public void sougou()
{
string cookie = "CXID=D44C63D34623066DA36D11B4B82C488C; SUV=1518337255419884; SMAPUVID=1518337255419884; SUV=1801190926013760; IPLOC=CN3213; sct=1; SNUID=D67B2A4373761425DCE0226F733E2FA7; ad=zMGYjlllll2zYIpclllllVr1fv6lllllGq6poyllll9llllljZlll5@@@@@@@@@@; SUID=A10659313565860A5A6291B800040AA1; wP_w=544ebe2c0329~HXcgwyvcH_c_5BDcHXULrZvNyX9XwbmSJXPsNNDN3XNBbbDJbdNhb; activecity=%u5BBF%u8FC1%2C13168867%2C3999623%2C12; ho_co=";
string url = "http://map.sogou.com/EngineV6/search/json";
string charset = "gb2312";
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)
{
//搜索的城市和关键词都需要两次url编码
string city1 = System.Web.HttpUtility.UrlEncode(city, System.Text.Encoding.GetEncoding("utf-8"));
string cityutf8 = System.Web.HttpUtility.UrlEncode(city1, System.Text.Encoding.GetEncoding("utf-8"));
foreach (string keyword in keywords)
{
string keyword1 = System.Web.HttpUtility.UrlEncode(keyword, System.Text.Encoding.GetEncoding("utf-8"));
string keywordutf8 = System.Web.HttpUtility.UrlEncode(keyword1, System.Text.Encoding.GetEncoding("utf-8"));
for (int i = 1; i <= pages; i++)
{
string postData = "what=keyword%3A" + keywordutf8 + "&range=bound%3A00000000.5%2C0000000.5%2C99999999.5%2C9999999.5%3A0&othercityflag=1&appid=1361&thiscity=" + cityutf8 + "&lastcity=" + cityutf8 + "&userdata=3&encrypt=1&pageinfo=" + i + "%2C10&locationsort=0&version=7.0&ad=0&level=12&exact=1&type=&attr=&order=&submittime=0&resultTypes=poi&sort=0&reqid=1526008949358471&cb=parent.IFMS.search";
string html = Method.PostUrl(url, postData, cookie, charset);
MatchCollection TitleMatchs = Regex.Matches(html, @"""dataid"":""([\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.sogou.com/poi/1_" + poid + ".htm";
string strhtml = GetUrl(Url1);
string title = @"""caption"":""([\s\S]*?)""";
string Rxg = @"""phone"":""([\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();
}
}