/* given the relation names it will auto build keys for common stats */
public StatisticsManager(List <string> relations)
{
m_stats = new Dictionary <string, int>();
scan sc = null;
foreach (string s in relations)
{
/* build generic statistics keys and values */
/* use string builders */
sc = new scan(s);
sc.open();
StringBuilder sb = new StringBuilder();
sb.Append("count.");
sb.Append(s);
int relationSize = sc.outSize();
/* add statistic */
m_stats.Add(sb.ToString(), relationSize);
DataTable sample = new DataTable();
sample = sc.cloneSchema();
int buildsample = 0;
while (sc.hasMore() && buildsample < 5)
{
sample.ImportRow(sc.next());
buildsample++;
}
List <string> values = new List <string>();
for (int j = 0; j < sample.Columns.Count; j++)
{
/* for each column calculate distinctness and store */
string colName = sample.Columns[j].ColumnName;
values.Clear();
/* 5 is the sample size; could make #define so you can change ir
* or maybe even user specified */
int sampleSize = (5 > sample.Rows.Count) ? sample.Rows.Count : 5;
for (int i = 0; i < sampleSize; i++)
{
//DataRow temp = sample.Rows[i];
object value = sample.Rows[i][j];
StringBuilder valPair = new StringBuilder();
valPair.Append(colName + value.ToString());
/* new value */
if (!values.Contains(valPair.ToString()))
{
values.Add(valPair.ToString());
}
}
StringBuilder newstat = new StringBuilder();
newstat.Append("distinct" + "." + s + "." + colName);
int newstatValue = values.Count;
m_stats.Add(newstat.ToString(), newstatValue);
}
sample.Clear();
sc.close();
} /* end for each relation specified */
}