PHP纯手写正则爬取星座屋网站星座运势数据

摘要：星座屋运势界面最终爬取数据结果展示在上的效果下面就是使用正则实现的代码，是自己一年多前花了半天时间写的。

星座屋（http://www.xzw.com/fortune/）运势界面：

最终爬取数据结果展示在APP上的效果：

下面就是使用正则实现的代码，是自己一年多前花了半天时间写的。现在想来，如果使用Scrapy或者phpspider只用几行代码就搞定了，不用这么费力气了~

 array("aries", "03/21-04/19"),
        "金牛座" => array("taurus", "04/20-05/20"),
        "双子座" => array("gemini", "05/21-06/21"),
        "巨蟹座" => array("cancer", "06/22-07/22"),
        "狮子座" => array("leo", "07/23-08/22"),
        "处女座" => array("virgo", "08/23-09/22"),
        "天秤座" => array("libra", "09/23-10/23"),
        "天蝎座" => array("scorpio", "10/24-11/22"),
        "射手座" => array("sagittarius", "11/23-12/21"),
        "魔羯座" => array("capricorn", "12/22-01/19"),
        "水瓶座" => array("aquarius", "01/20-02/18"),
        "双鱼座" => array("pisces", "02/19-03/20")
);*/

$constellation = isset($_REQUEST["xingzuo"]) && !empty($_REQUEST["xingzuo"]) ? trim($_REQUEST["xingzuo"]) : null;
$category = isset($_REQUEST["category"]) && !empty($_REQUEST["category"]) ? intval($_REQUEST["category"]) : null;

$all_xingzuo = array(
    "aries",
    "taurus",
    "gemini",
    "cancer",
    "leo",
    "virgo",
    "libra",
    "scorpio",
    "sagittarius",
    "capricorn",
    "aquarius",
    "pisces"
    );
$all_category = array(0,1,2,3,4,5);
if(!in_array($constellation, $all_xingzuo) || !in_array($category, $all_category)){
    exit("Params error");
}

$domain = "http://www.xzw.com/fortune/";
$apiUrl = "";
if($category){
    $apiUrl = $domain.$constellation."/".$category.".html";
} else {
    $apiUrl = $domain.$constellation."/";
}

header("Content-type: text/html; charset=utf-8");
function getFortuneData($url){
    $fortune_data = array();
    $data = file_get_contents($url);
    $data = mb_convert_encoding($data, "utf-8", "gbk");
    preg_match("/
(.*)
/ism", $data, $div_c_main);
    preg_match("/
(.*?)
/ism", $div_c_main[1], $dl);
    preg_match("/
(.*?)
/ism", $dl[1], $dd);
    preg_match("/
(.*?)
/ism", $dd[1], $ul);
    $ul = str_replace("", "{label}", $ul[1]);
    $ul = preg_replace("/]*?>/ism", "", $ul);
    $ul = preg_replace("/]*>/ism", "", $ul);
    $ul = preg_replace("//ism", "", $ul);
    $ul = preg_replace("//ism", "", $ul);
    //$ul = preg_replace("/s+/","",$ul);

    $ul_arr = explode("{label}", $ul);
    array_shift($ul_arr);

    foreach ($ul_arr as $key => &$li) {
        //preg_match_all("/([x81-xfe][x40-xfe])+/", $li, $matches);//转换编码 $matches[1][0]表示"："
        preg_match("//ism", $li, $width);
        
        if(!empty($width)){
            $li = explode("：",$li);
            $li["label"] = preg_replace("/]*?>/ism", "", $li[0]);
            $li["value"] = sprintf("%0.2f", floatval($width[1]/80));
            unset($li[0]);
            unset($li[1]);
              
            //$val[1] = $width[1]/16;
        } else {
            $li = explode("：",$li);
            $li["label"] = $li[0];
            $li["value"] = $li[1];
            unset($li[0]);
            unset($li[1]);
        }
    }

    $fortune_data["ul"] = $ul_arr;

    //获取c_cont
    preg_match("/
(.*?)
/ism", $data, $cont);
    $p_cont = preg_replace("/]*?>/ism", "", $cont[1]);
    $p_cont = str_replace("", "{span}", $p_cont);
    $p_cont = str_replace("
", "{p}", $p_cont);
    $p_cont = preg_replace(""<[/!]*?[^<>]*?>"si","",$p_cont);
    $p_cont = preg_replace("/s+/","",$p_cont);
    //$p_cont = preg_replace(""([rn])[s]+"","",$p_cont);
    $p_cont = str_replace("
", "", $p_cont);

    $p_cont_arr = explode("{p}",$p_cont);
    array_shift($p_cont_arr);

    foreach ($p_cont_arr as $key => $val) {
          $temp = explode("{span}", $val);
          $temp_arr["label"] = $temp[0];
          $temp_arr["value"] = $temp[1]; 
          $fortune_data["cont"][] = $temp_arr;
          unset($temp);
    }
    
    return $fortune_data;
}

$write_result = "";//写入状态 默认为空表示不写入
$local_data = "";
$result = array();
$fileName = !empty($category) ? $constellation."-".$category.".php" : $constellation.".php";
$fortune_data_path = "fortune_data/".$fileName;

if(file_exists($fortune_data_path)){
    $local_data = @file_get_contents("fortune_data/".$fileName);
}

if (!empty($local_data)) {
    $filemtime = filemtime($fortune_data_path);
    //判断缓存时间是否在当天内产生
    $todayStart = mktime(0, 0, 0, date("m"), date("d"), date("Y"));

    if($filemtime < $todayStart){ //缓存过期
        $data = getFortuneData($apiUrl);
        $write_result = write_fortune_cache($data, $fileName);
        if(empty($data)){
            $result["result"] = -1;
            $result["msg"] = "数据抓取失败！"; 
            $result["write_result"] = $write_result;
            $result["data"] = array();
            exit($json->encode($result));
        }
    } else {
        $data = unserialize($local_data);
    }
    $result["result"] = 0;
    $result["msg"] = "success";
    $result["write_result"] = $write_result;
    $result["data"] = $data;
    exit($json->encode($result));
} else {
    $data = getFortuneData($apiUrl);
    
    if(!empty($data)){
        $write_result = write_fortune_cache($data, $fileName);

        $result["result"] = 0;
        $result["msg"] = "success"; 
        $result["write_result"] = $write_result;
        $result["data"] = $data;
        exit($json->encode($result));
    } else {
        $result["result"] = -1;
        $result["msg"] = "数据抓取失败！"; 
        $result["write_result"] = $write_result;
        $result["data"] = array();
        exit($json->encode($result));
    }
}

function write_fortune_cache($data, $fileName){
    $fp = fopen("./fortune_data/".$fileName, "w+") or die("fortune_data/".$fileName."不存在！");
    $fw = fwrite($fp, serialize($data));
    
    if($fw){
        $write_result = "success";
    } else {
        $write_result = "fail";
    }
    fclose($fp);

    return $write_result;
}
?>