You're right. I assumed there was a corresponding ALTER TABLE foo SET ROW
FORMAT ...
But I found the answer in the archives. Modify the SERDE properties, e.g. SET
SERDEPROPERTIES ('field.delim' = '|');
http://osdir.com/ml/hive-user-hadoop-apache/2009-12/msg00109.html
On May 11, 2012, at 8:15 PM, Igor Tatarinov wrote:
> Thanks but that requires fixing the table schema. Actually, I haven't found a
> way to change the delimiters of an existing table (created with a LIKE
> statement). I did find a workaround.
>
> While I don't know the schema of the data, I do know the number of columns,
> so I am going to create a table with N string columns and my preferred
> delimiter and then insert into it. Hive doesn't seem to mind inserting
> numeric data into a string column so that should work regardless of the
> expressions I will be "inserting".
>
>
> On Fri, May 11, 2012 at 5:07 PM, David Kulp <[email protected]> wrote:
> Here is the default textfile. Substitute delimiters as necessary.
>
> CREATE TABLE ...
> ROW FORMAT DELIMITED
> FIELDS TERMINATED BY '\001' COLLECTION ITEMS TERMINATED BY '\002' MAP KEYS
> TERMINATED BY '\003'
> LINES TERMINATED BY '\n'
> STORED AS TEXTFILE;
>
>
> On May 11, 2012, at 5:58 PM, Igor Tatarinov wrote:
>
>> Is that possible?
>>
>> What I am trying to do is create an S3 table using CTAS. Since CTAS doesn't
>> allow specifying a location, I have to create a managed table first:
>>
>> CREATE TABLE T AS
>> SELECT ...;
>>
>> (I don't want to fix T's schema because the list of selected expressions is
>> dynamically generated and can change.)
>>
>> Then, I want to create an s3 table like T:
>>
>> CREATE EXTERNAL TABLE S LIKE S
>> LOCATION ...;
>>
>> Unfortunately, I can't specify a different delimiter there ('\t' instead of
>> the default one). Is there another way to do that before INSERTing into the
>> S3 table?
>>
>> Thanks!
>> igor
>> decide.com
>>
>
>
