PHP可以灵活配置使用的采集器
发布时间：2022-10-21 15:22:38 所属栏目：PHP教程 来源：
导读： if (strpos($val, 'http:') === false)
{
if (substr($val, 0, 1) == '/')
$val = WEB_HOST.$val;
else
$val = WEB_HOST.'/'.$val;
}
{
if (substr($val, 0, 1) == '/')
$val = WEB_HOST.$val;
else
$val = WEB_HOST.'/'.$val;
}
|
if (strpos($val, 'http:') === false) { if (substr($val, 0, 1) == '/') $val = WEB_HOST.$val; else $val = WEB_HOST.'/'.$val; } $web_content = $this->get($val); if (empty($web_content)) { $this->write('抓取的内容页为空,所以过滤掉'); continue; } $web_content = str_replace("\r", '', $web_content); $web_content = str_replace("\n", '【】', $web_content); $sql = "INSERT INTO ".TABLE_NAME."(".implode(', ', array_keys($table_mapping)).")VALUES("; foreach ($table_mapping as $field => $reg) $sql .= ':'.$field.','; $sql = substr($sql ,0, -1); $sql .= ')'; if (IS_DEBUG) $this->write('执行SQL '.$sql); $dsn = 'mysql:dbname='.DB_NAME.';host='.DB_HOST; try { $dbh = new PDO($dsn, DB_USER, DB_PWD); } catch (PDOException $e) { $this->write( 'Connection failed: ' . $e->getMessage(), true); } $dbh->query("set names 'utf8'"); $sth = $dbh->prepare($sql); foreach ($table_mapping as $field => $reg) { if (substr($reg, 0, 1) != '/') { $$field = $reg; } else { if (!preg_match($reg, $web_content, $tmp_match)) { $this->write('对不起,匹配字段：'.$field.'失败，过滤此记录'); continue 2; } $$field = $tmp_match[1]; $$field = $this->closetags($$field); //删除javascript脚本 $$field = preg_replace('/ （编辑：航空爱好网） 【声明】本站内容均来自网络，其相关言论仅代表作者个人观点，不代表本站立场。若无意侵犯到您的权利，请及时与联系站长删除相关内容! |
站长推荐