Fetching CLOB via fetch_df_all raises DPY-3030 but it's documented as supported

[mauropagano]

I'm not sure this is really an issue vs me forgetting what works and what doesn't but I seem to be stuck.

As per docs, CLOBs should be converted to LARGE_STRING in Arrow

When fetching from a table with CLOB though a very fatal oracledb.exceptions.NotSupportedError: DPY-3030: conversion from Oracle Database type DB_TYPE_CLOB to Apache Arrow format is not supported is raised.

Should fetching CLOBs work?

[cjbj]

It should work. Some related tests are test test_8019, test_8053 and test_8054 in tests/test_8000_dataframe.py. Can you open an issue?

[mauropagano]

Thanks Chris, that's what I remembered.

Btw I omitted one bit and that's what causes it, apologies about it.
The column is a CLOB storing JSON via IS JSON constraint, DB is 19c so no OSON.

import oracledb

conn = oracledb.connect("...")

with conn.cursor() as c:
    c.execute("create table mauro_clob_isjson (body clob, check (body is json))")
    c.execute("insert into mauro_clob_isjson values ('{\"a\":1}')")
    conn.commit()

odf = conn.fetch_df_all(f"select body from mauro_clob_isjson")

Want me to file or is this not supported?

[anthony-tuininga]

Please file an issue. I just tried your example and it fails for me. As far as I know it should work so why it is not working is something I will investigate. Note that it fails for me with both 19c and 26ai.