RogerBW's Blog

I've been doing the Perl Weekly Challenges. This one dealt with months and word-wrapping.

The first one was to find months with 5 weekends, specified as 5 sequences of Friday, Saturday and Sunday all falling in the same month. (Actually the specification didn't say "sequences", it just said "5 Friday, 5 Saturday and 5 Sunday"; if you're allowed to start with Saturday and Sunday, and end with a Friday, that'll end up being rather more months.)

This is susceptible to a significant optimisation: it can only happen in a 31-day month, and only if the first day of that month is a Friday. And it will happen in every such month.

use Time::Local;
use POSIX qw(strftime);

foreach my $y (1900..2019) {

Pick just the 31-day months.

  foreach my $m (1,3,5,7,8,10,12) {

I don't actually care about the unixtime, so I'll throw it away. But the output of gmtime (localtime's neglected sibling) is what I need later for strftime.

    my @d=gmtime(timegm(0,0,0,1,$m-1,$y));

If the first day of the month is a Friday… (with the looser specification, Friday, Saturday or Sunday, so 5, 6 and 0):

    if ($d[6]==5) {
      print strftime('%B %Y',@d),"\n";
    }
  }
}

The other one is a paragraph wrapper. Obviously we already have Text::Wrap in CPAN, and indeed Text::LineFold.

use Getopt::Std;

my %o=(w => 72);
getopts('w:',\%o);

my $s=$o{w};
my @w;
while (<>) {
  chomp;

I've extended the specification a bit. If we get a blank line, count that as end of paragraph, and dump any remaining words before it.

if ($_ eq '') {
    if (@w) {
      print join(' ',@w),"\n";
      @w=();
      $s=$o{w};
    }
    print "\n";

Otherwise, it's the standard greedy algorithm: if the word will fit, fit it, otherwise dump the line buffer and start a new one.

  } else {
    foreach my $w (split ' ',$_) {
      my $lw=length($w);
      if ($lw+1 > $s) {
        print join(' ',@w),"\n";
        @w=($w);
        $s=$o{w}-$lw;
      } else {
        push @w,$w;
        $s-=($lw+1);
      }
    }
  }
}

If there's something left at the end, dump that too.

if (@w) {
  print join(' ',@w),"\n";
}

In production I'd use a function for the buffer dump, because otherwise I'd have three separate places to change the code when I needed to modify it.

I also wrote my first Perl6 programs, to do the same tasks, with much help from the Perl 5 to 6 in a nutshell guide to get me through the various syntax changes. The first challenge becomes simpler, because there's a built-in Date type, though it's not immediately obvious how to access strftime natively (do I really need to pull in DateTime::Format?) so the output is numeric for now:

for 1990..2019 -> $y {
  for 1,3,5,7,8,10,12 -> $m {
    if Date.new($y,$m,1).day-of-week == 5 {
      say "$m $y";
    }
  }
}

The second looks very similar to the way it was before, basically just with syntax changes, though command-line options are also in external modules and I'm trying to keep this basic.

my $width=72;

my $s=$width;
my @w;
for lines() {
  .chomp;
  if ($_ eq '') {
    if (@w) {
      print join(' ',@w),"\n";
      @w=();
      $s=$width;
    }
    print "\n";
  } else {
    for split ' ',$_ -> $w {
      my $lw=chars($w);
      if ($lw+1 > $s) {
        print join(' ',@w),"\n";
        @w=($w);
        $s=$width-$lw;
      } else {
        push @w,$w;
        $s-=($lw+1);
      }
    }
  }
}
if (@w) {
  print join(' ',@w),"\n";
}