check if NSNumber is empty

Question

How do I check if a NSNumber object is nil or empty?

OK nil is easy:

NSNumber *myNumber;
if (myNumber == nil)
    doSomething

But if the object has been created, but there is no value in it because an assignment failed, how can I check this? Use something like this?

if ([myNumber intValue]==0)
   doSomething

Is there a general method for testing objects on emptiness like for NSString available (see this post)?

Example 1

NSMutableDictionary *dict = [[NSMutableDictionary alloc] init];
[dict setValue:@"" forKey:@"emptyValue"];
NSNumber *emptyNumber = [dict objectForKey:@"emptyValue"];

Which value does emptyNumber contain? How can I check if emptyNumber is empty?

Example 2

NSMutableDictionary *dict = [[NSMutableDictionary alloc] init];
[dict setValue:@"" forKey:@"emptyValue"];
NSString *myString = [dict objectForKey:@"emptyValue"];
if (myString == nil || [myString length] == 0)
    // got an empty value
    NSNumber *emptyNumber=nil;

What happens if I use this after emptyNumber was set to nil?

[emptyNumber intValue]

Do I get zero?

Example 3

NSMutableDictionary *dict = [[NSMutableDictionary alloc] init];
[dict setValue:@"" forKey:@"emptyValue"];
NSNumber *myEmptyValue = [dict objectForKey:@"emptyValue"];
if (myEmptyValue == nil)
    // NSLog is never called
    NSLog(@"It is empty!");

Like this way NSLog is never called. myEmptyValue is not nil and not NSNull. So it contains an arbitrary number?

Answer 1

NSNumber is either nil, or it contains a number, nothing in between. “Emptiness” is a notion that depends on the semantics of the particular object and therefore it makes no sense to look for a general emptiness check.

As for your examples, there are several things going on:

NSMutableDictionary *hash = [NSMutableDictionary dictionary];
[hash setObject:@"" forKey:@"key"];
NSNumber *number = [hash objectForKey:@"key"];
NSLog(@"%i", [number intValue]);

The NSLog will print 0 here, but only because there’s an intValue method in NSString. If you change the message to something that only NSNumber can do, the code will fail:

NSLog(@"%i", [number unsignedIntValue]);

This will throw:

-[NSCFString unsignedIntValue]: unrecognized selector sent to instance 0x303c

Which means you are not getting some general “empty” value back from the hash, you just get the NSString you stored there.

When you have an empty (== nil) NSNumber and send it a message, the result will be zero. That’s simply a language convention that simplifies code:

 (array != nil && [array count] == 0)
 (someNumber == nil ? 0 : [someNumber intValue])

Will turn into this:

 ([array count] == 0)
 ([someNumber intValue])

Hope this helps.

Answer 2

NSValue, NSNumber, ... are supposed to be created from a value and to always hold one. Testing for a specific value like 0 only works if it isn't in the range of valid values you are working with.

In the rare case where code is more straight-forward to work with if you have a value that represents "invalid" or "not set" and you can't use nil (e.g. with the standard containers) you can use NSNull instead.

In your first example this could be:

[dict setValue:[NSNull null] forKey:@"emptyValue"];

if ([dict objectForKey:@"emptyValue"] == [NSNull null]) {
    // ...
}

But note that you can simply not insert (or remove) that value unless you need to differentiate nil (i.e. not in the container) and, say, "invalid":

if ([dict objectForKey:@"nonExistent"] == nil) {
    // ...
}

As for the second example, -intValue gives you 0 - but simply because sending messages to nil returns 0. You could also get 0 e.g. for a NSNumber whose intValue was set to 0 before, which could be a valid value.
As i already wrote above, you can only do something like this if 0 is not a valid value for you. Note the for you, what works best completely depends on what your requirements are.

Let me try to summarize:

Option #1:

If you don't need all values from the numbers range, you could use one (0 or -1 or ...) and -intValue / ... to specifically represent "empty". This is apparently not the case for you.

Option #2:

You simply don't store or remove the values from the container if they are "empty":

// add if not empty:
[dict setObject:someNumber forKey:someKey];    
// remove if empty:
[dict removeObjectForKey:someKey];
// retrieve number:
NSNumber *num = [dict objectForKey:someKey];
if (num == nil) {
    // ... wasn't in dictionary, which represents empty
} else {
    // ... not empty
}

This however means that there is no difference between keys that are empty and keys that never exist or are illegal.

Option #3:

In some rare cases its more convenient to keep all keys in the dictionary and represent "empty" with a different value. If you can't use one from the number range we have to put something differently in as NSNumber doesn't have a concept of "empty". Cocoa already has NSNull for such cases:

// set to number if not empty:
[dict setObject:someNumber forKey:someKey];
// set to NSNull if empty:
[dict setObject:[NSNull null] forKey:someKey];
// retrieve number:
id obj = [dict objectForKey:someKey];
if (obj == [NSNumber null]) {
    // ... empty
} else { 
    // ... not empty
    NSNumber *num = obj;
    // ...
}

This option now allows you to differentiate between "empty", "not empty" and "not in the container" (e.g. illegal key).

Answer 3

NSNumbers are immutable and can only be created with either a factory method or initial method that gives them some numeric value. As far as I know it is not possible to end up with an 'empty' NSNumber, unless you count 0.