Pig join用法举例

jnd = join a by f1, b by f2;

 

join操作默认的是内连接，只有两边都匹配才会保留

 

需要用null补位的那边需要知道它的模式：

如果是左外连接，需要知道右边的数据集的模式，不匹配的字段用null补位

如果是右外连接，需要知道左边的数据集的模式，不匹配的字段用null补位

如果是全外连接，需要知道两边的数据集的模式，不匹配的字段用null补位

 

触发reduce阶段

 

基本用法

a = load 'input1';
b = load 'input2';
jnd = join a by $0, b by $1;

　　 

多字段连接

a = load 'input1' as (username, age, city);
b = load 'input2' as (orderid, user, city);
jnd = join a by (username, city), b by (user, city);

　　 

:: join后的字段引用

a = load 'input1' as (username, age, address);
b = load 'input2' as (orderid, user, money;
jnd = join a by username, b by user;
result = foreach jnd generate a::username, a::age, address, b::orderid;

　　 

多数据集连接

a = load 'input1' as (username, age);
b = load 'input2' as (orderid, user);
c = load 'input3' as (user, acount);
jnd = join a by username, b by user, c by user;

　　 

外连接 仅限两个数据集

a = load 'input1' as (username, age);
b = load 'input2' as (orderid, user);
jnd = join a by username left outer, b by user;
jnd = join a by username right, b by user;
jnd = join a by username full, b by user;

　　

自连接 需要加载自身数据集两次，使用不同的别名

a = load 'data' as (node, parentid, name);
b = load 'data' as (node, parentid, name);
jnd = join a by node, b by parentid;