|
@@ -15,31 +15,26 @@ class Dataoke
|
|
|
{
|
|
{
|
|
|
$ql = QueryList::html($content);
|
|
$ql = QueryList::html($content);
|
|
|
|
|
|
|
|
-// 元数据采集规则
|
|
|
|
|
- $rules = [
|
|
|
|
|
- 'link' => ["a", "href"]
|
|
|
|
|
- ];
|
|
|
|
|
-// 切片选择器
|
|
|
|
|
$rt = $ql->find("a")->attrs('href');
|
|
$rt = $ql->find("a")->attrs('href');
|
|
|
$data = $rt->all();
|
|
$data = $rt->all();
|
|
|
$data = array_unique($data);
|
|
$data = array_unique($data);
|
|
|
- foreach ($data as $url) {
|
|
|
|
|
|
|
+ foreach ($data as $k => $url) {
|
|
|
// if (strstr($url, "tao.php") === false) {
|
|
// if (strstr($url, "tao.php") === false) {
|
|
|
// continue;
|
|
// continue;
|
|
|
// }
|
|
// }
|
|
|
- $query=parse_url($url);
|
|
|
|
|
|
|
+ $query = parse_url($url);
|
|
|
|
|
|
|
|
$t = parse_query($query['query']);
|
|
$t = parse_query($query['query']);
|
|
|
if ($t['kw']) {
|
|
if ($t['kw']) {
|
|
|
- $q=[];
|
|
|
|
|
- if ($t['cid']){
|
|
|
|
|
- $q['cid']=$t['cid'];
|
|
|
|
|
|
|
+ $q = [];
|
|
|
|
|
+ if ($t['cid']) {
|
|
|
|
|
+ $q['cid'] = $t['cid'];
|
|
|
}
|
|
}
|
|
|
- if ($t['pid']){
|
|
|
|
|
- $q['pid']=$t['pid'];
|
|
|
|
|
|
|
+ if ($t['pid']) {
|
|
|
|
|
+ $q['pid'] = $t['pid'];
|
|
|
}
|
|
}
|
|
|
- if ($t['bids']){
|
|
|
|
|
- $q['bids']=$t['bids'];
|
|
|
|
|
|
|
+ if ($t['bids']) {
|
|
|
|
|
+ $q['bids'] = $t['bids'];
|
|
|
}
|
|
}
|
|
|
|
|
|
|
|
//elseif ($t['cid']){
|
|
//elseif ($t['cid']){
|
|
@@ -48,20 +43,22 @@ class Dataoke
|
|
|
// $d=$t;
|
|
// $d=$t;
|
|
|
// unset($d['kw']);
|
|
// unset($d['kw']);
|
|
|
// }
|
|
// }
|
|
|
- if (count($q)>0){
|
|
|
|
|
- $s="?".http_build_query($q);
|
|
|
|
|
- }else{
|
|
|
|
|
- $s="";
|
|
|
|
|
|
|
+ if (count($q) > 0) {
|
|
|
|
|
+ $s = "?" . http_build_query($q);
|
|
|
|
|
+ } else {
|
|
|
|
|
+ $s = "";
|
|
|
}
|
|
}
|
|
|
- $new_url="/kw/" . $t['kw'] . ".html".$s;
|
|
|
|
|
|
|
+ $new_url = "/kw/" . $t['kw'] . ".html" . $s;
|
|
|
|
|
|
|
|
- $content = str_replace($url, $new_url, $content);
|
|
|
|
|
- }else{
|
|
|
|
|
|
|
+ $content = str_replace('"' . $url . '"', '"' . $new_url . '"', $content);
|
|
|
|
|
+ } else {
|
|
|
if ($t['cid']) {
|
|
if ($t['cid']) {
|
|
|
- $content = str_replace($url, "/category/" . $t['cid'] . ".html", $content);
|
|
|
|
|
- }else{
|
|
|
|
|
|
|
+ $new_url = "/category/" . $t['cid'] . ".html";
|
|
|
|
|
+ $content = str_replace('"' . $url . '"', '"' . $new_url . '"', $content);
|
|
|
|
|
+ } else {
|
|
|
if ($t['id']) {
|
|
if ($t['id']) {
|
|
|
- $content = str_replace($url, "/product/" . $t['id'] . ".html", $content);
|
|
|
|
|
|
|
+ $new_url = "/product/" . $t['id'] . ".html";
|
|
|
|
|
+ $content = str_replace('"' . $url . '"', '"' . $new_url . '"', $content);
|
|
|
}
|
|
}
|
|
|
}
|
|
}
|
|
|
|
|
|
|
@@ -73,13 +70,13 @@ class Dataoke
|
|
|
// }
|
|
// }
|
|
|
|
|
|
|
|
}
|
|
}
|
|
|
- $content=str_replace('"/tao.php?r=p"', '"/pages/fengqiang.html"', $content);
|
|
|
|
|
- $content=str_replace('"/tao.php?r=nine"', '"/pages/nine.html"', $content);
|
|
|
|
|
- $content=str_replace('"/tao.php?r=ddq"', '"/pages/dongdong.html"', $content);
|
|
|
|
|
- $content=str_replace('"/tao.php?r=l"', '"/pages/haohuo.html"', $content);
|
|
|
|
|
- $content=str_replace('"/tao.php?r=index/index&u=1350274"', '"/"', $content);
|
|
|
|
|
- $content=str_replace('href="/tao.php?r=index/index"', 'href="/"', $content);
|
|
|
|
|
- $content=str_replace('https://cmsstatic.ffquan.cn//js/new_home.js', "/static/new_home.js", $content);
|
|
|
|
|
|
|
+ $content = str_replace('"/tao.php?r=p"', '"/pages/fengqiang.html"', $content);
|
|
|
|
|
+ $content = str_replace('"/tao.php?r=nine"', '"/pages/nine.html"', $content);
|
|
|
|
|
+ $content = str_replace('"/tao.php?r=ddq"', '"/pages/dongdong.html"', $content);
|
|
|
|
|
+ $content = str_replace('"/tao.php?r=l"', '"/pages/haohuo.html"', $content);
|
|
|
|
|
+ $content = str_replace('"/tao.php?r=index/index&u=1350274"', '"/"', $content);
|
|
|
|
|
+ $content = str_replace('href="/tao.php?r=index/index"', 'href="/"', $content);
|
|
|
|
|
+ $content = str_replace('https://cmsstatic.ffquan.cn//js/new_home.js', "/static/new_home.js", $content);
|
|
|
// $content=str_replace('index.php', "tao.php", $content);
|
|
// $content=str_replace('index.php', "tao.php", $content);
|
|
|
return $content;
|
|
return $content;
|
|
|
}
|
|
}
|