public void run(int item)
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i < 100; i++)
{
String Url = "https://" + city + ".nuomi.com/" + item + "-page" + i;
string html = Method.GetUrl(Url); //定义的GetRul方法 返回 reader.ReadToEnd()
MatchCollection TitleMatchs = Regex.Matches(html, @"<li class=""shop-infoo-list-item clearfix"">([\s\S]*?)<a href=""([\s\S]*?)""", RegexOptions.IgnoreCase | RegexOptions.Multiline);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add("https:" + NextMatch.Groups[2].Value);
}
if (lists.Count == 0)
{
break;
}
foreach (string list in lists)
{
int index = this.skinDataGridView1.Rows.Add();
string strhtml = Method.GetUrl(list); //定义的GetRul方法 返回 reader.ReadToEnd()
string Rxg = @"name:'([\s\S]*?)'";
string Rxg1 = @"address:'([\s\S]*?)'";
string Rxg2 = @"phone:'([\s\S]*?)'";
Match name = Regex.Match(strhtml, Rxg);
Match addr = Regex.Match(strhtml, Rxg1);
Match tel = Regex.Match(strhtml, Rxg2);
this.skinDataGridView1.Rows[index].Cells[0].Value = name.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[2].Value = addr.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[1].Value = tel.Groups[1].Value;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[i].Cells[0]; //让datagridview滚动到当前行
Application.DoEvents();
System.Threading.Thread.Sleep(800);
}
}
}
private void button1_Click(object sender, EventArgs e)
{
#region 通用验证
bool value = false;
string html = Method.GetUrl("http://acaiji.com/success/ip.php");
string localip = Method.GetIP();
MatchCollection ips = Regex.Matches(html, @"<td style='color:red;'>([\s\S]*?)</td>", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match ip in ips)
{
if (ip.Groups[1].Value.Trim() == "9.9.9.9")
{
value = true;
break;
}
}
if (value == true)
{
Thread thread = new Thread(new ThreadStart(run));
thread.Start();
Control.CheckForIllegalCrossThreadCalls = false;
}
else
{
MessageBox.Show("IP不符");
}
#endregion
}
public int getcityId(string cityName)
{
try
{
String Url = "https://map.baidu.com/?newmap=1&reqflag=pcmap&biz=1&from=webmap&da_par=direct&pcevaname=pc4.1&qt=s&da_src=searchBox.button&wd=" + cityName + "&c=289&src=0&wd2=&pn=0&sug=0&l=12&b=(13461858.87,3636969.979999999;13584738.87,3670185.979999999)&from=webmap&biz_forward={%22scaler%22:1,%22styles%22:%22pl%22}&sug_forward=&tn=B_NORMAL_MAP&nn=0&u_loc=13166533,3998088&ie=utf-8";
string html = Method.GetUrl(Url);
MatchCollection Matchs = Regex.Matches(html, @"""code"":([\s\S]*?),", RegexOptions.IgnoreCase);
int cityId = Convert.ToInt32(Matchs[0].Groups[1].Value);
return(cityId);
}
catch (System.Exception ex)
{
ex.ToString();
return(1);
}
}
public void Che()
{
try
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "http://" + city + ".ganji.com/ershouche/a1o" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"id=""puid-([\s\S]*?)""", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match NextMatch in TitleMatchs)
{
textBox14.Text += "https://3g.ganji.com/" + city + "_ershouche/" + NextMatch.Groups[1].Value + "x" + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
Application.DoEvents();
System.Threading.Thread.Sleep(500); //网址获取时间间隔 固定不变
string[] lines = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines.Length - 1; j++)
{
this.index = this.skinDataGridView1.Rows.Add();
String Url1 = lines[j];
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string rxg = @"title>([\s\S]*?)</title>";
string Rxg = @"联系人</span>([\s\S]*?)</span>";
string Rxg1 = @"data-phone=""([\s\S]*?)""";
string Rxg2 = @"province=([\s\S]*?);city=([\s\S]*?);";
Match titles = Regex.Match(strhtml, rxg);
Match contacts = Regex.Match(strhtml, Rxg);
Match tell = Regex.Match(strhtml, Rxg1);
Match region = Regex.Match(strhtml, Rxg2);
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
string temp = Regex.Replace(contacts.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[1].Value = temp1;
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[3].Value = region.Groups[2].Value;
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(time); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
//MessageBox.Show(ex.ToString());
}
}
public void ershoufang()
{
try
{
string[] headers = { "标题", "联系人", "电话", "地区", "小区", "价格", "网址" };
setDatagridview(skinDataGridView1, 7, headers);
