hive 使用php腳本來完成map/reduce

hive sql有個比較強大的功能就是可以使用外界腳本來完成map/reduce。用法如下：
TRANSFORM(….) USING ‘….’ AS (….)。
USING裡面可以使用php腳本來處理，具體用方法看下面這個範例。

①、建立一個表格：

<code>hive> CREATE TABLE <span>`member`</span>(
    >   <span>`id`</span><span>int</span>,
    >   <span>`user_name`</span><span>string</span>,
    >   <span>`passwd`</span><span>string</span>
    > )
    > row <span>format</span> delimited
    > fields terminated by <span>"\t"</span><span>//这句很必要，不然下面的文件导入会导致所有的值都变成null </span>
    > stored as textfile;</code>

②、準備以下資料/tmp/member.dat

<code>1     zhangsan     zs1024
2     lisi     ls1991
3     wangwu     ww2001
4     liumang     lm1234
5     linxing     lx1990</code>

③、匯入資料

<code>hive> <span>load</span> data <span>local</span> inpath <span>'/tmp/member.dat'</span> into <span>table</span> member;</code>

可以看到：

<code>hive> <span>select</span> * <span>from</span> member;
OK
<span>1</span>       zhangsan        zs1024
<span>2</span>       lisi    ls1991
<span>3</span>       wangwu  ww2001
<span>4</span>       liumang lm1234
<span>5</span>       linxing lx1990</code>

列做一次md5加密。我們使用php腳本來處理，php腳本（/tmp/changePasswd.php）程式碼如下：

<code><span><span><?php </span><span>while</span>(!feof(STDIN)){
     <span>$line</span> = rtrim(fgets(STDIN), <span>"\n"</span>);      <span>//逐行读取</span><span>if</span> (<span>empty</span>(<span>$line</span>)) <span>continue</span>;     <span>//空，跳出当次循环</span><span>$data</span> = explode(<span>"\t"</span>, <span>$line</span>);    <span>//将切分出来的数组保存起来，下面判断使用</span><span>foreach</span>(<span>$data</span><span>as</span> &<span>$val</span>){
          <span>$val</span> = md5(<span>$val</span>);
     }
     <span>unset</span>(<span>$val</span>);
     <span>echo</span> implode(<span>"\t"</span>, <span>$data</span>) . <span>"\n"</span>;
}</span></span></code>

④、將php腳本加入hive。

<code>hive> <span>add</span> file /tmp/changePasswd<span>.php</span><span>;</span></code>

⑤、使用php腳本來執行map/reduce

<code>hive> insert overwrite table member
    > select TRANSFORM(<span>`i</span>d<span>`,</span><span>`u</span>ser_name<span>`,</span><span>`p</span>asswd<span>`)</span> using <span>"/usr/bin/php changePasswd.php"</span>
    > as (<span>`i</span>d<span>`,</span><span>`u</span>ser_name<span>`,</span><span>`p</span>asswd<span>`)</span> from member<span>;</span></code>

最後，我們可以看到，passwd那一列的資料已經改變：

<code>hive> <span>select</span> * <span>from</span> member;
OK
<span>1</span>       zhangsan        d03eed89429cc3006cc279322c2800c5
<span>2</span>       lisi    <span>063401506</span>c9d9f0e49a706e3779b7428
<span>3</span>       wangwu  ac5a8109dbbb46c9f69ffd5fc93c11f8
<span>4</span>       liumang fda8b97fd723bdbf6a754812b5ecec27
<span>5</span>       linxing <span>4035378</span>ace8936e93d95aa77e7e224d4</code>

版權聲明：本文為部落客原創文章，轉載請註明出處。

以上就介紹了hive 使用php腳本來完成map/reduce，包含了方面的內容，希望對PHP教學有興趣的朋友有幫助。