知名网站新闻评论数评论链接抓取

最新推荐文章于 2023-04-04 22:03:57 发布

igaojie

最新推荐文章于 2023-04-04 22:03:57 发布

阅读量3.2k

点赞数

分类专栏： php 文章标签： url 2010 file path 测试 class

本文链接：https://blog.csdn.net/Alex_Best/article/details/6067406

版权

php 专栏收录该内容

54 篇文章 0 订阅

订阅专栏

/** * 测试网易新闻 */ function test_163(){ $url = "http://news.163.com/10/1118/01/6LO1PPDN00014AED.html"; $url = "http://2010.163.com/10/1121/07/6M0ESTRN00864IPP.html"; $url = 'http://lady.163.com/10/1125/23/6MCDSBEH00264IJ3.html'; $url = 'http://tech.163.com/10/1129/00/6MK82NQ4000915BE.html'; $url = 'http://ent.163.com/10/1129/08/6ML3ERFF00031H2L.html'; $url = 'http://auto.163.com/10/1126/17/6MECILD900082V1D.html'; $url = 'http://money.163.com/10/1130/09/6MNQHAMV002529IR.html'; $url = 'http://baby.163.com/10/0823/14/6EPF829700262USS.html'; $url = 'http://kids.163.com/10/1102/11/6KFU4HK4002941P8.html'; $url = 'http://gongyi.163.com/10/1203/09/6MVJDN7900933KBT.html'; $url = 'http://travel.163.com/10/1202/12/6MT966TA00063KU6.html'; $url = 'http://fashion.163.com/10/1202/18/6MTTLMU3002639VF.html'; $url = 'http://sports.163.com/10/1202/23/6MUGP50F00051C8M.html'; $url = 'http://expo.163.com/10/1031/23/6KC2PF7K00943RJS.html'; $url = 'http://daxue.163.com/10/1203/09/6MVHAV0M00913KM9.html'; $url = 'http://edu.163.com/10/1203/09/6MVH05PI00294IIO.html'; $sourcehtml = file_get_contents($url); $pattern = '|/$/("#replycounttop, #replycount001, #replycountbottom"/)/.attr/("innerHTML", (.*?)/);|ims'; if(preg_match($pattern,$sourcehtml,$match)){ $data['comment_num'] = $match[1]; } if (preg_match('|HotTieArea/(/$/("#tieArea"/), "(.*?)", "(.*?)"|ims',$sourcehtml,$out)) { $data['comment_url'] = "http://comment.news.163.com/{$out[2]}/{$out[1]}.html"; } pr($data); } /** * 新浪新闻测试 */ function test_sina(){ echo "<pre>"; $url = "http://news.sina.com.cn/s/2010-11-18/121221492927.shtml"; $url = "http://mil.news.sina.com.cn/2010-11-18/0930619674.html"; $url = 'http://finance.sina.com.cn/china/hgjj/20101129/01349021897.shtml'; $url = 'http://tech.sina.com.cn/t/3g/2010-11-29/00054917255.shtml'; $url = 'http://ent.sina.com.cn/j/2010-11-29/03063161010.shtml'; $url = 'http://auto.sina.com.cn/car/2010-11-29/0749682428.shtml'; $url = 'http://finance.sina.com.cn/roll/20101130/01549028210.shtml'; $url = 'http://news.sina.com.cn/o/2010-12-01/103121564537.shtml'; $url = 'http://edu.sina.com.cn/official/2010-12-05/1519277540.shtml'; $url = 'http://travel.sina.com.cn/world/2010-12-02/0939148379.shtml'; $url = 'http://sports.sina.com.cn/k/2010-12-06/10125353046.shtml'; $url = 'http://sh.sina.com.cn/news/s/2010-12-06/0803164846.html'; $url = 'http://gongyi.sina.com.cn/gyzx/2010-11-23/151521926.html'; $sourcehtml = file_get_contents($url); $pattern = '|channel:"(.*?)",newsid:"(.*?)"|ims'; if(preg_match($pattern,$sourcehtml,$match)){ //pr($match); $channel = $match[1]; $newsid = $match[2]; $data['comment_url'] = "http://comment4.news.sina.com.cn/comment/skin/default.html?channel={$channel}&newsid={$newsid}&style=0"; $cmt_num_url = "http://comment4.news.sina.com.cn/cgi-bin/comment/page_xml.cgi?type=P_TYPE_CMSG_JS&channel={$channel}&newsid={$newsid}&page=1"; $cmt_num_content = file_get_contents($cmt_num_url); if(preg_match('@Count.c_total.*?=.*?"