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "https://" + city + ".58.com/ershoufang/0/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"esf_id:([\s\S]*?),", RegexOptions.IgnoreCase | RegexOptions.Multiline);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add("https://" + city + ".58.com/ershoufang/" + NextMatch.Groups[1].Value + "x.shtml");
//滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
foreach (string url in lists)
{
this.index = this.skinDataGridView1.Rows.Add();
string strhtml = Method.GetUrl(url);
Match titles = Regex.Match(strhtml, @"<h1 class=""c_333 f20"">([\s\S]*?)</h1>");
Match contacts = Regex.Match(strhtml, @"""linkman"":""([\s\S]*?)"""); //unicode
Match tell = Regex.Match(strhtml, @"<p class='phone-num'>([\s\S]*?)<");
Match area = Regex.Match(strhtml, @"quyu'\)"">([\s\S]*?)<");
Match xiaoqu = Regex.Match(strhtml, @"xiaoqu_name'\)"">([\s\S]*?)<");
Match price = Regex.Match(strhtml, @"MinPrice':'([\s\S]*?)'");
textBox14.Text = contacts.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[1].Value = Unicode2String(contacts.Groups[1].Value.Trim());
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[3].Value = label14.Text + "" + area.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[4].Value = xiaoqu.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[5].Value = price.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[6].Value = url;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(1000); //内容获取间隔，可变量
}
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
}
}
public void chewei()
{
try
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "http://" + city + ".58.com/cheku/0/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<h2 class='title'>([\s\S]*?)<a href=""([\s\S]*?)""", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match NextMatch in TitleMatchs)
{
textBox14.Text += NextMatch.Groups[2].Value + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
Application.DoEvents();
System.Threading.Thread.Sleep(500); //网址获取时间间隔 固定不变
string[] lines = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines.Length - 1; j++)
{
this.index = this.skinDataGridView1.Rows.Add();
String Url1 = lines[j];
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string title = @"<title>([\s\S]*?)</title>";
string Rxg = @"<span class='f14 c_333 jjrsay'>([\s\S]*?)</span>";
string Rxg1 = @"<p class='phone-num'>([\s\S]*?)</p>";
string Rxg2 = @"province=([\s\S]*?);";
Match titles = Regex.Match(strhtml, title);
Match contacts = Regex.Match(strhtml, Rxg);
Match tell = Regex.Match(strhtml, Rxg1);
Match region = Regex.Match(strhtml, Rxg2);
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[1].Value = contacts.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[3].Value = region.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[4].Value = lines[j];
this.skinDataGridView1.Columns[2].FillWeight = 40; //设置列宽
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(time); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
//MessageBox.Show(ex.ToString());
}
}
public void run()
{
string city = Method.Get58pinyin(comboBox1.SelectedItem.ToString());
try
{
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i < 71; i++)
{
String Url = "http://" + city + ".58.com/ershoufang/pn" + i + "/" + "?key=" + textBox1.Text.Trim();
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<li logr=""([\s\S]*?)_([\s\S]*?)_([\s\S]*?)_([\s\S]*?)_", RegexOptions.IgnoreCase | RegexOptions.Multiline);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add("http://" + city + ".58.com/ershoufang/" + NextMatch.Groups[4].Value + "x.shtml");
//滚动到光标处
}
if (lists.Count < 0)
{
return;
}
for (int j = 0; j < lists.Count; j++)
{
this.index = this.dataGridView1.Rows.Add();
String Url1 = lists[j].ToString();
// str = str.Substring(str.Length - i) 从右边开始取i个字符
string Url2 = "http://m.58.com/" + city + "/ershoufang/" + Url1.Substring(Url1.Length - 21); //获取二手房手机端的网址
textBox2.Text = Url2;
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string strhtml2 = Method.GetUrl(Url2); //请求手机端网址
string rxg = @"<h1 class=""c_333 f20"">([\s\S]*?)</h1>"; //标题
string Rxg = @"<h2 class=""agent-title"">([\s\S]*?)</h2>"; //手机端正则匹配联系人
string Rxg1 = @"<p class='phone-num'>([\s\S]*?)<"; //电话
Match titles = Regex.Match(strhtml, rxg);
Match contacts = Regex.Match(strhtml2, Rxg); //手机端正则匹配联系人
Match tell = Regex.Match(strhtml, Rxg1);
