Different ways of comparing NSDecimalNumber

Question

For example, with primitive, I'll do this

if ( (x >= 6000) && (x <= 20000) )
    // do something here 

and with NSDecimalNumber, this is what I have

if ( (([x compare:[NSNumber numberWithInt:6000]] == NSOrderedSame) || 
        ([x compare:[NSNumber numberWithInt:6000]] == NSOrderedDescending))
    && (([x compare:[NSNumber numberWithInt:20000]] == NSOrderedSame) || 
        ([x compare:[NSNumber numberWithInt:6000]] == NSOrderedAscending)) )
{
    // do something here
}

Is there any other ways (easier and more elegant) to this comparison? If I convert the value to primitive, what primitive do I use? I don't want to use CGFloat, float or double, as I'm handling with currency here. Or if I do convert them to those mentioned above, can someone verify / explain about the precision?

Potts

Little tip: if (([x compare:[NSNumber numberWithInt:6000]] != NSOrderedAscending) && ([x compare:[NSNumber numberWithInt:20000]] != NSOrderedDescending)) gives same result. – Blida

Answer 1

My understanding is that you can only compare NSDecimalNumber and NSNumber objects using the compare: method. Super frustrating, but I believe it stems from Objective-C not supporting operator overloading.

If it's becoming really difficult to read, you could always add a category with some helper methods to try and make it a little more readable, something like this perhaps?

// NSNumber+PrimativeComparison.m

- (NSComparisonResult) compareWithInt:(int)i{
    return [self compare:[NSNumber numberWithInt:i]]
}

- (BOOL) isEqualToInt:(int)i{
    return [self compareWithInt:i] == NSOrderedSame;
}

- (BOOL) isGreaterThanInt:(int)i{
    return [self compareWithInt:i] == NSOrderedDescending;
}

- (BOOL) isGreaterThanOrEqualToInt:(int)i{
    return [self isGreaterThanInt:i] || [self isEqualToInt:i];
}

- (BOOL) isLessThanInt:(int)i{
    return [self compareWithInt:i] == NSOrderedAscending;
}

- (BOOL) isLessThanOrEqualToInt:(int)i{
    return [self isLessThanInt:i] || [self isEqualToInt:i];
}

Then things become a little more human-readable:

if([x isGreaterThanOrEqualToInt:6000] && [x isLessThanOrEqualToInt:20000]){
    //...
}

Edit I just noticed that you'd also asked about why using NSDecimalNumber is optimal in currency scenarios. This answer gives a pretty good run down on why floats (and doubles) are not precise enough when working with currency. Furthermore, Apple's documentation for NSDecimalNumber recommends its use whenever you're doing base-10 arithmetic.

Ure

"Greater than or equal" is equal to "not less". – Blida

Thanks! I was thinking about something similar to, but honestly, I still don't really like how hard it is to do this simple operation. My question about precision is more like, if I do convert them to primitive for comparison purposes, is there any one that can prove me that it is safe precision-wise. CGFloat is represented using binary format, which wouldn't be really precise when dealing with money. Thanks! – Potts

Answer 2

NSDecimalNumber+Comparison Category Code

@interface NSDecimalNumber (Comparison)

- (BOOL)isLessThan:(NSDecimalNumber *)decimalNumber;
- (BOOL)isLessThanOrEqualTo:(NSDecimalNumber *)decimalNumber;
- (BOOL)isGreaterThan:(NSDecimalNumber *)decimalNumber;
- (BOOL)isGreaterThanOrEqualTo:(NSDecimalNumber *)decimalNumber;
- (BOOL)isEqualToDecimalNumber:(NSDecimalNumber *)decimalNumber;

@end

@implementation NSDecimalNumber (Comparison)

- (BOOL)isLessThan:(NSDecimalNumber *)decimalNumber
{
    return [self compare:decimalNumber] == NSOrderedAscending;
}

- (BOOL)isLessThanOrEqualTo:(NSDecimalNumber *)decimalNumber
{
    return [self compare:decimalNumber] != NSOrderedDescending;
}

- (BOOL)isGreaterThan:(NSDecimalNumber *)decimalNumber
{
    return [self compare:decimalNumber] == NSOrderedDescending;
}

- (BOOL)isGreaterThanOrEqualTo:(NSDecimalNumber *)decimalNumber
{
    return [self compare:decimalNumber] != NSOrderedAscending;
}

- (BOOL)isEqualToDecimalNumber:(NSDecimalNumber *)decimalNumber
{
    return [self compare:decimalNumber] == NSOrderedSame;
}

