my $searchfor = shift;
$/="%%%\n"
open(IN, 'datafile.txt') or die "Could not open: $!\n";
while (<IN>) {
  print if /\Q$searchfor\E/i;
}
[download]

my $searchfor = shift;
$/="%%%\n";
[download]

Works EXACTLY how I need. With a few modifications this will run perfectly, thank you! Do you think you could explain what $/="%%%\n"; is doing? I've never seen or used $/ before.

In case the many replies using this approach have not made it clear, you can get more information about $/ from running "perldoc perlvar", and scanning down till you see this one described (with the long name "$INPUT_RECORD_SEPARATOR"). Normally -- i.e. by default -- $/ = "\n" (which in reality would be "\r\n" on MS-DOS/Windows systems, "\r" on "Classic" macs, and truly "\n" on unixes).

It's the character pattern that is removed from the end of a string when you "chomp" the string, and it is the pattern that the diamond operator looks for when reading data from a file handle into a scalar, to know when to stop. If set to "undef", it causes a single read operation to absorb the entire file and assign it all to a single scalar.

Then check your search criteria against the appropriate fields. Your next step would be to push the entire record into an array if it matches all of the search criteria.

This, of course, is just one possibility. If you have specific issues trying to implement something like this follow up here with code-related questions that we can sink our teeth into.

#!perl -w
use strict;
use DBI;
my @cols  = qw( Name Phone Fax Address Email Info );
my @table = (\@cols);
for my $record(split /\s*%%%\n*/,join '',<DATA>) {
    my @new_rec;
    my @fields = split /\n/, $record;
    for my $field(@fields) {
        $field =~ s/^.*:\s*//;
        push @new_rec, $field;
    }
    push @table, \@new_rec;
}
my $dbh  = DBI->connect('dbi:AnyData:');
$dbh->ad_import('t','ARRAY',\@table);
my $sth = $dbh->prepare("
    SELECT name,email FROM t WHERE address LIKE '%Anywhere%'
");
$sth->execute;
print $sth->dump_results;
__DATA__
Name: Robert Johnson
Phone: (555)555-1111
Fax: (555)555-1112
Address: 12345 Anywhere St.
Email: robertjohnson@someisp.com
Info: more info
%%%

Name: Robert Goor
Phone: (555)555-1113
Fax: (555)555-1114
Address: 12345 Somewhere St.
Email: robert@cs.someu.edu
Info: Some more info
%%%
[download]

This might not work as I expect it to, but . . .

local $/ = '%%%';
# Assume 'FH' is input file, opened elswhere
my @records;
while(my $line = <FH>) {
    $line =~ s/%%%//;

    # Remove any beginning or trailing whitespace
    # NOTE: Using \A and \z instead of ^ and $ is critical
    $line =~ s/\A \s*  (.*?)  \s* \z/$1/x;

    my %rec = split /[:\n]/, $line;
    push @records, \%rec;
}
[download]

@records should now hold an array-of-hashes containing the name/value pairs of your data. This may fail for some data, such as if a ':' appears anywhere besides splitting the name/value pairs in the file. It depends on the fact that the flattend list comming out of split can be coerced into a hash.

This is a problem that is screaming to be solved by a simple search engine. Check out Plucene and Plucene::Simple. There is also a nice tutorial on it at Perl.com. Yes, the article talks about a web-based search engine, but putting it into a script should be just as easy.

In your situation, you'll need an initial parse on the file to create the indexes, but after that the lookups should be quick. It will also return the full record or records related to your search.

How big is the file? If it's small enough (less than a megabyte, say), you could just slurp it in, make each whole record into a single array element, and simply search over them with

@hits=grep(/$term/,@directory);
[download]

And if it's more than a megabyte, do it with MySQL and just don't tell them :-)

Update:Aagh, what was I thinking - of course there is no need to slurp the whole file into memory - unless you wanted to make a search server :-)