[HIVE-5573] sql result messes up if char '0x0d ' in the database data file

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 0.10.0
• Component/s: Database/Schema

Environment

CDH4.3

• Tags: hive sql 0x0d
• Assignee: Unassigned
• Reporter: a bc
• Created: 17/Oct/13 05:28
• Updated: 17/Oct/13 05:50

Description

I didn't test this bug on 0.11.0, only on 0.10.

With select statement, the returned result is not correct, totally messed up!

================================
Test case:
for table, we have:
hive> desc tblfoo;
OK
sessionid string
userid string
groupid int
docid string
channeltype string
licensetype string
issuitelicense string
activationstatus string
licensedproduct string

The wrong result when select this table :
hive> select * from tblfoo;
OK
7B179D5F-6D1A-4B00-89CC-5FE9F64A6D74 9EF52B8C-10D2-4C79-9460-6C295C9D5E7A 90 SUBSCRIPTION Trial SUITE NULL NULL
Trial NULL NULL NULL NULL NULL NULL NULL

and the right result should be:
hive> select * from tblfoo;
OK
7B179D5F-6D1A-4B00-89CC-5FE9F64A6D74 9EF52B8C-10D2-4C79-9460-6C295C9D5E7A 90 SUBSCRIPTION Trial SUITE Trial NOVALUE

If I remove the char '0x0d' in the database data file, then I can get the correct result.

================================
Steps to reproduce this bug:
1. create table:
hive> create table tblFoo (
> sessionid string,
> userid string,
> groupid int ,
> docid string,
> channeltype string,
> licensetype string,
> issuitelicense string,
> activationstatus string,
> licensedproduct string);

2. hive> load data local inpath '/tmp/li.dat' overwrite into table tblFoo;
3. hive> select * from tblfoo;

================================
And I will attach some example files:
1. li.dat is the database file with a '0x0d' char in file
-bash-4.1$ od -x li.dat
0000000 4237 3731 4439 4635 362d 3144 2d41 4234
0000020 3030 382d 4339 2d43 4635 3945 3646 4134
0000040 4436 3437 3901 4645 3235 3842 2d43 3031
0000060 3244 342d 3743 2d39 3439 3036 362d 3243
0000100 3539 3943 3544 3745 0141 0139 0130 5553
0000120 5342 5243 5049 4954 4e4f 5401 6972 6c61
0000140 5301 4955 4554 01*0d* 7254 6169 016c 4f4e
0000160 4156 554c 0a45
0000166

2. li2.dat is the database file without a '0x0d' char in file.
-bash-4.1$ od -x li2.dat
0000000 4237 3731 4439 4635 362d 3144 2d41 4234
0000020 3030 382d 4339 2d43 4635 3945 3646 4134
0000040 4436 3437 3901 4645 3235 3842 2d43 3031
0000060 3244 342d 3743 2d39 3439 3036 362d 3243
0000100 3539 3943 3544 3745 0141 0139 0130 5553
0000120 5342 5243 5049 4954 4e4f 5401 6972 6c61
0000140 5301 4955 4554 5401 6972 6c61 4e01 564f
0000160 4c41 4555 000a

-bash-4.1$ ls -l
-rwxr-xr-x 1 hdfs hdfs 118 Oct 17 10:32 li.dat
rw-rr- 1 hdfs hdfs 117 Oct 17 11:08 li2.dat