ERR: disallow read_hdf mode='w'

[jkokorian]

Passing mode='w' to pd.read_hdf erases the target file immediately.

I believe this behavior is undesired, because deleting data is a side effect that no method called read_anything should have. The fact that a mode argument is even accepted is not documented anywhere. I just lost some experimental data this way, no backup...

Code Sample

import pandas as pd
import numpy as np
df = pd.DataFrame(data=np.random.rand(10))

#write some data to a new hdf5 file
df.to_hdf('test.hdf','test')

#read back the data with mode='w'
df = pd.read_hdf('test.hdf','test',mode='w')

Expected Output

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-4-5659d158fdb7> in <module>()
----> 1 df = pd.read_hdf('test.hdf','test',mode='w')

C:\Anaconda2\lib\site-packages\pandas\io\pytables.pyc in read_hdf(path_or_buf, key, **kwargs)
    328                                  'multiple datasets.')
    329             key = keys[0]
--> 330         return store.select(key, auto_close=auto_close, **kwargs)
    331     except:
    332         # if there is an error, close the store

C:\Anaconda2\lib\site-packages\pandas\io\pytables.pyc in select(self, key, where, start, stop, columns, iterator, chunksize, auto_close, **kwargs)
    660         group = self.get_node(key)
    661         if group is None:
--> 662             raise KeyError('No object named %s in the file' % key)
    663 
    664         # create the storer and axes

KeyError: 'No object named test in the file'

output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 2.7.11.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

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 23.0.0
Cython: 0.24
numpy: 1.11.1
scipy: 0.17.1
statsmodels: 0.6.1
xarray: None
IPython: 4.2.0
sphinx: 1.4.1
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: 1.0.0
tables: 3.2.2
numexpr: 2.6.0
matplotlib: 1.5.1
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.2
lxml: 3.6.0
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.40.0
pandas_datareader: None

[jreback]

this is just a passed thru argument (not in the doc-string). No checking is done on mode. Addtl args are passed thru to PyTables to support various forms of opening stores (e.g. in-memory).

I suppose it could be checked, though this will be quite tricky. Stores are allowed to be opened in multiple processes in append/read-only mode. So I guess 'w' could be disallowed if mode is supplied. But passing tests will be a bit tricky.

a pull-request would be appreciated.

[jkokorian]

I'll be happy to give it a try. However, this is the first time that I'm contributing to pandas, so I might need some guidance.

Am I correct that the check for mode='w' should be added to the read_hdf function in pandas/io/pytables.py (line 268)? How this function ends up as an instance method of DataFrame is still a mystery to me.

[jreback]

add as another named argument in read_hdf itself. you need to just disallow w if its passed, otherwise pass it thru. Need to test for passing r, r+, a (a) is the default.

contributing docs are here