Fetching of rows containing empty data still fail

[jussihi]

Describe the bug

Even after #205 was fixed, it still seems to be problematic to fetch empty data in a row.

Exception message: `RuntimeError: Buffer length for fetch (30) is smaller, & data to be retrieved is longer (31). ColumnID - 11, datatype - 12`
Stack trace:

Traceback (most recent call last):
  File "/home/jussi/GIT/**/**.py", line 131, in <module>
    MSSQL.fetch_manyrows()
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
  File "/home/jussi/GIT/**/classMSSQL.py", line 121, in fetch_manyrows
    rows = cursor.fetchall()
  File "/home/jussi/GIT/**/.venv/lib/python3.13/site-packages/mssql_python/cursor.py", line 1033, in fetchall
    raise e
  File "/home/jussi/GIT/**/.venv/lib/python3.13/site-packages/mssql_python/cursor.py", line 1017, in fetchall
    ret = ddbc_bindings.DDBCSQLFetchAll(self.hstmt, rows_data)
RuntimeError: Buffer length for fetch (30) is smaller, & data to be retrieved is longer (31). ColumnID - 11, datatype - 12

To reproduce

Fetch many rows that contain at least one column containing an emptry string/empty data

cursor = conn.cursor()
cursor.execute(query)
rows = cursor.fetchall()

Expected behavior

Return the fetched row successfully

Further technical details

Python version: 3.13.7, mssql-python 0.10.0
SQL Server version: SQL Server 2022
Operating system: WSL

Additional context

I also stumbled to this error message when I tried to patch the C++ bindings myself for bug described in #205 .

[sumitmsft]

Hi @jussihi - Thanks for reporting this issue.

We'll take a look at this issue on priority and validate why is it still failing. We will revert if we need more information from you.

Thanks again.

[bewithgaurav]

@jussihi - could you please help me with your usecase? e.g. relevant column info / usage and data sample which is being fetched? Also, just FYI - we have now released 0.11.0 of our driver last week with streaming support, could you retry the same with the newer release and share if the issue prevails? Thanks for your support!

[jussihi]

@bewithgaurav thanks for quick responses! With 0.11.0 I get the following error:

Traceback (most recent call last):
  File "/home/jussi/GIT/.../....py", line 131, in <module>
    MSSQL.getmany()
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
  File "/home/jussi/GIT/.../classMSSQL.py", line 121, in nouda_laskuttamattomat
    rows = cursor.fetchall()
  File "/home/jussi/GIT/.../.venv/lib/python3.13/site-packages/mssql_python/cursor.py", line 1832, in fetchall
    raise e
  File "/home/jussi/GIT/.../.venv/lib/python3.13/site-packages/mssql_python/cursor.py", line 1816, in fetchall
    ret = ddbc_bindings.DDBCSQLFetchAll(self.hstmt, rows_data)
RuntimeError: Error fetching LOB for column 11, cType=1, loop=1, SQLGetData return=-1

The result in Navicat (With same query copied from my python code) with a couple of first rows (attention: these are not the first columns of the result):

The problematic column seems to be one with either NULL or non-null values (now it's not failing to non-null empty any more?)

[jussihi]

EDIT: I don't quite know if the column 11 in the error is referring to is from the query result, but if it is, then the problematic column's data type is either varchar(30) or numeric(4,1) (no idea if the indexing starts from 0 or 1)

[bewithgaurav]

Thanks for the further details!
However to do a repro of this issue we'll need help with some more details, since we don't expect the driver to go to LOB fetching for varchar(30) or numeric(4,1)

1. Logging: could you please enable logging mode at the top of your import and share logs just prior to breaking point? this will clarify a bit more regarding which type/details is column 11:
   IMP: this will be a heavy log if you're working on a large data, recommended to filter out with the breaking data points only

from mssql_python import setup_logging
setup_logging('stdout')  # or `file` for file output inside `logs` dir of mssql_python package in your env

2. Schema of the table you're using: If possible, could you please share the below schema output:

   SELECT 
       ORDINAL_POSITION,
       COLUMN_NAME, 
       DATA_TYPE, 
       CHARACTER_MAXIMUM_LENGTH,
       NUMERIC_PRECISION,
       NUMERIC_SCALE,
       IS_NULLABLE,
       COLUMN_DEFAULT
   FROM INFORMATION_SCHEMA.COLUMNS 
   WHERE TABLE_NAME = 'your_tablename'
   ORDER BY ORDINAL_POSITION;

3. Repro details
   A minimal snippet of the executing code and the SQL query you’re running.

[bewithgaurav]

Hi @jussihi -
just checking in to get an update on the above details if possible..
FYI, we have since done a few more releases since 0.10.0, requesting you to please update the package to the latest version (0.13.0) using:
pip install --upgrade mssql-python