1 #!/usr/bin/env perl
2 # SPDX-License-Identifier: GPL-2.0
3 #
4 # Clean a text file -- or directory of text files -- of stealth whitespace.
5 # WARNING: this can be a highly destructive operation. Use with caution.
6 #
7
8 use warnings;
9 use bytes;
10 use File::Basename;
11
12 # Default options
13 $max_width = 79;
14
15 # Clean up space-tab sequences, either by removing spaces or
16 # replacing them with tabs.
17 sub clean_space_tabs($)
18 {
19 no bytes; # Tab alignment depends on characters
20
21 my($li) = @_;
22 my($lo) = '';
23 my $pos = 0;
24 my $nsp = 0;
25 my($i, $c);
26
27 for ($i = 0; $i < length($li); $i++) {
28 $c = substr($li, $i, 1);
29 if ($c eq "\t") {
30 my $npos = ($pos+$nsp+8) & ~7;
31 my $ntab = ($npos >> 3) - ($pos >> 3);
32 $lo .= "\t" x $ntab;
33 $pos = $npos;
34 $nsp = 0;
35 } elsif ($c eq "\n" || $c eq "\r") {
36 $lo .= " " x $nsp;
37 $pos += $nsp;
38 $nsp = 0;
39 $lo .= $c;
40 $pos = 0;
41 } elsif ($c eq " ") {
42 $nsp++;
43 } else {
44 $lo .= " " x $nsp;
45 $pos += $nsp;
46 $nsp = 0;
47 $lo .= $c;
48 $pos++;
49 }
50 }
51 $lo .= " " x $nsp;
52 return $lo;
53 }
54
55 # Compute the visual width of a string
56 sub strwidth($) {
57 no bytes; # Tab alignment depends on characters
58
59 my($li) = @_;
60 my($c, $i);
61 my $pos = 0;
62 my $mlen = 0;
63
64 for ($i = 0; $i < length($li); $i++) {
65 $c = substr($li,$i,1);
66 if ($c eq "\t") {
67 $pos = ($pos+8) & ~7;
68 } elsif ($c eq "\n") {
69 $mlen = $pos if ($pos > $mlen);
70 $pos = 0;
71 } else {
72 $pos++;
73 }
74 }
75
76 $mlen = $pos if ($pos > $mlen);
77 return $mlen;
78 }
79
80 $name = basename($0);
81
82 @files = ();
83
84 while (defined($a = shift(@ARGV))) {
85 if ($a =~ /^-/) {
86 if ($a eq '-width' || $a eq '-w') {
87 $max_width = shift(@ARGV)+0;
88 } else {
89 print STDERR "Usage: $name [-width #] files...\n";
90 exit 1;
91 }
92 } else {
93 push(@files, $a);
94 }
95 }
96
97 foreach $f ( @files ) {
98 print STDERR "$name: $f\n";
99
100 if (! -f $f) {
101 print STDERR "$f: not a file\n";
102 next;
103 }
104
105 if (!open(FILE, '+<', $f)) {
106 print STDERR "$name: Cannot open file: $f: $!\n";
107 next;
108 }
109
110 binmode FILE;
111
112 # First, verify that it is not a binary file; consider any file
113 # with a zero byte to be a binary file. Is there any better, or
114 # additional, heuristic that should be applied?
115 $is_binary = 0;
116
117 while (read(FILE, $data, 65536) > 0) {
118 if ($data =~ /\0/) {
119 $is_binary = 1;
120 last;
121 }
122 }
123
124 if ($is_binary) {
125 print STDERR "$name: $f: binary file\n";
126 next;
127 }
128
129 seek(FILE, 0, 0);
130
131 $in_bytes = 0;
132 $out_bytes = 0;
133 $blank_bytes = 0;
134
135 @blanks = ();
136 @lines = ();
137 $lineno = 0;
138
139 while ( defined($line = <FILE>) ) {
140 $lineno++;
141 $in_bytes += length($line);
142 $line =~ s/[ \t\r]*$//; # Remove trailing spaces
143 $line = clean_space_tabs($line);
144
145 if ( $line eq "\n" ) {
146 push(@blanks, $line);
147 $blank_bytes += length($line);
148 } else {
149 push(@lines, @blanks);
150 $out_bytes += $blank_bytes;
151 push(@lines, $line);
152 $out_bytes += length($line);
153 @blanks = ();
154 $blank_bytes = 0;
155 }
156
157 $l_width = strwidth($line);
158 if ($max_width && $l_width > $max_width) {
159 print STDERR
160 "$f:$lineno: line exceeds $max_width characters ($l_width)\n";
161 }
162 }
163
164 # Any blanks at the end of the file are discarded
165
166 if ($in_bytes != $out_bytes) {
167 # Only write to the file if changed
168 seek(FILE, 0, 0);
169 print FILE @lines;
170
171 if ( !defined($where = tell(FILE)) ||
172 !truncate(FILE, $where) ) {
173 die "$name: Failed to truncate modified file: $f: $!\n";
174 }
175 }
176
177 close(FILE);
178 }
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.