extract link từ url

Mình share snippest code dùng để get tất cả các link từ 1 url

CS:

            List<string> allUrls = new List<string>();

            WebClient client = new WebClient();

            string content = client.DownloadString("http://code2code.info");

            string pattern = @"(?i)(?s)<a[^>]+?href=""?(?<url>[^""]+)""?>(?<innerHtml>.+?)</a\s*>";

            MatchCollection result = Regex.Matches(content, pattern);

            foreach (Match match in result)

            {

                string url = match.Groups["url"].Value;

                if (url.IndexOf("http://") != -1)

                {

                    allUrls.Add(url);

                }

                Console.WriteLine(url);

            }

            Console.Read();

        }

    }

}

VB:

Imports System

Imports System.Collections.Generic

Imports System.Net

Imports System.Text.RegularExpressions

Module Module1

    Sub Main()

        Dim allUrls As New List(Of String)

        Dim client As New WebClient()

        Dim content As String = client.DownloadString("http://code2code.info")

        Dim pattern As String = "(?i)(?s)<a[^>]+?href=""?(?<url>[^""]+)""?>(?<innerHtml>.+?)</a\s*>"

        Dim resut As MatchCollection = Regex.Matches(content, pattern)

        For Each Match As Match In resut

            Dim url As String = Match.Groups("url").Value

            If url.IndexOf("http://") <> -1 Then

                allUrls.Add(url)

            End If

            Console.WriteLine(url)

        Next

        Console.WriteLine(allUrls.Count)

        Console.Read()

    End Sub

End Module

Reference

Check out my code: https://code2code.googlecode.com/svn/trunk/ExtractLinkFromUrl