I have in my dataframe one column with cx_Oracle.LOB object [2D polygon type], made with a SDO_UTIL.TO_WKTGEOMETRY SQL function and I want to create a geodataframe, and set my geometry column the geometry returned from the SDO_UTIL.TO_WKTGEOMETRY function to a valid geodataframe listlike geometry.

The Geodataframe function return error when i set the geometry column with the WKTGEOMETRY made with SDO_UTIL.TO_WKTGEOMETRY function. I work with PyCharm application.

my dataframe:

dfparc = pd.DataFrame(dfparc.values, columns=['PAR_ID', 'PAR_NUM', 'geometry', 'AREA_HA', 'DCF'])


class 'cx_Oracle.LOB'>


POLYGON ((225215.22 411994.91, 225214.02 411912.51, 225339.62 411916.91, 225339.62 411995.31, 225279.22 411995.31, 225215.22 411994.91))

My Geodataframe:

geoparc = gp.GeoDataFrame(dfparc, crs=102160, geometry=dfparc.geometry)

returned errors:

...Python35-32\lib\site-packages\geopandas\geodataframe.py", line 50, in __init__
     self.set_geometry(geometry, inplace=True)
...Python35-32\lib\site-packages\geopandas\geodataframe.py", line 149, in set_geometry
    raise TypeError("Input geometry column must contain valid geometry objects.") TypeError: Input geometry column must contain valid geometry objects.

I suppose the problem could be in the type of object returned from SDO_UTIL.TO_WKTGEOMETRY function that is not available for geodataframe geometry data type.

  • We might need to see your code populating dfparc with the SQL used. As I don't have the environment to test, my guess is converting the CLOB to a string (VARCHAR) might help. You can do this in the SQL statement or with Python. – Richard Morgan Feb 26 '19 at 21:22
  • here is my sql statement made to a oracle spatial (client profile): queryparcel = "select PAR_ID, PAR_NUM, SDO_UTIL.TO_WKTGEOMETRY(PAR_POLIGONO) as geometry, round(par_area/10000,3) as AREA_HA, DCF from sig_parcela_vw where PAR_NUM IN ('4117654001', '4126925016')" – Paulo Adão Feb 27 '19 at 12:51
  • then i set the SQL result to cursor: result = cursor.execute(queryparcel) cursor.arraysize = 500 – Paulo Adão Feb 27 '19 at 12:53
  • then i create a dataframe with the cursor: dfparc = pd.DataFrame(result) – Paulo Adão Feb 27 '19 at 12:53
  • after that i set the Dataframe columns names: dfparc = pd.DataFrame(dfparc.values, columns=['PAR_ID', 'PAR_NUM', 'geometry', 'AREA_HA', 'DCF']) – Paulo Adão Feb 27 '19 at 12:55

I hesitate to give this answer as I cannot test it, however, one way to avoid dealing with CLOB data types is to convert the result to VARCHAR2.

From the help page on SDO_UTIL.FROM_WKTGEOMETRY (you are calling the TO_ but the usage notes on the FROM_ page reference the TO_):

You can use the SQL function TO_CHAR to convert the resulting CLOB to VARCHAR2 type.

This would make your SQL look something like:

select PAR_ID, PAR_NUM, 
  round(par_area/10000,3) as AREA_HA, DCF 
from sig_parcela_vw 
where PAR_NUM IN ('4117654001', '4126925016')
  • Now i have the BLOB objet converted to VARCHAR. GREAT! Thanks! – Paulo Adão Feb 27 '19 at 13:41
  • But the geometry column construction in Geodataframe still return the error "TypeError: Input geometry column must contain valid geometry objects." – Paulo Adão Feb 27 '19 at 13:42
  • Iam triyng to find out where is the problem with the geometry column in Geodataframe. – Paulo Adão Feb 27 '19 at 13:44
  • @PauloAdão You don't have to accept my answer if it didn't help. People may not read your question if this answer is accepted. Feel free to remove the accepted check mark and update your original question. – Richard Morgan Feb 27 '19 at 14:10
  • 1
    Your answer fulfill the object type conversion. That was great!. One step at the time... :) – Paulo Adão Feb 27 '19 at 14:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.