Re: Formatting a long decimal into Gb, Mb, or Kb (using String.format)
Piotr Kobzda wrote:
There is still a mistake in your implementation. As pointed out by
Eric, "K" means something different than "k".
A base for "K" it is 1024, not 1000!
I was wondering whether anybody would call me on this! Switching to
capital 'B' was an easy decision. For the letter prefixes though I
decided to go with common usage and consistency (GB, MB, kB???)
incorrect as it may be. Gi, Mi, etc., is simply too pedantic for
my audience, who are other programmers here at work who under-
stand the difference between powers of 10 and powers of 2 so I'm
unlikely to invite lawsuits for misleading information! Users first!
The following bit of code is wonderful! I never get to fiddle bits
in my programs and any chance to use "<<" turns me on. With a
little head scratching I actually understand it. But whoever
inherits my code after me would curse my name if I swapped this
in for the few lines of code I now have. Clever programmers. Bah!
I've only been programming Java for about a month now and your
complete implementation is dripping with Java-y goodness. It'll
be great to refer to for tips on good programming and best
practices. I've never ever seen a class with a signature beginning
with "public enum"!
Consider also using the following approach for your purposes:
public enum StorageUnit {
BYTE ( "B", 1L),
KILOBYTE ("KB", 1L << 10),
MEGABYTE ("MB", 1L << 20),
GIGABYTE ("GB", 1L << 30),
TERABYTE ("TB", 1L << 40),
PETABYTE ("PB", 1L << 50),
EXABYTE ("EB", 1L << 60);
public static final StorageUnit BASE = BYTE;
private final String symbol;
private final long divider; // divider of BASE unit
StorageUnit(String name, long divider) {
this.symbol = name;
this.divider = divider;
}
public static StorageUnit of(final long number) {
final long n = number > 0 ? -number : number;
if (n > -(1L << 10)) {
return BYTE;
} else if (n > -(1L << 20)) {
return KILOBYTE;
} else if (n > -(1L << 30)) {
return MEGABYTE;
} else if (n > -(1L << 40)) {
return GIGABYTE;
} else if (n > -(1L << 50)) {
return TERABYTE;
} else if (n > -(1L << 60)) {
return PETABYTE;
} else { // n >= Long.MIN_VALUE
return EXABYTE;
}
}
public String format(long number) {
return nf.format((double)number / divider) + " " + symbol;
}
private static java.text.NumberFormat nf
= java.text.NumberFormat.getInstance();
static {
nf.setGroupingUsed(false);
nf.setMinimumFractionDigits(0);
nf.setMaximumFractionDigits(1);
}
}
Typical usage is as follows:
StorageUnit.of(number).format(number);
Note that there are still some units of information defined by the SI
not handled here, i.e. bits (and its derived units), zettabytes (ZB),
and yottabytes (YB). Handling them all together requires a bit more
advanced approach (likely something near to the JSR-275 proposals), but
I think you don't need to do that.
piotr