(/d+)"@ims',$cmt_num_content,$out)){ $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } } print_r($data); } /** * 腾讯新闻测试 */ function test_qq(){ echo "<pre>"; $url = "http://news.qq.com/a/20101118/000311.htm"; $url = 'http://finance.qq.com/a/20101129/000043.htm'; $url = 'http://tech.qq.com/a/20101129/000021.htm'; $url = 'http://ent.qq.com/a/20101128/000078.htm'; $url = 'http://lady.qq.com/a/20101129/000033.htm'; $url = 'http://sports.qq.com/a/20101128/000897.htm'; $url = 'http://auto.qq.com/a/20101206/000026.htm'; $url = 'http://edu.qq.com/a/20101206/000007.htm'; $url = 'http://xian.qq.com/a/20101206/000005.htm'; $url = 'http://cd.qq.com/a/20101206/000027.htm'; $url = 'http://hb.qq.com/a/20101206/000686.htm'; $url = 'http://2010.qq.com/a/20101031/000205.htm'; $url = 'http://games.qq.com/a/20101206/000005.htm'; $url = 'http://worldcup.qq.com/a/20100712/002214.htm'; $sourcehtml = file_get_contents($url); $pattern = '|cmt_id = (/d+)|ims'; if(preg_match($pattern,$sourcehtml,$match)){ $cmt_id = $match[1]; } $cmt_num_url = "http://sum.comment.gtimg.com.cn/php_qqcom/gsum.php?site=news&c_id=".$cmt_id; $cmt_num_page = @file_get_contents($cmt_num_url); if( preg_match("|_cbSum/((/d+),|i",$cmt_num_page,$out)){ $data['comment_num'] = $out[1]; } $data['comment_url'] = "http://comment5.news.qq.com/comment.htm?site=news&id=".$cmt_id; print_r($data); } /** * 人民网新闻测试 */ function test_people(){ $url = "http://society.people.com.cn/GB/41158/13274129.html"; //$url = "http://ccnews.people.com.cn/GB/13281986.html"; //$url = "http://energy.people.com.cn/GB/13282125.html"; //$url = "http://media.people.com.cn/GB/40606/13257934.html"; $url = "http://finance.people.com.cn/GB/13282644.html"; $url = "http://env.people.com.cn/GB/13274621.html"; $url = "http://society.people.com.cn/GB/42733/13235258.html"; $url = "http://opinion.people.com.cn/GB/13280563.html"; $sourcehtml = file_get_contents($url); $pattern = '|var link="(.*?)";|ims'; if(preg_match($pattern,$sourcehtml,$match)){ $data['comment_url'] = $match[1]; }elseif(preg_match('|<td width="30"><a target=_blank href="(.*?)" mce_href="(.*?)" >留言</a>|ims',$sourcehtml,$match)){ $data['comment_url'] = $match[1]; } $comment_content = @file_get_contents($data['comment_url']); if (preg_match('|(/d+)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; } pr($data); } /** * 测试xinhuanet */ function test_xinhuanet() { $comment_url = "http://view.home.news.cn/comment?url=http://news.xinhuanet.com/politics/2010-11/22/c_12800354.htm"; $url = "http://news.xinhuanet.com/politics/2010-11/22/c_12800354.htm"; $url = "http://news.xinhuanet.com/politics/2010-11/22/c_12802434.htm"; $url = "http://news.xinhuanet.com/world/2010-11/22/c_12802205.htm"; $url = "http://news.xinhuanet.com/mil/2010-11/22/c_12801955.htm"; $url = "http://news.xinhuanet.com/world/2010-11/04/c_13591259.htm"; $url = "http://news.xinhuanet.com/legal/2010-11/22/c_12800225.htm"; $url = "http://news.xinhuanet.com/finance/2010-11/22/c_12802425.htm"; $url = "http://news.xinhuanet.com/lady/2010-11/22/c_12801667.htm"; $url = "http://news.xinhuanet.com/fortune/2010-11/19/c_12793392.htm"; $data['comment_url'] = "http://view.home.news.cn/comment?url=".