this.dataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value;
this.dataGridView1.Rows[index].Cells[1].Value = contacts.Groups[1].Value;
this.dataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0];
while (this.zanting == false)
{
Application.DoEvents();//如果loader是false表明正在加载,,则Application.DoEvents()意思就是处理其他消息。阻止当前的队列继续执行。
}
if (this.status == false)
{
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(1000); //内容获取间隔，可变量
}
}
}
catch (System.Exception ex)
{
MessageBox.Show(ex.ToString());
}
}
public void gaode()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 200;
foreach (string city in citys)
{
int citycode = gaodeCityId(city);
foreach (string keyword in keywords)
{
string keywordutf8 = System.Web.HttpUtility.UrlEncode(keyword, System.Text.Encoding.GetEncoding("utf-8"));
for (int i = 0; i <= pages; i++)
{
String Url = "http://m.amap.com/service/poi/keywords.json?pagenum=" + i + "&user_loc=undefined&geoobj=&city=" + citycode + "&keywords=" + keywordutf8;
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"""diner_flag"":([\s\S]*?),""id"":""([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[2].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
Application.DoEvents();
Thread.Sleep(100);
foreach (string poid in lists)
{
int index = this.dataGridView1.Rows.Add();
string Url1 = "https://www.amap.com/detail/" + poid + "?citycode=" + citycode;
string strhtml = Method.GetUrl(Url1);
string title = @"<h4 class=""detail_title"">([\s\S]*?)</h4>";
string Rxg = @"""telephone"":""([\s\S]*?)""";
string Rxg1 = @"""address"":""([\s\S]*?)""";
MatchCollection titles = Regex.Matches(strhtml, title);
MatchCollection tell = Regex.Matches(strhtml, Rxg);
MatchCollection address = Regex.Matches(strhtml, Rxg1);
if (visualButton2.Text == "已停止")
{
return;
}
foreach (Match match in titles)
{
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = match.Groups[1].Value;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0];
}
foreach (Match match in tell)
{
this.dataGridView1.Rows[index].Cells[2].Value += match.Groups[1].Value.Trim() + ",";
}
foreach (Match match in address)
{
this.dataGridView1.Rows[index].Cells[3].Value = match.Groups[1].Value;
}
this.dataGridView1.Rows[index].Cells[4].Value = keyword;
this.dataGridView1.Rows[index].Cells[5].Value = city;
Application.DoEvents();
Thread.Sleep(500); //内容获取间隔，可变量
}
}
}
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
}
}
public void tengxun()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 250;
foreach (string city in citys)
{
string cityutf8 = System.Web.HttpUtility.UrlEncode(city, System.Text.Encoding.GetEncoding("utf-8"));
foreach (string keyword in keywords)
{
string keywordutf8 = System.Web.HttpUtility.UrlEncode(keyword, System.Text.Encoding.GetEncoding("utf-8"));
for (int i = 0; i <= pages; i++)
{
String Url = "http://map.qq.com/m/place/result/city=" + cityutf8 + "&word=" + keywordutf8 + "&bound=&page=" + i + "&cpos=&mode=list";
textBox3.Text = Url;
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"poid=([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[1].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
Application.DoEvents();
Thread.Sleep(100);
foreach (string poid in lists)
{
int index = this.dataGridView1.Rows.Add();
string Url1 = "http://map.qq.com/m/detail/poi/poid=" + poid;
string strhtml = Method.GetUrl(Url1);
string title = @"<div class=""poiDetailTitle "">([\s\S]*?)</div>";
string Rxg = @"<a href=""tel:([\s\S]*?)""";
string Rxg1 = @"span class=""poiDetailAddrTxt"">([\s\S]*?)</span>";
Match titles = Regex.Match(strhtml, title);
Match tell = Regex.Match(strhtml, Rxg);
Match address = Regex.Match(strhtml, Rxg1);
if (visualButton2.Text == "已停止")
{
return;
}
if (tell.Groups[1].Value.Trim() != "")
{
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = titles.Groups[1].Value;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0];
this.dataGridView1.Rows[index].Cells[2].Value += tell.Groups[1].Value.Trim() + ",";
this.dataGridView1.Rows[index].Cells[3].Value = address.Groups[1].Value;
this.dataGridView1.Rows[index].Cells[4].Value = keyword;
this.dataGridView1.Rows[index].Cells[5].Value = city;
}
Application.DoEvents();
Thread.Sleep(500); //内容获取间隔，可变量
}
}
}
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
}
}
public void txMap()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 200;
