SPSS: variable_value_label becomes None if any string value is longer than 8 characters

[wilfredrial]

Hi, not sure how to tell if this is an issue in the c library or pyreadstat. Apologies if it is the former.

As the title says, the "Values" column becomes "None" for a String variable in SPSS if any value is longer than 8 characters.

To reproduce:

import pandas as pd
import pyreadstat

var_values = {'mycol':{'-111':'err1'}}

# I thought setting var_format might have an effect but it seems it does not
# just leaving this here in case you want to compare
var_format = {} # {'mycol':'A50'} 

pyreadstat.write_sav(df=pd.DataFrame({'mycol':['-111', '123456789']}),  # 2nd value is 9 chars
                     dst_path='var_values_is_none.sav', 
                     variable_value_labels=var_values,
                     variable_format=var_format)


pyreadstat.write_sav(df=pd.DataFrame({'mycol':['-111', '12345678']}), # 2nd value is 8 chars
                     dst_path='var_values_is_NOT_none.sav', 
                     variable_value_labels=var_values,
                     variable_format=var_format)

Relevant versions:
python 3.12.3
pyreadstat 1.2.7
pandas 2.2.2

[ofajardo]

Thanks for the perfectly reproducible issue!
It looks similar to this. If it is indeed the same issue, the author of the C library Readstat says that it has not been implemented yet. I do not currently see an issue about this topic in the Readstat repo, so probably it would be good to file one there.
Once fixed there, the fix should cascade to here and the problem should disappear.