Dynamic variable names from a list of strings in Python

Question

I'm coming from Matlab and need to reimplement some data processing in python/numpy. I have data files which are in a specific format, but it can vary which variables are listed in the file.

In matlab I could write functions to stick the data arrays in handy structs, like this:

s = data_to_struct(filename)
s.altitude
s.time
s.latitude
s.longitude
s.density
s.temperature

which I could then pass around to various plotting or filtering functions. I want to do the same in Python, using an object in place of the struct. I know the Proper™ way is to use a dict, but for this very specific and limited case the dict syntax is uglier in my eyes: s.temperature vs s['temperature'], as well as being more cumersome to type. What is the best way to read a string of variable names from file, and then create variables or object members with those names?

I will do a lot of interactive plotting and data handling, and I want to make the typing and tabbing as easy as possible.

Thanks, all. I'll go with setattr() for now, and make a note of NamedTuple and Pandas for future projects. — Åsmund
– Åsmund, Commented Feb 8, 2014 at 13:07

knitti · Accepted Answer · 2014-01-30 23:15:26Z

2

>>> from collections import namedtuple
>>> Something = namedtuple('Something', 'a b c')
>>> a = Something(1, 2, 4)
>>> a.c
4

namedtuple documentation

this comes especially handy, if you've already deserialized your data and it exists as a collection of multiple tuples of the same length.

answered Jan 30, 2014 at 23:15

knitti

7,03335 silver badges43 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

unutbu · Accepted Answer · 2014-01-30 23:21:42Z

2

Consider using Pandas. It has fast and flexible functions for reading data from various file formats, such as read_table, and DataFrame columns can be referenced by attribute lookup:

In [168]: df = pd.DataFrame(np.arange(12).reshape(4,3), columns=['foo', 'bar', 'baz'])

In [169]: df
Out[169]: 
   foo  bar  baz
0    0    1    2
1    3    4    5
2    6    7    8
3    9   10   11

[4 rows x 3 columns]

In [170]: df.bar
Out[170]: 
0     1
1     4
2     7
3    10
Name: bar, dtype: int32

answered Jan 30, 2014 at 23:21

unutbu

886k197 gold badges1.9k silver badges1.7k bronze badges

Comments

jdotjdot · Accepted Answer · 2014-01-30 23:28:35Z

The one problem I see with namedtuple is that you can't edit the values of the object after it's been instantiated, which I don't know if you need to do or not.

While it might be a bit convoluted, you could create an object using setattr and basically create a factory that will generate objects with your custom attributes.

An example would look like this:

def data_to_struct(*info):
    class O(object):
        def __init__(self, *args):
            for a in args:
                setattr(self, a, None)
    return O(*info)

Then, using it, you'd pass data_to_struct a list of strings that will be your attribute names and get back an object that has all of those as empty attributes.

>>> g = data_to_struct('a', 'b', 'c')
>>> g.a
>>> g.b = 3
>>> g.b
3

You could of course always choose to use kwargs also or something else to provide values to the various attributes upon instantiation.

Brave Sir Robin · Accepted Answer · 2014-01-30 23:13:26Z

0

Instead of obj.attr = value you can also write setattr(obj, 'attr', value) which takes the attribute name as a string.

answered Jan 30, 2014 at 23:13

Brave Sir Robin

1,0466 silver badges9 bronze badges

Collectives™ on Stack Overflow

Dynamic variable names from a list of strings in Python

4 Answers 4

Comments

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

Comments

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related