/// <summary>
/// function as a getter to the type of the term
/// </summary>
/// <param name="line">the line of the term toString</param>
/// <returns>the type</returns>
private term.Type GetType(string line)
{
string type = line.Split('\t')[2];
term.Type e = (term.Type)Enum.Parse(typeof(term.Type), type, true);
return(e);
}
public string CreatePhrase(string[] words, int idx, out int j, out term.Type type)
{
type = term.Type.time;
string output = null;
if (idx + 1 < words.Length)
{
if (times.Contains(words[idx + 1]))
{
output = words[idx] + " " + words[idx + 1];
j = idx + 1;
return(output);
}
}
j = idx + 2;
output = words[idx] + " " + words[idx + 1] + " " + words[idx + 2];
return(output);
}
public string CreatePhrase(string[] words, int idx, out int j, out term.Type type)
{
j = idx + 1;
type = term.Type.distance;
return(words[idx] + " " + words[idx + 1]);
}
/// <summary>
/// this method efficiently merge the newly created sorted files into posting files and
/// creates the index in the desired index path
/// </summary>
/// <param name="fileCount">number of files</param>
/// <param name="files">files array</param>
private void merge(int fileCount, string[] files)
{
bool MoreToRead = false;
string[] firstLines = new string[fileCount];
string[] sortedFirstLines;
StreamReader[] sr = new StreamReader[fileCount];
Dictionary <string, StreamWriter> writers = new Dictionary <string, StreamWriter>();
for (int i = 0; i < fileCount; i++)
{
sr[i] = new StreamReader(path + "\\index" + i + "sorted.txt");
}
for (char c = 'A'; c <= 'Z'; c++)
{
writers.Add(c + "", new StreamWriter(ipath + "\\" + c + ".txt"));
}
foreach (term.Type t in Enum.GetValues(typeof(term.Type)))
{
if (t == term.Type.word)
{
continue;
}
writers.Add(t.ToString(), new StreamWriter(ipath + "\\" + t.ToString() + ".txt"));
}
writers.Add("other", new StreamWriter(ipath + "\\other.txt"));
writers.Add("index", new StreamWriter(ipath + "\\index.txt"));
for (int i = 0; i < fileCount; i++)
{
firstLines[i] = sr[i].ReadLine();
if (firstLines[i] == null)
{
firstLines[i] = "\0";
}
if (firstLines[i] == "")
{
i--;
}
if (firstLines[i] != null)
{
MoreToRead = true;
}
}
int lastIndex = 0;
StringBuilder minLine = new StringBuilder();
StringBuilder minPhrase = new StringBuilder();
while (lastIndex < firstLines.Length)
{
minPhrase.Clear();
minLine.Clear();
int i = lastIndex;
sortedFirstLines = firstLines.OrderBy(s => s.Split(new string[] { "\t" }, StringSplitOptions.None)[0]).ToArray();
for (i = 0; i < sortedFirstLines.Length; i++)
{
if (sortedFirstLines[i].Equals("\0"))
{
continue;
}
else
{
break;
}
}
if (i >= sortedFirstLines.Length)
{
break;
}
minPhrase.Append(GetPhrase(sortedFirstLines[i]));
bool Cap = true;
double icf = 0, idf = 0;
term.Type type = GetType(sortedFirstLines[i]);
for (i = 0; i < sortedFirstLines.Length; i++)
{
if (sortedFirstLines[i].Equals("\0"))
{
continue;
}
if (string.Compare(minPhrase.ToString(), GetPhrase(sortedFirstLines[i]), true) == 0)
{
string[] splitted = sortedFirstLines[i].Split('\t');
if (type != (term.Type)Enum.Parse(typeof(term.Type), splitted[2], true))
{
continue;
}
Cap &= splitted[1].Equals("T") ? true : false;
minLine.Append(splitted[3]);
icf += double.Parse(splitted[4]);
idf += double.Parse(splitted[5]);
}
else
{
break;
}
}
string termPhrase = "";
if (Cap)
{
termPhrase = minPhrase.ToString().ToUpper();
}
else
{
termPhrase = minPhrase.ToString().ToLower();
}
if (type == term.Type.word)
{
if (char.IsLetter(minPhrase[0]))
{
writers[Char.ToUpper(minPhrase[0]).ToString()].WriteLine(termPhrase + "\t" + minLine.ToString());
if (Cap)
{
q.Enqueue(termPhrase + "\t" + minLine.ToString());
q_list.Release(1);
}
}
else
{
writers["other"].WriteLine(termPhrase + "\t" + minLine.ToString());
}
}
else
{
writers[type.ToString()].WriteLine(termPhrase + "\t" + minLine.ToString());
}
for (i = 0; i < fileCount; i++)
{
if (string.Compare(minPhrase.ToString(), GetPhrase(firstLines[i]), true) == 0)
{
if (type != GetType(firstLines[i]))
{
continue;
}
if (firstLines[i].Equals("\0"))
{
continue;
}
firstLines[i] = sr[i].ReadLine();
if (firstLines[i] == null)
{
firstLines[i] = "\0";
}
while (firstLines[i] == "")
{
firstLines[i] = sr[i].ReadLine();
if (firstLines[i] == null)
{
firstLines[i] = "\0";
}
}
}
}
writers["index"].WriteLine(termPhrase + "\t" + (int)type + "\t" + icf + "\t" + idf);
}
for (int i = 0; i < fileCount; i++)
{
sr[i].Close();
File.Delete(path + "\\index" + i + "sorted.txt");
}
foreach (KeyValuePair <string, StreamWriter> entry in writers)
{
entry.Value.Close();
}
}
public indexTerm(string phrase, term.Type type)
{
this.phrase = phrase;
this.type = type;
}
