function many_list_get_page($rules_arr, $start_url = '')
{
extract($rules_arr);
//print_r($rules_arr);
//exit();
$url = $start_url ? $start_url : $test;
$rules = stripslashes($rules);
$content = get_contents($url, array('login_cookie' => $login_cookie, 'cache' => -1));
if ($type == 1) {
//dom
$link_arr = dom_page_link($content, array('page_link_rules' => $rules, 'url_page_range' => $url));
} else {
$link_arr = string_page_link($content, $rules, $url);
}
return $link_arr;
}
function evo_get_pagelink($content, $url, $list = array())
{
$list = $list ? $list : $url;
$rules_info = match_rules($url, $content, 4, 0);
if ($rules_info && is_array($rules_info)) {
if ($rules_info['page_get_type'] == 1) {
$link_arr = dom_page_link($content, array('page_link_rules' => $rules_info['page_link_rules'], 'url' => $url));
} else {
if ($rules_info['page_get_type'] == 2) {
$link_arr = string_page_link($content, trim($rules_info['page_link_rules']), $url);
}
}
}
if ($link_arr) {
return $link_arr;
}
$base_url = get_base_url($content);
$base_url = $base_url ? $base_url : $url;
$link_arr = _striplinks($content, $base_url);
if (!$link_arr) {
return array();
}
foreach ((array) $link_arr as $k => $v_url) {
if (!check_fastpick_viewurl($v_url, $url)) {
unset($link_arr[$k]);
continue;
}
$c_arr[$k] = strlen($v_url);
}
$value_count_arr = array_count_values($c_arr);
arsort($value_count_arr);
$value_count_arr = array_keys($value_count_arr);
$view_lenth = array_shift($value_count_arr);
$link_arr = array_resolve($link_arr);
foreach ($link_arr as $k => $v) {
if (abs(strlen($v) - $view_lenth) > 5) {
unset($link_arr[$k]);
}
}
$link_arr = array_filter($link_arr, 'filter_url_callback');
return $link_arr;
}
function get_content_page($content)
{
if ($this->p_arr['content_page_get_type'] == 1) {
$html = get_htmldom_obj($content);
if (!$html) {
return false;
}
foreach ($html->find($this->p_arr['content_page_rules']) as $v) {
$a_url = $this->format_url($v->attr['href']);
if (!$a_url || $a_url == '#' || $v->innertext == milu_lang('up_page')) {
continue;
}
$item[] = _expandlinks($a_url, $this->base_url);
$re_arr = sarray_unique($item);
}
$html->clear();
unset($html);
} else {
$re_arr = string_page_link($content, $this->p_arr['content_page_rules'], $this->now_url);
//字符串
}
return $re_arr;
}