foreach (string city in citys)
{
int cityid = getcityId(city + "市"); //获取 citycode;
foreach (string keyword in keywords)
{
for (int i = 0; i <= pages; i++)
{
String Url = "https://map.baidu.com/?newmap=1&reqflag=pcmap&biz=1&from=webmap&da_par=direct&pcevaname=pc4.1&qt=con&from=webmap&c=" + cityid + "&wd=" + keyword + "&wd2=&pn=" + i + "&nn=" + i + "0&db=0&sug=0&addr=0&&da_src=pcmappg.poi.page&on_gel=1&src=7&gr=3&l=13&tn=B_NORMAL_MAP&u_loc=13167420,3999298&ie=utf-8";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"""primary_uid"":""([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[1].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
JsonParser jsonParser = JsonConvert.DeserializeObject <JsonParser>(html);
foreach (Content content in jsonParser.Content)
{
int index = this.dataGridView1.Rows.Add();
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = content.name;
this.dataGridView1.Rows[index].Cells[2].Value = content.tel;
this.dataGridView1.Rows[index].Cells[3].Value = content.addr;
this.dataGridView1.Rows[index].Cells[4].Value = keyword.Trim();
this.dataGridView1.Rows[index].Cells[5].Value = city;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (visualButton2.Text == "已停止")
{
return;
}
}
Application.DoEvents();
Thread.Sleep(100); //内容获取间隔，可变量
}
}
}
}
catch (System.Exception ex)
{
textBox1.Text = ex.ToString();
}
}
public void run()
{
try
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
string item = label1.Text;
if (item == "")
{
MessageBox.Show("请选择分类！");
return;
}
int page = 50;
for (int i = 1; i <= page; i++)
{
String Url = "http://b2b.huangye88.com/" + city + "/" + item + "/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<h4><a href=""([\s\S]*?)""", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match NextMatch in TitleMatchs)
{
textBox14.Text += NextMatch.Groups[1].Value + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox1.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox1.SelectionStart = textBox1.Text.Length; //设定光标位置
textBox1.ScrollToCaret();
Application.DoEvents();
System.Threading.Thread.Sleep(500); //网址获取时间间隔 固定不变
string[] lines = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines.Length - 1; j++)
{
int index = this.skinDataGridView1.Rows.Add();
String Url1 = lines[j];
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string rxg = @"<dl class=""bottom"">([\s\S]*?)名称：([\s\S]*?)<"; //公司
string Rxg = @"<dl class=""bottom"">([\s\S]*?)手机：([\s\S]*?)<"; //电话
string Rxg1 = @"联系人：([\s\S]*?)rel=""nofollow"">([\s\S]*?)</a>"; //联系人
string Rxg2 = @"<meta name=""Description"" content=""([\s\S]*?)"""; //介绍
MatchCollection company = Regex.Matches(strhtml, rxg);
MatchCollection tel = Regex.Matches(strhtml, Rxg);
MatchCollection contacts = Regex.Matches(strhtml, Rxg1);
MatchCollection introduction = Regex.Matches(strhtml, Rxg2);
foreach (Match NextMatch in company)
{
this.skinDataGridView1.Rows[index].Cells[0].Value = NextMatch.Groups[2].Value;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
}
foreach (Match NextMatch in tel)
{
this.skinDataGridView1.Rows[index].Cells[1].Value = NextMatch.Groups[2].Value;
}
foreach (Match NextMatch in contacts)
{
this.skinDataGridView1.Rows[index].Cells[2].Value = NextMatch.Groups[2].Value;
}
foreach (Match NextMatch in introduction)
{
this.skinDataGridView1.Rows[index].Cells[3].Value = NextMatch.Groups[1].Value;
}
if (skinButton2.Text == "已停止")
{
skinButton1.Enabled = false;
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(Convert.ToInt32(500)); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
MessageBox.Show(ex.ToString());
}
}
//string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
#region 生意转让、商铺出租、商铺出售
public void shangpu(object item)
{
string[] headers = { "标题", "联系人", "电话", "地区", "网址" };
setDatagridview(skinDataGridView1, 5, headers);
try
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "http://" + city + ".58.com/" + item.ToString() + "/0/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<li logr=""([\s\S]*?)_0_([\s\S]*?)_([\s\S]*?)_", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match NextMatch in TitleMatchs)
{
textBox14.Text += "https://" + city + ".58.com/shangpu/" + NextMatch.Groups[3].Value + "x.shtml" + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
Application.DoEvents();
