You need a look ahead assertion. In this case a negative look ahead assertion (?!to\b|from\b):

use strict;
use warnings;

while (<DATA>) {
  print "Matched $_" if /train times (?!to\b|from\b)\w/;
}
__DATA__
train times to xyz
train times from xyz
train times including xyz
train times tomorrow xyz
[download]

Prints:

Matched train times including xyz
Matched train times tomorrow xyz
[download]

Update: fixed missing \b bug

Actually this brings me to another question along the same lines which is...if I wanted to (as I do) check for records where there may be a keyword before "train times" that may mean the record is to be rejected i.e. a negative lookbehind assertion I assume
e.g.


help train times from xyz
load train times including xyz
help train times including xyz
book train times at 1234
[download]

I need to reject records where the word immediately before "train times" is help. If help is used in the text it would always be the first word in the record. So only record two & four in my example would be printed. I tried doing this

print if /(?<!help).*train times/; #doesn't work
print if /^(!help).*train times/; # doesn't work either
[download]

Based on other comments I've read it seems that the use of negative lookbehind assertion should be discouraged but I'm not sure how to handle this without doing an assertion ?

A negative look ahead assertion at the start of the line (not a look back) is what you want:

use strict;
use warnings;

while (<DATA>) {
  print "Matched: $_" if /^(?!help).*?train times (?!to\b|from\b)\w/;
}
__DATA__
help train times from xyz
load train times including xyz
help train times including xyz
book train times at 1234
[download]

Prints:

Matched: load train times including xyz
Matched: book train times at 1234
[download]

---

DWIM is Perl's answer to Gödel

One happy customer, thank you

The square brackets ([]) delimit a character class, which essentially means "match any of these characters". So in your example, the regex ([^to ]|[^from ]) would match any character that is not (because of the ^ character which negates the character class) one of the characters f,m,o,r,t or a space.

You can use simple alternation to match either of these words:

if $line !~ /train times (?:to|from)\b/ {
    # do something with line
}
[download]

Update: added word boundary \b, thanks to GrandFather for pointing this out :-)

Try it with the string "train times tomorrow xyz" :)

---

DWIM is Perl's answer to Gödel

#!/usr/bin/perl -w

my $str1 = "train times tomorrow xyz";
my $str2 = "train times to xyz";
my $str3 = "train times from xyz";
my $str4 = "train times defrom xyz";
my $str5 = "train times fromar xyz";
my $str6 = "train times from";
my $str7 = "train from somewhere";

if($str1 !~ /^(\w+)?\s+(\w+)?\s+(to|from)([^\w]|\z)/) {
  print "does not match. Now we can do something\n";
}
[download]