| public SetByIndex ( int index, Object value ) : void | ||
| index | int | |
| value | Object | |
| return | void | |
public override void SetByIndex(int index, object value)
{
lock (host.SyncRoot)
{
host.SetByIndex(index, value);
}
}
[SuppressMessage("Microsoft.Contracts", "CC1055")] // Skip extra error checking to avoid *potential* AppCompat problems.
public override void SetByIndex(int index, Object value)
{
lock (_root)
{
_list.SetByIndex(index, value);
}
}
public virtual Object this[int index] {
get {
return(sortedList.GetByIndex(index));
}
set {
sortedList.SetByIndex(index, value);
}
}
static public int SetByIndex(IntPtr l)
{
try {
System.Collections.SortedList self = (System.Collections.SortedList)checkSelf(l);
System.Int32 a1;
checkType(l, 2, out a1);
System.Object a2;
checkType(l, 3, out a2);
self.SetByIndex(a1, a2);
pushValue(l, true);
return(1);
}
catch (Exception e) {
return(error(l, e));
}
}
public void TestSetByIndex ()
{
SortedList sl1 = new SortedList (24);
for (int i = 49; i >= 0; i--) sl1.Add (100 + i, i);
try {
sl1.SetByIndex (-1, 77);
Assert.Fail ("#A");
} catch (ArgumentOutOfRangeException) {
}
try {
sl1.SetByIndex (100, 88);
Assert.Fail ("#B");
} catch (ArgumentOutOfRangeException) {
}
for (int i = 5; i < 25; i++)
sl1.SetByIndex (i, -1);
for (int i = 0; i < 5; i++)
Assert.AreEqual (i, sl1 [100 + i], "#C1");
for (int i = 5; i < 25; i++)
Assert.AreEqual (-1, sl1 [100 + i], "#C2");
for (int i = 25; i < 50; i++)
Assert.AreEqual (i, sl1 [100 + i], "#C3");
}
/* record leaving of a function */
void LeaveFunction(SortedList functions, int functionId)
{
int index = functions.IndexOfKey(functionId);
if(index != -1)
{
int newValue = (int)functions.GetByIndex(index) - 1;
if(newValue <= 0)
{
functions.RemoveAt(index);
}
else
{
functions.SetByIndex(index, newValue);
}
}
}
/* record call to a function */
void EnterFunction(SortedList functions, int functionId)
{
int index = functions.IndexOfKey(functionId);
if(index == -1)
{
functions.Add(functionId, 1);
}
else
{
/* if in the list, add 1 to its counter (need to keep keys unique) */
functions.SetByIndex(index, 1 + (int)functions.GetByIndex(index));
}
}
private void btn_p5_termList_Click(object sender, EventArgs e)
{
string line = "", Term = "", POS = "", DocNO_tag = ""; //資料列, 詞彙, 詞性標記
int index = 0, value = 0, pt = 0;
StringBuilder SB5 = new StringBuilder();
List<string> temp_list = new List<string>();
SortedList Term_DF_slist = new SortedList(); //整個文件集之"詞彙列表" 及詞會出現次數 DF
List<KeyValuePair<string, int>> Doctermtf_list = new List<KeyValuePair<string, int>>(); //-記錄每篇文件出現之詞彙集詞頻 -
// if ((openFileDialog1.ShowDialog() == DialogResult.OK) && ((txt_p5_fileName.Text = openFileDialog1.FileName) != null)) //開啟Window開啟檔案管理對話視窗
