C#分析検索エンジンurl検索キーワードを得る
27759 ワード
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Web;
namespace HbTui.Portal
{
public class SearchKeyWord
{
private string[][] _Enginers = new string[][]
{
new string[]{"google","utf8","q"},
new string[]{"baidu","gb2312","wd"},
new string[]{"yahoo","utf8","p"},
new string[]{"yisou","utf8","search"},
new string[]{"live","utf8","q"},
new string[]{"tom","gb2312","word"},
new string[]{"163","gb2312","q"},
new string[]{"iask","gb2312","k"},
new string[]{"soso","gb2312","w"},
new string[]{"sogou","gb2312","query"},
new string[]{"zhongsou","gb2312","w"},
new string[]{"3721","gb2312","p"},
new string[]{"openfind","utf8","q"},
new string[]{"alltheweb","utf8","q"},
new string[]{"lycos","utf8","query"},
new string[]{"youdao","utf8","q"},
new string[]{"onseek","utf8","q"},
new string[]{"jike","utf8","q"}
};
//
private string _EngineName = string.Empty;
public string EngineName
{
get
{
return _EngineName;
}
}
//
private string _Coding = "utf8";
public string Coding
{
get
{
return _Coding;
}
}
//
private string _RegexWord = "";
public string RegexWord
{
get
{
return _RegexWord;
}
}
private string _Regex = @"(";
//
//
public void EngineRegEx(string myString)
{
for (int i = 0, j = _Enginers.Length; i < j; i++)
{
if (myString.Contains(_Enginers[i][0]))
{
_EngineName = _Enginers[i][0];
_Coding = _Enginers[i][1];
_RegexWord = _Enginers[i][2];
_Regex += _EngineName + @".+.*[?/&]" + _RegexWord + @"[=:])(?<key>[^&]*)";
break;
}
}
}
//
public string SearchKey(string myString)
{
EngineRegEx(myString.ToLower());
if (_EngineName != "")
{
Regex myReg = new Regex(_Regex, RegexOptions.IgnoreCase);
Match matche = myReg.Match(myString);
myString = matche.Groups["key"].Value;
// +
myString = myString.Replace("+", " ");
if (_Coding == "gb2312")
{
myString = GetUTF8String(myString);
}
else
{
myString = Uri.UnescapeDataString(myString);
}
}
return myString;
}
//
public string GetUTF8String(string myString)
{
Regex myReg = new Regex("(?<key>%..%..)", RegexOptions.IgnoreCase);
MatchCollection matches = myReg.Matches(myString);
string myWord;
for (int i = 0, j = matches.Count; i < j; i++)
{
myWord = matches[i].Groups["key"].Value.ToString();
myString = myString.Replace(myWord, GB2312ToUTF8(myWord));
}
return myString;
}
// GB2312 UTF8 URL
public string GB2312ToUTF8(string myString)
{
string[] myWord = myString.Split('%');
byte[] myByte = new byte[] { Convert.ToByte(myWord[1], 16), Convert.ToByte(myWord[2], 16) };
Encoding GB = Encoding.GetEncoding("GB2312");
Encoding U8 = Encoding.UTF8;
myByte = Encoding.Convert(GB, U8, myByte);
char[] Chars = new char[U8.GetCharCount(myByte, 0, myByte.Length)];
U8.GetChars(myByte, 0, myByte.Length, Chars, 0);
return new string(Chars);
}
// ,
public string isCrawler(string SystemInfo)
{
string[] BotList = new string[] { "Google", "Baidu", "yisou", "MSN", "Yahoo", "live", "tom", "163", "TMCrawler", "iask", "Sogou", "soso", "youdao", "zhongsou", "3721", "openfind", "alltheweb", "lycos", "bing", "118114" };
foreach (string Bot in BotList)
{
if (SystemInfo.ToLower().Contains(Bot.ToLower()))
{
return Bot;
}
}
return "null";
}
public bool IsSearchEnginesGet(string str)
{
string[] strArray = new string[] { "Google", "Baidu", "yisou", "MSN", "Yahoo", "live", "tom", "163", "TMCrawler", "iask",
"Sogou", "soso", "youdao", "zhongsou", "3721", "openfind","alltheweb", "lycos", "bing", "118114"};
str = str.ToLower();
for (int i = 0; i < strArray.Length; i++)
{
if (str.IndexOf(strArray[i].ToLower()) >= 0)
{
return true;
}
}
return false;
}
}
}