Thread.Sleep(500); //网址获取时间间隔 固定不变
string[] lines = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines.Length - 1; j++)
{
this.index = this.skinDataGridView1.Rows.Add();
String Url1 = lines[j];
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string title = @"<h1 class=""c_000 f20"">([\s\S]*?)</h1>";
string Rxg = @"<span class='f14 c_333 jjrsay'>([\s\S]*?)</span>";
string Rxg1 = @"<p class='phone-num'>([\s\S]*?)</p>";
string Rxg2 = @"详细地址:</span>([\s\S]*?)</span>";
Match titles = Regex.Match(strhtml, title);
Match contacts = Regex.Match(strhtml, Rxg);
Match tell = Regex.Match(strhtml, Rxg1);
Match region = Regex.Match(strhtml, Rxg2);
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[1].Value = contacts.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
string temp = Regex.Replace(region.Groups[1].Value, "<[^>]*>", "");
this.skinDataGridView1.Rows[index].Cells[3].Value = temp.Trim().Replace(" ", "").Replace(" ", "");
this.skinDataGridView1.Rows[index].Cells[4].Value = lines[j];
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
Thread.Sleep(time); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
ex.ToString();
}
}
public void ershoufang()
{
try
{
string[] headers = { "标题", "联系人", "电话", "地区", "小区", "面积", "价格", "网址" };
setDatagridview(skinDataGridView1, 8, headers);
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "http://" + city + ".58.com/ershoufang/0/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<li logr=""([\s\S]*?)_0_([\s\S]*?)_([\s\S]*?)_", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (Match NextMatch in TitleMatchs)
{
textBox14.Text += "http://" + city + ".58.com/ershoufang/" + NextMatch.Groups[3].Value + "x.shtml" + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
string[] lines = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines.Length - 1; j++)
{
this.index = this.skinDataGridView1.Rows.Add();
String Url1 = lines[j];
// str = str.Substring(str.Length - i) 从右边开始取i个字符
string Url2 = "http://m.58.com/" + city + "/ershoufang/" + Url1.Substring(Url1.Length - 21); //获取二手房手机端的网址
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string strhtml2 = Method.GetUrl(Url2); //请求手机端网址
string title = @"<h1 class=""c_333 f20"">([\s\S]*?)</h1>"; //标题
string Rxg = @"<h2 class=""agent-title"">([\s\S]*?)</h2>"; //手机端正则匹配联系人
string Rxg1 = @"<p class='phone-num'>([\s\S]*?)<"; //电话
string Rxg2 = @"<li class=""address-info"">([\s\S]*?) -"; //手机端地区
string Rxg3 = @"小区：([\s\S]*?)</h2>"; //手机端小区
string Rxg4 = @"面积</p>([\s\S]*?)</p>"; //手机端面积去除标签
string Rxg5 = @"售价</p>([\s\S]*?)</p>"; //手机端售价去除标签
Match titles = Regex.Match(strhtml, title);
Match contacts = Regex.Match(strhtml2, Rxg); //手机端正则匹配联系人
Match tell = Regex.Match(strhtml, Rxg1);
Match area = Regex.Match(strhtml2, Rxg2);
Match xiaoqu = Regex.Match(strhtml2, Rxg3);
Match mianji = Regex.Match(strhtml2, Rxg4);
Match price = Regex.Match(strhtml2, Rxg5);
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[1].Value = contacts.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[3].Value = label14.Text + "" + area.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[4].Value = xiaoqu.Groups[1].Value;
string temp = Regex.Replace(mianji.Groups[1].Value, "<[^>]*>", "");
this.skinDataGridView1.Rows[index].Cells[5].Value = temp.Trim();
string temp1 = Regex.Replace(price.Groups[1].Value, "<[^>]*>", "");
this.skinDataGridView1.Rows[index].Cells[6].Value = temp1.Trim();
this.skinDataGridView1.Rows[index].Cells[7].Value = lines[j];
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(time); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
}
}
//string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
#region 生意转让、商铺出租、商铺出售
public void shangpu(object item)
{
string[] headers = { "标题", "联系人", "电话", "地区", "网址", "面积" };
setDatagridview(skinDataGridView1, 6, headers);
try
{
string city = label6.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
for (int i = 1; i <= page; i++)
{
String Url = "https://" + city + ".58.com/" + item.ToString() + "/0/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"https://[a-z]+.58.com/[a-z]+/[0-9]+x.shtml", RegexOptions.IgnoreCase | RegexOptions.Multiline);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
if (!lists.Contains(NextMatch.Groups[0].Value))
{
lists.Add(NextMatch.Groups[0].Value);
}
