first row in the file does not match the FirstRowExpression");
}
Match m = FirstRowExpression.Match(Line);
foreach (string sGroup in FirstRowExpression.GetGroupNames())
{
if (sGroup != DefaultGroup)
{
DataTable.Columns[sGroup].ExtendedProperties.Add(NewName, m.Groups[sGroup].Value);
}
}
}
else if (!(IsFirstLine && SkipFirstRow) && ContentExpression.IsMatch(Line))
{
Match m = ContentExpression.Match(Line);
DataRow newRow = DataTable.NewRow();;
foreach (string sGroup in ContentExpression.GetGroupNames())
{
if (sGroup != DefaultGroup)
{
if (newRow.Table.Columns[sGroup].DataType == typeof(int))
{
newRow[sGroup] = Convert.ToInt32(m.Groups[sGroup].Value);
}
else if (newRow.Table.Columns[sGroup].DataType == typeof(double))
{
newRow[sGroup] = Convert.ToDouble(m.Groups[sGroup].Value);
}
else if (newRow.Table.Columns[sGroup].DataType == typeof(DateTime))
{
newRow[sGroup] = Convert.ToDateTime(m.Groups[sGroup].Value);
}
else
{
newRow[sGroup] = m.Groups[sGroup].Value;
}
}
}
DataTable.Rows.Add(newRow);
}
else if (!(IsFirstLine && SkipFirstRow))
{
AddMisRead(Line);
}
Line = sr.ReadLine();
IsFirstLine = false;
}
if (UseFirstRowNamesAsColumnNames)
{
foreach (DataColumn column in DataTable.Columns)
{
if (column.ExtendedProperties.ContainsKey(NewName))
{
column.ColumnName = column.ExtendedProperties[NewName].ToString();
}
}
}
}
protected virtual void ParseAndLoadLines(string lines)
{
foreach (string readLine in lines.Split(new string[] { Environment.NewLine }, StringSplitOptions.RemoveEmptyEntries))
{
bool bImportRow = false;
bool bLineParsed = false;
if ((ContentExpression != null) && ContentExpression.IsMatch(readLine))
{
var m = ContentExpression.Match(readLine);
bImportRow = true;
rowDict = new Dictionary <string, object>();
foreach (var sGroup in ContentExpression.GetGroupNames())
{
if ((sGroup != DefaultGroup) && (!Int16.TryParse(sGroup, out short groupNum)))
{
RegexColumn curRegexColumn = _regexColumns.Find(r => r.ColumnName == sGroup);
if (!String.IsNullOrWhiteSpace(curRegexColumn.ValueMatchingCondition) && (!Regex.IsMatch(m.Groups[sGroup].Value, curRegexColumn.ValueMatchingCondition)))
{
bImportRow = false;
break;
}
string fieldValue = m.Groups[sGroup].Value;
fieldValue = fieldValue.Trim('\"');
if (DataTable.Columns[sGroup] != null)
{
if (DataTable.Columns[sGroup].DataType == typeof(int))
{
rowDict[sGroup] = Convert.ToInt32(fieldValue);
}
else if (DataTable.Columns[sGroup].DataType == typeof(double))
{
rowDict[sGroup] = Convert.ToDouble(fieldValue);
}
else if (DataTable.Columns[sGroup].DataType == typeof(DateTime))
{
rowDict[sGroup] = Convert.ToDateTime(fieldValue);
}
else
{
rowDict[sGroup] = fieldValue;
}
}
}
}
if (bImportRow)
{
DataRow newRow = DataTable.NewRow();
PopulateDictionaryToRow(newRow);
//foreach (KeyValuePair<string, object> kvPair in rowDict)
//{
// newRow[kvPair.Key] = kvPair.Value;
//}
DataTable.Rows.Add(newRow);
PopulateRowToDictionary(DataTable.Rows[DataTable.Rows.Count - 1]);
bLineParsed = true;
}
}
if (!bLineParsed)
{
foreach (ConditionalRegexParser crp in Parsers)
{
if (!bLineParsed && crp.ConditionRegex.IsMatch(readLine))
{
bImportRow = true;
DataTable crpDataTable = Tables[crp.TableName];
var m = crp.parseRegex.Match(readLine);
foreach (var sGroup in crp.parseRegex.GetGroupNames())
{
if ((sGroup != DefaultGroup) && (!Int16.TryParse(sGroup, out short groupNum)))
{
RegexColumn curRegexColumn = crp.RegexColumns.Find(r => r.ColumnName == sGroup);
if (!String.IsNullOrWhiteSpace(curRegexColumn.ValueMatchingCondition) && (!Regex.IsMatch(m.Groups[sGroup].Value, curRegexColumn.ValueMatchingCondition)))
{
bImportRow = false;
break;
}
string fieldValue = m.Groups[sGroup].Value;
fieldValue = fieldValue.Trim('\"');
if (crpDataTable.Columns[sGroup] != null)
{
if (crpDataTable.Columns[sGroup].DataType == typeof(int))
{
rowDict[sGroup] = Convert.ToInt32(fieldValue);
}
else if (crpDataTable.Columns[sGroup].DataType == typeof(double))
{
rowDict[sGroup] = Convert.ToDouble(fieldValue);
}
else if (crpDataTable.Columns[sGroup].DataType == typeof(DateTime))
{
rowDict[sGroup] = Convert.ToDateTime(fieldValue);
}
else
{
rowDict[sGroup] = fieldValue;
}
}
}
}
if (bImportRow)
{
DataRow newRow = crpDataTable.NewRow();
PopulateDictionaryToRow(newRow);
crpDataTable.Rows.Add(newRow);
PopulateRowToDictionary(crpDataTable.Rows[crpDataTable.Rows.Count - 1]);
bLineParsed = true;
}
}
}
}
if (!bLineParsed)
{
AddMisRead(readLine);
}
}
}