sqoop import SDO_GEOMETRY type failing

[#1]

I am trying to import a SDO_GEOMETRY data type from oracle DB to hadoop but its populating null if I

import with option --map-column-java SHAPE=String. is there any other data type I can use for it? or is there a better way to import into hadoop/hive?

[#2]

I used spark for this job and basically I tried to create temporary columns and then union ther values at the end of the job as one column.

Here is how I basically created the temporary columns:

select *
from (with source_tbl as (
    select a.*, dbms_lob.getlength(SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(a.shape, 4326))) shape_len
    from shape_tbl a)
      select source_tbl.*,
             TO_CHAR(SUBSTR(SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(shape, 4326)), 0, 4000))             shape_part1,
             case
                 when shape_len > 4000 then TO_CHAR(
                         SUBSTR(SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(shape, 4326)), 4001, 4000)) end  shape_part2,
             case
                 when shape_len > 8000 then TO_CHAR(
                         SUBSTR(SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(shape, 4326)), 8001, 4000)) end  shape_part3,
             case
                 when shape_len > 12000 then TO_CHAR(
                         SUBSTR(SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(shape, 4326)), 12001, 4000)) end shape_part4
             -- other cases keep increasing the length condition by 4000
      from source_tbl);

I see this more as a workaround since usually these SDO_GEOMETRY objects are really big and trying to cover them with strings of length up to 4000 can lead to hundreds of columns.

[#3]

Actually you can do it better, and import it as a clob, and in spark works perfect:

SELECT SDO_UTIL.TO_WKTGEOMETRY(SDO_CS.TRANSFORM(shape, 4326)) AS SHAPE, FROM oracle_tbl