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集第" + i + "页\r\n";
textBox3.SelectionStart = textBox3.Text.Length; //设定光标位置
textBox3.ScrollToCaret();
Application.DoEvents();
Thread.Sleep(500); //网址获取时间间隔 固定不变
foreach (string list in lists)
{
this.index = this.skinDataGridView1.Rows.Add();
String Url1 = list;
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
Match titles = Regex.Match(strhtml, @"<title>([\s\S]*?)-");
Match contacts = Regex.Match(strhtml, @"<p class=""poster-name"">([\s\S]*?)</p>");
Match tell = Regex.Match(strhtml, @"""phone"":""([\s\S]*?)""");
Match region = Regex.Match(strhtml, @"域：</span>([\s\S]*?)</p>");
Match mianji = Regex.Match(strhtml, @"<span class=""up"">([\s\S]*?)</span>");
this.skinDataGridView1.Rows[index].Cells[0].Value = titles.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[1].Value = contacts.Groups[1].Value.Trim();
this.skinDataGridView1.Rows[index].Cells[2].Value = tell.Groups[1].Value;
string temp = Regex.Replace(region.Groups[1].Value, "<[^>]*>", "");
this.skinDataGridView1.Rows[index].Cells[3].Value = temp.Trim().Replace(" ", "").Replace(" ", "");
this.skinDataGridView1.Rows[index].Cells[4].Value = list;
this.skinDataGridView1.Rows[index].Cells[5].Value = mianji.Groups[1].Value.Trim();
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (skinButton13.Text == "已停止")
{
return;
}
Application.DoEvents();
Thread.Sleep(time); //内容获取间隔，可变量
}
textBox14.Text = "";
}
}
catch (System.Exception ex)
{
ex.ToString();
}
}
public void map_360()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 100;
foreach (string city in citys)
{
string cityutf8 = System.Web.HttpUtility.UrlEncode(city, System.Text.Encoding.GetEncoding("utf-8"));
foreach (string keyword in keywords)
{
string keywordutf8 = System.Web.HttpUtility.UrlEncode(keyword, System.Text.Encoding.GetEncoding("utf-8"));
for (int i = 1; i <= pages; i++)
{
String Url = "https://ditu.so.com/app/pit?jsoncallback=jQuery18308131636402501483_1525852464213&keyword=" + keywordutf8 + "&cityname=" + cityutf8 + "&batch=" + i + "%2c" + (i + 1) + "%2c" + (i + 2) + "%2c" + (i + 3) + "%2c" + (i + 4) + "&number=10";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"""pguid"":""([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[1].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
Application.DoEvents();
Thread.Sleep(100);
foreach (string poid in lists)
{
int index = this.dataGridView1.Rows.Add();
string Url1 = "https://m.map.so.com/onebox/?type=detail&id=" + poid + "&mso_x=&mso_y=&d=mobile&src=map_wap&fields=movies_all";
string strhtml = Method.GetUrl(Url1);
string title = @"data_poi_name = ""([\s\S]*?)""";
string Rxg = @"href=""tel:([\s\S]*?)""";
string Rxg1 = @"data_poi_address = ""([\s\S]*?)""";
MatchCollection titles = Regex.Matches(strhtml, title);
MatchCollection tell = Regex.Matches(strhtml, Rxg);
MatchCollection address = Regex.Matches(strhtml, Rxg1);
if (visualButton2.Text == "已停止")
{
return;
}
foreach (Match match in titles)
{
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = match.Groups[1].Value;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0];
}
foreach (Match match in tell)
{
this.dataGridView1.Rows[index].Cells[2].Value += match.Groups[1].Value.Trim() + ",";
}
foreach (Match match in address)
{
this.dataGridView1.Rows[index].Cells[3].Value = match.Groups[1].Value;
}
this.dataGridView1.Rows[index].Cells[4].Value = keyword;
this.dataGridView1.Rows[index].Cells[5].Value = city;
Application.DoEvents();
Thread.Sleep(500); //内容获取间隔，可变量
}
}
}
}
}
catch (System.Exception ex)
{
textBox3.Text = ex.ToString();
}
}
public void Run()
{
try
{
int page = 100;
string[] keywords = skinTextBox2.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
string city = label1.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
foreach (string keyword in keywords)
{
//string keywordtogb2312 = System.Web.HttpUtility.UrlEncode(keyword, System.Text.Encoding.GetEncoding("gb2312"));
for (int i = 1; i <= page; i++)
{
string Url = "https://s.hc360.com/company/search.html?kwd=" + keyword + "&k=0&z=" + city + "&pnum=" + i;
textBox14.Text = Url;
//String Url = "https://s.hc360.com/?w=" + keywordtogb2312 + "&mc=enterprise&ee="+i+"&z=" + citycode;
string strhtml = Method.GetUrl(Url); //定义的GetRul方法 返回 reader.ReadToEnd()
string Rxg = @"<h3><a data-exposurelog=""([\s\S]*?)"" href=""([\s\S]*?)""";
