Python function to map True, False, and None to symbols for display

Question

I have a use case in which I need to check whether a given value is explicitly True or False:

def stringify(value): """Returns the string representation of the value.""" if value is None: return '-' if value is True: return '✓' if value is False: return '✗' return str(value) 

I know, that in most cases, where you just need the truthiness, you'd just do if value:. But here this will not suffice, since I want zero ints and floats to be displayed in their decimal form as well as empty lists, sets dicts as such. Is the above code acceptable or should I rather check using isinstance(value, bool) first and then do if value:?

What's the least surprising way to go?

Usage Context
The function is part of an API that converts database records to terminal output:
MariaDB → peewee → API → terminal

Example output:

$ termutil ls -T 1000 ID TID CID VID OS IPv4 Address Deployed Testing Tainted Address Annotation 214 1 1000 - HIDSL latest XX.Y.Z.75 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 215 2 1000 - HIDSL latest XX.Y.Z.76 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 216 3 1000 - HIDSL latest XX.Y.Z.77 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 217 4 1000 - HIDSL latest XX.Y.Z.78 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 218 5 1000 - HIDSL latest XX.Y.Z.79 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 219 6 1000 - HIDSL latest XX.Y.Z.80 - ✓ ✗ Musterstraße 42, 123456 Irgendwo - 330 7 1000 399 HIDSL latest XX.Y.Z.182 - ✗ ✗ Musterstraße 42, 123456 Irgendwo - 508 8 1000 - HIDSL latest XX.Y.Z.189 - ✗ ✗ N/A - 

Answer 1

Because the code snippet is so small and missing any other context, it's hard to identify any problems in scaling / that may arise elsewhere in the code. Thus if you want a more tailored answer, you should add more code.

Yes, this is fine.

Regarding the isinstance() check, no. That only adds an extra level of indentation without justification.

Answer 2

Personally I'd use a dictionary, as they're pretty much the standard switch statement. If you just use a normal dictionary with just True as the key, then it has the down side of returning '✓' for both True and 1. And so you can make a dictionary with the key being a tuple of the value and the type.

_stringify_values = { (type(None), None): '-', (bool, True): '✓', (bool, False): '✗' } def stringify(value): try: return _stringify_values[(type(value), value)] except (KeyError, TypeError): return str(value) 

You can also simplify the input if you want to:

_stringify_values = { None: '-', True: '✓', False: '✗' } _stringify_values = {(type(k), k): v for k, v in _stringify_values.items()} 

---

If you plan on using this a lot, you can also make a simple class to contain all the special logic too. By using collections.abc.Mapping:

import collections class TypedMapping(collections.abc.Mapping): def __init__(self, values): self._dict = { (type(k), k): v for k, v in values } def __getitem__(self, key): try: return self._dict[(type(key), key)] except TypeError as e: raise KeyError(str(e)) from e def __len__(self): return len(self._dict) def __iter__(self): return (k for _, k in self._dict) stringify = TypedMapping([ (None, '-'), (True, '✓'), (False, '✗'), (1, 'one') ]) for i in (None, True, False, 1, 'True', {}, ()): print(stringify.get(i, str(i))) 

Answer 3

Even though return cuts the chain short if one matches, I would prefer to use an elif chain here, because it is more obvious, but YMMV (especially with automatic linters):

def stringify(value): """Returns the string representation of the value.""" if value is None: return '-' elif value is True: return '✓' elif value is False: return '✗' return str(value) 

Apart from that there is no real alternative, because all other approaches will fail on the difference between 0 and False, which you can only get with is or on unhashable types (like list).

The only other possibility I see is replacing the values for True, False, None after the conversion to str, unsing str.translate, but this would be unable to differentiate None and "None".

Answer 4

I might use nested conditional expressions, which can be formatted rather cleanly:

def stringify(value): return ('-' if value is None else '✓' if value is True else '✗' if value is False else str(value)) 

For a small set of hard-coded checks, the readability outweighs the linear search to find the matching case.

Answer 5

A remark about the function name and the docstring:

def stringify(value): """Returns the string representation of the value.""" 

• stringify: May or may not be okay, possibly to_string would be nicer.
• value: What value? Does it have to be a numeric value? A value of any kind, but no object? The variable name does not make this fully clear

Returns the string representation of the value.

This does not document what the function does, because after reading I do not know ...

• What values are allowed? Are there any restrictions?
• What defines a "string representation" of a value? for floats alone there are a lot of possible representations. Nested structures, objects and so on can have many possible representations as well, from [object] up to a full serialization.
• Why should I use this function instead of str/unicode? What is the difference?
• Is there any special handling for some data types? I do not know, that your representation of True will be an unicode character and not "True", "1", or "true" which would follow the principle of least surprise.

In addition it mixes a special handling for bool values with a default handling for everything else. Possibly it should be bool_to_unicodesymbol and raise a ValueError if the input is no boolean value.

Answer 6

To remove the ambiguity resulting from inputs like '-' and '[1, 2]', I suggest replacing your str(value) call with a repr(value) call.

If you wish to extend this to the point where a linear search might cause unnecessary delays, you will need to use the id of these singletons to avoid issues when stringifying unhashable types:

try: # Using the id() of singletons to support unhashable types return appropriately_named_lookup_table[id(value)] except KeyError: return repr(value) 

Make sure you document this if you end up doing it, and keep in mind that although it might seem like it works for integers in range(-5, 256) this is a CPython implementation detail only.