user-tools/untabularize

   1 #!/usr/bin/env perl
   2
   3 =pod
   4
   5 =head1 NAME
   6
   7 untabularize - Revert the formatting done by tabularize(1)
   8
   9 =head1 SYNOPSIS
  10
  11 untabularize [I<OPTIONS>]
  12
  13 =head1 DESCRIPTION
  14
  15 =head1 OPTIONS
  16
  17 =over 4
  18
  19 =back
  20
  21 =head1 LIMITATIONS
  22
  23 =head1 SEE ALSO
  24
  25 tabularize(1)
  26
  27 =cut
  28
  29
  30 use Data::Dumper;
  31 use Getopt::Long qw/:config no_ignore_case no_bundling no_getopt_compat no_auto_abbrev require_order/;
  32 use List::MoreUtils qw/all any none/;
  33 use Pod::Usage;
  34 use POSIX;
  35 no if ($] >= 5.018), 'warnings' => 'experimental::smartmatch';
  36 use utf8;
  37 use open ':std', ':encoding(UTF-8)';
  38
  39 GetOptions(
  40         'help' => sub { pod2usage(-exitval=>0, -verbose=>99); },
  41         '<>' => sub { unshift @ARGV, @_[0]; die '!FINISH'; },
  42 ) or pod2usage(-exitval=>2, -verbose=>99);
  43
  44
  45 while(<STDIN>)
  46 {
  47         chomp;
  48         if(/^(┌|├|└|\+)(.)/ and not @cellWidths)
  49         {
  50                 my $linechar = $2;
  51                 s/^[^\Q$linechar\E]//;
  52                 s/[^\Q$linechar\E]$//;
  53                 @cellWidths = map {length} split /[^\Q$linechar\E]/;
  54                 next;
  55         }
  56         if(/^(\||│)/)
  57         {
  58                 my $gridchar = $1;
  59                 s/^\Q$gridchar\E//;
  60                 s/\Q$gridchar\E$//;
  61                 if(not @cellWidths)
  62                 {
  63                         if($gridchar eq '|')
  64                         {
  65                                 # pipe char may occur in the cell data as well,
  66                                 # so can not confidentally take it as a cell separator.
  67                                 die "$0: can not determine the columns in --ascii --no-horizontal mode\n";
  68                         }
  69                         @cellWidths = map {length} split /\Q$gridchar\E/;
  70                 }
  71                 my $record = $_;
  72                 my $prev_cell_endpos = 0;
  73                 my @cells = map { s/\s+$// or s/^\s+//;  $_ }
  74                         map { my $cell = substr $record, $prev_cell_endpos, $_; $prev_cell_endpos += $_ + length $gridchar; $cell }
  75                         @cellWidths;
  76                 print join("\t", @cells) . "\n";
  77         }
  78 }