#!/usr/bin/perl

# https://perlmonks.org/?node_id=1220510

use strict;
use warnings;

while( <DATA> )
  {
  s/(^|    \K)(?!\d)/0/g;
  print;
  }

__DATA__
1    2            2        5                4
4    4    4            4                3    
        4    4            1                
    1    5    6        4
[download]

Hello, tybalt89, Thank you so much for your great help.

The data file is

1    2            2        5                4
4    4    4            4                3    
        4    4            1                
    1    5    6        4
[download]

It comes out

1    2            2        5                4
4    4    4            4                3    
0        4    4            1                
0    1    5    6        4                    



#!/usr/bin/perl

# https://perlmonks.org/?node_id=1220510

use strict;
use warnings;


open(DATA,"DATA")||die"cannot open the file: $!\n";


while( <DATA> )
  {
  s/(^|    \K)(?!\d)/0/g;
  print;
  }
[download]

You have too old a perl, try

#!/usr/bin/perl

# https://perlmonks.org/?node_id=1220510

use strict;
use warnings;

while( <DATA> )
  {
  s/(^|    )(?!\d)/$&0/g;
  print;
  }

__DATA__
1    2            2        5                4
4    4    4            4                3    
        4    4            1                
    1    5    6        4
[download]

Update: Unfortunately, neither of these approaches handles leading four-blank groups properly. Oh, well... (Update: Also, it seems to me that at some point the OPed example data changed from having groups of ASCII 0x20 blanks to having tabs. Is this really so? Only yueli711 can know. :)

One way (assuming a "space" is a blank, i.e., ASCII 0x20):

c:\@Work\Perl>perl -wMstrict -le
"my $s = '1    2            2        5                4';
 $s =~ s{ ([ ]{4}) (?! \d) }{${1}0}xmsg;
 print qq{'$s'};
 ;;
 my $t = '1    2    0    0    2    0    5    0    0    0    4';
 die 'bad transformation' unless $s eq $t;
"
'1    2    0    0    2    0    5    0    0    0    4'
[download]

Update: A slight variation if you have Perl version 5.10+ and the  \K regex operator:

c:\@Work\Perl\monks>perl -wMstrict -le
"my $s = '1    2            2        5                4';
 $s =~ s{ [ ]{4} \K (?! \d) }{0}xmsg;
 print qq{'$s'};
 ;;
 my $t = '1    2    0    0    2    0    5    0    0    0    4';
 die 'bad transformation' unless $s eq $t;
"
'1    2    0    0    2    0    5    0    0    0    4'
[download]

I recommend first splitting the columns and then replacing space/undef with 0.

A hex dump of the download of the OPed example data seems to show that these are tab-separated CSV records in which you need to expand null fields to a single  '0' character. If that's the case, the best advice is to save yourself a headache | many headaches and to use the flexible and reliable Text::CSV module (which will automatically install its cool, fast big brother Text::CSV_XS if it possibly can).

However, if a "pure" regex solution is needed, here's my approach. (A split/fixup/join approach might be easier to understand, but I'm too tired right now to take that on. :) Tested under Perl versions 5.8.9 and 5.10.1.4, but some alternate versions of the critical substitution regex need version 5.10+ for the  \K operator.

G'day yueli711,

Update: This approach has issues with some of the other data. See ++AnomalousMonk's feedback re these problems.

I believe applying this regex to each line will provide what you want.

s/\t(?=\t)/\t0/g
[download]

My test (just using the data from the first line you posted):

$ perl -le 'my $x = "1\t2\t\t\t2\t\t5\t\t\t\t4"; print $x; $x =~ s/\t(
+?=\t)/\t0/g; print $x'
1       2                       2               5                     
+          4
1       2       0       0       2       0       5       0       0     
+  0       4
[download]

Note: As you can see, my test used just tabs. The output lined up as shown above, both from my command line and when pasted into the textarea; however, the browser made a mess of it when rendered. I've actually changed all the output tabs to a series of spaces: what I've now posted is just a true representation of what the output looked like, not a verbatim copy of the actual output itself.

s/\t(?=\t)/\t0/g

Unfortunately, that regex doesn't work with other of the OPed example data strings that have leading and/or trailing tabs, e.g., "4\t4\t4\t\t\t4\t\t\t\t3\t" and "\t\t4\t4\t\t\t1\t\t\t\t". (It took me a while, but I finally figured out that all that whitespace was due to tab characters. I suspect that yueli711 changed the posted example data, too. Oh, well... :)

---

Give a man a fish:  <%-{-{-{-<

G'day AnomalousMonk,

++ Thanks for advising of that problem. I've updated my post.

— Ken