[jira] [Comment Edited] (HIVE-21452) Loss of query condition when exist exists

["zengxl (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-21452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16816852#comment-16816852 ] 

zengxl edited comment on HIVE-21452 at 4/13/19 6:59 AM:
--------------------------------------------------------

I found the problem in the genSQJoinTree method of the SmanticAnalyr class.
 The problem is that only the first element of the right connection is taken, and the rest is not。
 the problem code:
 {color:#f79232}leftAliases[leftChildAliases.length] = leftTree.getRightAliases()[0];{color}

{color:#333333}If there are many connections and the condition is placed in where, then the latter condition will be lost and only the where condition of the first connection will be obtained.{color}

{color:#333333}{color:#f79232}this is my change code{color}:{color}
{code:java}
if (!joinTree.getNoSemiJoin()){
 String[] leftChildAliases = leftTree.getLeftAliases();
 String[] rightChildAliases = leftTree.getRightAliases();
 String leftAliases[] = new String[leftChildAliases.length + rightChildAliases.length];
 for (int i = 0; i < leftChildAliases.length; i++) { leftAliases[i] = leftChildAliases[i]; }
 for (int i = 0; i < rightChildAliases.length; i++) { leftAliases[leftChildAliases.length + i] = rightChildAliases[i]; }
 joinTree.setLeftAliases(leftAliases);
 }else{
 String[] leftChildAliases = leftTree.getLeftAliases();
 String leftAliases[] = new String[leftChildAliases.length + 1];
 for (int i = 0; i < leftChildAliases.length; i++) { leftAliases[i] = leftChildAliases[i]; }
 leftAliases[leftChildAliases.length] = leftTree.getRightAliases()[0];
 joinTree.setLeftAliases(leftAliases);
 }{code}


was (Author: zengxl):
I found the problem in the genSQJoinTree method of the SmanticAnalyr class.
 The problem is that only the first element of the right connection is taken, and the rest is not。
 the problem code:
 {color:#f79232}leftAliases[leftChildAliases.length] = leftTree.getRightAliases()[0];{color}

{color:#333333}If there are many connections and the condition is placed in where, then the latter condition will be lost and only the where condition of the first connection will be obtained.{color}

{color:#333333}this is my change code:{color}

{color:#333333}if (!joinTree.getNoSemiJoin()){
 String[] leftChildAliases = leftTree.getLeftAliases();
 String[] rightChildAliases = leftTree.getRightAliases();
 String leftAliases[] = new String[leftChildAliases.length + rightChildAliases.length];
 for (int i = 0; i < leftChildAliases.length; i++) {
 leftAliases[i] = leftChildAliases[i];
 }
 for (int i = 0; i < rightChildAliases.length; i++) {
 leftAliases[leftChildAliases.length + i] = rightChildAliases[i];
 }
 joinTree.setLeftAliases(leftAliases);
}else{
 String[] leftChildAliases = leftTree.getLeftAliases();
 String leftAliases[] = new String[leftChildAliases.length + 1];
 for (int i = 0; i < leftChildAliases.length; i++) {
 leftAliases[i] = leftChildAliases[i];
 }
 leftAliases[leftChildAliases.length] = leftTree.getRightAliases()[0];
 joinTree.setLeftAliases(leftAliases);
}{color}

