public void translateTest()
{
GoogleTranslator g = new GoogleTranslator();
string s = g.Analytical(g.Translate("apple", "auto"));
Assert.IsTrue(s == "苹果");
}
private void FinalData_Load(object sender, EventArgs e)
{
//this.textBox1.Text = translate(_title);
string str = "", str2 = "";
foreach (string s in MainWindow._tags)
{
str += s + "；";
}
this.TagsText.Text = str;
this.TitleText.Text = MainWindow._title;
this.AuthorText.Text = MainWindow._author;
this.PostDateText.Text = MainWindow._postdate;
this.ContentText.Text = MainWindow._denoisingdata;
str = "";
foreach (string s in MainWindow.keywords)
{
str += s + "；";
}
GoogleTranslator gt = new GoogleTranslator();
str2 = gt.Analytical(gt.Translate(str, "auto", "auto"));
if (str != null && str.Length > 0 && GoogleTranslator.isChineseWord(str[0]))
{
this.KeywordsText.Text = str;
this.KeywordsTransText.Text = str2;
}
else
{
this.KeywordsText.Text = str2;
this.KeywordsTransText.Text = str;
}
}
private static string translate(string str)
{
if (str == null || str == "")
{
return(str);
}
GoogleTranslator googletranslator = new GoogleTranslator();
str = googletranslator.Analytical(googletranslator.Translate(str, "auto"));
return(str);
}
private static string translate(string str)
{
if (str == null || str == "")
{
return(str);
}
/*if (!isChineseWord(str[0]))
* {
* GoogleTranslator googletranslator = new GoogleTranslator();
* //str = str + "(" + googletranslator.Analytical(googletranslator.Translate(str, "zh")) + ")";
* }*/
GoogleTranslator googletranslator = new GoogleTranslator();
str = googletranslator.Analytical(googletranslator.Translate(str, "auto"));
return(str);
}
/**
* 一开始本来打算用TF-IDF算法，这样可以明显提高算法的准确度
* 但是考虑到时间上的原因，我们简单的采用频率计算的方式
*
*/
private void getKeyWords()
{
/*int N=0,n=0;
* SqlConnection con = Connection.instance("10.2.26.60", "XueBa", "crawler", "aimashi2015");
* SqlCommand cmd = con.CreateCommand();
* string countsql = "select count(distinct wid) from XueBa.dbo.WordInverseTable;";
* cmd.CommandText = countsql;
* SqlDataReader reader = cmd.ExecuteReader();
* if (reader.Read())
* {
* N = reader.GetInt32(0);
* }
* con.Close();
* string[] list = new string[MainWindow.wordFreq.Count];
* float[] valuelist = new float[MainWindow.wordFreq.Count];
* MainWindow.wordFreq.Keys.CopyTo(list,0);
* MainWindow.wordFreq.Values.CopyTo(valuelist, 0);
*
* float sum = 0;
* foreach (float num in valuelist)
* {
* sum += num;
* }
* int i = 0;
* foreach (string word in list)
* {
* int num = (int)MainWindow.wordFreq[word];
* float freq = num / sum;
* string sqlstr = string.Format("INSERT INTO XueBa.dbo.WordInverseTable(word,wid,freq) values('{0}', '{1}', '{2}')",word,MainWindow.curwid,freq);
* con = Connection.instance("10.2.26.60", "XueBa", "crawler", "aimashi2015");
* cmd = con.CreateCommand();
* cmd.CommandText = sqlstr;
* cmd.ExecuteNonQuery();
* con.Close();
*
* countsql = string.Format("select count(distinct wid) from XueBa.dbo.WordInverseTable where word = '{0}';",word);
* con = Connection.instance("10.2.26.60", "XueBa", "crawler", "aimashi2015");
* cmd = con.CreateCommand();
* cmd.CommandText = countsql;
* reader = cmd.ExecuteReader();
* if (reader.Read())
* {
* n = reader.GetInt32(0);
* }
* con.Close();
* valuelist[i++] = freq*(float)Math.Log(1.0 * (N + 2) / n);
* }
* Array.Sort(valuelist,list);*/
string[] list = new string[MainWindow.wordFreq.Count];
int[] valuelist = new int[MainWindow.wordFreq.Count];
MainWindow.wordFreq.Keys.CopyTo(list, 0);
MainWindow.wordFreq.Values.CopyTo(valuelist, 0);
Array.Sort(valuelist, list);
if ((list.Length + 9) > MainWindow.maxKeyWordsNo * 10)
{
for (int j = 1, k = 1; k <= MainWindow.maxKeyWordsNo && j <= list.Length; j++)
{
if (MainWindow._stopWords.Contains(list[list.Length - j]) || Regex.IsMatch(list[list.Length - j], "^[0-9]+(\\.[0-9]+)?$"))
{
continue;
}
else
{
MainWindow.keywords.Add(list[list.Length - j]);
GoogleTranslator googletranslator = new GoogleTranslator();
MainWindow.keywords.Add(googletranslator.Analytical(googletranslator.Translate(list[list.Length - j], "auto")));
//MainWindow.keywords.Add(Translator.translate(list[list.Length - j]));
k++;
}
}
}
else
{
for (int j = 1, k = 1; k <= (list.Length + 9) / 10 && j <= list.Length; j++)
{
if (MainWindow._stopWords.Contains(list[list.Length - j]) || Regex.IsMatch(list[list.Length - j], "^[0-9]+(\\.[0-9]+)?$"))
{
continue;
}
else
{
MainWindow.keywords.Add(list[list.Length - j]);
GoogleTranslator googletranslator = new GoogleTranslator();
MainWindow.keywords.Add(googletranslator.Analytical(googletranslator.Translate(list[list.Length - j], "auto")));
k++;
}
}
}
}
public FinalData()
{
InitializeComponent();
googletranslator = new GoogleTranslator();
}