using?System;
using?System.Collections.Generic;
using?System.IO;
using?System.Net;
using?System.Text;
using?System.Text.Regularexpressions;
using?System.Web;

///?

///?????網頁采集輔助類
///?

public?static?class?Collectionhelper
{
????///?

????///?????取得字符里的Dom元素?不包含元素屬性
????///?

????///?
????///?
????///?
????public?static?List?GetDomElem（string?source?string?domElem）
????{
????????var?matchList?=?new?List（）;
????????string?regStr?=?string.Format（“<{0}[^>]*?>[\\s\\S]+?<\\/{0}>“?domElem）;
????????try
????????{
????????????var?regex?=?new?Regex（regStr?RegexOptions.Compiled?|?RegexOptions.IgnoreCase）;
????????????MatchCollection?matches?=?regex.Matches（source）;
????????????foreach?（Match?match?in?matches）
????????????{
????????????????matchList.Add（match.Value）;
????????????}
????????}
????????catch?（Exception?ex）
????????{
????????????matchList.Add（ex.Message）;
????????}
????????return?matchList;
????}

????///?

????///?????取得字符里的Dom元素?包含元素屬性?如：class=“aa“
????///?

????///?
????///?
????///?
????///?
????public?static?List?GetDomElemByAttr（string?source?string?tagName?string?tagValue）
????{
????????var?matchList?=?new?List（）;
????????string?regStr?=
????????????string.Format（
????????????????@“<（?[\w]+）[^>]*\s{0}[\s]*?=[\s]*?（?[““‘]?）{1}（?（Quote）\k）[““‘]?[^>]*>（（?<\k[^>]*>）|>（?<-Nested>）|[\s\S]*?）*>“
????????????????tagName.ToLower（）?tagValue）;
????????try
????????{
????????????var?regex?=?new?Regex（regStr?RegexOptions.Compiled?|?RegexOptions.IgnoreCase）;
????????????var?matches?=?regex.Matches（source）;
????????????foreach?（Match?match?in?matches）
????????????{
????????????????matchList.Add（match.Value）;
????????????}
????????}
????????catch?（Exception?ex）
????????{
????????????matchList.Add（ex.Message）;
????????}
????????return?matchList;
????}

????///?

????///?????取得字符里的A元素鍵值對??[name，url]
????///?

????///?
????///?
????public?static?Dictionary?GetDomElem_A（string?source）
????{
????????var?matchList?=?new?Dictionary（）;
????????const?string?pattern?=?“]*??href=[\“‘]（?[^\“‘]*?）[\“‘][^>]*?>（?[\\w\\W]*?）“;
????????try
????????{
????????????var?regex?=?new?Regex（pattern?RegexOptions.Compiled?|?RegexOptions.IgnoreCase）;
????????????MatchCollection?matches?=?regex.Matches（source）;

????????????foreach?（Match?match?in?matches）
????????????{
????????????????string?key?=?RemoveHtml（match.Value）;
????????????????if?（!matchList.ContainsKey（key））
????????????????{
????????????????????matchList.Add（key?GetUrlArray（matc

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----

?????文件???????9891??2016-06-29?17:44??省市區街道\Collectionhelper.cs

?????文件????????460??2018-11-06?16:50??省市區街道\data\Default.aspx

?????文件??????10026??2018-11-08?14:54??省市區街道\data\Default.aspx.cs

?????文件????????535??2018-11-08?10:16??省市區街道\Whir_Cmn_Area.cs

?????文件???32440950??2018-11-08?14:57??省市區街道\截止2017年10月31日.sql

?????目錄??????????0??2018-11-08?15:05??省市區街道\data

?????目錄??????????0??2018-11-08?15:05??省市區街道

-----------?---------??----------?-----??----

?????????????32461862????????????????????7