> Loss of query condition when exist exists
> -----------------------------------------
>
>                 Key: HIVE-21452
>                 URL: https://issues.apache.org/jira/browse/HIVE-21452
>             Project: Hive
>          Issue Type: Bug
>    Affects Versions: 1.1.0
>            Reporter: zengxl
>            Priority: Major
>
> In our production environment, there are four tables to do association queries. There are exists in the conditions, and we found that the first two conditions were lost.
> The following two conditions are missing：
> {color:#f79232}t2.cust_no is null and t4.cust_level not in ('4','5'){color}
>   
>  In the test environment, I prepared the data of the following four tables, and a condition was lost in the simulation.
> tables:
> test_table1
>  cust_no,name
>  60001,lisa
>  60002,tina
>  60003,kylin
>  60004,jeny
>  60005,john
>  60006,jamse
> test_table2
>  cust_no,acct_type
>  60001,1
>  60001,1
>  60001,2
>  60002,1
>  60003,2
>  60003,3
> test_table3
>  cust_no
>  60001
>  60002
>  60003
>  60004
>  60005
>  60007
> test_table4
>  cust_no,cust_level
>  60001,1
>  60002,2
>  60003,3
>  60004,4
>  60005,5
>  
> create table tmp.test_table1(cust_no string,name string);
>  create table tmp.test_table2(cust_no string,acct_type string);
>  create table tmp.test_table3(cust_no string);
>  create table tmp.test_table4(cust_no string,cust_level string);
> insert into tmp.test_table1 select '60001','lisa';
>  insert into tmp.test_table1 select '60002','tina';
>  insert into tmp.test_table1 select '60003','kylin';
>  insert into tmp.test_table1 select '60004','jeny';
>  insert into tmp.test_table1 select '60005','john';
>  insert into tmp.test_table1 select '60006','jamse';
> insert into tmp.test_table2 select '60001','1';
>  insert into tmp.test_table2 select '60001','1';
>  insert into tmp.test_table2 select '60001','2';
>  insert into tmp.test_table2 select '60002','1';
>  insert into tmp.test_table2 select '60003','2';
>  insert into tmp.test_table2 select '60002','3';
> insert into tmp.test_table3 select '60001';
>  insert into tmp.test_table3 select '60002';
>  insert into tmp.test_table3 select '60003';
>  insert into tmp.test_table3 select '60004';
>  insert into tmp.test_table3 select '60005';
>  insert into tmp.test_table3 select '60007';
> insert into tmp.test_table4 select '60001','1';
>  insert into tmp.test_table4 select '60002','2';
>  insert into tmp.test_table4 select '60003','3';
>  insert into tmp.test_table4 select '60004','4';
>  insert into tmp.test_table4 select '60005','5';
>   
>  Here is my query SQL And shut down mapjoin:
> set hive.auto.convert.join=false;
> select t1.cust_no as cust_no,t2.cust_no as custNO,t1.name from tmp.test_table1 t1
>  left join tmp.test_table2 t2 on t1.cust_no=t2.cust_no
>  and t2.acct_type='1'
>  left join tmp.test_table4 t4 on t1.cust_no=t4.cust_no
>  where t2.cust_no is null and t4.cust_level not in ('4','5') and exists (select 1 from tmp.test_table3 t3 where t1.cust_no=t3.cust_no)
>  
> All I want is to include cust_no for 6003，But the result is inclusive 6004 and 6005，this wrong 。{color:#f79232}In my production environment, 6001 came out。Loss of condition because cust_no is  null。{color}
> {color:#f6c342}View the execution plan, t4.cust_level not in ('4','5') condition missing{color}
> *{color:#f6c342}Explain:{color}*
> STAGE DEPENDENCIES:
>  Stage-1 is a root stage
>  Stage-2 depends on stages: Stage-1
>  Stage-0 depends on stages: Stage-2
> STAGE PLANS:
>  Stage: Stage-1
>  Map Reduce
>  Map Operator Tree:
>  TableScan
>  alias: t1
>  Statistics: Num rows: 12 Data size: 2128 Basic stats: COMPLETE Column stats: NONE
>  Filter Operator
>  predicate: cust_no is not null (type: boolean)
>  Statistics: Num rows: 6 Data size: 1064 Basic stats: COMPLETE Column stats: NONE
>  Reduce Output Operator
>  key expressions: cust_no (type: string)
>  sort order: +
>  Map-reduce partition columns: cust_no (type: string)
>  Statistics: Num rows: 6 Data size: 1064 Basic stats: COMPLETE Column stats: NONE