$url; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|评论共有(/d+)条|ims',$comment_content,$out)) { $data['comment_num'] = $out[1];; } pr($data); } /** * 测试新民网 */ function test_china(){ $url = "http://news.china.com/zh_cn/focus/shglhz/11088936/20101122/16254234.html"; $url = "http://news.china.com/zh_cn/domestic/945/20101122/16254203.html"; $url = "http://news.china.com/zh_cn/history/all/11025807/20101122/16255408.html"; $url = 'http://digital.china.com/a/50054231.shtml'; $url = 'http://auto.china.com/zh_cn/gouche/wg/11026346/20101203/16275704.html'; $url = 'http://game.china.com/zh_cn/onlinegame/news/10003534/20101203/16276958.html'; $url = 'http://military.china.com/zh_cn/important/11052771/20101203/16276490.html'; $url = 'http://tech.china.com/zh_cn/news/tel/domestic/11066131/20101203/16275605.html'; $url = 'http://sports.china.com/zh_cn/football/na/other/11062529/20101203/16276317.html'; $url = 'http://culture.china.com/zh_cn/history/wenwu/11022845/20101203/16276689.html'; $url = 'http://ent.china.com/zh_cn/star/news/11052670/20101129/16266438.html'; $url = 'http://baobao.china.com/zh_cn/parent/jy/jrtt1/11077099/20101203/16276225.html'; $sourcehtml = file_get_contents($url); if (preg_match('|<a href="(.*?)" mce_href="(.*?)"|ims',$sourcehtml,$match)) { //pr($match); $data['comment_url'] = $match[1]; } if (preg_match("|includeCommentnum/('(.*?)'/);|ims",$sourcehtml,$out)) { //pr($out); $newid = $out[1]; $syspath = substr($newid,0,3); $path = substr($newid,3); //echo "newid:::{$newid}syspath::{$syspath}path:::{$path}"; $l = strlen($newid); for ($i = 8; $i > $l; $i--) { $path = "0".$path; } //http://pl.news.china.com/data/cms/162/54/23/4num.js $js = "http://pl.news.china.com/data/{$syspath}/".substr($path,0,3).'/'.substr($path,3,2).'/'.substr($path,5,2).'/'.substr($path,7).'num.js'; $js_content = file_get_contents($js); if(preg_match('|"commentnum":(/d+),|ims',$js_content,$outnum)){ $data['comment_num'] = $outnum[1];; }else{ $data['comment_num'] = 0;; } } pr($data); } /** * 测试中国新闻网 */ function test_chinanews() { $url = "http://www.chinanews.com.cn/gn/2010/11-22/2671995.shtml"; $url = "http://www.chinanews.com.cn/gj/2010/11-22/2671636.shtml"; $url = "http://www.chinanews.com.cn/ty/2010/11-21/2670470.shtml"; $url = "http://www.chinanews.com.cn/it/2010/11-22/2670544.shtml"; $sourcehtml = file_get_contents($url); if (preg_match('|<a class=noline href="(.*?)" mce_href="(.*?)">|ims',$sourcehtml,$match)) { $data['comment_url'] = $match[1]; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|(/d+)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; } } pr($data); } /** * 测试天极网 */ function test_yesky() { $url = "http://dc.yesky.com/475/11659475.shtml"; $url = "http://notebook.yesky.com/312/11670312.shtml"; $url = "http://cpu.yesky.com/466/11671466.shtml"; $url = "http://dc.yesky.com/415/11670415.shtml"; $url = 'http://news.yesky.com/477/11677477.shtml'; $sourcehtml = file_get_contents($url); $pattern = "|var site = '(.*?)';.*?type = '(.*?)';.*?articletitle = '(.*?)';.*?referId = (/d+);.