mapreduce join总结(多对多内连接,左外连接,右外连接,全连接)

在reduce端做join,以TextPair做key,TextPair.getSecond为源文件类型区分,reduce端提取出来赋值给变量bz

 

a join b

//a放进内存list
while(hasNext())
 if bz=0
 list.add()
 else 

{
 for clist :list
  write  clist next 

}

a left outer join b

int count=0;
//a放进内存list
while(hasNext())
 if bz=0
 list.add()
 else 

{
count++; //记录该key在b表的记录数
 for clist :list
  write  clist next 

}

if count==0  //如果b表没有记录，则输出null
for clist :list
 write clist null





//b放进内存list的写法

while(hasNext())
 if bz=0
 list.add()                   //b表的bz=0,b表放进list内存
 else 

{
 if list.size==0                 //如果b表没有对应该key的记录，则输出null
write next null
else                              //有则循环输出
for clist:list
 write next clist

}

a full outer join b

int count=0;
//a放进内存list
while(hasNext())
 if bz==0
 list.add()
 else 

{
count++; //记录该key在b表的记录数

if list.size==0  //如果a表没有记录
 write null next
else 
 for clist :list
  write  clist next 

}

if count==0  //如果b表没有记录，则输出null
for clist :list
 write clist null

a right outer join b

int count=0;
//a放进内存list
while(hasNext())
 if bz=0
 list.add()
 else 

{
count++; //记录该key在b表的记录数

if list.size==0
write null next
else
 for clist :list
  write  clist next