Re: Perl pattern matching question

Use \b for word boundaries and character classes instead of alternatives of single characters:

#!/usr/bin/perl
use strict;
use warnings;

my @strings = ('A dog ate an apple',
               'A apple ate a dog',
               'A apple ate a oak',
              );
for my $string (@strings)  {
    $string =~ s/\b([Aa])(\s+[aeiou])/$1n$2/g;
    print "$string\n";
}
[download]

What would you use before the following words, though?

heir user XML yak ylem
[download]

You need pronunciation, not spelling. Update: check Text::Phonetic and related modules for that.

لսႽ† ᥲᥒ⚪⟊Ⴙᘓᖇ Ꮅᘓᖇ⎱ Ⴙᥲ𝇋ƙᘓᖇ

Comment on Re: Perl pattern matching question Select or Download Code

Replies are listed 'Best First'.
Re^2: Perl pattern matching question by AnomalousMonk (Archbishop) on May 06, 2015 at 18:05 UTC
Update: Note failure in the case of 'an log'. `c:\@Work\Perl\monks\MadhAric>perl -wMstrict -le "my $string = 'A log and a axe, an log and an axe'; print qq{'$string'}; ;; $string =~ s/\b([Aa])(\s+[aeiou])/$1n$2/g; print qq{'$string'}; " 'A log and a axe, an log and an axe' 'A log and an axe, an log and an axe'` [download] Even ignoring the subtleties of pronunciation you pointed out, the OPed problem statement is tricky. Some cases need 'n' deleted, so I don't see how it can be done (in a single regex) without decision-making in the replacement clause, i.e., use of the `/e` modifier or equivalent trick. Give a man a fish: `<%-(-(-(-<`	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^2: Perl pattern matching question
by AnomalousMonk (Archbishop) on May 06, 2015 at 18:05 UTC

Update: Note failure in the case of 'an log'.

c:\@Work\Perl\monks\MadhAric>perl -wMstrict -le
"my $string = 'A log and a axe, an log and an axe';
 print qq{'$string'};
 ;;
 $string =~ s/\b([Aa])(\s+[aeiou])/$1n$2/g;
 print qq{'$string'};
"
'A log and a axe, an log and an axe'
'A log and an axe, an log and an axe'
[download]

Even ignoring the subtleties of pronunciation you pointed out, the OPed problem statement is tricky. Some cases need 'n' deleted, so I don't see how it can be done (in a single regex) without decision-making in the replacement clause, i.e., use of the /e modifier or equivalent trick.

Give a man a fish: <%-(-(-(-<

[reply]
[d/l]
[select]