related.php 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277
  1. <?php
  2. set_time_limit(0);
  3. /**
  4. * Created by PhpStorm.
  5. * User: chendeben
  6. * Date: 2019-01-16
  7. * Time: 17:06
  8. */
  9. require_once('vendor/autoload.php');
  10. use PhpOffice\PhpSpreadsheet\Spreadsheet;
  11. use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet;
  12. use PhpOffice\PhpSpreadsheet\Writer\Xlsx;
  13. use QL\QueryList;
  14. //$ql = QueryList::get('https://www.baidu.com/s?wd=QueryList');
  15. //$titles = $ql->find('h3>a')->texts(); //获取搜索结果标题列表
  16. //$links = $ql->find('h3>a')->attrs('href'); //获取搜索结果链接列表
  17. //print_r($titles);
  18. //print_r($links);
  19. //die();
  20. $wd = $_REQUEST['wd'];
  21. if (!$wd) {
  22. ?>
  23. <!DOCTYPE html>
  24. <html lang="zh-cn">
  25. <head>
  26. <meta charset="utf-8">
  27. <meta http-equiv="X-UA-Compatible" content="IE=edge">
  28. <meta name="viewport" content="width=device-width, initial-scale=1">
  29. <!-- ZUI 标准版压缩后的 CSS 文件 -->
  30. <link rel="stylesheet" href="//cdn.bootcss.com/zui/1.8.1/css/zui.min.css">
  31. <!-- ZUI Javascript 依赖 jQuery -->
  32. <script src="//cdn.bootcss.com/zui/1.8.1/lib/jquery/jquery.js"></script>
  33. <!-- ZUI 标准版压缩后的 JavaScript 文件 -->
  34. <script src="//cdn.bootcss.com/zui/1.8.1/js/zui.min.js"></script>
  35. </head>
  36. <body>
  37. <br>
  38. <br>
  39. <br>
  40. <br>
  41. <form method="post" action="">
  42. <div class="container">
  43. <div class="row">
  44. <div class="col-md-9">
  45. <textarea style="height: 300px;" class="form-control" rows="3" placeholder="请输入关键词,一行一个" name="wd"></textarea>
  46. </div>
  47. </div>
  48. <div class="row" style="margin-top: 5px;">
  49. <div class="col-md-2 col-md-offset-8">
  50. <input type="submit" class="btn btn-primary btn-lg" value="提交"/>
  51. </div>
  52. </div>
  53. </div>
  54. </form>
  55. </body>
  56. </html>
  57. <?php
  58. } else {
  59. $origin_str = [';', ';', ',', ',', ' ',"\r\n"];
  60. $des_str = [',', ',', ',', ',', ',',','];
  61. $wd = str_replace($origin_str, $des_str, $wd);
  62. $keys = explode(",", $wd);
  63. $baidu = baidu($keys);
  64. $shenma=shenma($keys);
  65. $sougou=sougou($keys);
  66. $so_360=so_360($keys);
  67. $spreadsheet = new Spreadsheet();
  68. $sheet = $spreadsheet->getActiveSheet();
  69. $sheet->setTitle("百度");
  70. $row = 1;
  71. foreach ($baidu as $key => $item) {
  72. $sheet->setCellValueByColumnAndRow(1, $row++, $item['keyword']);
  73. foreach ($item['related_keywords'] as $related_keyword) {
  74. $sheet->setCellValueByColumnAndRow(2, $row++, $related_keyword);
  75. }
  76. $row++;
  77. }
  78. $shenma_sheet=new Worksheet();
  79. $row = 1;
  80. foreach ($shenma as $key => $item) {
  81. $shenma_sheet->setCellValueByColumnAndRow(1, $row++, $item['keyword']);
  82. foreach ($item['related_keywords'] as $related_keyword) {
  83. $shenma_sheet->setCellValueByColumnAndRow(2, $row++, $related_keyword);
  84. }
  85. $row++;
  86. }
  87. $shenma_sheet->setTitle("神马");
  88. $spreadsheet->addSheet($shenma_sheet);
  89. $sougou_sheet=new Worksheet();
  90. $row = 1;
  91. foreach ($sougou as $key => $item) {
