用正则表达式解析BT网页到本地数据集

教育网的5Q网站有很多的资源，其中包括很多讲座学习资料，天天上觉得不方便，于是写了一个小软件来解析它。加入XtraGrid控件，查看方便，希望大家多给意见！
waxwork3@163.com



正则解析源码：

try 
            {
                this.m_WaitDialogForm.Caption = "Geting page";
                this.GetPage();

                this.m_WaitDialogForm.Caption = "Parsing data";
                string regex = @"(?<PublishTime>\d\d-\d\d\s\d\d:\d\d)
                                    .*\n.*0066CC>
                                    (?<Catalog>[^<]*)
                                    .*\n.*red>
                                    (?<Seed>[^<]*)
                                    .*\n.*green>
                                    (?<Connection>[^<]*)
                                    .*\n.*?""
                                    (?<SeedLink>[^""]*)
                                    .*?
                                    (?<Content>【.*?)
                                    </.*\n.*?""
                                    (?<Seed_1>[^""]*)
                                    "".*?""
                                    (?<Seed_2>[^""]*)
                                    "".*?""
                                    (?<Seed_3>[^""]*)
                                    "".*?""
                                    (?<Detail>[^""]*)
                                    .*\n.*>
                                    (?<Capacity>.*\..*?)
                                    <.*\n.*>
                                    (?<Completed>
                                    .*)<
                                    ";
                System.Text.RegularExpressions.RegexOptions options = ((System.Text.RegularExpressions.RegexOptions.IgnorePatternWhitespace | System.Text.RegularExpressions.RegexOptions.Multiline) 
                    | System.Text.RegularExpressions.RegexOptions.IgnoreCase);
                System.Text.RegularExpressions.Regex reg = new System.Text.RegularExpressions.Regex(regex, options);

                Match MatchResults = reg.Match(this.m_SubjectString);
                while (MatchResults.Success) 
                {
                    DataRow drw = this.m_dsBT.BT.Rows.Find(MatchResults.Groups["Content"].Value.GetHashCode());
                    if(drw != null)
                    {
                        drw["Seed"] = Convert.ToInt32(MatchResults.Groups["Seed"].ToString());
                        drw["Connection"] = Convert.ToInt32(MatchResults.Groups["Connection"].ToString());;
                        MatchResults = MatchResults.NextMatch();
                        continue;
                    }

                    dsBT.BTRow row = tbl.NewBTRow();
                    try
                    {
                        row.BeginEdit();
                    
                        row.Content = MatchResults.Groups["Content"].ToString();
                        row.ContentHashCode = MatchResults.Groups["Content"].Value.GetHashCode();
                        row.PublishTime = Convert.ToDateTime(DateTime.Now.Year + "-" + MatchResults.Groups["PublishTime"].ToString());
                        row.Catalog = MatchResults.Groups["Catalog"].ToString();
                        row.Seed = Convert.ToInt32(MatchResults.Groups["Seed"].ToString());
                        row.Connection = Convert.ToInt32(MatchResults.Groups["Connection"].ToString());
                        row.SeedLink = MatchResults.Groups["SeedLink"].ToString();
                        row.Detail = MatchResults.Groups["Detail"].ToString();
                        row.Completed = Convert.ToInt32(MatchResults.Groups["Completed"].ToString());
                        row.History = 0;

                        strTemp = MatchResults.Groups["Capacity"].ToString();
                        if(strTemp.IndexOf("M") - 1 > 0)
                        {
                            strTemp = strTemp.Substring(0, strTemp.Length - 3 > 0 ? strTemp.Length - 3 : 0);
                            row.Capacity = (int)Convert.ToSingle(strTemp);
                        }
                        else
                        {
                            strTemp = strTemp.Substring(0, strTemp.Length - 3 > 0 ? strTemp.Length - 3 : 0);
                            row.Capacity = (int)(Convert.ToSingle(strTemp) * 1024);
                        }

                        row.EndEdit();
                        tbl.Rows.Add(row);
                    }
                    catch(Exception ex)
                    {
                        this.comboBox1.SelectedIndex = this.comboBox1.SelectedIndex < 4 ? this.comboBox1.SelectedIndex + 1 : 0;
                        frmMessage message = new frmMessage(ex.Message);
                        message.Show();
                        //                        MessageBox.Show(ex.Message);
                    }
                    MatchResults = MatchResults.NextMatch();
                } 
                
                if(tbl.Rows.Count > 0)
                {
                    foreach(DataRow row in this.m_dsBT.BT.Rows)
                    {
                        row["History"] = (int)row["History"] + 1;
                    }
                    this.Text = "上次更新时间：" + DateTime.Now.ToString() + " | 导入" + tbl.Rows.Count + "条数据";
                    this.m_WaitDialogForm.Close();
                    this.m_dsBT.Merge(tbl);
                    this.gridView1.BestFitColumns();
                }
                else
                {
                    this.Text = "上次更新时间：" + DateTime.Now.ToString();
                    this.m_WaitDialogForm.Close();                    
                }
            } 
            catch (Exception ex) 
            {
                this.comboBox1.SelectedIndex = this.comboBox1.SelectedIndex < 4 ? this.comboBox1.SelectedIndex + 1 : 0;
                this.m_WaitDialogForm.Close();
                frmMessage message = new frmMessage(ex.Message);
                message.Show();
                //                MessageBox.Show(ex.Message);
            }

程序下载

转载于:https://www.cnblogs.com/waxwork3/archive/2006/02/24/336649.html