转自:http://www.aboutyun.com/thread-6713-1-1.html
pig简介
pig是hadoop上层的衍生架构,与hive类似。对比hive(hive类似sql,是一种声明式的语言),pig是一种过程语言,类似于存储过程一步一步得进行数据转化。pig简单操作
1.从文件导入数据
1)Mysql (Mysql需要先创建表).
CREATE TABLE TMP_TABLE(USER VARCHAR(32),AGE INT,IS_MALE BOOLEAN);
CREATE TABLE TMP_TABLE_2(AGE INT,OPTIONS VARCHAR(50)); -- 用于Join
LOAD DATA LOCAL INFILE '/tmp/data_file_1' INTO TABLE TMP_TABLE ;
LOAD DATA LOCAL INFILE '/tmp/data_file_2' INTO TABLE TMP_TABLE_2;
2)Pig
tmp_table = LOAD '/tmp/data_file_1' USING PigStorage('\t') AS (user:chararray, age:int,is_male:int);
tmp_table_2= LOAD '/tmp/data_file_2' USING PigStorage('\t') AS (age:int,options:chararray);
2.查询整张表
1)Mysql
SELECT * FROM TMP_TABLE;
2)Pig
DUMP tmp_table;
3. 查询前50行
1)Mysql
SELECT * FROM TMP_TABLE LIMIT 50;
2)Pig
tmp_table_limit = LIMIT tmp_table 50;
DUMP tmp_table_limit;
4.查询某些列
1)Mysql
SELECT USER FROM TMP_TABLE;
2)Pig
tmp_table_user = FOREACH tmp_table GENERATE user;
DUMP tmp_table_user;
5. 给列取别名
1)Mysql
SELECT USER AS USER_NAME,AGE AS USER_AGE FROM TMP_TABLE;
2)Pig
tmp_table_column_alias = FOREACH tmp_table GENERATE user AS user_name,age AS user_age;
DUMP tmp_table_column_alias;
6.排序
1)Mysql
SELECT * FROM TMP_TABLE ORDER BY AGE;
2)Pig
tmp_table_order = ORDER tmp_table BY age ASC;
DUMP tmp_table_order;
7.条件查询
1)Mysql
SELECT * FROM TMP_TABLE WHERE AGE>20;
2) Pig
tmp_table_where = FILTER tmp_table by age > 20;
DUMP tmp_table_where;
8.内连接Inner Join
1)Mysql
SELECT * FROM TMP_TABLE A JOIN TMP_TABLE_2 B ON A.AGE=B.AGE;
2)Pig
tmp_table_inner_join = JOIN tmp_table BY age,tmp_table_2 BY age;
DUMP tmp_table_inner_join;
9.左连接Left Join
1)Mysql
SELECT * FROM TMP_TABLE A LEFT JOIN TMP_TABLE_2 B ON A.AGE=B.AGE;
2)Pig
tmp_table_left_join = JOIN tmp_table BY age LEFT OUTER,tmp_table_2 BY age;
DUMP tmp_table_left_join;
10.右连接Right Join
1)Mysql
SELECT * FROM TMP_TABLE A RIGHT JOIN TMP_TABLE_2 B ON A.AGE=B.AGE;
2)Pig
tmp_table_right_join = JOIN tmp_table BY age RIGHT OUTER,tmp_table_2 BY age;
DUMP tmp_table_right_join;
11.全连接Full Join
1)Mysql
SELECT * FROM TMP_TABLE A JOIN TMP_TABLE_2 B ON A.AGE=B.AGE
UNION SELECT * FROM TMP_TABLE A LEFT JOIN TMP_TABLE_2 B ON A.AGE=B.AGE
UNION SELECT * FROM TMP_TABLE A RIGHT JOIN TMP_TABLE_2 B ON A.AGE=B.AGE;
2)Pig
tmp_table_full_join = JOIN tmp_table BY age FULL OUTER,tmp_table_2 BY age;
DUMP tmp_table_full_join;
12.同时对多张表交叉查询
1)Mysql
SELECT * FROM TMP_TABLE,TMP_TABLE_2;
2)Pig
tmp_table_cross = CROSS tmp_table,tmp_table_2;
DUMP tmp_table_cross;
13.分组GROUP BY
1)Mysql
SELECT * FROM TMP_TABLE GROUP BY IS_MALE;
2)Pig
tmp_table_group = GROUP tmp_table BY is_male;
DUMP tmp_table_group;
14.分组并统计
1)Mysql
SELECT IS_MALE,COUNT(*) FROM TMP_TABLE GROUP BY IS_MALE;
2)Pig
tmp_table_group_count = GROUP tmp_table BY is_male;
tmp_table_group_count = FOREACH tmp_table_group_count GENERATE group,COUNT($1);
DUMP tmp_table_group_count;
15.查询去重DISTINCT
1)MYSQL
SELECT DISTINCT IS_MALE FROM TMP_TABLE;
2)Pig
tmp_table_distinct = FOREACH tmp_table GENERATE is_male;
tmp_table_distinct = DISTINCT tmp_table_distinct;
DUMP tmp_table_distinct;
---------------------------------------------------------------------------------------------------------------------------------------------------
上面简单操作,下面需要进一步了解pig支持的内容:
