public override void LoadProfileSettings(XmlNode xNode)
{
if (xNode == null)
{
return;
}
bool hasHeaderRow = false;
bool skipFirstRow = false;
string separator = "";
string TableStructureType = String.Empty;
//Table level attributes...
foreach (XmlAttribute xAttr in xNode.Attributes)
{
switch (xAttr.Name.ToUpper())
{
case "SEPARATOR":
separator = xAttr.Value;
break;
case "TABLENAME":
TableName = xAttr.Value;
break;
case "TABLESTRUCTURETYPE":
TableStructureType = xAttr.Value;
break;
case "HASHEADER":
hasHeaderRow = Boolean.Parse(xAttr.Value);
break;
case "SKIPFIRSTROW":
skipFirstRow = Boolean.Parse(xAttr.Value);
break;
}
}
UseFirstRowNamesAsColumnNames = hasHeaderRow;
SkipFirstRow = skipFirstRow;
RegexColumnBuilder rcb = new RegexColumnBuilder(separator);
XmlNode tableNode = xNode;
if (!String.IsNullOrWhiteSpace(TableStructureType))
{
tableNode = Configuration.GetDataTableNode(TableStructureType);
}
if (tableNode.HasChildNodes)
{
foreach (XmlNode childNode in tableNode.ChildNodes)
{
ParseColumnOrParser(rcb, childNode, separator);
}
}
ColumnBuilder = rcb;
BuildRegexSchemaIntoDataSet();
}
public RegexDataSet(string fileName = "", string fieldSeparator = "", string tableName = "Table1", bool useFirstRowNamesAsColumns = true, bool skipFirstRow = false, params string[] columnNames)
{
TableName = tableName;
UseFirstRowNamesAsColumnNames = useFirstRowNamesAsColumns;
SkipFirstRow = skipFirstRow;
if (useFirstRowNamesAsColumns)
{
if ((!String.IsNullOrWhiteSpace(fileName)) && (File.Exists(fileName)))
{
using (StreamReader sr = new StreamReader(fileName))
{
string firstRow = sr.ReadLine();
if (String.IsNullOrWhiteSpace(fieldSeparator))
{
Regex regexSeparator = new Regex("^([a-zA-Z0-9_\"]*)(?<Separator>.)");
Match match = regexSeparator.Match(firstRow);
if (match.Success)
{
fieldSeparator = match.Groups["Separator"].ToString();
}
}
columnNames = firstRow.Split(fieldSeparator[0]);
}
UseFirstRowNamesAsColumnNames = false;
}
else
{
FirstRowExpression = new Regex(@"([^" + fieldSeparator + @"\n]*)[" + fieldSeparator + @"\w]");
}
SkipFirstRow = true;
}
if (columnNames.Length > 0)
{
ColumnBuilder = new RegexColumnBuilder(fieldSeparator, columnNames);
Fill(fileName);
}
else
{
ColumnBuilder = null;
}
}
public override void Fill(Stream textFile)
{
UseFirstRowNamesAsColumnNames = _hasHeaderRow;
if (_hasHeaderRow)
{
StreamReader sr = new StreamReader(textFile, Encoding.UTF8, true, 1024, true);
string firstRow = sr.ReadLine();
sr.Close();
if (String.IsNullOrWhiteSpace(_delimiter))
{
Regex regexSeparator = new Regex("^([a-zA-Z0-9_\"]*)(?<Separator>.)");
Match match = regexSeparator.Match(firstRow);
if (match.Success)
{
_delimiter = match.Groups["Separator"].ToString();
}
}
string[] columnNames = firstRow.Split(_delimiter[0]);
ColumnBuilder = new RegexColumnBuilder(_delimiter, columnNames);
SkipFirstRow = true;
}
base.Fill(textFile);
}
/// <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 override void Fill()
{
if (TextFile == null)
{
throw new ApplicationException("No stream available to convert to a DataSet");
}
TextFile.Seek(0, SeekOrigin.Begin);
using (StreamReader sr = new StreamReader(TextFile))
{
var readLine = sr.ReadLine();
var isFirstLine = true;
int lineNumber = 1;
SendMessageToCallingApplicationHandler(0, "Loading First Line");
if (UseFirstRowNamesAsColumnNames && (FirstRowExpression == null) && ((_regexColumns == null) || (_regexColumns.Count == 0)))
