//按行读取输入文件并统计
public WordCalculate Input(WordCalculate datanumber, WordTrie wtrie)
{
FileStream fs = null;
StreamReader sr = null;
String dataline = String.Empty;
try
{
fs = new FileStream(this.pathIn, FileMode.Open);
sr = new StreamReader(fs);
while ((dataline = sr.ReadLine()) != null)
{
datanumber.Calculate(dataline, wtrie); //按行统计数据
}
}
catch { Console.WriteLine("wrong！"); }
finally
{
if (sr != null)
{
sr.Close();
}
if (fs != null)
{
fs.Close();
}
}
return(datanumber);
}
public static Result Maintest()
{
WordIO io = new WordIO();
WordCalculate datanumber = new WordCalculate();
WordTrie wtrie = new WordTrie();
Result res = new Result();
io.pathIn = "F:\\Demo.txt";
io.pathOut = "F:\\Result.txt";
datanumber = io.Input(datanumber, wtrie); //按行读取文件并统计
io.Output(datanumber, wtrie);
res.charactersnumber = datanumber.charactersnumber;
res.wordsnumber = datanumber.wordsnumber;
res.linesnumber = datanumber.linesnumber;
return(res);
}
public long linesnumber = 0; //统计数据：行数
//数据统计
public void Calculate(string dataline, WordTrie wtrie)
{
if (string.IsNullOrEmpty(dataline))
{
return;
}
string word = null;
for (int i = 0, len = dataline.Length; i < len; i++)
{
char unit = dataline[i];
if (unit >= 65 && unit <= 90)
{
unit = (char)(unit + 32);
} //大写字母转换成小写
if ((unit >= 48 && unit <= 57) || (unit >= 97 && unit <= 122))
{
word = String.Concat(word, unit);
}
else
{
if (!string.IsNullOrEmpty(word)) //判断是否为词尾后的字符
{
if ((word[0] >= 97 && word[0] <= 122))
{
wtrie.Insert(word);
}
word = null;
}
}
}
if (!string.IsNullOrEmpty(word))
{
if ((word[0] >= 97 && word[0] <= 122))
{
wtrie.Insert(word);
}
word = null;
}
this.linesnumber++; //统计行数
this.wordsnumber = wtrie.CountSum; //统计单词数
this.charactersnumber += dataline.Length; //统计字符数
}
//将统计数据输出并写到输出文件
public void Output(WordCalculate datanumber, WordTrie wtrie)
{
FileStream fs = null;
StreamWriter sw = null;
List <WordTrie.ListUnit> WordList = new List <WordTrie.ListUnit>();
try
{
fs = new FileStream(this.pathOut, FileMode.Create);
sw = new StreamWriter(fs);
WordList = wtrie.Sort();
sw.WriteLine("字符总数为：{0}", datanumber.charactersnumber);
sw.WriteLine("单词总数为：{0}", datanumber.wordsnumber);
sw.WriteLine("有效行数为：{0}", datanumber.linesnumber);
sw.WriteLine("\n词频\t单词\n");
Console.WriteLine("字符总数为：{0}", datanumber.charactersnumber);
Console.WriteLine("单词总数为：{0}", datanumber.wordsnumber);
Console.WriteLine("有效行数为：{0}", datanumber.linesnumber);
Console.WriteLine("\n词频\t单词\n");
for (int i = 0; (i < 10 && i < WordList.Count); i++)
{
sw.WriteLine("{0}\t{1}", WordList[i].WordNum, WordList[i].Word);
Console.WriteLine("{0}\t{1}", WordList[i].WordNum, WordList[i].Word);
}
}
catch { Console.WriteLine("文档写入失败！"); }
finally
{
if (sw != null)
{
sw.Close();
}
if (fs != null)
{
fs.Close();
}
}
}
