function fetch_web($schemas, $test = null)
{
global $_user, $_channels;
$updated_file = array();
foreach ($schemas as $k => $v) {
if (!$v->channel && $v->url == '' || $v->keywords == '') {
continue;
}
if ($v->channel) {
$url_arr = $_channels[$v->channel]['url'];
} else {
$url_arr = explode(' ', $v->url);
}
foreach ($url_arr as $url_key => $url) {
$url = mkurl($v->channel, $url_key, $url, $v->url);
$filename = md5($url) . '.tmp';
if (in_array($filename, $updated_file)) {
$html = file_get_contents(CACHE_PATH . $filename);
} else {
$file = '';
$context = $v->channel ? mkfp_context($url_key) : null;
$html = file_get_contents($url, false, $context);
$html = format_script($html);
file_put_contents(CACHE_PATH . $filename, $html);
$updated_file[] = $filename;
}
//处理关键词
$keywords = $v->keywords;
$encoding = mb_detect_encoding($html, array('ASCII', 'GB2312', 'GBK', 'UTF-8'));
if ($encoding == 'EUC-CN') {
$encoding = 'GB2312';
$keywords = iconv('UTF-8', 'GBK', $keywords);
} elseif ($encoding == 'CP936') {
$keywords = iconv('UTF-8', $encoding, $keywords);
}
$keywords = explode(' ', $keywords);
$first_key = array_shift($keywords);
$lpos = 0;
$rpos = NULL;
$len = strlen($first_key);
for ($lpos = strpos($html, $first_key, $lpos); $lpos !== false; $lpos = strpos($html, $first_key, $lpos + $len)) {
$rpos = find_key($html, $keywords, $lpos + $len);
if ($rpos) {
$html_len = strlen($html);
//得到关键词上下文
for ($j = 1; $j < 6 && $lpos !== false; $j++) {
$lpos = strrpos($html, '</', $lpos - $html_len - 2);
}
if (!$lpos) {
$lpos = 0;
}
for ($i = 1; $i < 5; $i++) {
$rpos = strpos($html, '</', $rpos + 2);
}
if (!$rpos) {
$rpos = $html_len;
}
$result = substr($html, $lpos, $rpos - $lpos);
break;
}
}
$result = trim(strip_tags($result, '<a>'));
//确定包含数字
if ($v->max_num != '' || $v->min_num != '') {
$result = find_num($result, $v->max_num, $v->min_num);
}
if ($result) {
$result = str_replace(array("\n", "\r", "\t", " "), '', $result);
if ($encoding != 'UTF-8') {
$result = substr($result, 0, 400);
$result = iconv('GBK', 'UTF-8', $result);
} else {
$result = mb_substr($result, 0, 400, 'utf8');
}
$website = isset($_channels[$v->channel]['website'][$url_key]) ? $_channels[$v->channel]['website'][$url_key] : $url;
$content = $result . "<a href='{$website}' target='_blank'>查看</a>";
if ($test) {
return $content;
} else {
$_user = user::get_one(array('user_id' => $v->user_id));
schema::update(array('status' => 'off', 'schema_id' => $v->schema_id));
msg::add(array('schema_id' => $v->schema_id, 'title' => $v->title, 'content' => $content, 'status' => 'new'));
$email = filter_var($_user->email, FILTER_VALIDATE_EMAIL);
if ($_user->email_notify == 'on' && $email) {
mail::send_cron($email, $v->title, $content);
}
if ($_user->app_notify == 'on' && $_user->baidu_uid) {
push::push_message($_user->baidu_uid, $v->title, $result);
}
}
$result = NULL;
break;
}
}
}
}
<?php
if ($_view == 'reactivate_schema') {
schema::update(array('schema_id' => $_GET['schema_id'], 'status' => 'on'));
}
include TEMPLATES_PATH . 'msg.html';
<?php
$template = 'schema.html';
if ($_view == 'edit') {
$template = 'schema_edit.html';
if ($_POST['submit_schema_activate']) {
$_POST['status'] = 'on';
$_POST['submit_schema'] = 1;
}
if ($_POST['submit_schema']) {
if (is_numeric($_POST['schema_id'])) {
schema::update($_POST);
$template = 'schema.html';
} else {
schema::add($_POST);
$template = 'schema.html';
}
}
} elseif ($_view == 'test') {
$keywords = str_replace('^^', ' ', $_REQUEST['keywords']);
$schema = array('channel' => $_REQUEST['channel'], 'url' => $_REQUEST['url'], 'keywords' => $keywords);
$result = fetch_web(array((object) $schema), true);
echo $result;
exit;
} elseif ($_view == 'del') {
schema::delete($_GET);
}
include TEMPLATES_PATH . $template;