Hi
    Hdfs/hive to any RDBMS , sqoop is the right tool to go for. It is 
exclusively meant for to and fro parallel data transfers between Rdbms and hdfs.
 
Regards
Bejoy KS

Sent from handheld, please excuse typos.

-----Original Message-----
From: wzc <wzc1...@gmail.com>
Date: Sat, 5 May 2012 02:18:39 
To: <user@hive.apache.org>
Reply-To: user@hive.apache.org
Subject: Re: how can I escape hive CLI output

hi Nicole,
  Thanks for your response.
  I will try your method and use '\001' as the separator(since there
contains tabs in the query results).

2012/5/5 Gesli, Nicole <nicole.ge...@memorylane.com>

>  What I'd do is write the query output into local directory like this:
>
>  INSERT OVERWRITE LOCAL DIRECTORY '/mydir'
> SELECT …
>
>  The output columns will be delimited with ^A (\001). If you have to have
> tab delimited format you can replace them like this:
>
>  cat /mydir/* | tr "\001" "\t" >> /mynewdir/myfile.dat
>
>  I hope this helps.
> -Nicole
>
>   From: wzc <wzc1...@gmail.com>
> Reply-To: <user@hive.apache.org>
> Date: Sat, 5 May 2012 01:20:15 +0800
> To: <user@hive.apache.org>
> Subject: how can I escape hive CLI output
>
>  Hi all:
> I am new to hive, and I try to run a query through hive CLI and load the
> result into mysql.
> I try to redirect the CLI output to a tmp file and load the tmp file into
> mysql table. The problem here is that some columns of our query result may
> contains special chars, such as tab(\t), new line(\\), backslash (\) and so
> on. The result in mysql table is corrupted by these chars because the CLI
> output isn't escaped properly.
> It seems what i want is mentioned in  
> hive-692<https://issues.apache.org/jira/browse/HIVE-692>,
> but i can't find any solutions there.
> A dirty way I think may be escaping the query result in HQL, such like:
>
>  regexp_replace(regexp_replace(regexp_replace(result_column_1, '\\\\',
> '\\\\\\\\'), '\t', '\\\\t'), '\n', '\\\\n'),
>
>  I would like to know how to escape hive CLI output, or any alternative
> way to accomplish such job.
> Any help is appreciated.
>
>  Thanks.
>
>

Reply via email to