Replacing words in VB code but not strings or comments

[Anonymous Monk]

I am writing a program to rename variables in VB code. I generate a hash of new names keyed by the current names. I want to replace all instances of the old name by the new name except where they appear in strings or comments.

Some notes on VB:

• In VB double quotes are put into a string by using two of them: "This is a doube quote""."
• Comments start at "'" and contine until the end of the line

The program builds up the list of renames as is reads the code line by line. The renames are performed on each line after it is processed. Is there a better way to perform the renaming?

use strict;
use warnings;

my $renames = {
    this_sub => 'that_sub',
    var_1    => 'var_a',
    var_2    => 'var_b',
    Ignore   => 'FAILED',
    nothing  => 'FAILED',
    and      => 'FILED',
    can      => 'FAILED',
    t        => 'FAILED',
    do       => 'FAILED',
    it       => 'FAILED',
};

my $line = qq/   this_sub("first", var_1, "Ignore ""can't""", var_2) '
+ do it\n/;
print $line;

my $matched_keys = '\b' . join( '|', keys %$renames ) . '\b';
my $match = qr/($matched_keys)/;
my $vb_string = qr/("[^"]*(?:""[^"]*)*")/;

my @parts = split $vb_string, $line;

my $do_code = 0;
for my $part (@parts) {
    $do_code = !$do_code;
    next unless $do_code;

    if ( $part =~ m/'/ ) {
        my ( $code, $comment ) = split /'/, $part, 2;
        $code =~ s/$match/$renames->{$1}/ge;
        $part = "$code'$comment";
        last;
    }
    else {
        $part =~ s/$match/$renames->{$1}/ge;
    }
}

print join( "", @parts )

Output:

   this_sub("first", var_1, "Ignore ""can't""", var_2) ' do it
   that_sub("first", var_a, "Ignore ""can't""", var_b) ' do it

[AnomalousMonk]

Here's an approach that works for 5.14+. (Still working on something for earlier versions.) I would add some more test cases. (BTW: I'm not familiar with VB, so I'm just going by what you have described of the syntax.)

Read more... (4 kB)

Output:

c:\@Work\Perl\monks\Anonymous Monk\1056009>perl parse_and_sub_vb_2.pl
#
#  testing function replace_5_14()
#
ok 1
ok 2
ok 3
ok 4
ok 5
ok 6
ok 7
ok 8 - no warnings
1..8

[Anonymous Monk]

That is the kind of solution I was looking for. Unfortunately I am targeting 5.8.8, the version installed by the vendor of the I am supporting. Thank you for you effort.

[AnomalousMonk]

Following tested under ActiveState 5.8.9 and Strawberries 5.10.1.5, 5.12.3.0 and 5.14.4.1. I'm sure this code can be simplified a bit. No Benchmark-ing whatsoever done.

Read more... (6 kB)

[AnomalousMonk]

Here's a variation for 5.8 and also for 5.10 that's simpler (for some definition of 'simpler') and that I think should run faster since it avoids  /e code evaluation during replacement, but again, I've done no benchmarking. Of course, the same approach applies as well to the 5.14 version code. Tested with the same Perl versions/vendors as before. Also to repeat: adding more of your own test cases would probably be a good idea.

Read more... (2 kB)