How to use pandas.read_sql but keep RowVersion formatting?

Question

I've been struggling since the morning with this. I have a SQL table with a RowVersion column, that I'm creating using the following code:

mycursor = cnxn.cursor()
        
s = """
CREATE TABLE table_name (
  PersonID int PRIMARY KEY,
  Name varchar(255),
  RowVersion rowversion);
  """
        
mycursor.execute(s)
        
sql = "INSERT INTO table_name (PersonID, Name) VALUES (?, ?)"
val = [
       (1, 'Andres'),
       (2, 'Carl'),
       (3, 'Tracy'),
       ]
        
        
mycursor.executemany(sql, val)
cnxn.commit()

Output table:

PersonID	Name	RowVersion
1	Andres	AAAAAAAAD70=
2	Carl	AAAAAAAAD74=
3	Tracy	AAAAAAAAD78=

But once I read the table using pd.read_sql or pd.read_sql_query the column changes to the following:

dfTable = pd.read_sql("SELECT * FROM table_name", cnxn, coerce_float=False)
dfTable.head()

PersonID	Name	RowVersion
1	Andres	b'\x00\x00\x00\x00\x00\x00\x0f\xba'
2	Carl	b'\x00\x00\x00\x00\x00\x00\x0f\xbb'
3	Tracy	b'\x00\x00\x00\x00\x00\x00\x0f\xbc''

What I would like to get is to keep the RowVersion format as before. According to the pd.read_sql documentation there is a way to avoid converting non-string values to floating points, but since it is a hex value pandas is reading as a string.

Is there a workaround to find a solution?

Thank you,

Answer 1

I ended up reading and altering the table again after uploading the Dataframe as table...

s = """
ALTER TABLE [dbo].[table_name] ADD rv rowversion NULL
"""

mycursor.execute(s)
cnxn.commit()

Please, feel free to answer if there is another solution, ty.