public override void Execute(DataTable table, IIssueCollector issueCollector, IProviderCollection providers)
{
var textColumns = table.Columns.Where(c => DataTypes.DataTypesLists.TextTypes().Contains(c.DataType));
var candidates = textColumns.Select(c => new Candidate(c)).ToList();
using (var rowEnumerable = table.GetTableRowEnumerable())
foreach (var row in rowEnumerable)
{
foreach (var candidate in candidates.ToArray())
{
Object val = row[candidate.Column.ColumnName];
if (val is DBNull || (val is String) == false)
{
continue;
}
String str = (String)val;
ValueType type = Classifier.Classify(str);
if (type != ValueType.Date && type != ValueType.Float && type != ValueType.Int)
{
candidates.Remove(candidate);
continue;
}
if (candidate.ValuesFound == 0)
{
candidate.Type = type;
}
if (type != candidate.Type)
{
candidates.Remove(candidate);
}
else
{
candidate.ValuesFound += 1;
}
}
if (candidates.Count == 0)
{
break;
}
}
foreach (var candidate in candidates)
{
if (candidate.ValuesFound < this.minValues.Value)
{
continue;
}
Issue i = new Issue(this, this.Severity);
i.Name = "Numbers or Dates Stored in Varchar Column";
i.Context = new ColumnContext(candidate.Column);
i.Description = new Description("The varchar column '{0}' is used to store values of type '{1}'", candidate.Column, candidate.Type.ToString());
issueCollector.ReportIssue(i);
}
}
public override void Execute(DataTable table, IIssueCollector issueCollector, IProviderCollection providers)
{
var dataTypes = new[] { DataType.CHAR, DataType.NCHAR, DataType.NVARCHAR, DataType.VARCHAR };
var informationContent = providers.GetProvider <InformationContent>();
var columnsToCheck = (from c in table.Columns
where dataTypes.Contains(c.DataType) && informationContent[c] < 3 // Avoid checking columns with more than 8 unique values
select c).ToArray();
if (columnsToCheck.Length == 0)
{
return;
}
var columnNotBooleanCount = DictionaryFactory.CreateColumnID <int>();
foreach (var col in columnsToCheck)
{
columnNotBooleanCount[col] = 0;
}
int rowCount = 0;
using (var rowEnumerable = table.GetTableRowEnumerable())
foreach (var row in rowEnumerable)
{
rowCount++;
foreach (var column in columnsToCheck)
{
var value = row[column.ColumnName];
if (value is DBNull || !Classifier.IsBool(value.ToString()))
{
columnNotBooleanCount[column] += 1;
}
}
// Foreach 128th row, check that all columns are likely to be boolean
if ((rowCount & 127) == 0)
{
var allowedDirtiness = rowCount * DirtinessFactor.Value / 100f;
columnsToCheck = columnsToCheck.Where(c => columnNotBooleanCount[c] < allowedDirtiness).ToArray();
if (columnsToCheck.Length == 0)
{
return;
}
}
}
foreach (var column in columnsToCheck)
{
var allowedDirtiness = rowCount * DirtinessFactor.Value / 100f;
if (columnNotBooleanCount[column] < allowedDirtiness)
{
issueCollector.ReportIssue(new Issue(this, this.Severity)
{
Name = "Text Column Used for Boolean Values",
Context = new ColumnContext(column),
Description = new Description("The column '{0}' contains boolean values. Consider using another data type", column),
Severity = this.Severity
});
}
}
}