得到页面内容

先引用 Interop.MSXML2.dll

MSXML2.XMLHTTP xmlhttp = new MSXML2.XMLHTTP();
            string url = "http://www.guge.cc/board.html?dt=" + date;
            xmlhttp.open("GET", url, false, null, null);
            xmlhttp.send("");
            Byte[] txt = (Byte[])xmlhttp.responseBody;
            string content = System.Text.Encoding.GetEncoding("GB2312").GetString(txt).Trim();

            int startIndex = content.IndexOf("<tbody>");
            int endIndex = content.IndexOf("</tbody>");
            string TableString = content.Substring(startIndex + 7, endIndex - startIndex);

            Regex regex1 = new Regex(@"<(|/)tr[^>]*>", RegexOptions.IgnoreCase);
            TableString = regex1.Replace(TableString, "");
            Regex regex2 = new Regex(@"<(|/)a[^>]*>", RegexOptions.IgnoreCase);
            TableString = regex2.Replace(TableString, "");
            Regex regex3 = new Regex(@"<(|/)font[^>]*>", RegexOptions.IgnoreCase);
            TableString = regex3.Replace(TableString, "");
            TableString = TableString.Replace("\"", "");
            TableString = TableString.Replace("</td>", "");
            TableString = TableString.Replace("\r\n        ", "");
            Regex regex4 = new Regex(@"align=\w*", RegexOptions.IgnoreCase);
            TableString = regex4.Replace(TableString, "");
            TableString = TableString.Replace("<td >", ",");
            TableString = TableString.Replace(" ", "");
            string[] AllData = TableString.Split(',');
            List<string> myData = new List<string>();
            for (int i = 1; i < AllData.Length; i = i + 13)
            {
                myData.Add(AllData[i] + "," + AllData[i + 11]);
            }
            foreach (string oneData in myData)
            {
                string TableName = "IndustryJLR";

                string[] data = oneData.Split(',');

                try
                {
                    if (SqlHelper.ExecuteDataTable("select jdate from " + TableName + " where jdate ='" + date + "'").Rows.Count > 0)
                    {
                        string sql = "update " + TableName + " set " + data[0] + "=" + data[1] + " where jdate ='" + date + "'";
                        SqlHelper.ExecuteNonQuery(sql);
                    }
                    else
                    {
                        string sql = "insert into " + TableName + "(jdate," + data[0] + ") values ('" + date + "'," + data[1] + ")";
                        SqlHelper.ExecuteNonQuery(sql);
                    }
                }
                catch (Exception ex)
                {
                    if (ex.ToString().IndexOf("列名") >= 0 && ex.ToString().IndexOf("无效") >= 0)
                    {
                        string sqlAdd = "alter table [dbo].[" + TableName + "] add " + data[0] + " decimal(18,2)";
                        SqlHelper.ExecuteNonQuery(sqlAdd);
                        if (SqlHelper.ExecuteDataTable("select jdate from " + TableName + " where jdate ='" + date + "'").Rows.Count > 0)
                        {
                            string sql = "update " + TableName + " set " + data[0] + "=" + data[1] + " where jdate ='" + date + "'";
                            SqlHelper.ExecuteNonQuery(sql);
                        }
                        else
                        {
                            string sql = "insert into " + TableName + "(jdate," + data[0] + ") values ('" + date + "'," + data[1] + ")";
                            SqlHelper.ExecuteNonQuery(sql);
                        }
                    }
                }

            }

转载于:https://www.cnblogs.com/server/archive/2009/09/04/1560041.html