【问题标题】:PIG: FLATTEN error猪:展平错误
【发布时间】:2015-08-21 17:56:39
【问题描述】:

我有一个关系X 与结构X: {group: chararray,inboundCount: {(name: chararray,inb: long)},outboundCount: {(name: chararray,out: long)}} 如下:

(IAD,{},{(IAD,25)})
(LAX,{},{(LAX,2)})
(ORD,{(ORD,27)},{})
(PDX,{},{(PDX,3)}) 
(SFO,{(SFO,3)},{})

我想要一个具有以下结构的输出 final: {airport: chararray,inbound: long,outbound: long}with out put:

(IAD,,25)
(LAX,,2)
(ORD,27,)
(PDX,,3)
(SFO,3,)

我尝试了以下代码,它给出了我想要的输出结构。但是什么都没有打印出来。是因为空值包的原因吗?

final = foreach X generate group as airport,FLATTEN(inboundCount.inb) as inbound,FLATTEN(outboundCount.out) as outbound;

请帮帮我。

编辑 我通过执行以下命令得到了这种关系x

A= load '/user/hduser/airline.csv' using PigStorage(',') as (year:int,month:int,dayofmonth:int,dayofweek:int,dep:int,CRS:int,Arr:int,CRSArr:int,UniqueCarrier:chararray,FlightNum:int,TailNum:chararray,ActualElapsedTime:int,CRSElapsed:int,AirTime:int,ArrDelay:int,DepDelay:int,Origin:chararray,Dest:chararray,Distance:int,TaxiIn:int,TaxiOut:int,Cancelled:int,CancelCode:chararray,Diverted:int,CarrierDelay:int,WeatherDelay:int,NASDelay:int,SecurityDelay:int,LateAircraft:int);
B= foreach A generate year,month,UniqueCarrier,FlightNum,TailNum,Origin,Dest;
inbound = group B by Dest;
inboundCount = foreach inbound generate group,COUNT(B.FlightNum) as inb;
outbound = group B by Origin;
outboundCount = foreach outbound generate group,COUNT(B.FlightNum) as out;
X = COGROUP inboundCount BY name, outboundCount BY name;

样本输入记录:

2008,1,31,4,1757,1155,2400,1758,UA,114,N845UA,243,243,217,362,362,LAX,ORD,1745,11,15,0,,0,0,0,362,0,0

【问题讨论】:

  • 好的。这只是关系 X。你能否给出真实的输入数据集和你为达到关系 X 而编写的相应 pig 语句
  • @SurenderRaja 我已经编辑了这个问题。希望对您有所帮助。
  • 好的。您似乎想了解每个机场起飞和降落的航班数量。正确吗?
  • @SurenderRaja 是的。没错。

标签: apache-pig flatten


【解决方案1】:

你快到了。请试试这个。只需应用 SUM 而不是展平

 A= load '/user/hduser/airline.csv' using PigStorage(',') as (year:int,month:int,dayofmonth:int,dayofweek:int,dep:int,CRS:int,Arr:int,CRSArr:int,UniqueCarrier:chararray,FlightNum:int,TailNum:chararray,ActualElapsedTime:int,CRSElapsed:int,AirTime:int,ArrDelay:int,DepDelay:int,Origin:chararray,Dest:chararray,Distance:int,TaxiIn:int,TaxiOut:int,Cancelled:int,CancelCode:chararray,Diverted:int,CarrierDelay:int,WeatherDelay:int,NASDelay:int,SecurityDelay:int,LateAircraft:int);

B= foreach A generate year,month,UniqueCarrier,FlightNum,TailNum,Origin,Dest;

inbound = group B by Dest;

inboundCount = foreach inbound generate group,COUNT(B.FlightNum) as inb;

outbound = group B by Origin;

outboundCount = foreach outbound generate group,COUNT(B.FlightNum) as out;

X = COGROUP inboundCount BY name, outboundCount BY name;

final_data = FOREACH X GENERATE group as airport, SUM(inboundCount.inb) as inb, SUM(outboundCount.out) as out;

dump final_data;

final_data 的转储将为您提供预期的结果。

(IAD,,25)
(LAX,,2)
(ORD,27,)
(PDX,,3)
(SFO,3,)

如果你愿意,你仍然可以将 NULL 计数替换为 0

 final_null_check = FOREACH final_data GENERATE airport,(inb is null ? 0 :inb) as inb_cnt, (out is null ? 0 : out) as out_cnt;

NULL 检查后,如果你转储 final_null_check 关系,你会得到如下输出

 (IAD,0,25)
 (LAX,0,2)
 (ORD,27,0)
 (PDX,0,3)
 (SFO,3,0)

【讨论】:

  • 谢谢。那效果很好。但是你能告诉我为什么它不能使用 FLATTEN 吗?
猜你喜欢
  • 2016-02-20
  • 1970-01-01
  • 2023-04-03
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 2013-03-08
  • 1970-01-01
相关资源
最近更新 更多