using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using ;
using System.IO;
namespace _07正则_匹配邮箱
{
class Program
{
static void Main(string[] args)
{
List listUrl = new List() {
new Uri("http://gb./gb/contactus.html"),new Uri("/help/faq"),new Uri("/"),new Uri("/home/joinUs/campus"),new Uri("/about/ad.aspx"),new Uri("/about/contactus.aspx"),new Uri("/company/statement.html"),new Uri("/job/dczp/index.htm")
};
List listMail = new List();
foreach (Uri ur in listUrl)
{
GetMails(ur,listMail);
}
cw(listMail);
Console.ReadKey();
}
private static void GetMails(Uri uri,List list)
{
try
{
WebClient wc = new WebClient();
Console.WriteLine("创建WebClient - [{0}]",uri.ToString());
Stream stream = wc.OpenRead(uri);
//Console.WriteLine("正在下载:{0}",uri.ToString());
StreamReader reader = new StreamReader(stream,Encoding.Default);
string input = reader.ReadToEnd();
string reg = @"(?[a-zA-Z0-9_]+@[a-zA-Z0-9]+(?:\.[a-zA-Z0-9]+)+)"
+ @"|((?[a-zA-Z0-9_]+#[a-zA-Z0-9]+(?:\.[a-zA-Z0-9]+)+))"
+ @"|((?[a-zA-Z0-9_]+\(at\)[a-zA-Z0-9]+(?:\.[a-zA-Z0-9]+)+))"
+ @"|((?[a-zA-Z0-9_]+@[a-zA-Z0-9]+(?:\.[a-zA-Z0-9]+)+))";
Regex regex = new Regex(reg);
Console.WriteLine(Regex.IsMatch(input,reg));
MatchCollection matches = regex.Matches(input);
for (int i = 0; i < matches.Count; i++)
{
Match match = matches[i];
//Console.WriteLine("match: {0}",match.Value);
//Console.WriteLine(match.Groups.Count);
for (int j = 1; j < match.Groups.Count; j++)
{
string mail = match.Groups[j].Value;
if (!string.IsNullOrEmpty(mail))
{
mail = Regex.Replace(mail,@"(.+)(?:@)(.+)","$1@$2");
mail = Regex.Replace(mail,"(.+)#(.+)",@"(.+)\(at\)(.+)","$1@$2");
if (!list.Contains(mail))
{
list.Add(mail);
}
}
//Console.WriteLine("group: {0}",match.Groups[j].Value);
}
}
}
catch (Exception e)
{
Console.WriteLine(e.Message);
}
}
static void cw(List list)
{
Console.WriteLine("长度为{0}",list.Count);
int i = 0;
foreach (string str in list)
{
i++;
Console.WriteLine("{0} - [{1}]",i,str);
}
Console.WriteLine("______________________");
}
}
}
