function parseVodPad($rs) { while ($row = @mysql_fetch_array($rs)) { $m_urltest = $row["m_urltest"]; $m_pid = $row["m_pid"]; $m_typeid = $row["m_typeid"]; $m_playfrom = $row["m_playfrom"]; $m_id = $row["m_id"]; $id = ''; if (!isN($m_urltest) && preg_match('/movie\\/(\\w.+?).htm/', $m_urltest)) { preg_match_all('/movie\\/(\\w.+?).htm/', $m_urltest, $match); $id = $match[1][0]; } else { if (!isN($m_urltest) && preg_match('/tv\\/(\\w.+?).htm/', $m_urltest)) { preg_match_all('/tv\\/(\\w.+?).htm/', $m_urltest, $match); $id = $match[1][0]; } else { if (!isN($m_urltest) && preg_match('/show\\/(\\w.+?).htm/', $m_urltest)) { preg_match_all('/show\\/(\\w.+?).htm/', $m_urltest, $match); $id = $match[1][0]; } else { if (!isN($m_urltest) && preg_match('/comic\\/(\\w.+?).htm/', $m_urltest)) { preg_match_all('/comic\\/(\\w.+?).htm/', $m_urltest, $match); $id = $match[1][0]; } } } } cjBaiduView($m_id, $id, $m_pid, $m_typeid, $m_playfrom); unset($rs); } }
function cjList() { global $db, $p_collect_type, $strListUrl, $p_coding, $count; global $p_playtype, $p_id, $idarr, $starringarr, $linkarr, $titlearr, $picarr, $areaarr, $yeararr, $typearr, $finisharr, $introarr, $directorarr, $durationarr; $strListCode = getPage($strListUrl, $p_coding); writetofile("crawel_auto_info.log", $p_id . '{=====}' . $strListUrl . "{=====}List===start"); if ($p_playtype === 'baidu') { if (isN($_SESSION["strListCodeCut"])) { $_SESSION["strListCodeCut"] = $strListCode; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } $baiduList = NBaidu::parseListByContent($strListCode, $p_coding, ''); $idarr = $baiduList['idarr']; $linkarr = $baiduList['linkarr']; $starringarr = $baiduList['starringarr']; $titlearr = $baiduList['titlearr']; $picarr = $baiduList['picarr']; $areaarr = $baiduList['areaarr']; $yeararr = $baiduList['yeararr']; $typearr = $baiduList['typearr']; $finisharr = $baiduList['finisharr']; $introarr = $baiduList['introarr']; $directorarr = $baiduList['directorarr']; $durationarr = $baiduList['durationarr']; } for ($i = 0; $i < count($idarr); $i++) { $idMo = $idarr[$i]; if ($p_playtype === 'baidu') { $count++; cjBaiduView($idMo, $i); } else { cjView($idMo, $i); } } clearSession(); }
function cjList() { global $db, $flag, $p_collect_type, $listnum, $strListUrl, $p_pagetype, $p_collecorder, $p_listcodestart, $p_listcodeend, $p_listlinkstart, $p_listlinkend, $p_starringstart, $p_starringend, $p_titlestart, $p_titleend, $p_picstart, $p_picend, $p_starringtype, $p_titletype, $p_pictype, $p_coding, $p_showtype, $viewnum, $p_ids, $sb, $cg, $p_savefiles, $p_pagebatchid2, $p_pagebatchid1; global $p_playtype, $reCollExistMovie, $p_playspecialtype, $starringarr, $titlearr, $picarr, $strdstate, $action, $p_pagebatchurl, $p_colleclinkorder, $p_id; // var_dump("ddd".$strListUrl); $strListCode = getPage($strListUrl, $p_coding); writetofile("crawel_auto_info.log", $p_id . '{=====}' . $strListUrl . "{=====}List===start"); $listnum = $listnum + 1; $tempStep = 1; if ($p_playtype === 'baidu') { if (isN($_SESSION["strListCodeCut"])) { $_SESSION["strListCodeCut"] = $strListCode; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } // var_dump("sssss".$strListCode); $baiduList = BaiduParse::parseMovieListByContent($strListCode, $p_code, ''); // var_dump("sssss".