@end

Unit Tests for good measure

@interface NSDecimalNumber_Comparison_Tests : XCTestCase

@end


@implementation NSDecimalNumber_Comparison_Tests

#pragma mark - isLessThan: tests

- (void)test_isLessThan_whenGreaterThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6779"];

    BOOL result = [decimalNumberA isLessThan:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isLessThan_whenLessThan_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6781"];

    BOOL result = [decimalNumberA isLessThan:decimalNumberB];

    XCTAssertTrue(result);
}

- (void)test_isLessThan_whenEqualTo_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6780"];

    BOOL result = [decimalNumberA isGreaterThan:decimalNumberB];

    XCTAssertFalse(result);
}

#pragma mark - isLessThanOrEqualTo: tests

- (void)test_isLessThanOrEqualTo_whenGreaterThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6779"];

    BOOL result = [decimalNumberA isLessThanOrEqualTo:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isLessThanOrEqualTo_whenLessThan_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6781"];

    BOOL result = [decimalNumberA isLessThanOrEqualTo:decimalNumberB];

    XCTAssertTrue(result);
}

- (void)test_isLessThanOrEqualTo_whenEqualTo_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6780"];

    BOOL result = [decimalNumberA isLessThanOrEqualTo:decimalNumberB];

    XCTAssertTrue(result);
}

#pragma mark - isGreaterThan: tests

- (void)test_isGreaterThan_whenGreaterThan_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6779"];

    BOOL result = [decimalNumberA isGreaterThan:decimalNumberB];

    XCTAssertTrue(result);
}

- (void)test_isGreaterThan_whenLessThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6781"];

    BOOL result = [decimalNumberA isGreaterThan:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isGreaterThan_whenEqualTo_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6780"];

    BOOL result = [decimalNumberA isGreaterThan:decimalNumberB];

    XCTAssertFalse(result);
}

#pragma mark - isGreaterThanOrEqualTo: tests

- (void)test_isGreaterThanOrEqualTo_whenGreaterThan_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6779"];

    BOOL result = [decimalNumberA isGreaterThanOrEqualTo:decimalNumberB];

    XCTAssertTrue(result);
}

- (void)test_isGreaterThanOrEqualTo_whenLessThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6781"];

    BOOL result = [decimalNumberA isGreaterThanOrEqualTo:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isGreaterThanOrEqualTo_whenEqualTo_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6780"];

    BOOL result = [decimalNumberA isGreaterThanOrEqualTo:decimalNumberB];

    XCTAssertTrue(result);
}

#pragma mark - isEqualToDecimalNumber: tests

- (void)test_isEqualToDecimalNumber_whenGreaterThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6779"];

    BOOL result = [decimalNumberA isEqualToDecimalNumber:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isEqualToDecimalNumber_whenLessThan_returnsNO
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6781"];

    BOOL result = [decimalNumberA isEqualToDecimalNumber:decimalNumberB];

    XCTAssertFalse(result);
}

- (void)test_isEqualToDecimalNumber_whenEqualTo_returnsYES
{
    NSDecimalNumber *decimalNumberA = [NSDecimalNumber decimalNumberWithString:@"4.6780"];
    NSDecimalNumber *decimalNumberB = [NSDecimalNumber decimalNumberWithString:@"4.6780"];

    BOOL result = [decimalNumberA isEqualToDecimalNumber:decimalNumberB];

    XCTAssertTrue(result);
}

@end

Goldfish

I wish I had four hands so I could give this answer four thumbs up. Thank you. – Rodent

Answer 3

compare method returns NSOrderedDescending, NSOrderedAscending or NSOrderedSame

Instead you can then easily write

if ( 
    [x compare:[NSNumber numberWithInt:6000]] != NSOrderedAscending && 
    [x compare:[NSNumber numberWithInt:20000]] != NSOrderedDescending
)
{
    // do something here
}

Which is slighty better readable.

Borgeson

Answer 4

if(([x doubleValue]>=6000.0f) && ([x doubleValue] <=20000.0f))

I think this will be rather preciese too.

Blida

I believe doubles only have precision of around 15 decimal places, whereas NSDecimalNumber is something like 38. Since he is dealing with currency, it's probably best to err on the side of caution and use the most precise numerical representation here. – Ure

Actually this answer better explains why floating point (and double) math is not precise enough for use in currency calculations. – Ure

In this example he compares with integers so i think [x intValue] will be fine too. There's all about what precision is needed. – Blida

If you need a plain old int why bother with the heavyweight NSDecimalNumber in the first place? – Resemble