{
string firstRow = readLine;
Regex regexSeparator = new Regex("^([a-zA-Z0-9_\"]*)(?<Separator>.)");
Match match = regexSeparator.Match(firstRow);
if (match.Success)
{
string fieldSeparator = match.Groups["Separator"].ToString();
string[] columnNames = firstRow.Split(fieldSeparator[0]);
ColumnBuilder = new RegexColumnBuilder(fieldSeparator, columnNames);
}
UseFirstRowNamesAsColumnNames = false;
SkipFirstRow = true;
}
SendMessageToCallingApplicationHandler(lineNumber, "Building Schema...");
BuildRegexSchemaIntoDataSet();
while (readLine != null)
{
if (isFirstLine && UseFirstRowNamesAsColumnNames && !SkipFirstRow)
{
SendMessageToCallingApplicationHandler(lineNumber, "Building First Row...");
if (FirstRowExpression == null)
{
throw new RegexDataSetException(
"FirstRowExpression is not set, but UseFirstRowNamesAsColumnNames is set to true");
}
if (!FirstRowExpression.IsMatch(readLine))
{
throw new RegexDataSetException(
"The first row in the file does not match the FirstRowExpression");
}
var m = FirstRowExpression.Match(readLine);
foreach (var sGroup in FirstRowExpression.GetGroupNames())
{
if (sGroup != DefaultGroup)
{
DataTable.Columns[sGroup].ExtendedProperties.Add(NewName, m.Groups[sGroup].Value);
}
}
}
else if (!(isFirstLine && SkipFirstRow))
{
ProcessRowObject(readLine);
}
SendMessageToCallingApplicationHandler(lineNumber, "Processed line");
readLine = sr.ReadLine();
lineNumber += 1;
isFirstLine = false;
}
}
if (!UseFirstRowNamesAsColumnNames)
{
return;
}
foreach (DataColumn column in DataTable.Columns)
{
if (column.ExtendedProperties.ContainsKey(NewName))
{
column.ColumnName = column.ExtendedProperties[NewName].ToString();
}
}
}
private void ParseColumnOrParser(RegexColumnBuilder columnBuilder, XmlNode childNode, string separator = "")
{
if (childNode.Name.ToUpper() == "IF")
{
string strCondition = String.Empty;
string strTableName = TableName;
string strTableStructureType = String.Empty;
//Conditional Table level attributes...
foreach (XmlAttribute xAttr in childNode.Attributes)
{
switch (xAttr.Name.ToUpper())
{
case "SEPARATOR":
separator = xAttr.Value;
break;
case "CONDITION":
strCondition = xAttr.Value;
break;
case "TABLENAME":
strTableName = xAttr.Value;
break;
case "TABLESTRUCTURETYPE":
strTableStructureType = xAttr.Value;
break;
}
}
RegexColumnBuilder conditionalRCB = new RegexColumnBuilder(separator);
XmlNode tableNode = childNode;
if (!String.IsNullOrWhiteSpace(strTableStructureType))
{
tableNode = Configuration.GetDataTableNode(strTableStructureType);
}
foreach (XmlNode subNode in tableNode.ChildNodes)
{
ParseColumnOrParser(conditionalRCB, subNode, separator);
}
ConditionalRegexParser crp = new ConditionalRegexParser()
{
ConditionRegex = new Regex(strCondition), TableName = strTableName, parseRegex = conditionalRCB.CreateRegularExpression(), RegexColumns = conditionalRCB.Columns
};
Parsers.Add(crp);
}
else
{
string prefix = "";
string suffix = "";
string strCondition = String.Empty;
bool hasDoubleQuotes = false;
bool bAutoIncrement = false;
Int32 intStartValue = 1;
Int32 intIncrement = 1;
bool bForeignKey = false;
bool bPrimaryKey = false;
string strExpression = String.Empty;
string strDisplayName = childNode.Name;
string strDescription = String.Empty;
string strDefault = String.Empty;
int columnLength = 0;
RegexColumnType rct = RegexColumnType.STRING;
//Column level attributes...