  92. $sougou_sheet->setCellValueByColumnAndRow(1, $row++, $item['keyword']);
  93. foreach ($item['related_keywords'] as $related_keyword) {
  94. $sougou_sheet->setCellValueByColumnAndRow(2, $row++, $related_keyword);
  95. }
  96. $row++;
  97. }
  98. $sougou_sheet->setTitle("搜狗");
  99. $spreadsheet->addSheet($sougou_sheet);
  100. $so_360_sheet=new Worksheet();
  101. $row = 1;
  102. foreach ($so_360 as $key => $item) {
  103. $so_360_sheet->setCellValueByColumnAndRow(1, $row++, $item['keyword']);
  104. foreach ($item['related_keywords'] as $related_keyword) {
  105. $so_360_sheet->setCellValueByColumnAndRow(2, $row++, $related_keyword);
  106. }
  107. $row++;
  108. }
  109. $so_360_sheet->setTitle("360");
  110. $spreadsheet->addSheet($so_360_sheet);
  111. $writer = new Xlsx($spreadsheet);
  112. header("Pragma: public");
  113. header("Expires: 0");
  114. header("Cache-Control:must-revalidate, post-check=0, pre-check=0");
  115. header("Content-Type:application/force-download");
  116. header("Content-Type:application/vnd.ms-execl");
  117. header("Content-Type:application/octet-stream");
  118. header("Content-Type:application/download");
  119. header('Content-Disposition:attachment;filename="相关搜索词.xlsx"');
  120. header("Content-Transfer-Encoding:binary");
  121. $writer->save('php://output');
  122. $spreadsheet->disconnectWorksheets();
  123. }
  124. function baidu($wd){
  125. $_url = "https://www.baidu.com/s?ie=UTF-8&wd=%s";
  126. $result=[];
  127. foreach ($wd as $key) {
  128. $url = sprintf($_url, $key);
  129. // $url="https://www.baidu.com/s?ie=UTF-8&wd=".$key;
  130. $curl = curl_init();
  131. curl_setopt_array($curl, array(
  132. CURLOPT_URL => $url,
  133. CURLOPT_RETURNTRANSFER => true,
  134. CURLOPT_ENCODING => "",
  135. CURLOPT_MAXREDIRS => 10,
  136. CURLOPT_TIMEOUT => 0,
  137. CURLOPT_FOLLOWLOCATION => true,
  138. CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  139. CURLOPT_CUSTOMREQUEST => "GET",
  140. CURLOPT_HTTPHEADER => array(
  141. "Connection: keep-alive",
  142. "Pragma: no-cache",
  143. "Cache-Control: no-cache",
  144. "DNT: 1",
  145. "Upgrade-Insecure-Requests: 1",
  146. "User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36",
  147. "Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9",
  148. "Sec-Fetch-Site: same-origin",
  149. "Sec-Fetch-Mode: navigate",
  150. "Sec-Fetch-User: ?1",
  151. "Sec-Fetch-Dest: document",
  152. "Referer: https://www.baidu.com/",
  153. "Accept-Language: zh-CN,zh;q=0.9,en;q=0.8",
  154. "Cookie: PSTM=1593415318; BIDUPSID=217E2F2D2D25A987334C6DA48D602863; BD_UPN=123253; BAIDUID=C223ACE43BE868524F32AAC535BEF108:FG=1; MCITY=-%3A; BDUSS=jhiS0tFcHlLZkw2bWdUV1Q3ZkZ-dzZVWVM5bGRPMkVleFp3S2kwWXlFU2ZnTkZmSUFBQUFBJCQAAAAAAAAAAAEAAABl-iwFx-np5NOwu-oAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAJ~zqV-f86lfR; BDUSS_BFESS=jhiS0tFcHlLZkw2bWdUV1Q3ZkZ-dzZVWVM5bGRPMkVleFp3S2kwWXlFU2ZnTkZmSUFBQUFBJCQAAAAAAAAAAAEAAABl-iwFx-np5NOwu-oAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAJ~zqV-f86lfR; ispeed_lsm=0; BDORZ=B490B5EBF6F3CD402E515D22BCDA1598; delPer=0; BD_CK_SAM=1; PSINO=7; ZD_ENTRY=baidu; BD_HOME=1; H_PS_PSSID=1433_33061_33098_33100_33198_33147; sugstore=1; H_PS_645EC=d2a6avOWAlxpBjqV3N7Iber%2F0MSibSPb1HevBV%2BstxJRDWyT9TLKdMWt8wo; BA_HECTOR=8184a0alaga1ah80in1fscafr0r; COOKIE_SESSION=2610_0_8_0_12_35_1_3_7_8_5_1_2482_0_21_0_1606737947_0_1606737926%7C9%230_1_1594003058%7C1"
