dataframe.eval() can't work with 'cash' column ?

[halleygithub]

>>> df
                    cash    sales  net_pft
STK_ID RPT_Date                           
600141 20120331   7.8427  21.0882   0.6560
       20120630   7.5140  46.9515   1.5912
       20120930   7.7667  74.0364   2.5173
       20121231  17.2350  96.1204   2.9458
       20130331  13.3643  25.0869   0.7267
       20130630  16.0832  56.6674   0.8088
       20130930  16.7898  86.3745   1.3081

[7 rows x 3 columns]
>>> df.dtypes
cash       float64
sales      float64
net_pft    float64
dtype: object
>>> df.eval('net_pft +  sales')
STK_ID  RPT_Date
600141  20120331    21.7442
        20120630    48.5427
        20120930    76.5537
        20121231    99.0662
        20130331    25.8136
        20130630    57.4762
        20130930    87.6826
dtype: float64
>>> df.eval('net_pft +  cash')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "D:\Python2.7\lib\site-packages\pandas\core\frame.py", line 1845, in eval
    return _eval(expr, **kwargs)
  File "D:\Python2.7\lib\site-packages\pandas\computation\eval.py", line 208, in eval
    ret = eng_inst.evaluate()
  File "D:\Python2.7\lib\site-packages\pandas\computation\engines.py", line 50, in evaluate
    self.pre_evaluate()
  File "D:\Python2.7\lib\site-packages\pandas\computation\engines.py", line 33, in pre_evaluate
    self.expr.check_name_clashes()
  File "D:\Python2.7\lib\site-packages\pandas\computation\expr.py", line 799, in check_name_clashes
    _check_disjoint_resolver_names(res_keys, lcl_keys, gbl_keys)
  File "D:\Python2.7\lib\site-packages\pandas\computation\expr.py", line 41, in _check_disjoint_resolver_names
    raise NameResolutionError(msg)
pandas.computation.common.NameResolutionError: resolvers and locals overlap on names ['cash']
>>> df.eval('sales +  cash')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "D:\Python2.7\lib\site-packages\pandas\core\frame.py", line 1845, in eval
    return _eval(expr, **kwargs)
  File "D:\Python2.7\lib\site-packages\pandas\computation\eval.py", line 208, in eval
    ret = eng_inst.evaluate()
  File "D:\Python2.7\lib\site-packages\pandas\computation\engines.py", line 50, in evaluate
    self.pre_evaluate()
  File "D:\Python2.7\lib\site-packages\pandas\computation\engines.py", line 33, in pre_evaluate
    self.expr.check_name_clashes()
  File "D:\Python2.7\lib\site-packages\pandas\computation\expr.py", line 799, in check_name_clashes
    _check_disjoint_resolver_names(res_keys, lcl_keys, gbl_keys)
  File "D:\Python2.7\lib\site-packages\pandas\computation\expr.py", line 41, in _check_disjoint_resolver_names
    raise NameResolutionError(msg)
pandas.computation.common.NameResolutionError: resolvers and locals overlap on names ['cash']
>>> pd.__version__
'0.13.0-203-g4518630'
>>> 

>>> df=df.rename(columns={'cash':'cas'})
>>> df
                     cas    sales  net_pft
STK_ID RPT_Date                           
600141 20120331   7.8427  21.0882   0.6560
       20120630   7.5140  46.9515   1.5912
       20120930   7.7667  74.0364   2.5173
       20121231  17.2350  96.1204   2.9458
       20130331  13.3643  25.0869   0.7267
       20130630  16.0832  56.6674   0.8088
       20130930  16.7898  86.3745   1.3081

[7 rows x 3 columns]
>>> df.eval('sales +  cas')
STK_ID  RPT_Date
600141  20120331     28.9309
        20120630     54.4655
        20120930     81.8031
        20121231    113.3554
        20130331     38.4512
        20130630     72.7506
        20130930    103.1643
dtype: float64
>>> df.eval('net_pft +  cas')
STK_ID  RPT_Date
600141  20120331     8.4987
        20120630     9.1052
        20120930    10.2840
        20121231    20.1808
        20130331    14.0910
        20130630    16.8920
        20130930    18.0979
dtype: float64
>>> 

[jreback]

the variable cash is defined in your sessino. do a del cash,then try again.

this is an API issue as it should just warn you and not raise (or optionally just ignore the error).

@cpcloud can you link the related issue

[halleygithub]

Oh, yes. I do have a cash variable and df has a cash column. So if I don't want to del cash variable and want to add the two columns by df.eval('sales + cash'), what is the valid expression ?

[jreback]

del cash delete the name from the local name space

its raising because it sees a name conflict. as I said above we are going to change this to a warning (or actually for query this would just work as it will preference the column name).

[jorisvandenbossche]

See #5987 for discussion on improving this. For now you will have to rename one of both I think.

[jreback]

closing this as already discussed in #5987