private static List <tuple> CreateTuples4Merging(List <string> fileNames)
{
var tuples = new List <tuple>(fileNames.Count);
for (var i = 0; i < fileNames.Count; i++)
{
var nfr = new ngram_filereader(fileNames[i]);
var tf = nfr.ReadNext();
tuples.Add(new tuple(tf, nfr));
}
return(tuples);
}
private static IEnumerable <TermFrequency> GroupByMerging_1(List <string> fileNames)
{
var current_tuples = new List <tuple>(fileNames.Count);
for (var i = 0; i < fileNames.Count; i++)
{
var nfr = new ngram_filereader(fileNames[i]);
var tf = nfr.ReadNext();
current_tuples.Add(new tuple(tf, nfr));
}
for ( ; current_tuples.Count != 0;)
{
current_tuples.Sort(tuple_tf_term_Comparer.Instance);
var tuple = current_tuples[0];
var tf = tuple.tf;
for (var i = 1; i < current_tuples.Count; i++)
{
var t = current_tuples[i];
if (tf.Term == t.tf.Term)
{
tf.Frequency += t.tf.Frequency;
t.tf = t.ngram_filereader.ReadNext();
if (t.tf == null)
{
t.Dispose();
current_tuples.RemoveAt(i);
i--;
}
}
else
{
break;
}
}
tuple.tf = tuple.ngram_filereader.ReadNext();
if (tuple.tf == null)
{
tuple.Dispose();
current_tuples.RemoveAt(0);
}
yield return(tf);
}
}
public tuple(TermFrequency _tf, ngram_filereader _ngram_filereader)
{
tf = _tf;
ngram_filereader = _ngram_filereader;
}
