c# 爬虫（三） 文件上传

在上一篇中，我们说了模拟登录，

下面我们说说附件上传。

据说，最早的http协议是不支持附件上传的，后来有添加了一个RFC 2045 协议，才支持附件上传，关于附件上传，请参见

http://www.cnblogs.com/greenerycn/archive/2010/05/15/csharp_http_post.html

 

好了，其实用C#模拟上传附件，主要的难点就在于如何构建此协议的格式以及编码。

作为备忘，我这里只放一段成品代码了。

参数：URL,上传控件的id，附件的本地全路径，传输的Key-value数据

public string HttpWebUpload(string url, string fileControlID, string fileFullPath, Dictionary<string, string> dic)
        {
            string fileContentType = "application/octet-stream";//"application/octet-stream";//"text/plain";//image/jpeg
         
            int index = fileFullPath.LastIndexOf('\\');
            string filename = fileFullPath.Substring(index+1);
            
            var boundary = DateTime.Now.Ticks.ToString("x");
            var beginBoundary = "--" + boundary;
            var reqContentType = "multipart/form-data; boundary=" + boundary;

            var memStream = new MemoryStream();
            var fileStream = new FileStream(fileFullPath, FileMode.Open, FileAccess.Read);
            int fsLen = (int)fileStream.Length;
            byte[] btfileValue = new byte[fsLen];
            int r = fileStream.Read(btfileValue, 0, btfileValue.Length);

            StringBuilder fileHeader = new StringBuilder();
            fileHeader.AppendLine(beginBoundary);
            fileHeader.AppendLine(string.Format("Content-Disposition: form-data; name=\"{0}\"; filename=\"{1}\"", fileControlID, filename));
            fileHeader.AppendLine(string.Format("Content-Type: {0}", fileContentType));
            fileHeader.AppendLine();
            var btfileKey = Encoding.UTF8.GetBytes(fileHeader.ToString());

            StringBuilder dicData = new StringBuilder();
            dicData.AppendLine();
            foreach (var item in dic)
            {

                dicData.AppendLine(beginBoundary);
                dicData.AppendLine(string.Format("Content-Disposition:form-data; name=\"{0}\"", item.Key));
                dicData.AppendLine();
                dicData.AppendLine(item.Value);
            }

            dicData.AppendLine(beginBoundary + "--");
            var btDic = Encoding.UTF8.GetBytes(dicData.ToString());

            HttpWebRequest req = (HttpWebRequest)WebRequest.Create(url);
            req.Method = "POST";
            req.AllowAutoRedirect = false;
            req.ContentType = reqContentType;
            if (!string.IsNullOrEmpty(UserAgent))
            {
                req.UserAgent = this.UserAgent;
            }

            if (!string.IsNullOrEmpty(Cookie))
            {
                req.CookieContainer = new CookieContainer();
                req.CookieContainer.SetCookies(req.RequestUri, this.Cookie);
            }

            req.ContentLength = btfileKey.Length + btfileValue.Length + btDic.Length;
            Stream postDataStream = req.GetRequestStream();
            postDataStream.Write(btfileKey, 0, btfileKey.Length);
            postDataStream.Write(btfileValue, 0, btfileValue.Length);
            postDataStream.Write(btDic, 0, btDic.Length);
            postDataStream.Close();

            HttpWebResponse resp = (HttpWebResponse)req.GetResponse();
            string html = new StreamReader(resp.GetResponseStream()).ReadToEnd();
            return html;
        } 

　　注意，这里如果需要cookie,则添加对应的cookie。

调用方法

            string controlid = "fupCert";
            string filefullPath = "f:\\3.pdf";
            string url = "http://10.35.3.240/NT_HandbookExtend/FrmCertificate.aspx?type=N&zdtype=J&DELEGATENO=SBWJ16100001&ID=SB20161018000001%2c3206960757&Action=Modify&manualRecordID=SL20161018000001&tm=0.17032586549994377";

            var hiddenVal = web.GetDoNetHiddenValuesByUrl(url);//获取__VIEWSTATE，与__EVENTVALIDATION隐藏域值

            Dictionary<string, string> dic = new Dictionary<string, string>();
            dic.Add("__EVENTARGUMENT", "");
            dic.Add("__EVENTTARGET", "btnUpload");
            dic.Add("__EVENTVALIDATION", hiddenVal.Eventvalidation);
            dic.Add("__PREVIOUSPAGE", "EncayjCF95BJXxMazWGgd9UdPYeLp64GjuJlZ-rvnQ5n34-y7KQYllE35nAdVFvk0");
            dic.Add("__VIEWSTATE", hiddenVal.Vievstate);
            dic.Add("__VIEWSTATEENCRYPTED", "");
            dic.Add("grdNavigatorPRE_EMS3_CUS_IMG$ctl06", "");
            dic.Add("grdNavigatorPRE_EMS3_CUS_IMG$ctl07", "15");
            dic.Add("grdNavigatorPRE_EMS3_CUS_IMG$labelRowCount", "8");
            dic.Add("hfERP", "http://www.nteport.gov.cn/cerp/platform/erp/documentManagement/documentDocking.jsp");
            dic.Add("hidFileType", "333");
            dic.Add("hidID", "");
            //web.UserAgent = "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0)";
            string html = web.HttpWebUpload(url, controlid, filefullPath, dic);

　　

其他注意事项：

1.如果不限制文件类型，请使用 ContentType = "application/octet-stream";这个我也没怎么懂，尴尬。。。

2. 编码很重要，由于传参时，都是以流的形式传参，所以文件编码使用默认的编码方式，和其他构建的RFC 2045 协议格式，使用UTF-8编码。