Perl 5.10, 392 384 372 235 369 (ASCII) / 234 (Unicode)
The shortest ASCII version of this is 369 characters long:
@_=(fly,spider,bird,cat,dog);$_="There was an old lady who!ed a";for$P("",",
That wrJand Jand jJinside her",",
How absurd&",",
Imagine that&",",
What a hog&"){$p=$c;$c=$".shift@_;$t=$p?"She!ed the$c to catch the$p,
$t":"I don't know why she!ed that$c,
Perhaps she'll die.
$_";$_.="$c$P.
$t";s/&/ to! a$c/}s/!/ swallow/g;s/J/iggled /g;say"$_ horse,
She died of course."
It started from this base program:
my @animals = qw(fly spider bird cat dog);
my $buf = "There was an old lady who swallowed a ";
for my $phrase ( "",
",\nThat wriggled and iggled and jiggled inside her",
",\nHow absurd&",
",\nImagine that&",
",\nWhat a hog&" ) {
$previous = $current;
$current = shift @animals;
$trail = $previous ? "She swallowed the $current to catch the $previous,\n$trail"
: "I don't know why she swallowed that $current,\n"
. "Perhaps she'll die.\n\n$buf";
$buf .= "$current$phrase.\n$trail";
$buf =~ s/&/ to swallow a $current/;
}
say "$buf horse,\nShe died of course.";
The core idea is to keep the end of the rhyme and the beginning of the next one in $trail, augmenting it as we go along. It's made non-trivial by the need of a special case for the first use, and the attempt to re-use the animal name variable even in the animal-specific phrase. Further optimizations include:
- one-character identifiers for everything
- using barewords instead of quoted strings for the animal list
- use of the accumulator
$_ for $buf to shorten most substitution operations even more (use of @_ is by force of habit and doesn't win anything more than any other character)
- including the preceding space directly inside the animal name variable (the space character taken from the
$" variable)
- regexp substitution to shorten the most common phrases:
' swallow' and 'iggled '
- no code spacing whatsoever and all
\n in string literals replaced with actual newlines
All but the last optimization yield this:
@_ = (fly, spider, bird, cat, dog);
$_ = "There was an old lady who!ed a";
for $P ( "",
",\nThat wrJand Jand jJinside her",
",\nHow absurd&",
",\nImagine that&",
",\nWhat a hog&" ) {
$p = $c;
$c = $" . shift @_;
$t = $p ? "She!ed the$c to catch the$p,\n$t"
: "I don't know why she!ed that$c,\nPerhaps she'll die.\n\n$_";
$_ .= "$c$P.\n$t";
s/&/ to! a$c/;
}
s/!/ swallow/g;
s/J/iggled /g;
say "$_ horse,\nShe died of course.";
Additionally, this golf is a victim of the underspecified encoding problem. As it—as of now—counts individual characters instead of bytes in a specified encoding, there's a big gain to be achieved by decoding the program source from UCS2 before starting. The final result isn't very readable anymore, but it's short all right. (234 characters, counted as a difference from perl -E'' as usual) (I had to include the trailing newline back to make it valid UCS2)
$ perl -MEncode=from_to -e'$_="䁟㴨晬礬獰楤敲Ɫ楲搬捡琬摯朩㬤弽≔桥牥⁷慳渠潬搠污摹⁷桯Ⅵ搠愢㭦潲⠢∬∬桡琠睲䩡湤⁊慮搠橊楮獩摥敲∬∬ੈ潷扳畲搦∬∬浡杩湥⁴桡琦∬∬桡琠愠桯朦∩笤瀽④㬤挽␢桩晴䁟㬤琽⑰㼢卨攡敤⁴桥④⁴漠捡瑣栠瑨攤瀬ਤ琢㨢䤠摯渧琠歮潷⁷桹桥Ⅵ搠瑨慴④Ⰺ健牨慰猠獨攧汬楥⸊ਤ弢㬤弮㴢④⸊⑴∻猯☯⁴漡④⽽猯ℯ睡汬潷⽧㭳⽊⽩杧汥搠⽧㭳慹∤张桯牳攬桥楥搠潦潵牳攮∊";from_to($_,utf8,ucs2);eval'
A good thing there was a lot to golf from before resorting to Unicode, or it wouldn't be much fun.
Edit: can't find a way to copy/paste the 234-character version to this browser, so I'm leaving the 235-character one. Will fix this evening, when I get my hands on a real UTF8-aware clipboard. found a way. Quasi-proof on ideone.
2Counted in UTF-8 encoded bytes? – J B – 2011-09-18T18:39:53.927
The golf clearly says "fewest characters", not bytes. Encode it how you want as long as it is still a valid program. – Thomas Eding – 2011-09-19T20:07:57.753
1For reference, the poem is 1208 characters long. – J B – 2011-09-21T08:50:44.740