H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 1 | #!/usr/bin/perl -w |
| 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 | |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 10 | # Default options |
| 11 | $max_width = 79; |
| 12 | |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 13 | # Clean up space-tab sequences, either by removing spaces or |
| 14 | # replacing them with tabs. |
| 15 | sub clean_space_tabs($) |
| 16 | { |
| 17 | no bytes; # Tab alignment depends on characters |
| 18 | |
| 19 | my($li) = @_; |
| 20 | my($lo) = ''; |
| 21 | my $pos = 0; |
| 22 | my $nsp = 0; |
| 23 | my($i, $c); |
| 24 | |
| 25 | for ($i = 0; $i < length($li); $i++) { |
| 26 | $c = substr($li, $i, 1); |
| 27 | if ($c eq "\t") { |
| 28 | my $npos = ($pos+$nsp+8) & ~7; |
| 29 | my $ntab = ($npos >> 3) - ($pos >> 3); |
| 30 | $lo .= "\t" x $ntab; |
| 31 | $pos = $npos; |
| 32 | $nsp = 0; |
| 33 | } elsif ($c eq "\n" || $c eq "\r") { |
| 34 | $lo .= " " x $nsp; |
| 35 | $pos += $nsp; |
| 36 | $nsp = 0; |
| 37 | $lo .= $c; |
| 38 | $pos = 0; |
| 39 | } elsif ($c eq " ") { |
| 40 | $nsp++; |
| 41 | } else { |
| 42 | $lo .= " " x $nsp; |
| 43 | $pos += $nsp; |
| 44 | $nsp = 0; |
| 45 | $lo .= $c; |
| 46 | $pos++; |
| 47 | } |
| 48 | } |
| 49 | $lo .= " " x $nsp; |
| 50 | return $lo; |
| 51 | } |
| 52 | |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 53 | # Compute the visual width of a string |
| 54 | sub strwidth($) { |
| 55 | no bytes; # Tab alignment depends on characters |
| 56 | |
| 57 | my($li) = @_; |
| 58 | my($c, $i); |
| 59 | my $pos = 0; |
| 60 | my $mlen = 0; |
| 61 | |
| 62 | for ($i = 0; $i < length($li); $i++) { |
| 63 | $c = substr($li,$i,1); |
| 64 | if ($c eq "\t") { |
| 65 | $pos = ($pos+8) & ~7; |
| 66 | } elsif ($c eq "\n") { |
| 67 | $mlen = $pos if ($pos > $mlen); |
| 68 | $pos = 0; |
| 69 | } else { |
| 70 | $pos++; |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | $mlen = $pos if ($pos > $mlen); |
| 75 | return $mlen; |
| 76 | } |
| 77 | |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 78 | $name = basename($0); |
| 79 | |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 80 | @files = (); |
| 81 | |
| 82 | while (defined($a = shift(@ARGV))) { |
| 83 | if ($a =~ /^-/) { |
| 84 | if ($a eq '-width' || $a eq '-w') { |
| 85 | $max_width = shift(@ARGV)+0; |
| 86 | } else { |
| 87 | print STDERR "Usage: $name [-width #] files...\n"; |
| 88 | exit 1; |
| 89 | } |
| 90 | } else { |
| 91 | push(@files, $a); |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | foreach $f ( @files ) { |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 96 | print STDERR "$name: $f\n"; |
| 97 | |
| 98 | if (! -f $f) { |
| 99 | print STDERR "$f: not a file\n"; |
| 100 | next; |
| 101 | } |
| 102 | |
| 103 | if (!open(FILE, '+<', $f)) { |
| 104 | print STDERR "$name: Cannot open file: $f: $!\n"; |
| 105 | next; |
| 106 | } |
| 107 | |
| 108 | binmode FILE; |
| 109 | |
| 110 | # First, verify that it is not a binary file; consider any file |
| 111 | # with a zero byte to be a binary file. Is there any better, or |
| 112 | # additional, heuristic that should be applied? |
| 113 | $is_binary = 0; |
| 114 | |
| 115 | while (read(FILE, $data, 65536) > 0) { |
| 116 | if ($data =~ /\0/) { |
| 117 | $is_binary = 1; |
| 118 | last; |
| 119 | } |
| 120 | } |
| 121 | |
| 122 | if ($is_binary) { |
| 123 | print STDERR "$name: $f: binary file\n"; |
| 124 | next; |
| 125 | } |
| 126 | |
| 127 | seek(FILE, 0, 0); |
| 128 | |
| 129 | $in_bytes = 0; |
| 130 | $out_bytes = 0; |
| 131 | $blank_bytes = 0; |
| 132 | |
| 133 | @blanks = (); |
| 134 | @lines = (); |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 135 | $lineno = 0; |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 136 | |
| 137 | while ( defined($line = <FILE>) ) { |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 138 | $lineno++; |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 139 | $in_bytes += length($line); |
| 140 | $line =~ s/[ \t\r]*$//; # Remove trailing spaces |
| 141 | $line = clean_space_tabs($line); |
| 142 | |
| 143 | if ( $line eq "\n" ) { |
| 144 | push(@blanks, $line); |
| 145 | $blank_bytes += length($line); |
| 146 | } else { |
| 147 | push(@lines, @blanks); |
| 148 | $out_bytes += $blank_bytes; |
| 149 | push(@lines, $line); |
| 150 | $out_bytes += length($line); |
| 151 | @blanks = (); |
| 152 | $blank_bytes = 0; |
| 153 | } |
H. Peter Anvin | cb3ed5b | 2007-05-25 17:58:26 -0700 | [diff] [blame] | 154 | |
| 155 | $l_width = strwidth($line); |
| 156 | if ($max_width && $l_width > $max_width) { |
| 157 | print STDERR |
| 158 | "$f:$lineno: line exceeds $max_width characters ($l_width)\n"; |
| 159 | } |
H. Peter Anvin | 12b3156 | 2007-03-12 12:16:30 -0700 | [diff] [blame] | 160 | } |
| 161 | |
| 162 | # Any blanks at the end of the file are discarded |
| 163 | |
| 164 | if ($in_bytes != $out_bytes) { |
| 165 | # Only write to the file if changed |
| 166 | seek(FILE, 0, 0); |
| 167 | print FILE @lines; |
| 168 | |
| 169 | if ( !defined($where = tell(FILE)) || |
| 170 | !truncate(FILE, $where) ) { |
| 171 | die "$name: Failed to truncate modified file: $f: $!\n"; |
| 172 | } |
| 173 | } |
| 174 | |
| 175 | close(FILE); |
| 176 | } |