private void keyWords_Click(object sender, EventArgs e)
{
string fileName = "";
string path = "";
String allTxt = "";
int num = 0;
if (filePath.Text == "")
{
MessageBox.Show("请选择文件！");
}
else
{
path = filePath.Text;
fileName = System.IO.Path.GetFileName(path);
}
if (NLPIR.Init(@"F:/Project/Visual Studio 2012/Projects/TextProcess/TextProcess/", 0))
{
string keywords = NLPIR.GetFileKeyWords(path, 6, false);
keyword.Text = keywords;
//string sql = string.Format("insert into KeyWords(filename,keywords) values('{0}','{1}')", fileName, keywords);
// num = DbHelperSQL.ExecuteSql(sql);
}
}
private void openFile_Click(object sender, EventArgs e)
{
OpenFileDialog ofp = new OpenFileDialog();
string keywords = "";
if (ofp.ShowDialog() == DialogResult.OK)
{
if (NLPIR.Init(@"F:/Project/Visual Studio 2012/Projects/TextProcess/TextProcess/", 0))
{
keywords = NLPIR.GetFileKeyWords(ofp.FileName, 6, false);
}
string fileName = System.IO.Path.GetFileName(ofp.FileName);
filePath.Text = ofp.FileName;
//string sql = string.Format("select * from KeyWords");
//DataTable keyWords = DbHelperSQL.Query(sql).Tables[0];
//var dataRows = keyWords.AsEnumerable().Where<DataRow>(C => C["filename"].ToString() == fileName && C["keywords"].ToString() == keywords);
//int row = dataRows.Count<DataRow>();
//if (row==1)
//{
// MessageBox.Show("该文章记录已存在,您可以从历史记录中查找！");
//}
keyword.Text = keywords;
}
}
List <String> resultList = new List <String>(); //接收最后结果
//#region 分词数相等
//public List<String> preSetence(String allTxt)
//{
// String[] setenceArray = allTxt.Split(new char[] { '，', '。', '？','r' });
// int n = 1;
// for (int m = 0; m < setenceArray.Length; m++)
// {
// if (m == 0)
// {
// continue;
// }
// if (setenceArray[m] == "\r" || setenceArray[m] == " ")
// {
// continue;
// }
// string words1 = setenceArray[m].ToString();
// words1 = words1.Trim();
// if (NLPIR.Init(@"F:/Project/Visual Studio 2012/Projects/TextProcess/TextProcess/", 0))
// {
// string words2 = setenceArray[m-1].ToString();
// words2 = words2.Trim();
// string s1 = NLPIR.ParagraphProcess(words1, true);
// string s2 = NLPIR.ParagraphProcess(words2, true);
// String[] splitDic1 = s1.Split(' ');
// String[] splitDic2 = s2.Split(' ');
// if (splitDic1.Length == splitDic2.Length)
// {
// s1 += " " + splitDic1.Length;
// s2 += " " + splitDic2.Length;
// if (resultList.Contains(setenceArray[m]))
// {
// int index = resultList.IndexOf(words1);
// resultList.Insert(index, words2);
// }
// else
// {
// resultList.Add(words1);
// resultList.Add(words2);
// }
// }
// n++;
// }
// }
// return resultList;
//}
//#endregion
#region 分词数相等，对应词性相近
public List <String> preSetence(String allTxt)
{
String[] setenceArray = allTxt.Split(new char[] { '，', '。', '？', 'r', '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
int n = 1;
for (int m = 0; m < setenceArray.Length - 1; m++)
{
if (setenceArray[m] == "\r" || setenceArray[m] == " " || setenceArray[m] == "\r\n")
{
continue;
}
string words1 = setenceArray[m].ToString();
words1 = words1.Trim();
if (NLPIR.Init(@"F:/Project/Visual Studio 2012/Projects/TextProcess/TextProcess/", 0))
{
string words2 = setenceArray[m + 1].ToString();
words2 = words2.Trim();
string s1 = NLPIR.ParagraphProcess(words1, true);
string s2 = NLPIR.ParagraphProcess(words2, true);
String[] splitDic1 = s1.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries);
String[] splitDic2 = s2.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries);
if (splitDic1.Length == splitDic2.Length)
{
int count = 0;
for (int p = 0; p < splitDic1.Length; p++)
{
if (splitDic1[p] == "" && splitDic2[p] == "")
{
continue;
}
string[] everysplitDic1 = splitDic1[p].ToString().Split(new string[] { "/" }, StringSplitOptions.RemoveEmptyEntries);
string[] everysplitDic2 = splitDic2[p].ToString().Split(new string[] { "/" }, StringSplitOptions.RemoveEmptyEntries);
if (everysplitDic1[1].Contains(everysplitDic2[1]) || everysplitDic2[1].Contains(everysplitDic1[1]))
{
count++;
}
}
if (count == splitDic1.Length)
{
if (resultList.Contains(setenceArray[m]))
{
int index = resultList.IndexOf(words1);
while (resultList[index + 1] != "")
{
index++;
}
resultList.Insert(index + 1, words2);
}
else
{
resultList.Add(words1);
resultList.Add(words2);
resultList.Add("");
}
}
}
n++;
}
}
return(resultList);
}