pig支持数据类型
double > float > long > int > bytearray
tuple|bag|map|chararray > bytearray
double float long int chararray bytearray都相当于pig的基本类型
tuple相当于数组 ,但是可以类型不一,举例('dirkzhang','dallas',41)
Bag相当于tuple的一个集合,举例{('dirk',41),('kedde',2),('terre',31)},在group的时候会生成bag
Map相当于哈希表,key为chararray,value为任意类型,例如['name'#dirk,'age'#36,'num'#41
nulls 表示的不只是数据不存在,他更表示数据是unkown
pig latin语法
1:load
LOAD 'data' [USING function] [AS schema];
例如:
load = LOAD 'sql://{SELECT MONTH_ID,DAY_ID,PROV_ID FROM zb_d_bidwmb05009_010}' USING com.xxxx.dataplatform.bbdp.geniuspig.VerticaLoader('oracle','192.168.6.5','dev','1522','vbap','vbap','1') AS (MONTH_ID:chararray,DAY_ID:chararray,PROV_ID:chararray);
Table = load ‘url’ as (id,name…..); //table和load之间除了等号外 还必须有个空格 不然会出错,url一定要带引号,且只能是单引号。
2:filter
alias = FILTER alias BY expression;
Table = filter Table1 by + A; //A可以是 id > 10;not name matches ‘’,is not null 等,可以用and 和or连接各条件
例如:
filter = filter load20 by ( MONTH_ID == '1210' and DAY_ID == '18' and PROV_ID == '010' );
3:group
alias = GROUP alias { ALL | BY expression} [, alias ALL | BY expression …] [USING 'collected' | 'merge'] [PARTITION BY partitioner] [PARALLEL n];
pig的分组,不仅是数据上的分组,在数据的schema形式上也进行分组为groupcolumn:bag
Table3 = group Table2 by id;也可以Table3 = group Table2 by (id,name);括号必须加
可以使用ALL实现对所有字段的分组
4:foreach
alias = FOREACH alias GENERATE expression [AS schema] [expression [AS schema]….];
alias = FOREACH nested_alias {
alias = {nested_op | nested_exp}; [{alias = {nested_op | nested_exp}; …]
GENERATE expression [AS schema] [expression [AS schema]….]
};
一般跟generate一块使用
Table = foreach Table generate (id,name);括号可加可不加。
avg = foreach Table generate group, AVG(age); MAX ,MIN..
在进行数据过滤时,建议尽早使用foreach generate将多余的数据过滤掉,减少数据交换
5:join
Inner join Syntax
alias = JOIN alias BY {expression|'('expression [, expression …]')'} (, alias BY {expression|'('expression [, expression …]')'} …) [USING 'replicated' | 'skewed' | 'merge' | 'merge-sparse'] [PARTITION BY partitioner] [PARALLEL n];
Outer join Syntax
alias = JOIN left-alias BY left-alias-column [LEFT|RIGHT|FULL] [OUTER], right-alias BY right-alias-column [USING 'replicated' | 'skewed' | 'merge'] [PARTITION BY partitioner] [PARALLEL n];
join/left join / right join
daily = load 'A' as (id,name, sex);
divs = load 'B' as (id,name, sex);
join
jnd = join daily by (id, name), divs by (id, name);
left join
jnd = join daily by (id, name) left outer, divs by (id, name);
也可以同时多个变量,但只用于inner join
A = load 'input1' as (x, y);
B = load 'input2' as (u, v);
C = load 'input3' as (e, f);
alpha = join A by x, B by u, C by e;
6: union
alias = UNION [ONSCHEMA] alias, alias [, alias …];
union 相当与sql中的union,但与sql不通的是pig中的union可以针对两个不同模式的变量:如果两个变量模式相同,那么union后的变量模式与 变量的模式一样;如果一个变量的模式可以由另一各变量的模式强制类型转换,那么union后的变量模式与转换后的变量模式相同;否则,union后的变量 没有模式。
A = load 'input1' as (x:int, y:float);
B = load 'input2' as (x:int, y:float);
C = union A, B;
describe C;
C: {x: int,y: float}
A = load 'input1' as (x:double, y:float);
B = load 'input2' as (x:int, y:double);
C = union A, B;
describe C;
C: {x: double,y: double}
A = load 'input1' as (x:int, y:float);
B = load 'input2' as (x:int, y:chararray);
C = union A, B;
describe C;
Schema for C unknown.