//{
using (StreamReader myTextFileReader = new StreamReader(@txt_p5_fileName.Text, Encoding.Default))
{
SortedList DocTermTF_slist = new SortedList(); //暫存每篇文件之"詞彙列表" 及詞會出現次數 TF
// ---- 處理文件集中的每一篇文件與文件中的所有詞彙 (詞彙列表 ,計算 TF , 計算 DF)----
while (myTextFileReader.EndOfStream != true) //非資料檔案結尾
{
line = (myTextFileReader.ReadLine().Trim()) + " ";
// ----- 若為一篇新文件的開始
if (line.Contains("<DocNo>"))
{
TotalDocNum++;
DocNO_tag = line.Substring(line.IndexOf("<DocNo>"), ((line.IndexOf("</DocNo>") + 8) - line.IndexOf("<DocNo>")));
//line = line.Remove(line.IndexOf("<DocNo>"), ((line.IndexOf("</DocNo>") + 8) - line.IndexOf("<DocNo>")));
// -------------------------------------- 計算詞彙之 DF ---------------------------------------------------
if (temp_list.Count != 0)
{
foreach (string str in temp_list) //複製與計算出現在每一文件之詞彙的"DF"
{
//-- 計算詞彙之 DF --
if ((index = Term_DF_slist.IndexOfKey(str)) != -1) //在termList中找尋特定的Term,若Term不存在怎則傳回-1
{
value = (int)Term_DF_slist.GetByIndex(index);
value++; //該term存在,則term的出現次數加1
Term_DF_slist.SetByIndex(index, value);
}
else
Term_DF_slist.Add(str, 1); //該詞彙(term)不存在於termList中,則加入該新詞彙(term)
}
temp_list.Clear(); //清除紀錄每一篇文件內之詞彙的SortedList
}
//--------------記錄每篇文件出現之詞彙集詞頻 -Doctermtf_list -----------
if (DocTermTF_slist.Count != 0)
{
foreach (DictionaryEntry obj in DocTermTF_slist)
{
KeyValuePair<string, int> x = new KeyValuePair<string, int>((string)obj.Key, (int)obj.Value);
Doctermtf_list.Add(x);
}
DocTermTF_slist.Clear();
}
KeyValuePair<string, int> x1 = new KeyValuePair<string, int>(DocNO_tag, 0); //加入文件編號標籤
Doctermtf_list.Add(x1);
continue;
}
// ---------------------------------- 記錄並產生整個文件集之"詞彙列表" 及詞會出現次數 TF------------------------------------------
string[] terms = line.Split(new char[] { '|' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in terms)
{
if (str.IndexOf("(") == -1 || str.IndexOf(")") == -1) //找不到'()'，就不處理/
continue;
Term = str.Substring(0, str.IndexOf("("));
POS = str.Substring(str.IndexOf("("));
if (Term.Length != 0) //若有"詞彙"存在
{
//--- 記錄每篇文件之詞彙及其出現次數 ---
if ((pt = DocTermTF_slist.IndexOfKey((Term + POS))) != -1)
{
value = (int)DocTermTF_slist.GetByIndex(pt);
value++; //該term存在,則term的出現次數加1
DocTermTF_slist.SetByIndex(pt, value); //紀錄整個文件集之"詞彙列表" 及詞會出現次數 TF
}
else
DocTermTF_slist.Add((Term + POS), 1); //該詞彙(term)不存在於termList中,則加入該新詞彙(term)
// --- 紀錄出現在每一文件之詞彙有哪些,用於計算詞彙之DF
if ((index = temp_list.IndexOf((Term + POS))) == -1)
temp_list.Add((Term + POS)); //紀錄出現在每一文件之詞彙有哪些,用於計算詞彙之DF
}
}
//TF詞頻比較
if (Doctermtf_list.Count != 0)
{
foreach (KeyValuePair<string, int> kvp in Doctermtf_list)
{
//KeyValuePair<string, int> x = new KeyValuePair<string, int>((string)obj.Key, (int)obj.Value);
}
}
}//End of while每一文件
