function getinfodetail(&$data) { global $model, $_root, $cid, $strreplace, $pregreplace; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); // file_put_contents('error_view.html',$html); //$html = iconv("GBK","UTF-8//TRANSLIT",$html) ; $html = mb_convert_encoding($html, "UTF-8", "GBK"); if (!$html) { echo "获取html失败"; exit; } $data['keyword'] = ''; $data['actor'] = @iconv("UTF-8", "UTF-8//TRANSLIT", $data['actor']); // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#<div class="introduction" itemprop="description"><p>(.+)</p></div>#Uis', $html, $match); $match[1] = isset($match[1]) ? $match[1] : ''; $match[1] = @iconv("UTF-8", "UTF-8//TRANSLIT", $match[1]); //echo $match[1],"\n"; $data['intro'] = strip_tags($match[1]); $data['intro'] = preg_replace('#&\\S+;#Uis', '', $data['intro']); $data['intro'] = mb_strlen($data['intro']) > 300 ? mb_substr($data['intro'], 0, 256, 'UTF-8') : $data['intro']; $data['intro'] = trim($data['intro']); $playhtml = getArticlePlayData($data['purl']); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; return 0; } $data['vols'] = $playhtml; unset($data['purl']); if (!$data['name'] || empty($data['vols'])) { echo "抓取失败 {$data['ourl']} \r\n"; exit; return false; } $data['ourl'] = str_replace($_root, '', $data['ourl']); # echo '<pre>';var_dump($data);exit; /**/ //在判断是否更新 $oname = $data['name']; $aid = checkArticleByOname($oname); if ($aid) { $vdata = array('name' => $data['name'], 'vols' => $data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } /**/ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }
function getinfodetail(&$data) { global $model, $_root, $cid, $strreplace, $pregreplace; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); // file_put_contents('error_view.html',$html); //$html = iconv("GBK","UTF-8//TRANSLIT",$html) ; $html = mb_convert_encoding($html, "UTF-8", "GBK"); if (!$html) { echo "获取html失败"; exit; } //kw $data['keyword'] = ''; // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#<DIV class=n_bd>(.+)</div>#Uis', $html, $match); $match[1] = @$match[1]; $match[1] = @iconv("UTF-8", "UTF-8//TRANSLIT", $match[1]); //echo $match[1],"\n"; $r = mstrip_tags($match[1]); $data['intro'] = $r['intro']; // $data['thum'] = $r['thum']; $playhtml = getArticlePlayData($data['purl']); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; return 0; } $data['vols'] = jsary2phpary($playhtml); unset($data['purl']); if (!$data['name'] || empty($data['vols'])) { echo "抓取失败 {$data['ourl']} \r\n"; return 1; exit; } $data['ourl'] = str_replace($_root, '', $data['ourl']); // echo '<pre>';var_dump($data);exit; //在判断是否更新 /* $oname = $data['name']; $aid = checkArticleByOname($oname); if($aid){ $vdata = array('name'=>$data['name'],'vols'=>$data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } */ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }
function getinfodetail(&$data) { global $model, $start_page, $_root, $cid, $strreplace, $pregreplace; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); // $html = mb_convert_encoding($html,"UTF-8","GBK"); if (!