$baiduList); // var_dum if (isN($_SESSION["linkarrcode"])) { $linkarrcode = $baiduList['linkarr']; $linkarrcode = implode('{Array}', $linkarrcode); } else { $linkarrcode = $_SESSION["linkarrcode"]; } if ($linkarrcode == false) { writetofile("crawel_auto_error.log", $p_id . '{=====}' . $strListUrl); return; } $linkarr = explode("{Array}", $linkarrcode); if ($p_starringtype == 1) { $starringarr = $baiduList['starringarr']; } if ($p_titletype == 1) { $titlearr = $baiduList['titlearr']; } if ($p_pictype == 1) { $picarr = $baiduList['picarr']; } } else { if (isN($_SESSION["strListCodeCut"])) { $strListCodeCut = getBody($strListCode, $p_listcodestart, $p_listcodeend); $_SESSION["strListCodeCut"] = $strListCodeCut; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } if (isN($_SESSION["linkarrcode"])) { $linkarrcode = getArray($strListCodeCut, $p_listlinkstart, $p_listlinkend); $_SESSION["linkarrcode"] = $linkarrcode; } else { $linkarrcode = $_SESSION["linkarrcode"]; } if ($p_starringtype == 1) { $starringarrcode = getArray($strListCodeCut, $p_starringstart, $p_starringend); } if ($p_titletype == 1) { $titlearrcode = getArray($strListCodeCut, $p_titlestart, $p_titleend); } if ($p_pictype == 1) { $picarrcode = getArray($strListCodeCut, $p_picstart, $p_picend); } if ($linkarrcode == false) { writetofile("crawel_auto_error.log", $p_id . '{=====}' . $strListUrl); return; } $linkarr = explode("{Array}", $linkarrcode); if ($p_starringtype == 1) { $starringarr = explode("{Array}", $starringarrcode); } if ($p_titletype == 1) { $titlearr = explode("{Array}", $titlearrcode); } if ($p_pictype == 1) { $picarr = explode("{Array}", $picarrcode); } } $viewcount = count($linkarr); if ($p_colleclinkorder == 1) { for ($i = $viewcount; $i >= 0; $i--) { $urlMo = getHrefFromLink($linkarr[$i]); if ($reCollExistMovie) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } else { $sql = "select m_id from {pre}cj_vod where m_urltest='" . $urlMo . "' order by m_id desc"; $rowvod = $db->getRow($sql); if (!$rowvod) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } } } } else { for ($i = 0; $i < count($linkarr); $i++) { // cjView(getHrefFromLink($linkarr[$i]),$i); $urlMo = getHrefFromLink($linkarr[$i]); if ($reCollExistMovie) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } else { $sql = "select m_id from {pre}cj_vod where m_urltest='" . $urlMo . "' order by m_id desc"; $rowvod = $db->getRow($sql); if (!$rowvod) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } } } } clearSession(); }
function cjList() { global $db, $reCollExistMovie, $p_collect_type, $flag, $listnum, $strListUrl, $p_pagetype, $p_collecorder, $p_listcodestart, $p_listcodeend, $p_listlinkstart, $p_listlinkend, $p_starringstart, $p_starringend, $p_titlestart, $p_titleend, $p_picstart, $p_picend, $p_starringtype, $p_titletype, $p_pictype, $p_coding, $p_showtype, $viewnum, $p_ids, $sb, $cg, $p_savefiles, $p_pagebatchid2, $p_pagebatchid1; global $p_playtype, $p_playspecialtype, $starringarr, $titlearr, $picarr, $strdstate, $action, $p_pagebatchurl, $p_colleclinkorder, $p_id; if (isN($_SESSION["strListCode"])) { $strListCode = getPage($strListUrl, $p_coding); $_SESSION["strListCode"] = $strListCode; } else { $strListCode = $_SESSION["strListCode"]; } if ($strListCode == false) { echo "<tr><td colspan=\"2\">在获取:" . $strListUrl . "网页源码时发生错误!