*?channelId = (/d+);|ims"; if (preg_match($pattern,$sourcehtml,$match)) { list($all,$site,$type,$articletitle,$referId,$channelId) = $match; $data['comment_url'] = "http://comment.yesky.com/more.htm?site={$site}&referId={$referId}&type={$type}&channelId={$channelId}&isValidate=1&articleUrl={$url}"; $comment_content = file_get_contents($data['comment_url']); if (preg_match("| <td align='right' nowrap='nowrap'>.*?共(/d+)条|ims",$comment_content,$out)) { $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } } pr($data); } function test_eval(){ $string = 'cup'; $name = 'coffee'; $str = 'This is a $string with my $name in it.'; echo $str. "/n"; eval("/$str = /"$str/";"); echo $str. "/n"; echo " "; $arr = $this->test_yesky(); print_r($arr); } /** * sohu新闻评论数 */ function test_sohu(){ set_time_limit(0); $url = 'http://news.sohu.com/20101126/n277939004.shtml'; $url = 'http://women.sohu.com/20101129/n277887754.shtml'; $url = 'http://money.sohu.com/20101129/n277974879.shtml'; $url = 'http://stock.sohu.com/20101129/n277974701.shtml'; $url = 'http://sports.sohu.com/20101129/n277975937.shtml'; $url = 'http://business.sohu.com/20101129/n277982607.shtml'; $url = 'http://fund.sohu.com/20101129/n277975935.shtml'; $url = 'http://it.sohu.com/20101129/n277972131.shtml'; $url = 'http://auto.sohu.com/20101203/n278069399.shtml'; $url = 'http://cul.sohu.com/20101203/n278068617.shtml'; $url = 'http://men.sohu.com/20101130/n278016966.shtml'; $url = 'http://learning.sohu.com/20101130/n278013946.shtml'; $url = 'http://gongyi.sohu.com/20101202/n278061618_1.shtml'; $url = 'http://health.sohu.com/20101203/n278062656.shtml'; $url = 'http://news.sohu.com/20101202/n278056978.shtml'; $url = 'http://yule.sohu.com/20101203/n278066902.shtml'; $url = 'http://chihe.sohu.com/20101201/n278033205.shtml'; $sourcehtml = file_get_contents($url); if (preg_match('|<div class="commentLink" collection="Y"><a href="(.*?)" mce_href="(.*?)" target="_blank">|ims',$sourcehtml,$match)) { $data['comment_url'] = $match[1]; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|/((/d+)条/)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } }elseif (preg_match('|/[<a href="(.*?)" mce_href="(.*?)" target="_blank">我来说两句</a>|ims',$sourcehtml,$match)){ $data['comment_url'] = $match[1]; if (preg_match('|<script language="javascript">comment_updateCountAll/(/'(/d+)/',|ims',$sourcehtml,$out)) { $data['comment_num'] = $out[1];; }else{ $data['comment_num'] = 0; } } pr($data); } /** * 凤凰网 */ function test_ifeng(){ $url = 'http://news.ifeng.com/world/special/chaoxianpaojihanguo/content-2/detail_2010_11/26/3242644_0.shtml'; $url = 'http://tech.ifeng.com/3g/detail_2010_11/29/3260862_0.shtml'; $url = 'http://house.ifeng.com/toutiao/detail_2010_11/29/3260887_0.shtml'; $url = 'http://finance.ifeng.com/news/20101129/2966340.shtml'; $url = 'http://ent.ifeng.com/idolnews/special/chaonvwangbei/content-1/detail_2010_11/29/3260477_0.shtml'; $url = 'http://finance.ifeng.com/stock/qsch/20101201/2982947.shtml'; $url = 'http://auto.ifeng.com/news/report/20101202/479834.shtml'; $url = 'http://phtv.ifeng.com/program/jqgcs/detail_2010_12/02/3309963_0.shtml'; $url = 'http://culture.ifeng.com/whrd/detail_2010_12/03/3331577_0.shtml'; $url = 'http://media.ifeng.com/news/tradition/tv/detail_2010_12/03/3330239_0.shtml'; $url = 'http://expo2010.ifeng.com/dongtai/detail_2010_11/01/2958589_0.shtml'; $url = 