$html) { echo "获取html失败"; exit; } //kw $data['keyword'] = ''; // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#<span class="more" txt="([^"]+)"\\s*>#Uis', $html, $match); $match[1] = isset($match[1]) ? $match[1] : ''; //echo $match[1],"\n";exit; $data['intro'] = strip_tags($match[1]); $data['intro'] = preg_replace('#&\\S+;#Uis', '', $data['intro']); $data['intro'] = mb_strlen($data['intro']) > 300 ? mb_substr($data['intro'], 0, 300, 'utf-8') : $data['intro']; $data['intro'] = str_replace('?', '', $data['intro']); $data['intro'] = trim($data['intro']); $data['intro'] = preg_replace("#(\r\n)+#is", "\r\n", $data['intro']); $data['intro'] = preg_replace("#\n+#is", "\n", $data['intro']); $playhtml = getArticlePlayData($data['purl']); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; return 0; } $data['vols'] = getParseVideoInfo($playhtml); unset($data['purl']); if (!$data['name'] || empty($data['vols'])) { echo "抓取失败 {$data['ourl']} \r\n"; return false; } $data['ourl'] = str_replace($_root, '', $data['ourl']); // echo '<pre>';var_dump($data);exit; /**/ //在判断是否更新 $oname = $data['name']; $aid = checkArticleByOname($oname); if ($aid) { $vdata = array('name' => $data['name'], 'vols' => $data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } /**/ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }
function getinfodetail(&$data) { global $model, $_root, $cid, $strreplace, $pregreplace, $movieCid; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); // file_put_contents('error_view.html',$html); if (!$html) { echo "获取html失败"; exit; } /**/ $data['keyword'] = ''; // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#<div class="detail-desc-cnt">(.+)</div>#Uis', $html, $match); $match[1] = isset($match[1]) ? $match[1] : ''; $match[1] = @iconv("UTF-8", "UTF-8//TRANSLIT", $match[1]); //echo $match[1],"\n"; $data['intro'] = strip_tags($match[1]); $data['intro'] = preg_replace('#&\\S+;#Uis', '', $data['intro']); $data['intro'] = trim($data['intro']); if (strlen($data['intro']) > 1024) { $data['intro'] = mb_substr($data['intro'], 0, 510, 'UTF-8') . '...'; } preg_match('#<p \\s*class="play-list"><a \\s*target="_blank" \\s*href="(/videos/\\d+vod-play-id-\\d+-sid-\\d+-pid-\\d+\\.html)">.+</a>#Uis', $html, $match); $purl = isset($match[1]) ? $match[1] : ''; //$purl = '/videos/57122vod-play-id-57122-sid-0-pid-15.html'; if (!$purl) { file_put_contents('play_url_html.html', $html); die("\n Ourl: {$data['ourl']} Get PlayUrl Error!\n"); } $data['purl'] = $_root . $purl; $playhtml = getArticlePlayData($data['purl']); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; //return 0; } $data['vols'] = $playhtml; unset($data['purl']); if (!$data['name']) { echo "抓取失败 {$data['ourl']} \r\n"; return false; } if (empty($data['vols'])) { $data['status'] = 5; echo "\nOurl: {$data['ourl']} Get PlayVols Empty!\n"; } $data['ourl'] = str_replace($_root, '', $data['ourl']); echo '<pre>'; var_dump($data); exit; /** //在判断是否更新 $oname = $data['name']; $aid = checkArticleByOname($oname); if($aid && !empty($data['vols'])){ $vdata = array('name'=>$data['name'],'vols'=>$data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } if($aid && empty($data['vols'])){ echo "{$aid}已存在更新!\r\n"; return 6; } /**/ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }
function getinfodetail(&$data) { global $model, $_root, $cid, $strreplace, $pregreplace; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); $html = mb_convert_encoding($html, "UTF-8", "GBK"); if (!$html) { echo "获取html失败"; exit; } //kw $data['keyword'] = ''; // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#剧情介绍:</h2>\\s*</div>\\s*<div style="[^"]+">(.