function test_window($get_type, $url_test, $is_fiter, $rules, $replace_rules, $filter_data, $show_type, $login_cookie, $filter_html_arr) { $url_test = rpc_str($url_test); $rules = rpc_str($rules); $replace_rules = rpc_str($replace_rules); $login_cookie = rpc_str(urlencode($login_cookie)); foreach ($filter_data as $k => $v) { if ($v) { $filter_data[$k][1] = rpc_str($v[1]); } } $filter_html_arr = sarray_unique($filter_html_arr); //去重 if ($show_type == 'title') { $show_name = milu_lang('title'); } else { if ($show_type == 'body') { $show_name = milu_lang('body'); } else { $show_name = milu_lang('reply'); } } $contents = get_contents($url_test, array('cookie' => $login_cookie)); $contents = dz_attach_format($url_test, $contents); $c_arr = format_article_imgurl($url_test, $contents); $contents = $c_arr['message']; if ($get_type == 1) { //dom if ($show_type == 'reply') { $result_data = dom_get_manytext($contents, $rules); } else { if ($show_type == 'title') { $dom_rules['title'] = $rules; } else { $dom_rules['content'] = $rules; } $re = dom_single_article($contents, $dom_rules); $result_data = $show_type == 'title' ? $re['title'] : $re['content']; } } else { if ($get_type == 2) { //字符串 if ($contents != -1) { if ($show_type == 'reply') { $rules = str_replace('[body]', '[reply]', $rules); $result_data = str_get_str($contents, $rules, $show_type, -1); unset($result_data[0]); } else { $result_data = str_get_str($contents, $rules, $show_type, 1); } } } else { //智能获取 if ($contents != -1) { $re = get_single_article($contents, $url_test); if ($show_type == 'title') { $result_data = $re['title']; } else { $result_data = $re['content']; } } } } if ($result_data == -1) { echo milu_lang('unable_pick'); return; } else { if ($result_data == -2) { echo milu_lang('get_time_out'); return; } } if (!$result_data) { echo milu_lang('no_get_data') . $show_name; return; } $format_args = array('is_fiter' => $is_fiter, 'show_type' => $show_type, 'result_data' => $result_data, 'replace_rules' => $replace_rules, 'filter_data' => $filter_data, 'test' => 1, 'filter_html' => $filter_html_arr); $result_data = filter_article($format_args); if ($show_type == 'reply') { $body = show_reply_output($result_data); } else { $body .= $result_data; } $body .= $notice; echo $body; }
function get_other_info($content, $args) { if (!$content) { return false; } extract($args); if (!$from_get_rules && !$author_get_rules && !$dateline_get_rules) { return false; } $html = get_htmldom_obj($content); if (!$html) { return false; } if ($from_get_rules) { if ($from_get_type == 1) { $re['from'] = dom_get_str($html, $from_get_rules); } else { $re['from'] = str_get_str($content, $from_get_rules, 'data'); } } if ($author_get_rules) { if ($author_get_type == 1) { $re['author'] = dom_get_str($html, $author_get_rules); } else { $re['author'] = str_get_str($content, $author_get_rules, 'data'); } } if ($dateline_get_rules) { if ($dateline_get_type == 1) { $re['article_dateline'] = dom_get_str($html, $dateline_get_rules); unset($div); } else { $re['article_dateline'] = str_get_str($content, $dateline_get_rules, 'data'); } } foreach ((array) $re as $k => $v) { $re[$k] = format_html($v); } $html->clear(); unset($html); return $re; }
function get_article($content, $url = '') { global $_G; $url = $url ? $url : $this->now_url; $pick_config = $_G['cache']['evn_milu_pick']['pick_config']; require_once libfile('function/home'); if ($this->temp_arr['have_page'] != 1) { $this->a++; $this->pick_cache_data['a'] = $this->a; } if ($this->p_arr['rules_type'] == 3) { //一键采集 $article_info = get_single_article($content, $url); } else { //先取标题 if ($this->p_arr['theme_get_type'] == 3) { //智能识别 $article_info = get_single_article($content, $url); } else { if ($this->p_arr['theme_get_type'] == 1) { //dom获取 $article_info = dom_single_article($content, array('title' => $this->p_arr['theme_rules'])); } else { if ($this->p_arr['theme_get_type'] == 2) { //字符串 $re = pregmessage($content, '<title>[title]</title>', 'title', -1); $article_info['other']['old_title'] = $re[0]; $article_info['title'] = str_get_str($content, $this->p_arr['theme_rules'], 'title', -1); } } } //标题要做额外的工作,清理掉多余的html $article_info['title'] = format_html($article_info['title']); if (!trim($article_info['title'])) { return $article_info; } //如果标题都取不到,不必浪费时间获取内容 //再取内容 if ($this->p_arr['content_get_type'] == 3) { //智能识别 if ($this->p_arr['theme_get_type'] != 3) { $info_arr = get_single_article($content, $url); $article_info['content'] = $info_arr['content']; } } else { if ($this->p_arr['content_get_type'] == 1) { //dom获取 $info_arr = dom_single_article($content, array('content' => $this->p_arr['content_rules'])); $article_info['content'] = $info_arr['content']; } else { if ($this->p_arr['content_get_type'] == 2) { //字符串 $article_info['content'] = str_get_str($content, $this->p_arr['content_rules'], 'body', -1); } } } if (!$article_info['content'] && $pick_config['ask_mode'] == 1) { $article_info['content'] = $article_info['title']; } //取其他内容 $other_arr = $this->get_article_other($content); $other_arr = $other_arr ? $other_arr : array(); $article_info = array_merge($article_info, $other_arr); } //附件 if ($this->p_arr['is_download_file'] == 1) { $article_info['content'] = attach_format($url, $article_info['content']); } //需要回复才能看到的内容 if (VIP) { $article_info['content'] = post($article_info['content'], array('cookie' => $this->p_arr['login_cookie'], 'page_url' => $url)); } $article_info['content'] = format_html($article_info['content']); $article_info['content'] = unhtmlentities($article_info['content']); $article_info['url'] = $url; $article_info['content'] = clear_ad_html($article_info['content']); $article_info['content'] = getstr($article_info['content'], 0, 1, 1, 0, 1); return $article_info; }