wantarray

Returns true if the context of the currently executing subroutine or eval is looking for a list value. Returns false if the context is looking for a scalar. Returns the undefined value if the context is looking for no value (void context).

FYI instead of $line =~ s/\s+$//; # rtrim() in Perl you can use AWK hack in split function:

($line)=split(' ',$line,1)
[download]

[0]  # time perl -e 'for (1..1000000) { $line="   aaa bbb ccc ddd eee 
+fff    "; $line =~ s/\s+$//;  $line =~ s/\s+$//;}'

real    0m3.526s
user    0m3.510s
sys     0m0.000s
[download]

so we are talking about real money here (without regex time is 0.14).

May be extending tr to stop after the first symbol which is not in set1 would also be beneficial and faster.

In my use of Perl I would prefer an option in the open statement that would do the job for all lines read, as this is the most common use case. Something like open(file,'<t(rln)','test')

FYI instead of $line =~ s/\s+$//; # rtrim() in Perl you can use AWK hack is split:

($line)=split(' ',$line,1)

Perhaps I am misunderstanding something but are you saying that $line =~ s/\s+$//; and ($line)=split(' ',$line,1); are equivalent? That doesn't seem to be the case, the latter appearing to implement the ltrim() you desire. This code

use strict;
use warnings;

my $line = q{   aaa bbb ccc ddd eee fff    };
( $line ) = split( ' ', $line, 1 );
print qq{-->$line<--\n};
[download]

produces

-->aaa bbb ccc ddd eee fff    <--
[download]

Perhaps you could clarify?

Cheers,

JohnGG

It removes only leading whitespace. There is no alternative to regex idiom for trailing whitespace that I know of.

Please also show the time of just populating the lines with no substitution.

|14:29 choroba@triangle 
|~ $ time perl -e 'for (1..1000000) { $line="   aaa bbb ccc ddd eee ff
+f    ";}'

real    0m0.041s
user    0m0.042s
sys     0m0.000s
|14:29 choroba@triangle 
|~ $ time perl -e 'for (1..1000000) { $line="   aaa bbb ccc ddd eee ff
+f    "; $line =~ s/^ +//; $line =~ s/ +$//;}'

real    0m0.747s
user    0m0.740s
sys     0m0.004s
[download]

map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]

split is faster, but also less versatile:

$ perl -MBenchmark=cmpthese -wE'my$x=join"    "=>"",("aaa")x5,"";say"s
+ourc: |$x|";sub splt{split" ",$x,1};sub rgx{$x=~s/^\s+//r};say "split
+: |",splt(),"|";say"regex: |",rgx(),"|";cmpthese(-2,{splt=>\&splt,rgx
+=>\&rgx})'
sourc: |    aaa    aaa    aaa    aaa    aaa    |
split: |aaa    aaa    aaa    aaa    aaa    |
regex: |aaa    aaa    aaa    aaa    aaa    |
          Rate  rgx splt
rgx  3382502/s   -- -58%
splt 7989872/s 136%   --
[download]

and split is not able to trim trailing space without help

$ perl -MBenchmark=cmpthese -wE'my$x=join"    "=>"",("abc")x5,"";say"s
+ourc: |$x|";sub splt{split" ",reverse((split" ",(reverse$x),1)[0]),1;
+};sub rgx{$x=~s/^\s+//r=~s/\s+$//r};say "split: |",splt(),"|";say"reg
+ex: |",rgx(),"|";cmpthese(-2,{splt=>\&splt,rgx=>\&rgx})'
sourc: |    abc    abc    abc    abc    abc    |
split: |abc    abc    abc    abc    abc|
regex: |abc    abc    abc    abc    abc|
          Rate  rgx splt
rgx  1047602/s   -- -64%
splt 2915089/s 178%   --
[download]

It is still faster, but I would pity the coder that has to maintain it.

So, concluding split (as a function) is faster, the next step is to profile your complete process and wonder if winning this little is worth the maintainability nightmare.

If this functionality only uses 0.01% of the complete time, I would not even think about bothering.

If it makes up 75% of the process time, and you are dealing with 4 Tb datastreams I would create an XS module that does exactly what I need and squeeze every CPU microsecond out of that optimized code.

I seriously doubt if that would be worth it. Seriously. There are millions of existing scripts that already use s{^\s+}{} and s{\s+$}{} (which removes all whitespace, so my functions may not be identical for some data), and people that use Perl, are familiar with it. Introducing functions like ltrim and rtrim or whatever they are to be called will for sure appeal to many users (unless the syntax is not obvious: as an excercise to the reader, please try to build split from scratch and implement all edge cases).

Is presented as an XS module, it may be used as playground, but then one has to ask themselves, does the time required to import the module weigh against the time gained in the function?


Enjoy, Have FUN! H.Merijn

[0]  # time perl -e 'for (1..1000000) { $line="   aaa bbb ccc ddd eee 
+fff    "; }'

real    0m0.140s
user    0m0.124s
sys     0m0.015s
[download]