Unicode issue with HDFS

[llllllllll]

When running , I faced Unicode issues:
odo(df_pd, 'hdfstore://F://data.h5::/data/')

In pandas, we can filter it and re-code it.

types = df_pd.apply(lambda x: pd.lib.infer_dtype(df_pd.values))
types[types=='unicode']
for col in types[types=='unicode'].index:
   df_pd[col] = df_pd[col].astype(str)

Is there a way to do it in Blaze ?

repost from: blaze/blaze#1607 from @arita37

[llllllllll]

What exactly is the issue you are seeing? Is this raising an exception? Basically, what is the expected outcome and what is happening instead?

[arita37]

This is linked to Unicode string (after we imported some CSV file).
Blaze behaviour: ERROR

   odo(df_pd, 'hdfstore://F://data.h5::/data/')

  File "site-packages\pandas\io\pytables.py", line 3460, in create_axes raise e
TypeError: [unicode] is not implemented as a table column

Pandas behaviour : OK, but Warning

df_pd.to_hdf('F:/_data/stock/intraday/q5min/usetf5.h5', key='data')

site-packages\pandas\core\generic.py:1101: PerformanceWarning: 
your performance may suffer as PyTables will pickle object types that it cannot
map directly to c-types [inferred_type->unicode,key->block1_values] [items->[u'symbol']]

Solution :
Need to convert all Unicode string / Other encoding string to str type in Python, like this kind of code:

types = df_pd.apply(lambda x: pd.lib.infer_dtype(df_pd.values))
types[types=='unicode']
for col in types[types=='unicode'].index:
   df_pd[col] = df_pd[col].astype(str)

Remark 2 :
odo has PRE-DEFINED parameters for conversion...., it would be good to add **args in odo so user can pass extra arguments like in dictionnary style/ tuple for the sub-routines in other packages (pandas, bcolz,...).

Otherwise, we need to come back to pandas based solution...... which limits the usage of odo / blaze....