  155. ),
  156. ));
  157. curl_setopt($curl, CURLOPT_SSL_VERIFYHOST, FALSE);
  158. curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
  159. $html = curl_exec($curl);
  160. $error=curl_error($curl);
  161. if ($error){
  162. print_r($error);
  163. }
  164. curl_close($curl);
  165. $ql=new QueryList();
  166. $ql->setHtml($html);
  167. // $ql = (new QL\QueryList)->get($url);
  168. $table = $ql->find("#rs > table");
  169. $tableRows = $table->find('tr')->map(function ($row) {
  170. return $row->find('th')->texts()->all();
  171. });
  172. $data = $tableRows->all();
  173. $_result = [];
  174. foreach ($data as $item) {
  175. foreach ($item as $value) {
  176. $_result[] = $value;
  177. }
  178. }
  179. $result[] = [
  180. 'keyword' => $key,
  181. 'related_keywords' => $_result
  182. ];
  183. }
  184. return $result;
  185. }
  186. function shenma($keys){
  187. $url="https://m.sm.cn/s?q=%s&from=smor&safe=1&snum=20";
  188. $result=[];
  189. foreach ($keys as $key) {
  190. $url = sprintf($url, $key);
  191. $_result = [];
  192. $ql = (new QL\QueryList)->get($url);
  193. $_data =$ql->find(".js-c-e-button-text")->texts()->all();
  194. $_result=array_merge($_result,$_data);
  195. $result[] = [
  196. 'keyword' => $key,
  197. 'related_keywords' => $_result
  198. ];
  199. }
  200. return $result;
  201. }
  202. function sougou($keys){
  203. $url="https://www.sogou.com/web?query=%s";
  204. $result=[];
  205. foreach ($keys as $key) {
  206. $_result = [];
  207. $url = sprintf($url, $key);
  208. $ql = (new QL\QueryList)->get($url);
  209. $_data =$ql->find("#float_uphint a")->texts()->all();
  210. $_result=array_merge($_result,$_data);
  211. $_data =$ql->find(".hint-mid a")->texts()->all();
  212. $_result=array_merge($_result,$_data);
  213. $table = $ql->find("#hint_container");
  214. $tableRows =$table->find("tr")->map(function ($row){
  215. return $row->find('a')->texts()->all();
  216. });
  217. $data = $tableRows->all();
  218. foreach ($data as $item) {
  219. $_result=array_merge($_result,$item);
  220. }
  221. $result[] = [
  222. 'keyword' => $key,
  223. 'related_keywords' => $_result
  224. ];
  225. }
  226. return $result;
  227. }
  228. function so_360($keys){
  229. $url="https://www.so.com/s?ie=utf-8&fr=none&src=360sou_newhome&q=%s";
  230. $result=[];
  231. foreach ($keys as $key) {
  232. $url = sprintf($url, $key);
  233. $ql = QueryList::get($url);
  234. $table = $ql->find("#rs > table");
  235. $tableRows = $table->find('tr')->map(function ($row) {
  236. return $row->find('a')->texts()->all();
  237. });
  238. $data = $tableRows->all();
  239. // var_dump($data);
  240. // die();
  241. $_result = [];
  242. foreach ($data as $item) {
  243. foreach ($item as $value) {
  244. $_result[] = $value;
  245. }
  246. }
  247. $result[] = [
  248. 'keyword' => $key,
  249. 'related_keywords' => $_result
  250. ];
  251. }
  252. return $result;
  253. }