RogerBW's Blog

I’ve been doing the Weekly Challenges. The latest involved splitting arrays and strings. (Note that this is open until 26 September 2021.)

Task 1: Consecutive Sub-Arrays

You are given a sorted list of unique positive integers.

Write a script to return list of arrays where the sub-arrays are consecutive integers.

Well, it's a little unexpected, but we can do that fairly cleanly. Here I'm using the Ruby version, since I haven't shown that for a while.

def csa(list)

The first sublist in the output holds the first value.

  o=[[list[0]]]

Skip that and proceed from the next value…

  1.upto(list.length-1) do |ni|

If it isn't the next expected value in a consecutive list, create a new empty sublist.

    unless o[-1][-1]+1 == list[ni] then
      o.push([])
    end

Then drop the value into the last sublist (either the newly created one or the previous one that already has content).

    o[-1].push(list[ni])
  end
  return o
end

It would be nice not to have to handle the indices directly, such as by using for n in lst, but then there'd be a conditional inside the loop to skip the first value. Or I could make a copy of the list and remove its first value. Neither of these is ideal. In the Rust and Python versions I explicitly define an iterator based on the list contents (which I hope should be implemented as a pointer rather than copying the whole list), call next() on it once to throw away the first value, then loop on what's left; I think Raku can do this too, with a bit more faff. The Rust version:

let mut i=list.iter();
i.next();
for n in i {

Ruby has things it calls "iterators", but they seem to be its term for method calls to iterate over a block, and I don't know if it has the gaps in it to let me get my pincers in between "define the things over which I will loop" and "loop over them".

Task 2: Find Pairs

You are given a string of delimiter pairs and a string to search.

Write a script to return two strings, the first with any characters matching the "opening character" set, the second with any matching the "closing character" set.

Key point here: the same character may occur in both "opening" and "closing" delimiters, in which case it's added to both output strings each time it occurs. Also there's no escaping or consideration of matched opening/closing pairs. So what I do is build a hash of "opening" and "closing" delimiter characters (a set, in the languages which support sets, which is all of them except Perl since I don't need to remove keys from the set this time), then check each character in the search string to see whether it appears in either (or both) sets.

In Perl/Raku one might reach for a regexp but punctuation characters are a pain, and one of the reasons I'm doing this in multiple languages is to get out of the habits formed by Perl where regexps and hash key lookups are ferociously optimised and one uses them for everything. Anyway, here's the Perl.

sub fp {
  my ($delims,$sample)=@_;

Iterate through the delimiter string noting opening and closing characters.

  my @d;
  foreach my $i (0..length($delims)-1) {
    $d[$i % 2]->{substr($delims,$i,1)}=1;
  }
  my @o;
  foreach my $s (split '',$sample) {

Test each character of the sample string for being in the openers and/or closers hashes, and push to a list if so.

    foreach my $x (0,1) {
      if (exists $d[$x]->{$s}) {
        push @{$o[$x]},$s;
      }
    }
  }

Turn the lists into strings.

  return [map {join('',@{$_})} @o];
}

The Raku is a bit uglier because I need to define the arrays explicitly so that they'll be mutable rather than default immutable Lists. In the Rust version, I'm especially happy with the char_indices method on the delims string, which returns a series of (index, character) tuples so that I don't have to manage the loop explicitly.

The PostScript is crude, with the open and close lists as separate variables (yeah, I should have put them in an array) and setting characters directly into fixed-length strings, but it works…

/fp {
    exch
    dup length
    dup dict /sopen exch def
    dict /sclose exch def
    /i 0 def
    {
        i 0 eq {
            sopen exch 1 put
        } {
            sclose exch 1 put
        } ifelse
        /i 1 i sub def
    } forall
    dup length
    dup string /copen exch def
    string /cclose exch def
    /iopen 0 def
    /iclose 0 def
    {
        dup
        dup sopen exch known {
            copen exch iopen exch put
            /iopen iopen 1 add def
        } {
            pop
        } ifelse
        dup sclose exch known {
            cclose exch iclose exch put
            /iclose iclose 1 add def
        } {
            pop
        } ifelse
    } forall
    cclose copen
} def

Full code on github.