</TD></TR>"; writetofile("crawel_error.log", $p_id . '{=====}' . $strListUrl); exit; } writetofile("crawel_info.log", $p_id . '{=====}' . $strListUrl . "{=====}List===start"); $listnum = $listnum + 1; $tempStep = 1; if ($p_playtype === 'tv_live') { if (isN($_SESSION["strListCodeCut"])) { $strListCodeCut = getBody($strListCode, $p_listcodestart, $p_listcodeend); $_SESSION["strListCodeCut"] = $strListCodeCut; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } if (isN($_SESSION["linkarrcode"])) { $linkarrcode = getArray($strListCodeCut, $p_listlinkstart, $p_listlinkend); $_SESSION["linkarrcode"] = $linkarrcode; } else { $linkarrcode = $_SESSION["linkarrcode"]; } if ($p_starringtype == 1) { $starringarrcode = getArray($strListCodeCut, $p_starringstart, $p_starringend); } if ($p_titletype == 1) { $titlearrcode = getArray($strListCodeCut, $p_titlestart, $p_titleend); } if ($p_pictype == 1) { $picarrcode = getArray($strListCodeCut, $p_picstart, $p_picend); } // writetofile("tv.log",$starringarrcode); // writetofile("tv.log",$titlearrcode); if ($p_starringtype == 1) { $starringarr = explode("{Array}", $starringarrcode); } if ($p_titletype == 1) { $titlearr = explode("{Array}", $titlearrcode); } if (is_array($titlearr)) { $count = count($titlearr); for ($i = 0; $i < $count; $i++) { writetofile("tv.log", 'name:' . $titlearr[$i] . ' code' . $starringarr[$i]); $db->query("insert into mac_tv(tv_name,tv_code,country,create_date) values('" . $titlearr[$i] . "','" . $starringarr[$i] . "','中国','" . date('Y-m-d H:i:s', time()) . "')"); } } return; } switch ($p_pagetype) { case 3: $strViewCode = $strListCode; $j = 1; if ($p_collecorder == 1) { $startnum = $p_pagebatchid2; $endnum = $p_pagebatchid1; } else { $startnum = $p_pagebatchid1; $endnum = $p_pagebatchid2; } if (!strpos($p_pagebatchurl, "{ID}")) { $startnum = 0; $endnum = 0; } wtablehead(); for ($i = $startnum; $i <= $endnum; $i++) { $UrlTest = replaceStr($p_pagebatchurl, "{ID}", $i); echo "<tr><td colspan=\"2\"></TD>正在采集列表:" . $UrlTest . "的数据 </TR>"; cjView($UrlTest, $i); $j = $j + 1; } wtablefoot(); if ($flag) { echo "<br> 此分页数据采集完毕 --- <script language=\"javascript\">setTimeout(\"makeNextPage();\",2000);function makeNextPage(){location.href='collect_vod_manage.php';}</script>"; } break; default: if ($p_playtype === 'baidu') { if (isN($_SESSION["strListCodeCut"])) { $_SESSION["strListCodeCut"] = $strListCodeCut; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } $baiduList = BaiduParse::parseMovieListByContent($strListCode, $p_code, ''); if (isN($_SESSION["linkarrcode"])) { $linkarrcode = $baiduList['linkarr']; $linkarrcode = implode('{Array}', $linkarrcode); $_SESSION["linkarrcode"] = $linkarrcode; } else { $linkarrcode = $_SESSION["linkarrcode"]; } if ($linkarrcode == false) { echo "<tr><td colspan=\"2\"></TD>在获取链接列表时出错!'.{$strListUrl}.’</TR>"; $sb = $sb + 1; writetofile("crawel_error.log", $p_id . '{=====}' . $strListUrl); return; } wBreakpoint("../../upload/vodbreakpoint", getUrl()); $linkarr = explode("{Array}", $linkarrcode); if ($p_starringtype == 1) { $starringarr = $baiduList['starringarr']; } if ($p_titletype == 1) { $titlearr = $baiduList['titlearr']; } if ($p_pictype == 1) { $picarr = $baiduList['picarr']; } } else { if (isN($_SESSION["strListCodeCut"])) { $strListCodeCut = getBody($strListCode, $p_listcodestart, $p_listcodeend); $_SESSION["strListCodeCut"] = $strListCodeCut; } else { $strListCodeCut = $_SESSION["strListCodeCut"]; } if (isN($_SESSION["linkarrcode"])) { $linkarrcode = getArray($strListCodeCut, $p_listlinkstart, $p_listlinkend); $_SESSION["linkarrcode"] = $linkarrcode; } else { $linkarrcode = $_SESSION["linkarrcode"]; } if ($p_starringtype == 1) { $starringarrcode = getArray($strListCodeCut, $p_starringstart, $p_starringend); } if ($p_titletype == 1) { $titlearrcode = getArray($strListCodeCut, $p_titlestart, $p_titleend); } if ($p_pictype == 1) { $picarrcode = getArray($strListCodeCut, $p_picstart, $p_picend); } if ($linkarrcode == false) { echo "<tr><td colspan=\"2\"></TD>在获取链接列表时出错!'