hexdump -C

[NERDVANA]

My goal is minification rather than obfuscation, but the more I try to minify it, the more it falls into this category :-)

I want to reproduce the output of the unix utility hexdump -C.

sub hexdump($data) {
   $data =~ s/\G(?|
       ( \0{16} )+ (?{ '*' })
     | ( .{1,16} ) (?{
            sprintf '%08X  %-50s|%-16s|', $-[1],
              join(' ', unpack('(H2)8a0(H2)8', $1)),
              $1 =~ tr{[\0-\x1F\x7F-\xFF]}{.}r
       }) 
   )/$^R\n/xgr;
}

Do any fellow monks see opportunities for additional savings? Obviously whitespace can be removed, but I'm leaving it in for ease of discussion.

[tybalt89]

Here's a couple of tweaks...

#!/usr/bin/perl

use strict; # https://perlmonks.org/?node_id=11166492
use warnings;
use Path::Tiny;
use feature 'signatures';

my $file = path('../tmp.tmp'); # FIXME for testing
system "hexdump -C $file; echo";
print hexdump( $file->slurp );

sub hexdump($data) {
   $data =~ s/\G(?|
       ( \0{16} )+ (?{ '*' })
     | ( .{1,16} ) (?{
            sprintf '%08X  %-50s|%s|', $-[1],
              "@{[unpack '(H2)8a0(H2)8', $1]}",
              $1 =~ y{ -~}{.}cr
       })
   )/$^R\n/xsgr . sprintf "%08x\n", length $data;
   }

Outputs:

00000000  6f 6e 65 0a 74 77 6f 0a  74 68 72 65 65 20 91 92  |one.two.t
+hree ..|
00000010  20 66 6f 75 72 0a 66 69  76 65 0a 6f 6e 65 0a 74  | four.fiv
+e.one.t|
00000020  77 6f 0a 74 68 72 65 65  20 2c 2c 20 66 6f 75 72  |wo.three 
+,, four|
00000030  0a 66 69 76 65 0a                                 |.five.|
00000036

00000000  6f 6e 65 0a 74 77 6f 0a  74 68 72 65 65 20 91 92  |one.two.t
+hree ..|
00000010  20 66 6f 75 72 0a 66 69  76 65 0a 6f 6e 65 0a 74  | four.fiv
+e.one.t|
00000020  77 6f 0a 74 68 72 65 65  20 2c 2c 20 66 6f 75 72  |wo.three 
+,, four|
00000030  0a 66 69 76 65 0a                                 |.five.|
00000036

[NERDVANA]

Nice! I forgot that y/// existed, and didn't know about the /c flag.

[ikegami]

The use of (?{}) doesn't help.

s/\G(?|(\0{16})+(?{'*'})|(.{1,16})(?{...}))/$^R\n/sgr  What you used.
s/\G(\0{16})|\G.{1,16}/($1?"*":...)."\n"/segr          Could be used.
s/\G.{1,16}/($&eq"\0"x16?"*":...)."\n"/segr            What I used.

Combining this change and tybalt89's changes, we get

#!/usr/bin/perl

use v5.36;
use Path::Tiny qw( path );

sub hexdump( $data ) {
   $data =~ s/\G.{1,16}/ 
      ( $& eq "\0"x16
      ? "*"
      : sprintf "%08X  %-50s|%s|",
           $-[0],
           "@{[unpack q{(H2)8a0(H2)8}, $&]}",
            $& =~ y{ -~}{.}cr
      ) . "\n"
   /segr . sprintf "%08x\n", length $data
}

print hexdump( path( $0 )->slurp );

[NERDVANA]

Actually it was solving the problem that the run of \0 lines needs to be replaced by a single '*', not one '*' per every 16 zeroes.

But, as I went to create an example for you, I realized I misunderstood the '*' behavior. It prints one full line of 00 00 00 ... and then the '*' means "repeat the previous line". I'd only ever seen it replace zeroes with '*' because that's the most likely repetition to appear in data files.

$ perl -E 'say "\0"x64 ."A"x64' | hexdump -C
00000000  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |.........
+.......|
*
00000040  41 41 41 41 41 41 41 41  41 41 41 41 41 41 41 41  |AAAAAAAAA
+AAAAAAA|
*
00000080  0a                                                |.|
00000081

Combining everyone's ideas, I now get:

sub hexdump($data) {
  $data =~ s/\G(.{1,16})(\1+)?/
    sprintf "%08X  %-50s|%s|\n%s",
        $-[0],
        "@{[unpack q{(H2)8a0(H2)8},$1]}",
        $1 =~ y{ -~}{.}cr,
        "*\n"x!!$+[2]
  /segr . sprintf "%08X", $+[0]
}

[harangzsolt33]

That is so cool! I modified it slightly so it also runs on TinyPerl 5.8:

sub hexdump
{
  my $s;
  $_[0] =~ s/\G([\0-\xff]{1,16})/ $s = $1; $s =~ y|\x20-\x7e|.|c; spri
+ntf("%08X  %-50s|%s|\n", $-[0], "@{[unpack q{(H2)8a0(H2)8}, $1]}", $s
+);/ge;
  return $_[0];
}

I don't understand why I had to do this: ([\0-\xff]{1,16}) instead of (.{16}) This latter one won't capture anything, which is weird. The second thing I don't understand is why this works:

"@{[unpack q{(H2)8a0(H2)8}, $1]}"