RogerBW's Blog

I’ve been doing the Weekly Challenges. The latest involved array comparisons and date calculations. (Note that this is open until 25 September 2022.)

Task 1: Unique Array

You are given list of arrayrefs.

Write a script to remove the duplicate arrayrefs from the given list.

Removing duplicates from a list is a trivial task, but raw arrays can't be used as hash keys. So I have to serialise them into a scalar variable.

In Perl this is very straightforward with a core module:

use Storable qw(freeze);

sub uniquearray($n) {
  my @o;
  my %oh;
  foreach my $i (@{$n}) {

Make the scalar version of the sublist.

    my $ti = freeze($i);

If we haven't seen it before,

    if (!exists $oh{$ti}) {

stick it on the output

      push @o,$i;

and note that we have now seen it.

      $oh{$ti} = 1;
    }
  }
  return \@o;
}

(And most of the other languages have sets rather than using hashes.)

In Raku I don't have the Storable module any more (as far as I can see), but I do have the .raku() method, which will produce a string representation of a variable (the equivalent of Perl's Data::Dumper); I suspect parsing it would be slow, but I'm only using this as a collision-free hash function, so I never have to retrieve a value.

In Python I can convert the list to a tuple. In Rust everyone says that serde is what the cool kids use, so I serialise to a JSON representation. Ruby has Marshal and JavaScript has JSON.Stringify. Which just leaves three languages in which I had to write my own serialiser: Kotlin has one but it's in the kotlinx libraries which I can't readily add on the command line (largely because documentation is very sparse; everyone assumes you're using an IDE, or at least Gradle, and I suppose I really should try the latter), and Lua and PostScript don't have them at all.

Task 2: Date Difference

You are given two dates, $date1 and $date2 in the format YYYY-MM-DD.

Write a script to find the difference between the given dates in terms of years and days only.

My approach here is to find the latest date with the month and day of date1 that falls prior to date2. The span of years from date1 to that interstitial date is the years component; the span of days from there to date2 is the days.

There is a problem with this algorithm: if date1 falls on 29 February and the interstitial date would fall in a non-leap-year, the program will generate an invalid interstitial date. The behaviour in this case is undefined anyway (what's the span from 2020-02-29 to 2021-03-01?), and fixing it would have taken effort, so I didn't.

I'm skipping Lua because of the lack of a native date library.

Rust with the chrono library:

fn datedifference(d1s: &str, d2s: &str) -> [usize; 2] {
    let mut d1 = NaiveDate::parse_from_str(d1s, "%Y-%m-%d").unwrap();
    let mut d2 = NaiveDate::parse_from_str(d2s, "%Y-%m-%d").unwrap();
    if d1 > d2 {
        let t = d1;
        d1 = d2;
        d2 = t;
    }

Get the year span assuming that d1's month and day are earlier in the year than d2's, and construct the interstitial date.

    let mut yd = d2.year() - d1.year();
    let mut inter = NaiveDate::from_ymd(d2.year(), d1.month(), d1.day());

If that date is later in the year than d2 (this could also be done as if inter > d2).

    if d1.month() > d2.month()
        || (d1.month() == d2.month() && d1.day() > d2.day())
    {

then build the previous year's version instead, and reduce the year span.

        inter = NaiveDate::from_ymd(d2.year() - 1, d1.month(), d1.day());
        yd -= 1;
    }
    [yd.try_into().unwrap(), (d2 - inter).num_days().try_into().unwrap()]
}

I can't help noticing a great variation in verbosity: in Kotlin it's

var inter = LocalDate.of(d2.getYear(), d1.getMonthValue(), d1.getDayOfMonth())

while Python has the much more compact

inter = date(d2.year, d1.month, d1.day)

Full code on github.