//public void analysisTest3()
//{
// string path = @"D:\文档\人工智能\语料\聊天机器人词库\纯化\";
// string[] files = Directory.GetFiles(path);
// string outfile = path + @"result.txt";
// Dictionary<string, string> dir = new Dictionary<string, string>();
// foreach(string file in files)
// {
// string[] res = File.ReadAllLines(file,Encoding.Default);
// List<string> reslist = new List<string>();
// for(int i = 0; i < res.Length; i++)
// {
// if (!string.IsNullOrWhiteSpace(res[i]))
// {
// reslist.Add(res[i]);
// }
// else
// {
// if (reslist.Count >= 2)
// {
// if (reslist[0].Contains("[") && reslist[0].Contains("]")) { reslist.Clear(); continue; }
// reslist[0] = reslist[0].Replace("[name]", "你").Replace("[cqname]", "我").Replace("[enter]"," ");
// reslist[1] = reslist[1].Replace("[name]", "你").Replace("[cqname]", "我").Replace("[enter]", " ");
// reslist[1] = Regex.Replace(reslist[1], @"\[.*?\]", "");
// if (dir.ContainsKey(reslist[0]))
// {
// //have it
// if (reslist[0].Length > dir[reslist[0]].Length)
// {
// if (reslist[1].Contains("http")) { reslist.Clear(); continue; }
// //change
// dir[reslist[0]] = reslist[1];
// }
// }
// else
// {
// //dont have
// dir[reslist[0]] = reslist[1];
// }
// }
// reslist.Clear();
// }
// }
// }
// using(FileStream fs=new FileStream(outfile, FileMode.Create))
// {
// using(StreamWriter sw=new StreamWriter(fs,Encoding.UTF8))
// {
// foreach (var k in dir)
// {
// sw.WriteLine(string.Format("{0}\r\n{1}\r\n",k.Key,k.Value));
// }
// }
// }
//}
//private void analysisTest4()
//{
// string file = @"D:\文档\人工智能\语料\聊天机器人词库\纯化\result.txt";
// //Dictionary<string, List<Pair>> dir = new Dictionary<string, List<Pair>>();
// string[] res = File.ReadAllLines(file, Encoding.UTF8);
// string resstr = "";
// for (int i = 0; i < res.Length; i++)
// {
// string key = res[i++];
// string value = res[i++];
// List<Pair> valueWords = WordCutTool.cut(value);
// resstr += string.Format("{0}\t{1}\t{2}\r\n", key, value, 1);
// foreach (var p in valueWords)
// {
// if (p.Flag.StartsWith("n") || p.Flag.StartsWith("v") || p.Flag.StartsWith("a"))
// {
// resstr += string.Format("{0}\t{1}\t{2}\r\n", value, p.Word, 2);
// }
// }
// //dir[key] = valueWords;
// if (i > 1000) break;
// }
// printChangeResult(resstr);
//}
private void analysis5()
{
var res = WordCutTool.cutSentence(textBox1.Text);
string output = "";
foreach (var sen in res)
{
foreach (var w in sen)
{
if (w.Flag.ToUpper().StartsWith("N") || w.Flag.ToUpper().StartsWith("R"))
{
output += "[" + w.Word + "]";
}
else if (w.Flag.ToUpper().StartsWith("V"))
{
output += "" + w.Word + "";
}
else if (w.Flag.ToUpper().StartsWith("U"))
{
;
}
else
{
output += w.Word + w.Flag;
}
}
output += "\r\n\r\n";
}
printChangeResult(output);
}
public void analysisTest2()
{
string text1 = textBox1.Text;
var res = WordCutTool.cut(text1);
List <WordPair> out1 = new List <WordPair>();
List <WordPair> verbs = WordCutTool.sumPair(res, "V");
foreach (var v in verbs)
{
for (int i = 1; i < res.Count - 1; i++)
{
if (res[i].Flag == v.Flag && res[i].Word == v.Word)
{
//verb
out1.Add(new WordPair(string.Format("{0}\t{1}\t{2}\r\n", res[i - 1].Flag, res[i].Word, res[i + 1].Flag), ""));
}
}
}
printChangeResult(WordCutTool.getString(out1));
}
public void analysisTest1()
{
string text1 = textBox1.Text;
var res = WordCutTool.cut(text1);
List <WordPair> out1 = new List <WordPair>();
foreach (var w in res)
{
//if (w.Flag.ToUpper().StartsWith("V")) out1.Add(new Pair("[ V " + w.Word + " ]", "v"));
//else if (w.Flag.ToUpper().StartsWith("N")) out1.Add(new Pair("[ N " + w.Word + " ]", "n"));
//else if (w.Flag.ToUpper().StartsWith("A")) out1.Add(new Pair("[ A " + w.Word + " ]", "a"));
// else if (w.Flag.ToUpper().StartsWith("D")) out1.Add(new Pair("[D" + w.Word + "]", "d"));
if (w.Flag == "x" && ("，。？！…—；,.?!()（）“”‘’\r\n\t ".Contains(w.Word) || w.Word.Length <= 0))
{
out1.Add(new WordPair(" [" + w.Word + "/" + w.Flag + "]\r\n\r\n", ""));
}
else
{
out1.Add(new WordPair(" [" + w.Word + "/" + w.Flag + "]", ""));
}
}
printChangeResult(WordCutTool.getString(out1));
}
private void workChangeWords()
{
string text1 = textBox1.Text;
string text2 = textBox3.Text;
print("开始分析文本1");
var res = WordCutTool.cut(text1);
var verbs = WordCutTool.sumPair(res, "V");
var nones = WordCutTool.sumPair(res, "N");
var adjs = WordCutTool.sumPair(res, "A");
print("开始分析文本2");
var res2 = WordCutTool.cut(text2);
var verbs2 = WordCutTool.sumPair(res2, "V");
var nones2 = WordCutTool.sumPair(res2, "N");
var adjs2 = WordCutTool.sumPair(res2, "A");
List <WordPair[]> changeres = new List <WordPair[]>();
for (int i = 0; i < verbs.Count; i++)
{
WordPair[] c = new WordPair[2];
c[0] = verbs[i];
if (verbs2.Count > i)
{
c[1] = verbs2[i];
}
else
{
c[1] = c[0];
}
changeres.Add(c);
}
for (int i = 0; i < nones.Count; i++)
{
WordPair[] c = new WordPair[2];
c[0] = nones[i];
if (nones2.Count > i)
{
c[1] = nones2[i];
}
else
{
c[1] = c[0];
}
changeres.Add(c);
}
for (int i = 0; i < adjs.Count; i++)
{
WordPair[] c = new WordPair[2];
c[0] = adjs[i];
if (adjs2.Count > i)
{
c[1] = adjs2[i];
}
else
{
c[1] = c[0];
}
changeres.Add(c);
}
List <WordPair> respair = new List <WordPair>();
foreach (var w in res)
{
bool change = false;
foreach (var wp in changeres)
{
if (w.Flag == wp[0].Flag && w.Word == wp[0].Word)
{
respair.Add(new WordPair(wp[1].Word, wp[1].Flag));
change = true;
break;
}
}
if (!change)
{
respair.Add(w);
}
}
printChangeResult(WordCutTool.getString(respair));
//print();
print("分析结束");
}
