private void visualButton4_Click(object sender, EventArgs e)
{
Method.Txt(dataGridView1);
}
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();
}
}
private void visualButton3_Click(object sender, EventArgs e)
{
Method.DataTableToExcel(Method.DgvToTable(dataGridView1), "Sheet1", true);
}
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 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 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 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 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();
}
}
