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