scripts/cleanfile

   1 #!/usr/bin/env perl
   2 #
   3 # Clean a text file -- or directory of text files -- of stealth whitespace.
   4 # WARNING: this can be a highly destructive operation.  Use with caution.
   5 #
   6
   7 use bytes;
   8 use File::Basename;
   9 use warnings;
  10
  11 # Default options
  12 $max_width = 79;
  13
  14 # Clean up space-tab sequences, either by removing spaces or
  15 # replacing them with tabs.
  16 sub clean_space_tabs($)
  17 {
  18     no bytes;                   # Tab alignment depends on characters
  19
  20     my($li) = @_;
  21     my($lo) = '';
  22     my $pos = 0;
  23     my $nsp = 0;
  24     my($i, $c);
  25
  26     for ($i = 0; $i < length($li); $i++) {
  27         $c = substr($li, $i, 1);
  28         if ($c eq "\t") {
  29             my $npos = ($pos+$nsp+8) & ~7;
  30             my $ntab = ($npos >> 3) - ($pos >> 3);
  31             $lo .= "\t" x $ntab;
  32             $pos = $npos;
  33             $nsp = 0;
  34         } elsif ($c eq "\n" || $c eq "\r") {
  35             $lo .= " " x $nsp;
  36             $pos += $nsp;
  37             $nsp = 0;
  38             $lo .= $c;
  39             $pos = 0;
  40         } elsif ($c eq " ") {
  41             $nsp++;
  42         } else {
  43             $lo .= " " x $nsp;
  44             $pos += $nsp;
  45             $nsp = 0;
  46             $lo .= $c;
  47             $pos++;
  48         }
  49     }
  50     $lo .= " " x $nsp;
  51     return $lo;
  52 }
  53
  54 # Compute the visual width of a string
  55 sub strwidth($) {
  56     no bytes;                   # Tab alignment depends on characters
  57
  58     my($li) = @_;
  59     my($c, $i);
  60     my $pos = 0;
  61     my $mlen = 0;
  62
  63     for ($i = 0; $i < length($li); $i++) {
  64         $c = substr($li,$i,1);
  65         if ($c eq "\t") {
  66             $pos = ($pos+8) & ~7;
  67         } elsif ($c eq "\n") {
  68             $mlen = $pos if ($pos > $mlen);
  69             $pos = 0;
  70         } else {
  71             $pos++;
  72         }
  73     }
  74
  75     $mlen = $pos if ($pos > $mlen);
  76     return $mlen;
  77 }
  78
  79 $name = basename($0);
  80
  81 @files = ();
  82
  83 while (defined($a = shift(@ARGV))) {
  84     if ($a =~ /^-/) {
  85         if ($a eq '-width' || $a eq '-w') {
  86             $max_width = shift(@ARGV)+0;
  87         } else {
  88             print STDERR "Usage: $name [-width #] files...\n";
  89             exit 1;
  90         }
  91     } else {
  92         push(@files, $a);
  93     }
  94 }
  95
  96 foreach $f ( @files ) {
  97     print STDERR "$name: $f\n";
  98
  99     if (! -f $f) {
 100         print STDERR "$f: not a file\n";
 101         next;
 102     }
 103
 104     if (!open(FILE, '+<', $f)) {
 105         print STDERR "$name: Cannot open file: $f: $!\n";
 106         next;
 107     }
 108
 109     binmode FILE;
 110
 111     # First, verify that it is not a binary file; consider any file
 112     # with a zero byte to be a binary file.  Is there any better, or
 113     # additional, heuristic that should be applied?
 114     $is_binary = 0;
 115
 116     while (read(FILE, $data, 65536) > 0) {
 117         if ($data =~ /\0/) {
 118             $is_binary = 1;
 119             last;
 120         }
 121     }
 122
 123     if ($is_binary) {
 124         print STDERR "$name: $f: binary file\n";
 125         next;
 126     }
 127
 128     seek(FILE, 0, 0);
 129
 130     $in_bytes = 0;
 131     $out_bytes = 0;
 132     $blank_bytes = 0;
 133
 134     @blanks = ();
 135     @lines  = ();
 136     $lineno = 0;
 137
 138     while ( defined($line = <FILE>) ) {
 139         $lineno++;
 140         $in_bytes += length($line);
 141         $line =~ s/[ \t\r]*$//;         # Remove trailing spaces
 142         $line = clean_space_tabs($line);
 143
 144         if ( $line eq "\n" ) {
 145             push(@blanks, $line);
 146             $blank_bytes += length($line);
 147         } else {
 148             push(@lines, @blanks);
 149             $out_bytes += $blank_bytes;
 150             push(@lines, $line);
 151             $out_bytes += length($line);
 152             @blanks = ();
 153             $blank_bytes = 0;
 154         }
 155
 156         $l_width = strwidth($line);
 157         if ($max_width && $l_width > $max_width) {
 158             print STDERR
 159                 "$f:$lineno: line exceeds $max_width characters ($l_width)\n";
 160         }
 161     }
 162
 163     # Any blanks at the end of the file are discarded
 164
 165     if ($in_bytes != $out_bytes) {
 166         # Only write to the file if changed
 167         seek(FILE, 0, 0);
 168         print FILE @lines;
 169
 170         if ( !defined($where = tell(FILE)) ||
 171              !truncate(FILE, $where) ) {
 172             die "$name: Failed to truncate modified file: $f: $!\n";
 173         }
 174     }
 175
 176     close(FILE);
 177 }