>  value expressions: name (type: string)
>  TableScan
>  alias: t2
>  Statistics: Num rows: 12 Data size: 2088 Basic stats: COMPLETE Column stats: NONE
>  Filter Operator
>  predicate: ((acct_type = '1') and cust_no is not null) (type: boolean)
>  Statistics: Num rows: 3 Data size: 522 Basic stats: COMPLETE Column stats: NONE
>  Reduce Output Operator
>  key expressions: cust_no (type: string)
>  sort order: +
>  Map-reduce partition columns: cust_no (type: string)
>  Statistics: Num rows: 3 Data size: 522 Basic stats: COMPLETE Column stats: NONE
>  TableScan
>  alias: t4
>  Statistics: Num rows: 10 Data size: 1740 Basic stats: COMPLETE Column stats: NONE
>  Filter Operator
>  predicate: cust_no is not null (type: boolean)
>  Statistics: Num rows: 5 Data size: 870 Basic stats: COMPLETE Column stats: NONE
>  Reduce Output Operator
>  key expressions: cust_no (type: string)
>  sort order: +
>  Map-reduce partition columns: cust_no (type: string)
>  Statistics: Num rows: 5 Data size: 870 Basic stats: COMPLETE Column stats: NONE
>  Reduce Operator Tree:
>  Join Operator
>  condition map:
>  Left Outer Join0 to 1
>  Left Outer Join0 to 2
>  keys:
>  0 cust_no (type: string)
>  1 cust_no (type: string)
>  2 cust_no (type: string)
>  outputColumnNames: _col0, _col1, _col5
>  Statistics: Num rows: 13 Data size: 2340 Basic stats: COMPLETE Column stats: NONE
>  Filter Operator
>  predicate: _col5 is null (type: boolean)
>  Statistics: Num rows: 6 Data size: 1080 Basic stats: COMPLETE Column stats: NONE
>  File Output Operator
>  compressed: false
>  table:
>  input format: org.apache.hadoop.mapred.SequenceFileInputFormat
>  output format: org.apache.hadoop.hive.ql.io.HiveSequenceFileOutputFormat
>  serde: org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe
> Stage: Stage-2
>  Map Reduce
>  Map Operator Tree:
>  TableScan
>  Reduce Output Operator
>  key expressions: _col0 (type: string)
>  sort order: +
>  Map-reduce partition columns: _col0 (type: string)
>  Statistics: Num rows: 6 Data size: 1080 Basic stats: COMPLETE Column stats: NONE
>  value expressions: _col1 (type: string)
>  TableScan
>  alias: t3
>  Statistics: Num rows: 12 Data size: 1068 Basic stats: COMPLETE Column stats: NONE
>  Filter Operator
>  predicate: cust_no is not null (type: boolean)
>  Statistics: Num rows: 6 Data size: 534 Basic stats: COMPLETE Column stats: NONE
>  Select Operator
>  expressions: cust_no (type: string)
>  outputColumnNames: _col1
>  Statistics: Num rows: 6 Data size: 534 Basic stats: COMPLETE Column stats: NONE
>  Group By Operator
>  keys: _col1 (type: string)
>  mode: hash
>  outputColumnNames: _col0
>  Statistics: Num rows: 6 Data size: 534 Basic stats: COMPLETE Column stats: NONE
>  Reduce Output Operator
>  key expressions: _col0 (type: string)
>  sort order: +
>  Map-reduce partition columns: _col0 (type: string)
>  Statistics: Num rows: 6 Data size: 534 Basic stats: COMPLETE Column stats: NONE
>  Reduce Operator Tree:
>  Join Operator
>  condition map:
>  Left Semi Join 0 to 1
>  keys:
>  0 _col0 (type: string)
>  1 _col0 (type: string)
>  outputColumnNames: _col0, _col1
>  Statistics: Num rows: 6 Data size: 1188 Basic stats: COMPLETE Column stats: NONE
>  Select Operator
>  expressions: _col0 (type: string), null (type: string), _col1 (type: string)
>  outputColumnNames: _col0, _col1, _col2
>  Statistics: Num rows: 6 Data size: 1188 Basic stats: COMPLETE Column stats: NONE
>  File Output Operator
>  compressed: false
>  Statistics: Num rows: 6 Data size: 1188 Basic stats: COMPLETE Column stats: NONE
>  table:
>  input format: org.apache.hadoop.mapred.SequenceFileInputFormat
>  output format: org.apache.hadoop.hive.ql.io.HiveSequenceFileOutputFormat
>  serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
> Stage: Stage-0
>  Fetch Operator
>  limit: -1
>  Processor Tree:
>  ListSink
>  
> {color:#f6c342}*result:*{color}
>  60003 NULL kylin
>  60004 NULL jeny
>  60005 NULL john
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)