Value transfered to next row when Null

[RomainDGrey]

Hello all,

When I try to fetchall the result from a select request, some rows, with null values, shows value from previous line.

Context:

• I am using a HFSQL Database and the driver of the same name, developed by the WinDev company for it.
• Python3.11
• pypyodbc version 1.3.6

Example:

import pypyodbc
DB_URI="DRIVER={HFSQL};Server Name=some.server;Server Port=4900; Database=some_database;UID=admin;PWD=some_pass"
conn = pypyodbc.connect(DB_URI)
cursor = conn.cursor()
result = cursor.execute("SELECT CODE from CWS_BASE WHERE id in ('66783')")
result.fetchall()
result = cursor.execute("SELECT CODE from CWS_BASE WHERE id in ('120140', '66783')")
result.fetchall() 

Output:

[('',)]
[('BLOCAGE',), ('BLOCAGE',)]

Is it linked to the driver or is it linked to the parsing feature of pypyodbc ?

[RomainDGrey]

After more investigation, it seems that it's not link to the driver as this powershell script sucessfully retrieve the information :

---
(venv) PS C:\Users\R\Documents\projects\test> $result =(new-Object Data.Odbc.OdbcCommand("SELECT CODE from CWS_BASE WHERE id in ('120140','66783')",$conn)).ExecuteReader()
(venv) PS C:\Users\R\Documents\projects\test> while ($result.Read()) {
>>     $value = $result.GetValue(0)
>>     Write-Output $value
>> }
BLOCAGE 

---

How can I check where in pypyodbc where the value is added if it contains nothing ?

[Replect]

Did you potentially find a solution for this?
I just realized we have the same issue (also HFSQL server, pypodbc version 1.3.6)... doesn't seem like there has been a new version to this library since...

It might happen somewhere within the fetchone function of pypodbc, based on what I've seen.

[RomainDGrey]

@Replect Test the 1.3.x branch, from what I remember it fixes my issue

[Replect]

You mean a previous version to 1.3.6?

EDIT: The branch 1.3.x didn't solve the issue. Or is there anything else to replace beside the pypyodbc.py?

[Replect]

Okay, I didn't see any version of the pypyodbc.py here to fix the issue, but I think we got it resolved via the following adjustment in the pypyodbc.py (version 1.3.6):

At line 1883, replace the following code:

if used_buf_len.value == SQL_NULL_DATA:
    value_list.append(None)
else:
    if raw_data_parts == []:

with this:

if used_buf_len.value <= SQL_NULL_DATA:
    value_list.append(None)
elif used_buf_len.value == 0:
    value_list.append('')
else:
    if raw_data_parts == []: