Is there a generic way to compute component values like year+month+day from days or miles+yards+feet+inches from inches?

Question

I know there must be a generic method but I don't even know the term to search for and I don't want to re-invent the wheel, badly.

'Separate components values' and 'composite values' don't give good results.

another example: degrees+minutes+seconds from seconds of arc, or tons+pounds+ounces from ounces.

I'm working in C but it shouldn't be too hard to translate from some other language.

The inputs would be the basic value, say total inches= 9534567, and the unit divisions, like feet=12, yards=feet*3, miles=yards*1760. The output would be 150 miles, 849 yards, 0 feet and 3 inches.

Surely this can be generalized to use some other set of unit divisions. And a whole library seems like overkill, doesn't it?

Answer 1

There's no generic way for every composite value. Each of the examples you gave are fairly radically different in the 'splits' for each one. The worst of course being dates due to the differences in length of months etc.

For the others, you probably could manage something along the lines of providing a list of the 'cutoffs' for each value and being returned a list of each one. Then you pretty much just have modulus and division operations to split the value down.

Answer 2

Mathematically this is integer division with remainder.

In C (and Java, ...) you can use the / and % operators to compute quotient and remainder, respectively.

Edit:

1249559005844: "6. August 2009 11:43:25 UTC"
1249559005844 % 1000 = 844
1249559005844 / 1000 = 1249559005
1249559005 % 60 = 25
1249559005 / 60 = 20825983
20825983 % 60 = 43
20825983 / 60 = 347099
347099 % 24 = 11
347099 / 24 = 14462

Above days it gets messy, because months and years have different lengths.

Answer 3

Converting between units mathematically merely involves a simple one or two step process of division or multiplication depending on which way you are going between units. A simple example is the logarithmic scale of the metric system: To go from metres to centimeters you multiply by 100 because there is 100 centimetres in a metre.

I don't know any C but all you really need to know is the conversion factor and whether you are going up or down in unit magnitude. This should work for well designed systems like the metric system where things go in a logarithmic scale but where the conversion factors differ as in the imperial system you will run in to trouble if you are trying to convert with a single factor. You could just have your program go recursively up or down different units. So you could go:

Metres  (n*100)-> cm (cm*10) -> mm (mm*1000) -> µm

And then division would just go in revers, keeping the conversion factors the same.

Moral of the story: don't use the imperial system.

Answer 4

In most cases the computations are pretty straight-forward, though it can be tedious to code all of the conversion factors.

However the computations involving duration are conceptually difficult. How many days are there in a year? Is it 365 ... or 366? How many seconds in a year?

Answer 5

As mentioned, this is more difficult when dealing with non-standard lengths of intermediate elements. However, given a standard set of breaks you could compute using the following:

void components (long value, long * parts, int parts_sz, long * result) {

    int i = 0;
    for (; i < parts_sz; ++i) {
        result[i] = value / parts[i];
        value %= parts[i];
    }
}

int main () {

    int i = 0;
    long parts[] = {10000, 1000,100,10,1}, result[] = {0,0,0,0,0};
    long value = 99853;

    components (value, parts, 5, result);

    for (; i < 5; ++i)
        printf ("There were %ld %lds\n", result[i], parts[i]);

    retrun 0;
}

Which yields:

There were 9 10000s
There were 9 1000s
There were 8 100s
There were 5 10s
There were 3 1s