how can i split a string to an array and convert it into regular expression

[Anonymous Monk]

hello monks,
I am newbee in perl i am actually using <STDIN> to take the input
when the person enter data to it it is taken as a string
i want to convert the string to an array
the $string appears like 'gat_12345'
i want to split it into an array becoz i want to convert the
user input into regular expression with some loops
is there any other way i could convert the input to a reguler expression. Thanks

[runrig]

Perhaps you want to take a look at the split function (to split the data) and the qr operator (to create regexes). If that doesn't help, perhaps you write a short example of how you are trying to split the data, what you want to split it into, and what sort of regular expressions you want, and how you want to use them?

[vitaly]

I do not understand exactly what string you want to convert, but in order to convert a string into array you can use

@array = split /delimeter/, $my_string;

So you need to figure out what delimeter is. By default it is space.
Vitaly

[Anonymous Monk]

The string i want to convert is $string = 'gat_i2345'
it has no delimiter

[ikegami]

What do you want the array to contain? ('gat', 'i2345')? Then _ is the delimiter. If you want to add the entire string to the array, simply use push @array, $string;.

[mwah]

i want to split it into an array becoz i want to convert the user input into regular expression with some loops

You don't need to split into an array of characters for performing the task you described. The "program" you'll need looks probably like:

 ...
 my $input = <STDIN>;
 chomp $input;

 my $regex;
 $regex = transform_string_to_optimal_matching_regex($input);

 if( $input =~ /$regex/ ) {
   print "regex |$regex| to input |$input| mapping succeeded\n";
 } 
 ...

Perl is really good at finding strings that match a given regular expression, but rather bad at finding regular expressions that *would match* given strings 'optimally'. The remaining task for you is therefore to enter your favorite algorithm to do that between the curly braces of the subroutine 'transform_string_to_optimal_matching_regex()':

sub transform_string_to_optimal_matching_regex {
 my $given_string = shift;
 my $regex = '';
 my $optimal_solution_found = 0;
 
 while(! $optimal_solution_found ) {
    #
    # construct optimal regex by some
    # stochastic algorithm, possibly
    # long running Monte Carlo-solver
    #
 }
  return $regex;   
}

This is really an interesting thing to do. Maybe there are some approaches already published (none of which I know from).

Regards

mwa

[CountZero]

If you could give us the contents of the array you wish 'gat_12345' to be split into, we can perhaps give you some useful answers. Now we are just second guessing you.

[Anonymous Monk]

the array should contain (g,a,t,_,1,2,3,4,5)
and the regex should be like /\b\w+_\d+\b/i
Thanks for your help.

[ikegami]

To seperate the string into individual characters, you can use

my @array = $string =~ /./g;

or

my @array = map /./g, $string;

or

my @array = split(//, $string);

[runrig]

You have not clearly described how you want the program to infer what regular expression to generate from the input.

[plobsing]

Your regex groups characters using the + operator. Splitting into a list isn't the first way I would try this. Instead, I might try:

my @groups = qw'[a-zA-Z]+ \d+ \s+';
my $matcher_pattern = join('|', (map {"($_)"} @groups, '.'));
my $pattern = '';
MATCH: while ($string =~ /$matcher_pattern/g) {
     no strict 'refs';
     for my $index (1.. scalar @groups) {
         if ($$index) {
             $pattern .= $groups[$index - 1];
             next MATCH;
         }
     }
     $pattern .= quotemeta ${scalar @groups + 1};
}
$regex = qr/$pattern/

Update:Fixed off by one error. Update 2:Fixed metachar issue

[plobsing]

If I understand correctly, every line on STDIN is a possible match in a regex you are building. Then one solution is this:

my @arry = <>;
my $pattern = join('|', @arry);
my $regex = eval("qr/$pattern/");

That of course assumes that your input isn't doing anything evil (like having a '/' in it).

[ikegami]

You forgot to strip the line endings and you needlessly invoke the Perl parser and compiler in addition to the regex parser and compiler when only the latter is needed.

# If we are reading in regexs.
chomp( my @arry = <> );
my $pattern = join('|', @arry);
my $regex = qr/$pattern/;

# If we are reading in text strings.
chomp( my @arry = <> );
my $pattern = join('|', map quotemeta, @arry);
my $regex = qr/$pattern/;

[plobsing]

I thought qr// was a compile time thing. I stand corrected.
Thank you.