foreach (XmlAttribute xAttr in childNode.Attributes)
{
switch (xAttr.Name.ToUpper())
{
case "SEPARATOR":
separator = xAttr.Value;
break;
case "PREFIX":
prefix = xAttr.Value;
break;
case "SUFFIX":
suffix = xAttr.Value;
break;
case "QUOTES":
hasDoubleQuotes = Boolean.Parse(xAttr.Value);
break;
case "LENGTH":
columnLength = Int16.Parse(xAttr.Value);
break;
case "TYPE":
rct = (RegexColumnType)Enum.Parse(typeof(RegexColumnType), xAttr.Value);
break;
case "CONDITION":
strCondition = xAttr.Value;
break;
case "AUTOINCREMENT":
bAutoIncrement = Boolean.Parse(xAttr.Value);
break;
case "STARTVALUE":
case "START":
case "SEED":
intStartValue = Int32.Parse(xAttr.Value);
break;
case "INCREMENT":
intIncrement = Int32.Parse(xAttr.Value);
break;
case "EXPRESSION":
strExpression = xAttr.Value;
break;
case "FOREIGNKEY":
bForeignKey = Boolean.Parse(xAttr.Value);
break;
case "UNIQUE":
case "PRIMARYKEY":
case "PRIMARY":
bPrimaryKey = Boolean.Parse(xAttr.Value);
break;
case "DISPLAYNAME":
case "CAPTION":
strDisplayName = xAttr.Value;
break;
case "DESCRIPTION":
strDescription = xAttr.Value;
break;
case "DEFAULT":
strDefault = xAttr.Value;
break;
}
}
bool bColumnAdded = false;
string strColumnName = childNode.Name;
if (strColumnName.Trim('_') == String.Empty)
{
strColumnName = String.Empty;
}
if (bAutoIncrement)
{
columnBuilder.AddColumn(strColumnName, bAutoIncrement, intStartValue, intIncrement);
bColumnAdded = true;
}
if (!bColumnAdded && !String.IsNullOrEmpty(strExpression))
{
columnBuilder.AddColumn(strColumnName, rct, strExpression);
bColumnAdded = true;
}
if ((!bColumnAdded) && (bForeignKey))
{
columnBuilder.AddColumn(strColumnName, bForeignKey);
bColumnAdded = true;
}
if (!bColumnAdded) //This is a regular column with regex... let us add this to the column builder...
{
if (!String.IsNullOrEmpty(separator))
{
if (hasDoubleQuotes)
{
columnBuilder.AddColumn('\"' + strColumnName + '\"', separator[0], rct);
}
else
{
if (childNode.NextSibling == null)
{
columnBuilder.AddColumn(strColumnName, ".*", rct);
}
else
{
columnBuilder.AddColumn(strColumnName, "[^" + columnBuilder.RegexFormattedOutput(separator[0]) + "\\n]*", prefix, suffix, rct);
}
}
}
else
{
if (columnLength > 0)
{
columnBuilder.AddColumn(strColumnName, columnLength, rct);
}
else
{
columnBuilder.AddColumn(strColumnName, ".*", rct);
}
}
RegexColumn addedColumn = columnBuilder.Columns[columnBuilder.Columns.Count - 1];
if (!String.IsNullOrWhiteSpace(strCondition))
{
//There is a condition to be matched with the value... let us set it to the last column added...
addedColumn.ValueMatchingCondition = strCondition;
}
if (bPrimaryKey)
{
addedColumn.IsUnique = bPrimaryKey;
}
if (strDisplayName != strColumnName)
{
addedColumn.DisplayName = strDisplayName;
}
if (!String.IsNullOrEmpty(strDescription))
{
addedColumn.Description = strDescription;
}
if (!String.IsNullOrEmpty(strDefault))
{
addedColumn.Default = strDefault;
}
}
}
}
public RegexDataSet(RegexColumnBuilder columnBuilder)
{
ColumnBuilder = columnBuilder;
}
