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