.{$strListUrl}.’</TR>"; $sb = $sb + 1; writetofile("crawel_error.log", $p_id . '{=====}' . $strListUrl); return; } wBreakpoint("../../upload/vodbreakpoint", getUrl()); $linkarr = explode("{Array}", $linkarrcode); if ($p_starringtype == 1) { $starringarr = explode("{Array}", $starringarrcode); } if ($p_titletype == 1) { $titlearr = explode("{Array}", $titlearrcode); } if ($p_pictype == 1) { $picarr = explode("{Array}", $picarrcode); } } $viewcount = count($linkarr); if ($p_showtype == 1) { if ($viewnum >= $viewcount) { clearSession(); if ($flag) { echo "<br> 此分页数据采集完毕 --- 暂停2秒后继续采集<script language=\"javascript\">setTimeout(\"makeNextPage();\",2000);function makeNextPage(){location.href='collect_vod_cj.php?p_id=" . $p_ids . "&listnum=" . $listnum . "&sb=" . $sb . "&cg=" . $cg . "&num=" . $num . "&action=" . $action . "';}</script>"; } } else { if ($p_savefiles == 1) { $strdstate = "false"; } else { $strdstate = "true"; } wtablehead(); if ($p_playtype === 'baidu') { cjBaiduView(getHrefFromLink($linkarr[$viewnum]), $viewnum); } else { cjView(getHrefFromLink($linkarr[$viewnum]), $viewnum); } wtablefoot(); if ($flag) { echo "数据采集完毕 --- 稍后继续采集<script language=\"javascript\">var dstate=" . $strdstate . ";setInterval(\"makeNextPage();\",500);function makeNextPage(){if(dstate){dstate=false;location.href='collect_vod_cj.php?p_id=" . $p_ids . "&listnum=" . ($listnum - 1) . "&sb=" . $sb . "&cg=" . $cg . "&num=" . $num . "&viewnum=" . ($viewnum + 1) . "&action=" . $action . "';}}</script>"; } exit; } } else { if ($p_colleclinkorder == 1) { for ($i = $viewcount; $i >= 0; $i--) { wtablehead(); if ($i == $viewcount) { echo "<tr><td colspan=\"2\"></TD>正在采集列表:" . $strListUrl . "的数据 </TR>"; } $urlMo = getHrefFromLink($linkarr[$i]); if ($reCollExistMovie) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } else { $sql = "select m_id from {pre}cj_vod where m_urltest='" . $urlMo . "' order by m_id desc"; $rowvod = $db->getRow($sql); if (!$rowvod) { cjView($urlMo, $i); } } wtablefoot(); } } else { for ($i = 0; $i < count($linkarr); $i++) { writetofile("crawel_infosss.log", $p_id . '{=====}' . $urlMo . "{=====}View===is collected."); wtablehead(); if ($i == 0) { echo "<tr><td colspan=\"2\"></TD>正在采集列表:" . $strListUrl . "的数据 </TR>"; } $urlMo = getHrefFromLink($linkarr[$i]); if ($reCollExistMovie) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } else { $sql = "select m_id from {pre}cj_vod where m_urltest='" . $urlMo . "' order by m_id desc"; $rowvod = $db->getRow($sql); if (!$rowvod) { if ($p_playtype === 'baidu') { cjBaiduView($urlMo, $i); } else { cjView($urlMo, $i); } } else { writetofile("crawel_info.log", $p_id . '{=====}' . $urlMo . "{=====}View===is collected."); } } wtablefoot(); } } clearSession(); if ($flag) { echo "<br> 此分页数据采集完毕 --- 暂停2秒后继续采集<script language=\"javascript\">setTimeout(\"makeNextPage();\",2000);function makeNextPage(){location.href='collect_vod_cj.php?p_id=" . $p_ids . "&listnum=" . $listnum . "&sb=" . $sb . "&cg=" . $cg . "&num=" . $num . "&action=" . $action . "';}</script>"; } } } }