Question

File I want to parse:

input Pattern;

input SDF;

input ABC

input Pattern;

output Pattern;

output XYZ;

In perl, usual operation is scan line by line.
I want to check that if
current line has output Pattern; and previous line (or all previous lines)has input Pattern;
then change all the previous lines matches to "input Pattern 2;" and current line to "output Pattern2;".

It is complicated ,I hope I have explained properly.
Is it possible in Perl to scan and change previous lines after they have been read?

Thanks

Answer 1

If this is your data:

my $sfile =
'input Pattern;
input SDF;
input ABC
input Pattern;
output Pattern;
output XYZ;' ;

then, the following snippet will read the whole file and change text accordingly:

open my $fh, '<', \$sfile or die $!;
local $/ = undef;                # set file input mode to 'slurp'
my $content = <$fh>;
close $fh;

$content =~ s{ (                   # open capture group
                input \s+ (Pattern); # find occurence of input pattern
                .+?                  # skip some text
                output \s+ \2        # find same for output
               )                   # close capture group
             }
             {                     # replace by evaluated expression
              do{                    # within a do block
                 local $_=$1;        # get whole match to $_
                 s/($2)/$1 2/g;      # substitute Pattern by Pattern 2
                 $_                  # return substituted text
                }                    # close do block
             }esgx;

Then, you may close your file and check the string:

print $content;

=>

input Pattern 2;
input SDF;
input ABC
input Pattern 2;
output Pattern 2;
output XYZ;

---

You may even include a counter $n which will be incremented after each successful match (by code assertion (?{ ... }):

our $n = 1;

$content =~ s{ (                   # open capture group
                input \s+ (Pattern); # find occurence of input pattern
                .+?                  # skip some text
                output \s+ \2        # find same for output
                )                  # close capture group
                (?{ $n++ })        # ! update match count 
             }
             {                     # replace by evaluated expression
              do{                    # within a do block
                 local $_=$1;        # get whole match to $_
                 s/($2)/$1 $n/g;     # substitute Pattern by Pattern and count
                 $_                  # return substituted text
                }                  # close do block
             }esgx;

The substitution will now start with input Pattern 2; und increment subsequently.