MatchCollection all = Regex.Matches(strhtml, Rxg);
foreach (Match NextMatch in all)
{
textBox14.Text += NextMatch.Groups[2].Value + "\r\n";
textBox14.SelectionStart = textBox14.Text.Length; //设定光标位置
textBox14.ScrollToCaret(); //滚动到光标处
}
if (textBox14.Text == "") //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox1.Text += tm1 + "-->正在采集" + label1.Text + "第" + i + "页" + keyword + "\r\n";
textBox1.SelectionStart = textBox1.Text.Length; //设定光标位置
textBox1.ScrollToCaret();
Application.DoEvents();
System.Threading.Thread.Sleep(1000);
string[] lines1 = textBox14.Text.Split(new string[] { "\r\n" }, StringSplitOptions.None);
for (int j = 0; j < lines1.Length - 1; j++)
{
int index = this.skinDataGridView1.Rows.Add(); //利用dataGridView1.Rows.Add()事件为DataGridView控件增加新的行，该函数返回添加新行的索引号，即新行的行号，然后可以通过该索引号操作该行的各个单元格，如dataGridView1.Rows[index].Cells[0].Value = "1"。这是很常用也是很简单的方法。
String Url1 = lines1[j].ToString();
string strhtml1 = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string Rxg0 = @"infoname=""([\s\S]*?)""";
string Rxg1 = @"联系人</span><span>：([\s\S]*?)</span>";
string Rxg2 = @"电话</span><span>：([\s\S]*?)</span>";
string Rxg3 = @"手机</span><span>：([\s\S]*?)</span>";
string Rxg4 = @"<p>地址：([\s\S]*?)&";
MatchCollection name = Regex.Matches(strhtml1, Rxg0);
MatchCollection contacts = Regex.Matches(strhtml1, Rxg1);
MatchCollection phone = Regex.Matches(strhtml1, Rxg2);
MatchCollection tell = Regex.Matches(strhtml1, Rxg3);
MatchCollection addr = Regex.Matches(strhtml1, Rxg4);
foreach (Match NextMatch in name)
{
string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[0].Value = temp1;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
}
foreach (Match NextMatch in contacts)
{
string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[1].Value = temp1;
}
foreach (Match NextMatch in phone)
{
string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[2].Value = temp1;
}
foreach (Match NextMatch in tell)
{
string temp = Regex.Replace(NextMatch.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[3].Value = temp1;
}
foreach (Match NextMatch in addr)
{
this.skinDataGridView1.Rows[index].Cells[4].Value = NextMatch.Groups[1].Value.Trim();
}
if (button3.Text == "已停止") //停止事件触发
{
button2.Enabled = false;
return;
}
Application.DoEvents();
System.Threading.Thread.Sleep(1000);
}
textBox14.Text = "";
}
}
}
catch (System.Exception ex)
{
textBox1.Text = ex.ToString();
}
}
public void baidu()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 10;
foreach (string city in citys)
{
int cityid = getcityId(city + "市"); //获取 citycode;
foreach (string keyword in keywords)
{
for (int i = 0; i <= pages; i++)
{
int j = i - 1 > 0 ? i - 1 :0;
String Url = "https://map.baidu.com/?newmap=1&reqflag=pcmap&biz=1&from=webmap&da_par=direct&pcevaname=pc4.1&qt=con&from=webmap&c=" + cityid + "&wd=" + keyword + "&wd2=&pn=" + i + "&nn=" + j + "0&db=0&sug=0&addr=0&&da_src=pcmappg.poi.page&on_gel=1&src=7&gr=3&l=13.2&auth=6GNgMOxNx%40CM2KzLPeYAvKP725L6c0z5uxHLVTTxHNNtBnlQADZZzy1uVt1GgvPUDZYOYIZuVt1cv3uztHee%40ewWvPWv3GuxtVwi04960vyACFIMOSU7ucEWe1GD8zv7u%40ZPuHt0A%3DH73uzCCyoET1jlBhlADM5ZYYDMJ7zlp55CKBvaaZyY&device_ratio=1&tn=B_NORMAL_MAP&u_loc=13167726,4000141&ie=utf-8";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"""primary_uid"":""([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[1].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
JsonParser jsonParser = JsonConvert.DeserializeObject <JsonParser>(html);
foreach (Content content in jsonParser.Content)
{
int index = this.dataGridView1.Rows.Add();
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = content.name;
this.dataGridView1.Rows[index].Cells[2].Value = content.tel;
this.dataGridView1.Rows[index].Cells[3].Value = content.addr;
this.dataGridView1.Rows[index].Cells[4].Value = keyword.Trim();
this.dataGridView1.Rows[index].Cells[5].Value = city;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (visualButton2.Text == "已停止")
{
return;
}
}
Application.DoEvents();
Thread.Sleep(10); //内容获取间隔，可变量
}
}
}
}
catch (System.Exception ex)
{
ex.ToString();
}
}
public void run()
{
try
{
string city = label6.Text;
string area = label10.Text;
if (city == "")
{
MessageBox.Show("请选择城市！");
return;
}
string item = label1.Text;
string item1 = label5.Text;
if (item == "")
{
MessageBox.Show("请选择分类！");
return;
}
int page = 70;
