CODEWITHSUNDEEP
pythonpython-3.xpython-3.3
Patt Mehta
Patt Mehta

Reputation: 4204

Using True/False as keys - how/why does this work?

I am comfortable using this simple syntax for initializing a dictionary 

d = {'a':'Apple','b':'Bat'};

Today, while reading a page, I encountered this weird piece of code 

{True:0, False:1}[True];

I was wondering why/how that could work? True and False are reserved keywords, and so, that crazy syntax should be meaningless (for the compiler), but it is not. 

>>> d = {True:0, False:1};
>>> d
{False: 1, True: 0}

And this gets crazier 

>>> d = dict(True = 0, False = 1);
SyntaxError: assignment to keyword
>>> d = dict(_True = 0, _False = 1);
>>> d
{'_False': 1, '_True': 0}

In dict() constructor, True keyword is not allowed! But...

Update

Python 3.3.2 (v3.3.2:d047928ae3f6, May 16 2013, 00:03:43) [MSC v.1600 32 bit (Intel)] on win32
Type "copyright", "credits" or "license()" for more information.
>>> import keyword
>>> keyword.iskeyword('print');
False
>>> keyword.iskeyword('else');
True
>>> keyword.iskeyword('True');
True
>>> keyword.iskeyword('False');
True

Upvotes: 15

Views: 25923

Answers (3)

Blckknght
Blckknght

Reputation: 104842

While they are keywords (in Python 3), True and False are still names of objects (they are, respectively bool(1) and bool(0)).

So you can use them anywhere a value makes sense. Since they're hashable, that includes using them as dictionary keys. You can do:

d = {}
d[True] = "True"
d[False] = "False"

You can get the same dictionary directly with curly brackets (d = {True: "True", False: "False"}), but not using the dict constructor with keyword arguments. The keyword version of the constructor only builds dicts with keys that are strings, and those strings must be legal Python identifiers. True and False are not strings, nor (since they're keywords) can they be used as identifiers.

You can of course use the sequence of key/value tuples constructor, if there's some reason you need to be calling dict rather than using brackets:

d = dict([(True, "True"), (False, "False")])

I would note that the code fragment you show is a bit silly. The bool type in Python is a subclass of int, so you can do mathematical operations on True or False if you want. The dictionary indexing code is equivalent to 1-True, which will be 0.

Upvotes: 16

mipadi
mipadi

Reputation: 411252

True and False aren't keywords -- they're actual objects. You can verify this in the Python interpreter (using 2.7 here, but the same holds in 3.x):

Python 2.7.6 (default, Jan 29 2014, 21:22:07) 
[GCC 4.2.1 Compatible Apple LLVM 5.0 (clang-500.2.79)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> type(True)
<type 'bool'>
>>> True.__class__.__name__
'bool'
>>> type(False)
<type 'bool'>
>>> False.__class__.__name__
'bool'
>>> hash(True)
1
>>> hash(False)
0
>>> True.__hash__
<method-wrapper '__hash__' of bool object at 0x100134da0>
>>> False.__hash__
<method-wrapper '__hash__' of bool object at 0x100134db8>

So they can be used as keys in a dict or similar structure.

In fact, prior to Python 3, you could use True and False as parameter names to dict():

>>> d = dict(True="true", False="false")
>>> d
{'False': 'false', 'True': 'true'}

However, to avoid confusion, the Python 3 interpreter prevents you from doing that, as it essentially treats them like keywords now -- but really, they're still objects. (I guess you could say they're both, sort of.)

Upvotes: 2

Daniel Roseman
Daniel Roseman

Reputation: 600041

Why should it be meaningless? True and False are values, not keywords. That's why you can compare other values with them.

You can't use integers in the dict constructor, either. That's a limitation of keyword arguments, not of dictionaries.

Upvotes: 4

Related Questions