注意:在pig 1.0中 执行不了最后一种union。
如果需要对两个具有不通列名的变量union的话,可以使用onschema关键字
A = load 'input1' as (w: chararray, x:int, y:float);
B = load 'input2' as (x:int, y:double, z:chararray);
C = union onschema A, B;
describe C;
C: {w: chararray,x: int,y: double,z: chararray}
join和union之后alias的别名会变
7:Dump
dump alias
用于在屏幕上显示数据。
8:Order by
alias = ORDER alias BY { * [ASC|DESC] | field_alias [ASC|DESC] [, field_alias [ASC|DESC] …] } [PARALLEL n];
A = order Table by id desc;
9:distinct
A = distinct alias;
10:limit
A = limit alias 10;
11:sample
SAMPLE alias size;
随机抽取指定比例(0到1)的数据。
some = sample divs 0.1;
13:cross
alias = CROSS alias, alias [, alias …] [PARTITION BY partitioner] [PARALLEL n];
将多个数据集中的数据按照字段名进行同值组合,形成笛卡尔积。
--cross.pig
daily = load 'NYSE_daily' as (exchange:chararray, symbol:chararray,date:chararray, open:float, high:float, low:float,
close:float, volume:int, adj_close:float);
divs = load 'NYSE_dividends' as (exchange:chararray, symbol:chararray,date:chararray, dividends:float);
tonsodata = cross daily, divs parallel 10;
15:split
Syntax
SPLIT alias INTO alias IF expression, alias IF expression [, alias IF expression …] [, alias OTHERWISE];
A = LOAD 'data' AS (f1:int,f2:int,f3:int);
DUMP A;
(1,2,3)
(4,5,6)
(7,8,9)
SPLIT A INTO X IF f1<7, Y IF f2==5, Z IF (f3<6 OR f3>6);
DUMP X;
(1,2,3)
(4,5,6)
DUMP Y;
(4,5,6)
DUMP Z;
(1,2,3)
(7,8,9)
16:store
Store … into … Using…
pig在别名维护上:
1、join
如e = join d by name,b by name;
g = foreach e generate $0 as one:chararray, $1 as two:int, $2 as three:chararray,$3 asfour:int;
他生成的schemal:
e: {d::name: chararray,d::position: int,b::name: chararray,b::age: int}
g: {one: chararray,two: int,three: chararray,four: int}
2、group
B = GROUP A BY age;
----------------------------------------------------------------------
| B | group: int | A: bag({name: chararray,age: int,gpa: float}) |
----------------------------------------------------------------------
| | 18 | {(John, 18, 4.0), (Joe, 18, 3.8)} |
| | 20 | {(Bill, 20, 3.9)} |
----------------------------------------------------------------------
(18,{(John,18,4.0F),(Joe,18,3.8F)})
pig udf自定义
pig支持嵌入user defined function,一个简单的udf 继承于evalFunc,通常用在filter,foreach中
----------------------------------------------------------------------------------------------------------------------------------------------------
如果我们知道nosql和传统数据库是有差别的,那么他们的都支持什么运算符和函数
4.6 Retional operators
首先编写两个数据文件A:
0,1,2
1,3,4
数据文件B:
0,5,2
1,7,8
运行pig:
xuqiang@ubuntu:~/hadoop/src/pig/pig-0.8.1/tutorial/pigtmp$ pig -x local
2011-06-05 18:46:54,039 [main] INFO org.apache.pig.Main - Logging error messages to: /home/xuqiang/hadoop/src/pig/pig-0.8.1/tutorial/pigtmp/pig_1307324814030.log
2011-06-05 18:46:54,324 [main] INFO org.apache.pig.backend.hadoop.executionengine.HExecutionEngine - Connecting to hadoop file system at: file:///
grunt>
加载数据A:
grunt> a = load 'A' using PigStorage(',') as (a1:int, a2:int, a3:int);
加载数据B:
grunt> b = load 'B' using PigStorage(',') as (b1:int, b2:int, b3:int);
求a,b的并集:
grunt> c = union a, b;
grunt> dump c;
(0,5,2)
(1,7,8)
(0,1,2)
(1,3,4)
将c分割为d和e,其中d的第一列数据值为0,e的第一列的数据为1($0表示数据集的第一列):
grunt> split c into d if $0 == 0, e if $0 == 1;
查看d:
grunt> dump d;
(0,1,2)
(0,5,2)
查看e:
(1,3,4)
(1,7,8)
选择c中的一部分数据:
grunt> f = filter c by $1 > 3;
查看数据f:
grunt> dump f;
(0,5,2)
(1,7,8)
对数据进行分组:
grunt> g = group c by $2;
查看g:
grunt> dump g;
(2,{(0,1,2),(0,5,2)})
(4,{(1,3,4)})
(8,{(1,7,8)})
当然也能够将所有的元素集合到一起:
grunt> h = group c all;
grunt> dump h;
(all,{(0,1,2),(1,3,4),(0,5,2),(1,7,8)})
查看h中元素个数:
grunt> i = foreach h generate COUNT($1);
查看元素个数:
grunt> dump i;
这里可能出现Could not resolve counter using imported: [, org.apache.pig.built in., org.apache.pig.impl.builtin. ]的情况,这是需要使用register命令来注册pig对应的jar版本。
接下俩试一下jon操作:
grunt> j = join a by $2, b by $2;
该操作类似于sql中的连表查询,这是的条件是$2 == $2。
取出c的第二列$1和$1 * $2,将这两列保存在k中:
grunt> k = foreach c generate $1, $1 * $2;
查看k的内容:
grunt> dump k;
(5,10)
(7,56)
(1,2)
(3,12)