for (int i = 1; i <= page; i++)
{
String Url = "http://" + city + ".58.com/" + area + "/" + item + "/pn" + i + "/";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"<a href=""https://qy.58.com([\s\S]*?)""", RegexOptions.IgnoreCase | RegexOptions.Multiline);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add("http://qy.m.58.com/m_detail/" + NextMatch.Groups[1].Value);
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
label3.Text = tm1 + "-->正在采集第" + i + "页\r\n";
Application.DoEvents();
System.Threading.Thread.Sleep(500); //网址获取时间间隔 固定不变
foreach (string list in lists)
{
int index = this.skinDataGridView1.Rows.Add();
String Url1 = list;
string strhtml = Method.GetUrl(Url1); //定义的GetRul方法 返回 reader.ReadToEnd()
string rxg = @"<h1>([\s\S]*?)</h1>"; //公司
string Rxg = @"<a href=""tel:([\s\S]*?)"""; //电话
string Rxg1 = @"</span><span>([\s\S]*?)</span>"; //联系人
string Rxg2 = @"<dt>公司地址：</dt>([\s\S]*?)</dd>";
string Rxg3 = @"<div class=""retTit""><strong>([\s\S]*?)</strong>";
Match company = Regex.Match(strhtml, rxg);
Match tel = Regex.Match(strhtml, Rxg);
Match contacts = Regex.Match(strhtml, Rxg1);
Match addr = Regex.Match(strhtml, Rxg2);
Match job = Regex.Match(strhtml, Rxg3);
this.skinDataGridView1.Rows[index].Cells[0].Value = company.Groups[1].Value;
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
this.skinDataGridView1.Rows[index].Cells[1].Value = tel.Groups[1].Value;
this.skinDataGridView1.Rows[index].Cells[2].Value = contacts.Groups[1].Value;
string temp = Regex.Replace(addr.Groups[1].Value, "<[^>]*>", ""); //去除所有标签去除所有标签
string temp1 = Regex.Replace(temp, "\\s+", ""); //去除所有标签去除空格
this.skinDataGridView1.Rows[index].Cells[3].Value = temp1;
this.skinDataGridView1.Rows[index].Cells[4].Value = job.Groups[1].Value;
if (button3.Text == "已停止")
{
return;
}
this.skinDataGridView1.CurrentCell = this.skinDataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
Application.DoEvents();
System.Threading.Thread.Sleep(1000); //内容获取间隔，可变量
}
}
}
catch (System.Exception ex)
{
label3.Text = ex.ToString();
}
}
public void baidu()
{
try
{
string[] citys = textBox1.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
string[] keywords = textBox2.Text.Trim().Split(new string[] { "\r\n" }, StringSplitOptions.None);
int pages = 200;
foreach (string city in citys)
{
int cityid = getcityId(city + "市"); //获取 citycode;
foreach (string keyword in keywords)
{
for (int i = 0; i <= pages; i++)
{
int j = i - 1 > 0 ? i - 1 :0;
String Url = "https://map.baidu.com/?newmap=1&reqflag=pcmap&biz=1&from=webmap&da_par=direct&pcevaname=pc4.1&qt=con&from=webmap&c=" + cityid + "&wd=" + keyword + "&wd2=&pn=" + i + "&nn=" + j + "0&db=0&sug=0&addr=0&&da_src=pcmappg.poi.page&on_gel=1&src=7&gr=3&l=9.557396536572428&auth=Y%3DyW9VVIUF45OV8VGZCZb85Tzf4HRUPIuxHENxTzTzztComRB199Ay1uVt1GgvPUDZYOYIZuVt1cv3uVtGccZcuVtPWv3GuNtZComRdXmB1F234Q6W89AcEWe1GD8zv7u%40ZPuNtkB1FAprGnrFHQNWHaJ9caEmqQQDICfy9GUIsxA2wEjjg2JK&device_ratio=1&tn=B_NORMAL_MAP&u_loc=13177040,3994297&ie=utf-8&b=(12730445.232221898,4765711.534422691;13205701.20474543,4916707.941505569)&t=1547192932734";
string html = Method.GetUrl(Url);
MatchCollection TitleMatchs = Regex.Matches(html, @"""primary_uid"":""([\s\S]*?)""", RegexOptions.IgnoreCase);
ArrayList lists = new ArrayList();
foreach (Match NextMatch in TitleMatchs)
{
lists.Add(NextMatch.Groups[1].Value);
}
if (lists.Count == 0) //当前页没有网址数据跳过之后的网址采集，进行下个foreach采集
{
break;
}
string tm1 = DateTime.Now.ToString(); //获取系统时间
textBox3.Text += tm1 + "-->正在采集" + city + "" + keyword + "第" + i + "页\r\n";
JsonParser jsonParser = JsonConvert.DeserializeObject <JsonParser>(html);
foreach (Content content in jsonParser.Content)
{
int index = this.dataGridView1.Rows.Add();
this.dataGridView1.Rows[index].Cells[0].Value = index;
this.dataGridView1.Rows[index].Cells[1].Value = content.name;
this.dataGridView1.Rows[index].Cells[2].Value = content.tel;
this.dataGridView1.Rows[index].Cells[3].Value = content.addr;
this.dataGridView1.Rows[index].Cells[4].Value = keyword.Trim();
this.dataGridView1.Rows[index].Cells[5].Value = city;
this.dataGridView1.CurrentCell = this.dataGridView1.Rows[index].Cells[0]; //让datagridview滚动到当前行
if (visualButton2.Text == "已停止")
{
return;
}
}
Application.DoEvents();
Thread.Sleep(10); //内容获取间隔，可变量
}
}
}
}
catch (System.Exception ex)
{
ex.ToString();
}
}