第詳解C#如何利用爬蟲技術(shù)實現(xiàn)快捷租房ListView.SelectedListViewItemCollectionselectItem=listView1.SelectedItems;

if(selectItem.Count0)

Process.Start(selectItem[0].SubItems[1].Text);

///summary

///獲取前100個討論組

////summary

///returns/returns

privateListstringGetGroupUrls()

ListstringgroupUrls=newListstring

intpageSize=20;//豆瓣每頁顯示20個討論組

stringgroupUrl=string.Empty;

for(intgroupNum=0;groupNum=80;groupNum+=pageSize)

if(isStop)break;

groupUrl=string.Format(douBanUrl+"q={0}start={1}",HttpUtility.UrlEncode(txt_city.Text+"租房",Encoding.UTF8),groupNum);

stringtext=HttpUtil.HttpGet(groupUrl);

if(!string.IsNullOrWhiteSpace(text))

Regexreg=newRegex(@"(is)(=div\sclass=""pic""[^]*).*(=/div

MatchCollectionmatchs=reg.Matches(text);

for(inti=0;imatchs.Count;i++)

if(matchs[i].Success)

RegexregHref=newRegex("(=href\\s*=\\s*\")\\S+(!\")");

stringhref=regHref.Match(matchs[i].Value).Value;

groupUrls.Add(href);

returngroupUrls;

///summary

///獲取討論組內(nèi)符合條件的房源

////summary

///paramname="urls"/param

privatevoidGetHouseInfo(Liststringurls)

for(intu=0;uurls.Count;u++)

if(isStop)break;

this.BeginInvoke(newAction(()=

progressBar1.Value++;

}));

stringtext=HttpUtil.HttpGet(urls[u]+"/discussionstart=");

Regexregex=newRegex(@"(is)(=tr\sclass=""""[^]*).*(=/tr

MatchCollectionmatchs=regex.Matches(text);

for(inti=0;imatchs.Count;i++)

if(matchs[i].Success)

#region匹配時間

RegexregDate=newRegex(@"(is)(=td\snowrap=""nowrap""\sclass=""time""[^]*).*(=/td

MatchmatchDate=regDate.Match(matchs[i].Value);

if(!matchDate.Success)

continue;

stringtime=matchDate.Value;

DateTimedtTemp=newDateTime();

if(!DateTime.TryParse(time,outdtTemp))

time=time.Insert(0,DateTime.Now.Year+"-");

dtTemp=Convert.ToDateTime(time);

if(dtTempdateTimePicker1.Value.Date)

continue;

#endregion

#region獲取標(biāo)題

RegexregTitle=newRegex("(=title\\s*=\\s*\")\\s*\\S*(!\")");

MatchmatchTitle=regTitle.Match(matchs[i].Value);

if(matchTitle.Success)

stringtitle=matchTitle.Value;

#region匹配標(biāo)題

boolisContain=false;

string[]keys=txt_keys.Text.Split('*');

foreach(stringkeyinkeys)

if(matchTitle.Value.Contains(key))

isContain=true;

#endregion

if(isContain)

//獲取鏈接

RegexregHref=newRegex("(=