// ----記錄並計算最後一篇文件之詞彙 DF --------
if (temp_list.Count != 0)
{
foreach (string str in temp_list) //複製與計算出現在每一文件之詞彙的"DF"
{
//-- 計算詞彙之 DF --
if ((index = Term_DF_slist.IndexOfKey(str)) != -1) //在termList中找尋特定的Term,若Term不存在怎則傳回-1
{
value = (int)Term_DF_slist.GetByIndex(index);
value++; //該term存在,則term的出現次數加1
Term_DF_slist.SetByIndex(index, value);
}
else
Term_DF_slist.Add(str, 1); //該詞彙(term)不存在於termList中,則加入該新詞彙(term)
}
temp_list.Clear(); //清除紀錄每一篇文件內之詞彙的SortedList
}
if (DocTermTF_slist.Count != 0)
{
foreach (DictionaryEntry obj in DocTermTF_slist)
{
KeyValuePair<string, int> x = new KeyValuePair<string, int>((string)obj.Key, (int)obj.Value);
Doctermtf_list.Add(x);
}
DocTermTF_slist.Clear();
}
}//End of using
// --- 顯示詞彙列表 "term-tf" (未過濾前) ---
StringBuilder SB = new StringBuilder(); //StringBuilder適合未知數目之字串動態串接,此處使用string物件則無法完成此大資料檔案的讀取串接
foreach (KeyValuePair<string, int> kvp in Doctermtf_list)
SB.AppendLine(kvp.Key + " " + kvp.Value);
rtb_p5_souce.Text = SB.ToString(); //顯示經詞彙挑選後資料
//}//End of if
// ============================================== 依詞彙出現次數挑選過濾詞彙 ==============================================
int tf = int.Parse(txt_p5_tf.Text); //過濾掉詞彙次數tf(txt_p5_tf.Text)以下之詞彙
List<KeyValuePair<string, int>> DoctermtfSelected_list = new List<KeyValuePair<string, int>>();
MaxTfValue = 0;
foreach (KeyValuePair<string, int> kvp in Doctermtf_list)
{
if (kvp.Value >= tf || kvp.Value == 0)
{
DoctermtfSelected_list.Add(kvp); //建立"挑選過濾後之詞彙-詞彙頻率列表"
if (kvp.Value > MaxTfValue) //找尋最大"詞彙頻率TF"值
MaxTfValue = kvp.Value;
}
if (kvp.Value > tf && Term_list.Contains(kvp.Key) == false)
Term_list.Add(kvp.Key); //建立"挑選過濾後之全部詞彙列表"
Term_list.Sort();
}
// --- 顯示詞彙列表 "term-tf" (過濾後) ---
Dictionary<string, int> termSelected_DF_dic = new Dictionary<string, int>(); //挑選過濾後之"詞彙總數"列表及詞彙之"文件頻df"
rtb_p5_selected.Clear();
StringBuilder SB1 = new StringBuilder(); //StringBuilder適合未知數目之字串動態串接,此處使用string物件則無法完成此大資料檔案的讀取串接
StringBuilder SB2 = new StringBuilder();
foreach (KeyValuePair<string, int> kvp in DoctermtfSelected_list)
{
SB1.AppendLine(kvp.Key + " " + kvp.Value.ToString());
if (((index = Term_DF_slist.IndexOfKey(kvp.Key)) != -1) && (termSelected_DF_dic.ContainsKey(kvp.Key) == false)) //詞彙挑選後之"term-TF" 與 "term-DF"相互配對對應
{
termSelected_DF_dic.Add((string)Term_DF_slist.GetKey(index), (int)Term_DF_slist.GetByIndex(index));
SB2.AppendLine((string)Term_DF_slist.GetKey(index) + " " + (int)Term_DF_slist.GetByIndex(index));
}
}
rtb_p5_selected.Text = SB1.ToString(); //顯示經詞彙挑選後資料
rtb_p5_termDF.Text = SB2.ToString(); //顯示經詞彙挑選後資料
/*/ ============================================= 計算詞彙的權重 "tf_idf" =============================================
StringBuilder SB3 = new StringBuilder();
