Perl替代：无法使用变量perl regex替代

Question

The following code reads defines.c file line-by-line and stores all the "#define" directives in a hash (with defines name and its replacement text). Then it replaces them if the usage is find on the consecutive lines.

I have an issue with the substitution if a "#define" directive is being used. It is storing all the #defines properly in the hash.

I am a beginner in perl and regex and unable to point out the silly thing as to why it's not working.
Any help?

#!/usr/bin/perl -w

use strict;
my %definesHash;

open(FILE, "defines.c") || die "Cannot open $!\n";
open(OUT, ">defines.i") || die "Cannot open $!\n";


while (<FILE>)
{
 my $line = $_;
 if ($line =~ /#define\s+/)
 {
   $line =~ s/#define\s+//g;
   if ($line =~ /\b([\w]+)\b\s+/)
   {
    my $define = $1;
    $line =~ s/\b[\w]+\b\s+//;
    $definesHash{$define} = "";
    if($line =~ /\s*(.*)\s*/)
    {
      $definesHash{$define} = $1;
    }
   }
   print OUT $_;
 }
 else
 { 
   my($def, $replace);
   while (($def, $replace) = each(%definesHash))
   {
     print " $def => $replace \n";
     if ($line =~ /$def/)
     {
      $line =~ s/$def/$replace/g; #****** Some Problem Here, But What? ********
     }
   }
   print OUT $line;
 }
}

close(FILE);
close(OUT);

defines.c

#include <stdio.h>

#define VALUE 10
#define PLACE (20 + 0)
#define NUM (VALUE + 10)
int main()
{
  int num;
  num = NUM + 25 + NUM + PLACE;
  return 0;
}

Expected Output: **defines.i**

#include <stdio.h>

#define VALUE 10
#define PLACE (20 + 0)
#define NUM (VALUE + 10)
int main()
{
  int num;
  num = (10 + 10) + 25 + (10 + 10) + (20 + 0);
  // or
  num = (VALUE + 10) + 25 + (VALUE + 10) + (20 + 0);
  return 0;
}  

I know the problem is in the substitution as I have pointed in the comment. I can see the original line becoming gibberish after that substitution whereas the contents in the hash seem to be correct.
Is it with the usage of variables in substitution?

Answer 1

It looks like your script depends on the order (or rather, disorder) of the hash.

#define VALUE 10
#define NUM (VALUE + 10)

The definition doesn't get updated because that replacement takes place in the else -- so replacement only happens on lines without #define .

However, you perform the replacements based on whatever order each gives you - so it could be trying to replace VALUE , and then trying to replace NUM .

I'd probably perform the initial replacements in each definition as they come, something like this:

#!/usr/bin/perl

use warnings;
use strict;
my %definesHash;

open(my $fh_in,  '<', "defines.c") || die "Cannot open $!\n";
open(my $fh_out, '>', "defines.i") || die "Cannot open $!\n";

while (<$fh_in>) {
    if (/^#define\s+(\w+)\s+(.*)\s*$/) {
        my ($define, $replacement) = ($1, $2);
        # perform existing replacements on
        # the current $replacement
        while (my ($def, $replace) = each %definesHash) {
            $replacement =~ s/$def/\Q$replace/g;
        }
        $definesHash{$define} = $replacement;
    }
    else {
        while (my ($def, $replace) = each(%definesHash)) {
            print " $def => $replace \n";
            s/$def/$replace/g;
        }
    }

    print $fh_out $_;
}

close($fh_in);
close($fh_out);

Or it could be the unescaped regex metacharacters, here escaped by \\Q .