Hello blaui,

You can just use a regex:

use strict;
use warnings;

for my $string ("BCInletTemperature = 90[C]", "BCInletTemperature = 33
+.56[C]")
{
    print "$1\n" if $string =~ / = \s* ([0-9.]+) \[ /x;
}
[download]

Output:

16:59 >perl 1906_SoPW.pl
90
33.56

17:03 >
[download]

The regex says: match an equals sign, followed by zero or more whitespace characters, followed by one or more digit-or-decimal-point characters, followed by a left square bracket; and capture the string of digit-or-decimal-point characters in $1. The /x modifier on the regex means to not match whitespace inside the regex: this just makes the regex easier to read.

Update: Note that it is necessary to escape the square bracket to tell the regex engine that this is a character to be matched literally, and not the beginning of a character class.

Hope that helps,

blaui:   Note that the regex  [0-9.]+ also matches substrings like '1.2.3.4' and '....'.

---

Give a man a fish:  <%-{-{-{-<

Maybe a better approach instead of split is to match the things you want?

    $temp =~ /BCInletTemperature = ([\d.]+)\[C\]/
        or die "Invalid temperature format in '$temp'";
    my( $inletTemperature ) = $1;
[download]

In general, both, split and the matching deal with regular expressions. See perlre on how to use them and on how to escape meta-characters.

Hello!

The text before the equal sign can differ from BCInletTemperature. It can be another word. But the word will be always low/capital charaters without a blank.


Can I get this with the above too?

Regards, blaui

Yes, see perlre on how "alternations" work:

    $temp =~ /(BCInletTemperature|someOtherTemperature|aThirdTemperatu
+reName) = ([\d.]+)\[C\]/
        or die "Invalid temperature format in '$temp'";
    my( $temperatureType, $temperatureValue ) = ($1,$2);
[download]

But maybe you want a more generic parser that will capture any word on the left hand side of the "=" sign and any number on the right hand side. Then you can use the \w character class or the [A-Za-z] character class (again, see perlre) for matching any word on the left hand side:

    $temp =~ /(\w+) = ([\d.]+)\[C\]/
        or die "Invalid temperature format in '$temp'";
    my( $temperatureType, $temperatureValue ) = ($1,$2);
[download]

first, a comment on the code that you've shown. This:

(undef,$tmp) = split(/=/,$tmp[5]);
[download]

Next, I definitely agree with my fellow monks above that using a regex to capture the numerical value between the = sign and the opening square bracket is the simplest solution.

If, however, you want to continue using split, take a look at this possible solution illustrated under the Perl debugger:

  DB<1> $temp = "BCInletTemperature = 90[C]";

  DB<2> (undef,$tmp) = split /=/, $temp;

  DB<3> print $tmp
 90[C]
  DB<4> temperature = (split /\[/, $tmp)[0];

  DB<5> p $temperature
 90
[download]

  DB<6> p "<$temperature>";
< 90>
[download]

It is quite easy to get rid of that leading space if needed, but the simplest way is to use a regex. So, why not start with a simple regex anyway?

Another regex approach, with broader matching and some negative cases. See Regexp::Common.

c:\@Work\Perl\monks>perl -wMstrict -le
"use Regexp::Common;
 ;;
 my @records = (
   'BCInletTemperature = 90[C]',
   'BCInletTemperature = 33.56[C]',
   'Whatever   =   -.012[C]',
   'Wrong = 90 [C]',
   'AlsoWrong = 9.0.1[C]',
   );
 ;;
 for my $record (@records) {
   printf qq{'$record' -> };
   my $got_temp =
   my ($temp) = $record =~ m{ = \s+ ($RE{num}{real}) \[ }xms;
   if ($got_temp) {
     print qq{temp '$temp'};
     }
   else {
     print 'no got temp';
     }
   }
"
'BCInletTemperature = 90[C]' -> temp '90'
'BCInletTemperature = 33.56[C]' -> temp '33.56'
'Whatever   =   -.012[C]' -> temp '-.012'
'Wrong = 90 [C]' -> no got temp
'AlsoWrong = 9.0.1[C]' -> no got temp
[download]

Hello!

Thanxs all a lot for the many suggestions. Because it seems the more felxible, I started to try with the above Regexp::Common.


I modified it a littel bit.

$tmp[5] = "BCInletTemperature = 90[C]";
print "$tmp[5] xx\n";
$temp = $tmp[5] =~ m{ = \s+ ($RE{num}{real}) \[ }xms;
print "xxx $temp xxx\n";
[download]

But running it gives:

Name "charnames::CARP_NOT" used only once: possible typo at /usr/lib/p
+erl5/5.18.2/Carp.pm line 437.
Name "Regexp::Common::delimited::CARP_NOT" used only once: possible ty
+po at /usr/lib/perl5/5.18.2/Carp.pm line 437.
      BCInletTemperature = 90[C] xx
xxx 1 xxx
[download]

Perl is

perl --version

This is perl 5, version 18, subversion 2 (v5.18.2) built for x86_64-li
+nux-thread-multi
[download]

at openSUse Leap 42.3.
Do I need a newer version?

Regards, buchi

Do I need a newer version?

No, but you do need to remember to put the brackets around $temp when you assign to it in order to ensure list context. You should also provide an SSCCE rather than 4 lines out of context as something could be wrong in the code you aren't showing too.

Name "charnames::CARP_NOT" used only once: possible typo at /usr/lib/p
+erl5/5.18.2/Carp.pm line 437.
Name "Regexp::Common::delimited::CARP_NOT" used only once: possible ty
+po at /usr/lib/perl5/5.18.2/Carp.pm line 437.
[download]

Those are warnings, not errors, and I can't seem to be able to reproduce them here (don't have 5.18.2 handy though). Are you perhaps using perl -w on the command line or on the shebang line? In that case, see What's wrong with -w and $^W and use warnings; instead.

The other issue with that code is that $string=~/(regex)/ won't return the match(es) unless you use it in list context, but my $temp = ... puts it in scalar context. To put the right-hand side into list context, use my ($temp) = $string =~ /(regex)/;.