hadoop - Apache Pig - 无法读取包

Question

我正在尝试使用 PIG 读取逗号分隔的数据，如下所示:

grunt> cat script/pig/emp_tuple1.txt
1,kirti,250000,{(100),(200)}
2,kk,240000,{(100),(300)}
3,kumar,200000,{(200),(400)}
4,shinde,290000,{(200),(500),(300),(100)}
5,shinde k y,260000,{(100),(300),(200)}
6,amol,255000,{(300)}
grunt> emp_t1 = load 'script/pig/emp_tuple1.txt' using PigStorage(',') as (empno:int, ename:chararray, salary:int, dlist:bag{});
grunt> dump emp_t1;
2015-11-23 12:26:44,450 [main] INFO  org.apache.pig.backend.hadoop.executionengine.util.MapRedUtil - Total input paths to process : 1
(1,kirti,250000,)   
(2,kk,240000,)
(3,kumar,200000,)
(4,shinde,290000,)
(5,shinde k y,260000,)
(6,amol,255000,{(300)})

在它之间显示警告:

2015-11-23 12:26:44,173 [LocalJobRunner Map Task Executor #0] WARN  org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigHadoopLogger - org.apache.pig.builtin.Utf8StorageConverter(FIELD_DISCARDED_TYPE_CONVERSION_FAILED): Unable to interpret value [123, 40, 49, 48, 48, 41] in field being converted to type bag, caught ParseException <Unexpect end of bag> field discarded

好像是遇到包里的逗号(,)就提示了。

现在我所做的是:将逗号更改为制表符(或任何其他分隔符)并且有效:

grunt> cat script/pig/emp_tuple2.txt;
1|kirti|250000|{(100),(200)}
2|kk|240000|{(100),(300)}
3|kumar|200000|{(200),(400)}
4|shinde|290000|{(200),(500),(300),(100)}
5|shinde k y|260000|{(100),(300),(200)}
6|amol|255000|{(300)}
grunt> emp_t2 = load 'script/pig/emp_tuple2.txt' using PigStorage('|') as (empno:int, ename:chararray, salary:int, dlist:bag{});
grunt> dump emp_t1;
2015-11-23 12:31:33,408 [main] INFO  org.apache.pig.backend.hadoop.executionengine.util.MapRedUtil - Total input paths to process : 1
(1,kirti,250000,{(100),(200)})
(2,kk,240000,{(100),(300)})
(3,kumar,200000,{(200),(400)})
(4,shinde,290000,{(200),(500),(300),(100)})
(5,shinde k y,260000,{(100),(300),(200)})
(6,amol,255000,{(300)})

所以我只是想知道如果您有逗号分隔的数据和用逗号分隔的包，它不会工作吗？

Answer 1

Lets go into details, 
 1. Data is being read as TextInputFormat 
 2. Line Record Reader is being used to read lines
 3. , is being used to separate columns. 

as "," occurs in the bag and is the delimeter across columns, bag is being split into multiple columns. 

There are various way to overcome this. 

 1. pre-process the input and replace first three "," in each row by some other delimeter.