Simplify long if-else chain with regex by comparing with array - arrays

Question

for (my $j = 0; $j <100000; $j++){
my $outcome = rand();
for (my $k = 0; $k < #cum_sum; $k++){
if ($cum_sum[$k] >= $outcome){
if ($keys[$k] =~ m/\"|\/|\<|\>|\\|\`|\~|\#|\#|\$|\%|\^|\*|[0-9]/) {
print $out "";
}
if ($keys[$k] =~ m/\s/){
print $out " ";
}
elsif ($keys[$k] =~ m/\&/){
print $out " and ";
}
elsif ($keys[$k] =~ m/\!/){
print $out "! ";
}
elsif ($keys[$k] =~ m/\:/){
print $out ": ";
}
elsif ($keys[$k] =~ m/\'/){
print $out "' ";
}
elsif ($keys[$k] =~ m/\./){
print $out ". ";
}
elsif ($keys[$k] =~ m/\;/){
print $out "; ";
}
elsif ($keys[$k] =~ m/\?/){
print $out "? ";
}
elsif ($keys[$k] =~ m/\,/){
print $out ", ";
}
else {
print $out "$keys[$k]";
}
last;
}
}
# print "$outcome\n";
}
I mostly need help with simplifying the long chain of elsif statements I have.
The logic in the outer for loops works.
#keys is an array of two character (digrams) strings.
I am trying to make the elsif statements more efficient by comparing the digrams from #key to an array of the punctuation marks #punctuation = qw(! : " ' ; ? , .)
Then, if the digram does contain one of the punctuation marks in the punctuation array, the digram gets changed to "punctuation_mark " e.g. "a!" -> "! "
The end result would be that I do not use regexes for any of #punctuation.
However, I am not sure on how to implement this change.
Thank you!

Answer 1

In order to simplify your code, it may have sense to use the Tie::RegexpHash CPAN module. The main idea is build a hash with regexpr as keys, so you easily find the related values by matching:
use Tie::RegexpHash;
my $rehash = Tie::RegexpHash->new();
$rehash->add( qr/\s/, " " );
$rehash->add( qr/\&/, " and " );
#...
my $value = $rehash->match( "&" ); # $value <-- " and "

Answer 2

Use alternation, just like in the first regex, but capture the match and use $1 to replace the word
my $re_punct = join '|', map { quotemeta } qw(& ! : ' . ; ? ,); #'
for my $j (0..99_999) {
my $outcome = rand();
for my $k (0..$#cum_sum) {
...
if ($keys[$k] =~ /($re_punct)/) {
if ($1 eq '&') { $keys[$k] = " and " }
else { $keys[$k] = "$1 " }
}
...
}
}
Comments
quotemeta escapes by \ all "ASCII non-"word" characters"
syntax: $#ary is the index of the last element in #ary, just right for looping over array index
for my $i ($beg .. $end) is much clearer than the equivalent C-style for loop†
The presented logic leaves a question: what if both characters are punctuation?
Note on your idea to do it "By Comparing With Array"
You could use List::MoreUtils::first_value, for instance. It would go like
use List::MoreUtils qw(first_value);
my #punc = map { quotemeta } qw(& ! : ' . ; ? ,); #'
foreach my $word (#words) {
if (my $match = first_value { $word =~ /$_/ } #punct) {
$word = $match;
}
}
The first_value (or firstval) returns the first element of #punct for which the block returns true, and undef if none do. The $word aliases the currently processed element of #words and changing it changes the array element; so you get your replacement.
However, you still have to deal with regex and escape (at least some of) punctuation. Thus I see no advantage in going to this trouble; the "straight-up" regex is far clearer in this case.
† Even compiled languages evolve this way. The C++11 introduced the range-based for loop
for (auto var: container) ... // (really, const auto&), or auto&, or auto&&
and a standard reference linked above says
Used as a more readable equivalent to the traditional for loop [...]
In Perl this is how things are done; just use it.

Answer 3

So, looking at it you have two cases:
One where you replace a set of values, with 'the value and a space'. And another where you replace with something different.
So how about creating a lookup table for each, and processing just two regexes:
#!/usr/bin/env perl.
use strict;
use warnings;
use Data::Dumper;
my %replace = (
'"' => "",
'/' => "",
'&' => " and ",
);
my #add_space = ( ',', '?', ';', '.', "'", ':', '!' );
my $search = join ( "|", map { quotemeta } keys %replace );
$search = qr/($search)/;
my $add_space_after = join "|", map {quotemeta} #add_space;
$add_space_after = qr/($add_space_after)/;
while ( <DATA> ) {
s/\s/ /g;
s/$search/$replace{$1}/;
s/$add_space_after/$1 /;
print;
}
__DATA__
Work:Work
cookies&milk;wordhere
why?are;you/so "sad"
This gives us what you want I think, and - hopefully - keeps the code pretty conscise.
Of importance is the quotemeta function here, because it ensures your metachars are escaped before regexing them.
Note - the only one of your examples this doesn't handle is the \s to " ". But that's IMO probably best to write separately for clarity, as obviously youc can't quotemeta it.