为每个文件添加一个额外的列，给出文件号，然后对修改后的数据集进行并集，然后根据给出“文件号”的列进行排序

A = LOAD 'A.txt' USING PigStorage() AS (a:int);
B = LOAD 'B.txt' USING PigStorage() AS (b:int);
A_mod = FOREACH A GENERATE a, 1 AS file_number;
B_mod = FOREACH A GENERATE b, 2 AS file_number;
unified_mod = UNION A_mod, B_mod;
output = SORT unified_mod BY file_number;

我尝试了经典的联合，对我来说数据保持有序。
但如果不成功，让我们试着强迫它：）
正如我在前面的评论中所说的那样，它效率不高，但它使工作更有效。

--In order to determine nbA you can run the following cmd in the shell : wc -l A.txt
%default nbA 3

A = LOAD 'A.txt' USING PigStorage() AS (a:int);
B = LOAD 'B.txt' USING PigStorage() AS (b:int);

A = RANK A;
B = RANK B;

--DESCRIBE B; 
B = FOREACH B GENERATE rank_B + $nbA, $1;

C= UNION B,A;
C= ORDER C BY $0;
C= FOREACH C GENERATE $1; --If you want to drop the first column
DUMP C;

输出：

(111)
(222)
(555)
(333)
(444)
(666)

哪里：
a、 文本

111
222
555

和b.txt：

333
444
666