'http://fashion.ifeng.com/art/fashion/detail_2010_12/03/3331196_0.shtml'; $url = 'http://health.ifeng.com/longevity/tieshi/job/detail_2010_12/03/3332816_0.shtml'; $url = 'http://edu.ifeng.com/news/detail_2010_12/03/3328672_0.shtml'; $url = 'http://yayun2010.ifeng.com/chinanews/detail_2010_11/27/3252198_0.shtml'; $url = 'http://games.ifeng.com/netgame/wow/hottopics/detail_2010_12/03/3330131_0.shtml'; $url = 'http://gongyi.ifeng.com/news/detail_2010_12/03/3330698_0.shtml'; $sourcehtml = file_get_contents($url); if (preg_match("|var commentURL.*?=.*?'(.*?)'|ims",$sourcehtml,$match)) { $data['comment_url'] = $match[1]; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|(/d+)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } } pr($data); } /** * it168 */ function test_it168(){ $url = 'http://notebook.it168.com/a2010/1119/1128/000001128255.shtml'; $url = 'http://tech.it168.com/a2010/1206/1134/000001134375.shtml'; $url = 'http://datacenter.it168.com/a2010/1203/1133/000001133938.shtml'; $url = 'http://cio.it168.com/a2010/1203/1133/000001133947.shtml'; $url = 'http://server.it168.com/a2010/1202/1133/000001133449.shtml'; $url = 'http://net.it168.com/a2010/1203/1134/000001134020.shtml'; $url = 'http://safe.it168.com/a2010/1202/1133/000001133508.shtml'; $url = 'http://vga.it168.com/a2010/1203/1133/000001133965.shtml'; $url = 'http://power.it168.com/a2010/1201/1132/000001132899.shtml'; $url = 'http://mb.it168.com/a2010/1203/1133/000001133921.shtml'; $url = 'http://virtual.it168.com/a2010/1110/1124/000001124296.shtml'; $url = 'http://lcd.it168.com/a2010/1203/1133/000001133746.shtml'; $url = 'http://software.it168.com/a2010/1205/1134/000001134142.shtml'; $url = 'http://oa.it168.com/a2010/1203/1133/000001133787.shtml'; $url = 'http://security.it168.com/a2010/1206/1134/000001134259.shtml'; $url = 'http://storage.it168.com/a2010/1206/1134/000001134417.shtml'; $url = 'http://digital.it168.com/a2010/1202/1133/000001133389.shtml'; $url = 'http://cloud.it168.com/a2010/1206/1134/000001134289.shtml'; $url = 'http://wireless.it168.com/a2010/1014/1113/000001113476.shtml'; $url = 'http://sound.it168.com/a2010/1202/1133/000001133109.shtml'; $sourcehtml = file_get_contents($url); if (preg_match('|评论：<a href="(.*?)" mce_href="(.*?)" target="_blank">|ims',$sourcehtml,$match)) { $data['comment_url'] = $match[1]; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|(/d+)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } } pr($data); } /** * 中关村在线 */ function test_zol(){ $url = 'http://mobile.zol.com.cn/203/2032041.html'; $url = 'http://news.zol.com.cn/196/1964372.html'; $url = 'http://nb.zol.com.cn/205/2054458.html'; $url = 'http://game.zol.com.cn/206/2064176.html'; $url = 'http://dcdv.zol.com.cn/205/2053179.html'; $url = 'http://mp3.zol.com.cn/205/2059590.html'; $url = 'http://power.zol.com.cn/205/2053230.html'; $sourcehtml = file_get_contents($url); if (preg_match('|/[<a href="(.*?)" mce_href="(.*?)" class="huei12i">|ims',$sourcehtml,$match)) { $data['comment_url'] = $match[1]; $comment_content = file_get_contents($data['comment_url']); if (preg_match('|全部评论/((/d+)条/)|ims',$comment_content,$out)) { $data['comment_num'] = $out[1]; }else{ $data['comment_num'] = 0; } } pr($data); }