获取网页内容，解决大部分乱码问题代码,C#代码库,德仔网

代码语言

代码分类

【C#】获取网页内容，解决大部分乱码问题

作者:泽惠 / 发布于2016/10/11/ 747

public static string GetHtml(string url)
        {
            string htmlCode;
            HttpWebRequest webRequest = (System.NET.HttpWebRequest)System.Net.WebRequest.Create(url);
            webRequest.Timeout = 30000;
            webRequest.Method = "GET";
            webRequest.UserAgent = "Mozilla/4.0";
            webRequest.Headers.Add("Accept-Encoding", "gzip, deflate");

            HttpWebResponse webResponse = (System.Net.HttpWebResponse)webRequest.GetResponse();
            //获取目标网站的编码格式
            string contentype = webResponse.Headers["Content-Type"];
            Regex regex = new Regex("charset\\s*=\\s*[\\W]?\\s*([\\w-]+)", RegexOptions.IgnoreCase);
            if (webResponse.ContentEncoding.ToLower() == "gzip")//如果使用了GZip则先解压
            {
                using (System.IO.Stream streamReceive = webResponse.GetResponseStream())
                {
                    using (var zipStream = new System.IO.Compression.GZipStream(streamReceive, System.IO.Compression.CompressionMode.Decompress))
                    {
                        //匹配编码格式
                        if (regex.IsMatch(contentype))
                        {
                            Encoding ending = Encoding.GetEncoding(regex.Match(contentype).Groups[1].Value.Trim());
                            using (StreamReader sr = new System.IO.StreamReader(zipStream, ending))
                            {
                                htmlCode = sr.ReadToEnd();
                            }
                        }
                        else
                        {
                            using (StreamReader sr = new System.IO.StreamReader(zipStream, Encoding.UTF8))
                            {
                                htmlCode = sr.ReadToEnd();
                            }
                        }
                    }
                }
            }
            else
            {
                using (System.IO.Stream streamReceive = webResponse.GetResponseStream())
                {
                    using (System.IO.StreamReader sr = new System.IO.StreamReader(streamReceive, Encoding.Default))
                    {
                        htmlCode = sr.ReadToEnd();
                    }
                }
            }
            return htmlCode;
        }

试试其它关键字

同语言下

可能有用的

泽惠贡献的其它代码(10)