Question

While writing script I noticed some very strange Perl behavior. The task is to trim last character of string stored in $temp1 variable, this can be done using either chop($temp1) or more complicated substr($temp1, 0, length($temp1) - 1). These both work as one-liners but in my script the substr solution does not work. Below is example from debugger:

1st PROBLEM:

1st SOLUTION – NOT WORKING:

main::(delmid_n.pl:24):     if (($state == 0) && ($_ !~ $musr) && ($_ !~ $lusr)) {
>> n
main::(delmid_n.pl:54):         $strprn=substr($temp1, 0, length($temp1) - 1);
>> p $temp1
                (-,user2,t-mobile.co.uk)\

>> p substr($temp1, 0, length($temp1) - 1) . "\n";
                (-,user2,t-mobile.co.uk)


  DB<4> 
main::(delmid_n.pl:55):     print $strprn . "\n";
  DB<4> 


main::(delmid_n.pl:56):         $temp1 = "";
  DB<4> 

As you can see in the $strprn variable, nothing is stored. If the same piece of code (which is stored into the $strprn variable) is printed via the ‘p’ command, the output is OK. This “bug” can be overcome using the mentioned chop() function (see code below):

2nd SOLUTION – WORKING:

main::(delmid_w.pl:24):     if (($state == 0) && ($_ !~ $musr) && ($_ !~ $lusr)) {
>> p $temp1
                (-,user2,t-mobile.co.uk)\

  DB<4> n
main::(delmid_w.pl:56):     chop ($temp1);
  DB<4> n
main::(delmid_w.pl:57):     print $temp1;
  DB<4> n
                (-,user2,t-mobile.co.uk)
main::(delmid_w.pl:58):         $temp1 = "";
  DB<4>

The above code is exactly the same as the first example, but the following two lines are from 1st example:

$strprn=substr($temp1, 0, length($temp1) - 1);
print $strprn . "\n";

are replaced with following two lines in the 2nd example:

chop ($temp1);
print $temp1;

What is wrong with the 2nd solution?

2nd PROBLEM:

This is a problem which I do not have a workaround for, so far.

 DB<1> 
main::(delbeg_n.pl:15): $state = 0;
  DB<1> 
main::(delbeg_n.pl:16): $muser = qr/\(-,user1,[^,]+\.co\.uk\)\\$/;
  DB<1> 
main::(delbeg_n.pl:19): line: while (<>) {
  DB<1> 
main::(delbeg_n.pl:20):     chomp;      # strip record separator
  DB<1> 
main::(delbeg_n.pl:21):     @Fld = split(/\s+/, $_,);
  DB<1> 
main::(delbeg_n.pl:23):     if (($state == 0) && ($Fld[1] =~ $muser)) {
  DB<1> p $Fld[1]        
(-,user1,one2one.co.uk)\

  DB<2> n
main::(delbeg_n.pl:43):         print $_;
  DB<2> p $_
netgroup1       (-,user1,one2one.co.uk)\

  DB<3> if ($Fld[1] =~ $muser) {print "TRUE"}
TRUE

As you can see after executing line 21 in the code, the next execution line is 43 (the else statement). Why is the following condition not evaluated as a true, allowing the code to continue with line 23, 24, 25?

if (($state == 0) && ($Fld[1] =~ $muser))

The following line was inserted as a demonstration that the condition should be evaluated as a true:

if ($Fld[1] =~ $muser) {print "TRUE"}

Thanks a lot.

Answer 1

Both problems are related. There is something strange at the end of the line.

First, it looks as if you have a newline character after the slash at the end of the line that you have not chomped off — at least that is what the debugger shows:

>> p $temp1
                (-,user2,t-mobile.co.uk)\

  DB<4> n

If there was no newline it would have been

>> p $temp1
                (-,user2,t-mobile.co.uk)\
  DB<4> n

So, there should be two characters: \ and newline.

Why the weird behavior — I don’t know without looking at your input file.

Also, instead of using string length, you can just do substr( $string, 0, -1 ) — that will also return all but the last character.

That is probably the reason for your second problem — I guess that \(-,user1,[^,]+\.co\.uk\)\\ matches and it is the end of line marker $ that causes the mismatch.

• use strict please
• check the file types (dos, unix, old mac)
• view the file in a hex editor