You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@phoenix.apache.org by "wuyang (JIRA)" <ji...@apache.org> on 2014/10/09 09:52:34 UTC
[jira] [Comment Edited] (PHOENIX-1334) Issue when LIKE expression
contains Chinese characters on Key column
[ https://issues.apache.org/jira/browse/PHOENIX-1334?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14164886#comment-14164886 ]
wuyang edited comment on PHOENIX-1334 at 10/9/14 7:51 AM:
----------------------------------------------------------
[~jaywong],
Yes, that works. But it may cause some performance problem :( and returning unexpected result like "xxx中xxxx"
was (Author: wuyang630):
[~jaywong],
Yes, that works. But it may cause some performance problem :(
> Issue when LIKE expression contains Chinese characters on Key column
> --------------------------------------------------------------------
>
> Key: PHOENIX-1334
> URL: https://issues.apache.org/jira/browse/PHOENIX-1334
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.1
> Environment: jdk 1.8 linux
> Reporter: wuyang
>
> When I use like expression in SELECT query. It works well when I put *Chinese* characters in LIKE expression on NONE PRIMARY KEY columns . BUT when I put them in LIKE expression on *PRIMARY KEY* , it occurs an Exception:
> select * from "test3" where PK like '中%';
> ||COLUMN_NAME||DATA_TYPE||TYPE_NAME
> |PK|12|VARCHAR
> |VAL|12|VARCHAR
> {quote}
> org.apache.phoenix.schema.IllegalDataException: CHAR types may only contain single byte characters (中)
> at org.apache.phoenix.schema.PDataType$2.toBytes(PDataType.java:216)
> at org.apache.phoenix.compile.WhereOptimizer$KeyExpressionVisitor.visitLeave(WhereOptimizer.java:829)
> at org.apache.phoenix.compile.WhereOptimizer$KeyExpressionVisitor.visitLeave(WhereOptimizer.java:349)
> at org.apache.phoenix.expression.LikeExpression.accept(LikeExpression.java:269)
> at
> ....
> {quote}
> the type of PRIMARY KEY and NONE PRIMARY KEY columns are all VARCHAR
> In the relative source code:
> {code}
> byte[] b = VARCHAR.toBytes(object);
> if (b.length != ((String) object).length()) {
> throw new IllegalDataException("CHAR types may only contain single byte characters (" + object + ")");
> }
> {code}
> actually, Chinese (or other non-Latin) characters will never meet the condition b.length == ((String) object).length() . Default encode method is UTF-8.
> *User following sentences to reappear:*
> create table "test_c" ( pk varchar primary key , val varchar);
> upsert into "test_c" values ('中文','中文');
> select * from "test_c" where VAL like '中%';
> _// it works well until now_
> select * from "test_c" where PK like '中%';
> _// oops..._
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)