HDFStore fails to read non-ascii characters

[FilipDusek]

When I try to save some non-ascii character like é and then load it again, I end up with UnicodeDecodeError. If you add some more data to the string (like 'aée'), the data gets stored and retrieved without error, but the result is missing the last character.

import pandas as pd

df = pd.DataFrame(columns=["A"])
toAppend = {"A": "é"}
df = df.append(toAppend, ignore_index = True)

store = pd.HDFStore(r'thiswillcrash.h5')
store.put('df', df, format='table', encoding="utf-8")
d = store["df"]
print(d)

store.close()

Versions

INSTALLED VERSIONS
------------------
commit: None
python: 3.4.3.final.0
python-bits: 64
OS: Windows
OS-release: 8
machine: AMD64
processor: Intel64 Family 6 Model 69 Stepping 1, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None

pandas: 0.16.2
nose: 1.3.4
Cython: 0.22
numpy: 1.9.3
scipy: 0.15.1
statsmodels: 0.6.1
IPython: 3.0.0
sphinx: 1.2.3
patsy: 0.3.0
dateutil: 2.4.2
pytz: 2015.6
bottleneck: None
tables: 3.2.2
numexpr: 2.4.4
matplotlib: 1.4.3
openpyxl: 1.8.5
xlrd: 0.9.3
xlwt: None
xlsxwriter: 0.6.7
lxml: 3.4.4
bs4: 4.3.2
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 0.9.9
pymysql: None
psycopg2: None

[jreback]

should be fixed by : pydata#10889

give a try with v0.17.0rc2

conda install pandas -c pandas

[FilipDusek]

No, unfortunately I still get the error

---------------------------------------------------------------------------
UnicodeDecodeError                        Traceback (most recent call last)
<ipython-input-3-3e6096eba1ca> in <module>()
      8 store = pd.HDFStore(r'iwillcrash30.h5')
      9 store.put('df', df, format='table', encoding="utf-8")
---> 10 d = store["df"]
     11 print(d)
     12 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in __getitem__(self, key)
    424 
    425     def __getitem__(self, key):
--> 426         return self.get(key)
    427 
    428     def __setitem__(self, key, value):

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in get(self, key)
    634         if group is None:
    635             raise KeyError('No object named %s in the file' % key)
--> 636         return self._read_group(group)
    637 
    638     def select(self, key, where=None, start=None, stop=None, columns=None,

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in _read_group(self, group, **kwargs)
   1271         s = self._create_storer(group)
   1272         s.infer_axes()
-> 1273         return s.read(**kwargs)
   1274 
   1275 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in read(self, where, columns, **kwargs)
   4004     def read(self, where=None, columns=None, **kwargs):
   4005 
-> 4006         if not self.read_axes(where=where, **kwargs):
   4007             return None
   4008 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in read_axes(self, where, **kwargs)
   3216         for a in self.axes:
   3217             a.set_info(self.info)
-> 3218             a.convert(values, nan_rep=self.nan_rep, encoding=self.encoding)
   3219 
   3220         return True

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in convert(self, values, nan_rep, encoding)
   2062         if _ensure_decoded(self.kind) == u('string'):
   2063             self.data = _unconvert_string_array(
-> 2064                 self.data, nan_rep=nan_rep, encoding=encoding)
   2065 
   2066         return self

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\io\pytables.py in _unconvert_string_array(data, nan_rep, encoding)
   4430 
   4431         if isinstance(data[0], compat.binary_type):
-> 4432             data = Series(data).str.decode(encoding).values
   4433         else:
   4434             data = data.astype(dtype, copy=False).astype(object, copy=False)

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\core\strings.py in decode(self, encoding, errors)
   1310     @copy(str_decode)
   1311     def decode(self, encoding, errors="strict"):
-> 1312         result = str_decode(self.series, encoding, errors)
   1313         return self._wrap_result(result)
   1314 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\core\strings.py in str_decode(arr, encoding, errors)
    979     """
    980     f = lambda x: x.decode(encoding, errors)
--> 981     return _na_map(f, arr)
    982 
    983 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\core\strings.py in _na_map(f, arr, na_result, dtype)
    119 def _na_map(f, arr, na_result=np.nan, dtype=object):
    120     # should really _check_ for NA
--> 121     return _map(f, arr, na_mask=True, na_value=na_result, dtype=dtype)
    122 
    123 

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\core\strings.py in _map(f, arr, na_mask, na_value, dtype)
    135         mask = isnull(arr)
    136         try:
--> 137             result = lib.map_infer_mask(arr, f, mask.view(np.uint8))
    138         except (TypeError, AttributeError):
    139             def g(x):

pandas\src\inference.pyx in pandas.lib.map_infer_mask (pandas\lib.c:61753)()

C:\Users\Filip\Anaconda3\lib\site-packages\pandas\core\strings.py in <lambda>(x)
    978     decoded : Series/Index of objects
    979     """
--> 980     f = lambda x: x.decode(encoding, errors)
    981     return _na_map(f, arr)
    982 

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc3 in position 0: unexpected end of data

Versions

INSTALLED VERSIONS
------------------
commit: None
python: 3.4.3.final.0
python-bits: 64
OS: Windows
OS-release: 8
machine: AMD64
processor: Intel64 Family 6 Model 69 Stepping 1, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None

pandas: 0.17.0rc2
nose: 1.3.7
pip: 7.1.2
setuptools: 18.3.2
Cython: 0.22.1
numpy: 1.9.3
scipy: 0.15.1
statsmodels: 0.6.1
IPython: 3.2.0
sphinx: 1.3.1
patsy: 0.3.0
dateutil: 2.4.2
pytz: 2015.6
blosc: None
bottleneck: 1.0.0
tables: 3.2.0
numexpr: 2.4.3
matplotlib: 1.4.3
openpyxl: 1.8.5
xlrd: 0.9.3
xlwt: 1.0.0
xlsxwriter: 0.7.3
lxml: 3.4.4
bs4: 4.3.2
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.5
pymysql: None
psycopg2: None

[TomAugspurger]

@jreback looks like we truncate the column to be length 1 since len(df.iloc[0, 0]) is 1.

This works though

In [19]: df = pd.DataFrame({'A': ['é']})

In [20]: store = pd.HDFStore(r'thiswillcrash.h5')

In [21]: store.put('df', df, format='table', min_itemsize={'A': 30})

In [22]: store.get('df')
Out[22]:
   A
0  é

Do you have a good idea where a fix would go?

[jreback]

https://github.com/pydata/pandas/blob/master/pandas/lib.pyx#L972

is where the width of the strings are determined
but it should work for unicode

[TomAugspurger]

Is it because the encoded length is different than the number of characters?

In [10]: x
Out[10]: 'é'

In [11]: len(x)
Out[11]: 1

In [12]: len(x.encode('utf-8'))
Out[12]: 2

[jreback]

yep should encode before we check and set the length

[jreback]

closed by #11240