Perl substitution using a hash - regex

Question

open (FH,"report");
read(FH,$text,-s "report");
$fill{"place"} = "Dhahran";
$fill{"wdesc:desc"} = "hot";
$fill{"dayno.days"} = 4;
$text =~ s/%(\w+)%/$fill{$1}/g;
print $text;
This is the content of the "report" template file
"I am giving a course this week in %place%. The weather is %wdesc:desc%
and we're now onto day no %dayno.days%. It's great group of blokes on the
course but the room is like the weather - %wdesc:desc% and it gets hard to
follow late in the day."
For reasons that I won't go into, some of the keys in the hash I'll be using will have dots (.) or colons (:) in them, but the regex stops working for these, so for instance in the example above only %place% gets correctly replaced. By the way, my code is based on this example.
Any help with the regex greatly appreciated, or maybe there's a better approach...

Answer 1

You could loosen it right up and use "any sequence of anything that isn't a %" for the replaceable tokens:
$text =~ s/%([^%]+)%/$fill{$1}/g;

Answer 2

Good answers so far, but you should also decide what you want to do with %foo% if foo isn't a key in the %fill hash. Plausible options are:
Replace it with an empty string (that's what the current solutions do, since undef is treated as an empty string in this context)
Leave it alone, so "%foo%" stays as it is.
Do some kind of error handling, perhaps printing a warning on STDERR, terminating the translation, or inserting an error indicator into the text.
Some other observations, not directly relevant to your question:
You should use the three-argument version of open.
That's not the cleanest way to read an entire file into a string. For that matter, for what you're doing you might as well process the input one line at a time.
Here's how I might do it (this version leaves unrecognized "%foo%" strings alone):
#!/usr/bin/perl
use strict;
use warnings;
my %fill = ( place => 'Dhahran',
'wdesc:desc' => 'hot',
'dayno.days' => 4 );
my $filename = 'report';
open my $FH,,'<', $filename or die "$filename: $!\n";
while (my $line = <$FH>) {
foreach my $key (keys %fill) {
$line =~ s/\Q%$key%/$fill{$key}/g;
}
print $line;
}
And here's a version that dies with an error message if there's an unrecognized key:
#!/usr/bin/perl
use strict;
use warnings;
my %fill = ( place => 'Dhahran',
'wdesc:desc' => 'hot',
'dayno.days' => 4 );
my $filename = 'report';
open my $FH,,'<', $filename or die "$filename: $!\n";
while (my $line = <$FH>) {
$line =~ s/%([^%]*)%/Replacement($1)/eg;
print $line;
}
sub Replacement {
my($key) = #_;
if (exists $fill{$key}) {
return $fill{$key};
}
else {
die "Unrecognized key \"$key\" on line $.\n";
}
}

Answer 3

http://codepad.org/G0WEDNyH
$text =~ s/%([a-zA-Z0-9_\.\:]+)%/$fill{$1}/g;
By default \w equates to [a-zA-Z0-9_], so you'll need to add in the \. and \:.