Unable to read Stata value_labels from .dta-file created by pandas (to_stata()) · Issue #16923 · pandas-dev/pandas (original) (raw)

Code Sample, a copy-pastable example if possible

import pandas as pd

d = {'A':['B','E','C','A','E']} df = pd.DataFrame(data=d) df['A'] = df['A'].astype('category') # Setting as categorical, similar to value_label in Stata

df.to_stata('test.dta') # Writing dataframe to Stata-file

dfs_fromstata = pd.read_stata('test.dta', iterator=True) # Creating StataReader-object

print(dfs_fromstata.value_labels()) # Printing value_labels

Traceback (most recent call last): File "", line 1, in print(dfs_fromstata.value_labels()) # Printing value_labels File "C:\Anaconda3\lib\site-packages\pandas\io\stata.py", line 1725, in value_labels self._read_value_labels() File "C:\Anaconda3\lib\site-packages\pandas\io\stata.py", line 1329, in _read_value_labels offset = self.nobs * self._dtype.itemsize AttributeError: 'NoneType' object has no attribute 'itemsize'

Problem description

It seems as if read_stata() is not able to read value_labels properly from a .dta-file created by to_stata(). If the file is created by Stata itself, the value_labels are read correctly. Also, if the .dta-file created by to_stata() is opened and saved by Stata, the value_labels are read correctly.

Expected Output

{'A': {0: 'A', 1: 'B', 2: 'C', 3: 'E'}}

Output of `pd.show_versions()`

INSTALLED VERSIONS

commit: None
python: 3.6.0.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 78 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None