Why does all columns get created as string when I use OpenCSVSerde in Hive?

Question

I am trying to create a table using the OpenCSVSerde and some integer and date columns. But the columns get converted to String. Is this an expected outcome? As a workaround, I do an explicit type-cast after this step (which makes the complete run slower)

hive> create external table if not exists response(response_id int,lead_id int,creat_date date ) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.OpenCSVSerde' WITH SERDEPROPERTIES ('quoteChar' = '"', 'separatorChar' = '\,', 'serialization.encoding'='UTF-8', 'escapeChar' = '~')   location '/prod/hive/db/response' TBLPROPERTIES ("serialization.null.format"="");
OK
Time taken: 0.396 seconds
hive> describe formatted response;
OK
# col_name              data_type               comment

response_id             string                  from deserializer
lead_id                 string                  from deserializer
creat_date              string                  from deserializer

Source Code that explains change of datatype to String.

Fugazy

Answer 1

This is known limitation of CSVSerDe serde. CSVSerDe treats all columns to be of type String. Even if you create a table with non-string column types using this SerDe, the DESCRIBE TABLE output would show string column type. The type information is retrieved from the SerDe. To convert columns to the desired type in a table, you can create a view over the table that does the CAST to the desired type.

See here: CSVSerde This confluence is about CSVSerDe, but it uses Open-CSV

Also see here: https://docs.aws.amazon.com/athena/latest/ug/csv.html

And here: Hive "OpenCSVSerde" Changes Your Table Definition

Nieshanieto

Thanks for sharing the links. Its very informative. – Fugazy

Is there a CSV serde that doesn't do this? – Expectant

@johnsullivan AFAIK LazySimpleSerDe. – Nieshanieto