You seem to have solved your original problem by keeping the chapters of the book in order to begin with, but here's another approach to the original problem using "pure" numeric sorting. I think you can extract chapter headers and bodies to a hash already, so I'll start from that point.
c:\@Work\Perl\monks>perl -wMstrict -le
"my %xlation = qw(
one 1 two 2 three 3 four 4 five 5
six 6 seven 7 eight 8 nine 9 ten 10
thirty 30 thirtyone 31 thirtytwo 32
);
;;
my %chapters = (
'Chapter One' => 'There were lots of monkeys ...',
'Chapter Nine' => 'This chapter has probably 1000 words.',
'Chapter two' => 'Here is the text in the second chapter...',
'Chapter Thirty-one' => 'Every chapter is of differing length.',
'Chapter Five' => qq{A chapter can have\nlots and lots\nof lines.}
+,
);
;;
my @ordered_chapters =
map undecorate($_),
sort
map decorate($_),
keys %chapters
;
;;
my @book = map qq{-: $_ :-\n$chapters{$_}}, @ordered_chapters;
;;
print for @book;
;;
exit;
;;
;;
sub decorate {
my ($chapter) = @_;
;;
my $rx_c_num = qr{ [[:alpha:]] [-[:alpha:]]+ [[:alpha:]] }xms;
;;
my ($n_chapt) =
$chapter =~ m{ \A Chapter \s+ ($rx_c_num) \z }xmso
or die qq{malformed chapter: '$chapter'};
;;
$n_chapt = normalize_n_chapt($n_chapt);
;;
exists $xlation{$n_chapt} or die qq{unnumbered: '$chapter'};
;;
return pack 'n a*', $xlation{$n_chapt}, $chapter;
}
;;
sub undecorate { return unpack 'x[n] a*', $_[0]; }
;;
sub normalize_n_chapt {
my ($n_chapt) = @_;
;;
$n_chapt =~ s{ - }{}xmsg;
return lc $n_chapt;
}
"
-: Chapter One :-
There were lots of monkeys ...
-: Chapter two :-
Here is the text in the second chapter...
-: Chapter Five :-
A chapter can have
lots and lots
of lines.
-: Chapter Nine :-
This chapter has probably 1000 words.
-: Chapter Thirty-one :-
Every chapter is of differing length.
Update: Note that the existing code can handle chapter numbers like 'ThirtyOne' or 'Thirtyone', but not 'Thirty One'. What changes would be needed to handle such numbers?
Give a man a fish: <%-{-{-{-<