Dictionary<string, double> termSelected_TDIDF_dic = new Dictionary<string, double>();
int TF = 0, DF = 0;
double idf = 0.0, tf_idf = 0.0;
foreach (KeyValuePair<string, int> kvp in DoctermtfSelected_list)
{
if (kvp.Key.Contains("<DocNo>") && (kvp.Value == 0))
{
KeyValuePair<string, double> tag = new KeyValuePair<string, double>(kvp.Key, 0.0);
DocTermtfidf_list.Add(tag);
tf_idf = 0.0;
}
TF = kvp.Value;
if (termSelected_DF_dic.TryGetValue(kvp.Key, out DF))
{
idf = 1 / (double)DF;
tf_idf = (TF / (double)MaxTfValue) * (Math.Log(1 / idf)); //基底=e
}
KeyValuePair<string, double> x = new KeyValuePair<string, double>(kvp.Key, tf_idf);
DocTermtfidf_list.Add(x);
SB3.AppendLine(kvp.Key + " " + tf_idf);
}
rtb_p5_tfidf.Text = SB3.ToString(); //顯示經詞彙挑選後之"詞彙與tfidf"
// ---- 顯示經詞彙出現次數挑選後出現在每一篇文件中的所有詞彙
StringBuilder SB4 = new StringBuilder();
foreach (string str in Term_list)
SB4.AppendLine(str);
rtb_p5_TermList.Text = SB4.ToString();
}//End 0f Event Fuction */
// ============================================= 詞彙的權重都設為1 =============================================
StringBuilder SB3 = new StringBuilder();
Dictionary<string, double> termSelected_TDIDF_dic = new Dictionary<string, double>();
int TF = 0, DF = 0;
double idf = 0.0, tf_idf = 0.0;
foreach (KeyValuePair<string, int> kvp in DoctermtfSelected_list)
{
if (kvp.Key.Contains("<DocNo>") && (kvp.Value == 0))
{
KeyValuePair<string, double> tag = new KeyValuePair<string, double>(kvp.Key, 0.0);
DocTermtfidf_list.Add(tag);
tf_idf = 0.0;
}
// TF = kvp.Value;
if (termSelected_DF_dic.TryGetValue(kvp.Key, out DF))
{
//idf = 1 / (double)DF;
tf_idf = 1; //基底=e
}
KeyValuePair<string, double> x = new KeyValuePair<string, double>(kvp.Key, tf_idf);
DocTermtfidf_list.Add(x);
SB3.AppendLine(kvp.Key + " " + tf_idf);
}
rtb_p5_tfidf.Text = SB3.ToString(); //顯示經詞彙挑選後之"詞彙與tfidf"
// ---- 顯示經詞彙出現次數挑選後出現在每一篇文件中的所有詞彙
StringBuilder SB4 = new StringBuilder();
foreach (string str in Term_list)
SB4.AppendLine(str);
rtb_p5_TermList.Text = SB4.ToString();
}//End 0f Event Fuction
public void TestSetByIndex() {
SortedList sl1 = new SortedList(24);
for (int i = 49; i>=0; i--) sl1.Add(100+i,i);
try {
sl1.SetByIndex(-1,77);
Fail("sl.SetByIndex: ArgumentOutOfRangeException not caught, when key is out of range");
} catch (ArgumentOutOfRangeException) {}
try {
sl1.SetByIndex(100,88);
Fail("sl.SetByIndex: ArgumentOutOfRangeException not caught, when key is out of range");
} catch (ArgumentOutOfRangeException) {}
for(int i=5; i<25; i++) sl1.SetByIndex(i,-1);
for(int i=0; i<5; i++)
AssertEquals("sl.SetByIndex: set failed(1)",sl1[100+i],i);
for(int i=5; i<25; i++)
AssertEquals("sl.SetByIndex: set failed(2)",sl1[100+i],-1);
for(int i=25; i<50; i++)
AssertEquals("sl.SetByIndex: set failed(3)",sl1[100+i],i);
}
