dataframe.query() select columns

[dingo9]

In function query@pandas/core/frame.py

I found dataframe return eval result, and use self.loc to return new dataframe, and I curious about in which situation dataframe.loc will raise ValueError.

        inplace = validate_bool_kwarg(inplace, 'inplace')
        if not isinstance(expr, compat.string_types):
            msg = "expr must be a string to be evaluated, {0} given"
            raise ValueError(msg.format(type(expr)))
        kwargs['level'] = kwargs.pop('level', 0) + 1
        kwargs['target'] = None
        res = self.eval(expr, **kwargs)

        try:
            new_data = self.loc[res]
        except ValueError:
            # when res is multi-dimensional loc raises, but this is sometimes a
            # valid query
            new_data = self[res]

        if inplace:
            self._update_inplace(new_data)
        else:
            return new_data

Problem description

Because I use Dataframe.query() as data_model api, and I wanna select columns from this api, so I do some googling, found nothing. After read code of query, I found interesting fragment in function query. If loc raise Exception, select column instead. BUT, it catch ValueError, not KeyError.

Version

In current pandas version query@pandas/core/frame.py ( https://github.com/pandas-dev/pandas/commit/20fda2223d5121be3f8204702b5ce1e6037e5b18)

[jorisvandenbossche]

@dingo9 Please provide a minimal, reproducible example (runnable code snippet) illustrating the problem you have.

[dingo9]

import pandas as pd
import numpy as np
df = pd.DataFrame(np.random.randn(6,4),columns=list('ABCD'))
df.query("['A']")

I expect it return columns 'A' in dataframe, but it raise KeyError at

        try:
            new_data = self.loc[res]
        except ValueError:
            # when res is multi-dimensional loc raises, but this is sometimes a
            # valid query
            new_data = self[res]

In such fragment, if it raise ValueError, it will return columns 'A' as I expect.

[jreback]

pls read the docs: http://pandas.pydata.org/pandas-docs/stable/indexing.html#the-query-method-experimental

In [7]: df = pd.DataFrame(np.random.randn(6,4),columns=list('ABCD'))
   ...: df.query("A>0")

Out[7]: 
          A         B         C         D
5  1.105172  1.219918 -1.555644  0.937555

In [8]: df.eval('A')
Out[8]: 
0   -1.534979
1   -0.328682
2   -0.181169
3   -1.026832
4   -0.984711
5    1.105172
dtype: float64