+)<p>#Uis', $html, $match); $match[1] = isset($match[1]) ? $match[1] : ''; //echo $match[1],"\n"; $data['intro'] = strip_tags($match[1]); $data['intro'] = preg_replace('#&\\S+;#Uis', '', $data['intro']); $data['intro'] = mb_strlen($data['intro']) > 300 ? mb_substr($data['intro'], 0, 300, 'utf-8') : $data['intro']; $data['intro'] = str_replace('?', '', $data['intro']); $data['intro'] = trim($data['intro']); $data['intro'] = preg_replace("#(\r\n)+#is", "\r\n", $data['intro']); $data['intro'] = preg_replace("#\n+#is", "\n", $data['intro']); $data['intro'] = preg_replace('#\\s\\s+#is', ' ', $data['intro']); $data['intro'] = @iconv("UTF-8", "UTF-8//TRANSLIT", $data['intro']); $data['intro'] = str_replace('?', '', $data['intro']); preg_match('#<li><a title=\'[^\']+\' href=\'(/.+/player-0-0\\.html)\' target="_blank">.+</a></li>#Uis', $html, $match); $data['purl'] = @$match[1]; $playhtml = getArticlePlayData($data['purl']); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; return 0; } $data['vols'] = getParseVideoInfo($playhtml); unset($data['purl']); if (!$data['name'] || empty($data['vols'])) { echo "抓取失败 {$data['ourl']} \r\n"; return false; } $data['ourl'] = str_replace($_root, '', $data['ourl']); # echo '<pre>';var_dump($data);exit; /**/ //在判断是否更新 $oname = $data['name']; $aid = checkArticleByOname($oname); if ($aid) { $vdata = array('name' => $data['name'], 'vols' => $data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } /**/ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }
function getinfodetail(&$data) { global $model, $_root, $cid, $strreplace, $pregreplace; echo $data['ourl'], "\n"; $html = getHtml($data['ourl']); // file_put_contents('error_view.html',$html); //$html = iconv("GBK","UTF-8//TRANSLIT",$html) ; $html = mb_convert_encoding($html, "UTF-8", "GBK"); if (!$html) { echo "获取html失败"; exit; } $data['keyword'] = ''; $data['actor'] = @iconv("UTF-8", "UTF-8//TRANSLIT", $data['actor']); // $data['ptime'] = time(); $data['utime'] = time(); preg_match('#<div class="description b mb">\\s*<h2>.+剧情介绍:</h2>(.+)</div>#Uis', $html, $match); $match[1] = isset($match[1]) ? $match[1] : ''; $match[1] = @iconv("UTF-8", "UTF-8//TRANSLIT", $match[1]); #echo $match[1],"\n";exit; $data['intro'] = strip_tags($match[1]); $data['intro'] = preg_replace('#&\\S+;#Uis', '', $data['intro']); $data['intro'] = preg_replace('#《[^》]+》全集在线观看由琪琪影院www.77vcd.com提供如果您觉得本站不错 请推荐给您的好友#Uis', '', $data['intro']); $data['intro'] = str_replace('琪琪影院', 'web_title', $data['intro']); $data['intro'] = str_replace('www.77vcd.com', 'web_domain', $data['intro']); $data['intro'] = mb_strlen($data['intro']) > 300 ? mb_substr($data['intro'], 0, 256, 'UTF-8') : $data['intro']; $data['intro'] = trim($data['intro']); #var_dump($data['intro']);exit; preg_match('#<ul><li><a title=\'[^\']*\' href=\'(/[^\']+/player-\\d+-\\d+\\.html)\' target="_blank">.*</a></li></ul>#Uis', $html, $match); $purl = isset($match[1]) ? $match[1] : ''; #var_dump($match);exit; $playhtml = getArticlePlayData($purl); if (empty($playhtml)) { echo "\n++ Ourl:{$data['ourl']} Purl:{$data['purl']} playdata vols decode error!++\n"; return 0; } $data['vols'] = $playhtml; unset($data['purl']); if (!$data['name'] || empty($data['vols'])) { echo "抓取失败 {$data['ourl']} \r\n"; exit; return false; } $data['ourl'] = str_replace($_root, '', $data['ourl']); echo '<pre>'; var_dump($data); exit; /* //在判断是否更新 $oname = $data['name']; $aid = checkArticleByOname($oname); if($aid){ $vdata = array('name'=>$data['name'],'vols'=>$data['vols']); $aid = addArticleVols($vdata); echo "{$aid}已存在更新!\r\n"; return 6; } /**/ $aid = addArticle($data); //echo '|',$aid,'|';exit; if (!$aid) { var_dump($data); echo "\r\n添加失败! {$data['ourl']} \r\n"; exit; return false; } echo "添加成功! {$aid} \r\n"; }