blob: 01f9bce74e8c061a3c1e5fa4e776d84ac4b7a8c8
1 | /* vi: set sw=4 ts=4: */ |
2 | /* fold -- wrap each input line to fit in specified width. |
3 | |
4 | Written by David MacKenzie, djm@gnu.ai.mit.edu. |
5 | Copyright (C) 91, 1995-2002 Free Software Foundation, Inc. |
6 | |
7 | Modified for busybox based on coreutils v 5.0 |
8 | Copyright (C) 2003 Glenn McGrath |
9 | |
10 | Licensed under GPLv2 or later, see file LICENSE in this source tree. |
11 | */ |
12 | //config:config FOLD |
13 | //config: bool "fold" |
14 | //config: default y |
15 | //config: help |
16 | //config: Wrap text to fit a specific width. |
17 | |
18 | //applet:IF_FOLD(APPLET_NOEXEC(fold, fold, BB_DIR_USR_BIN, BB_SUID_DROP, fold)) |
19 | |
20 | //kbuild:lib-$(CONFIG_FOLD) += fold.o |
21 | |
22 | //usage:#define fold_trivial_usage |
23 | //usage: "[-bs] [-w WIDTH] [FILE]..." |
24 | //usage:#define fold_full_usage "\n\n" |
25 | //usage: "Wrap input lines in each FILE (or stdin), writing to stdout\n" |
26 | //usage: "\n -b Count bytes rather than columns" |
27 | //usage: "\n -s Break at spaces" |
28 | //usage: "\n -w Use WIDTH columns instead of 80" |
29 | |
30 | #include "libbb.h" |
31 | #include "unicode.h" |
32 | |
33 | /* This is a NOEXEC applet. Be very careful! */ |
34 | |
35 | /* Must match getopt32 call */ |
36 | #define FLAG_COUNT_BYTES 1 |
37 | #define FLAG_BREAK_SPACES 2 |
38 | #define FLAG_WIDTH 4 |
39 | |
40 | /* Assuming the current column is COLUMN, return the column that |
41 | printing C will move the cursor to. |
42 | The first column is 0. */ |
43 | static int adjust_column(unsigned column, char c) |
44 | { |
45 | if (option_mask32 & FLAG_COUNT_BYTES) |
46 | return ++column; |
47 | |
48 | if (c == '\t') |
49 | return column + 8 - column % 8; |
50 | |
51 | if (c == '\b') { |
52 | if ((int)--column < 0) |
53 | column = 0; |
54 | } |
55 | else if (c == '\r') |
56 | column = 0; |
57 | else { /* just a printable char */ |
58 | if (unicode_status != UNICODE_ON /* every byte is a new char */ |
59 | || (c & 0xc0) != 0x80 /* it isn't a 2nd+ byte of a Unicode char */ |
60 | ) { |
61 | column++; |
62 | } |
63 | } |
64 | return column; |
65 | } |
66 | |
67 | /* Note that this function can write NULs, unlike fputs etc. */ |
68 | static void write2stdout(const void *buf, unsigned size) |
69 | { |
70 | fwrite(buf, 1, size, stdout); |
71 | } |
72 | |
73 | int fold_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
74 | int fold_main(int argc UNUSED_PARAM, char **argv) |
75 | { |
76 | char *line_out = NULL; |
77 | const char *w_opt = "80"; |
78 | unsigned width; |
79 | smallint exitcode = EXIT_SUCCESS; |
80 | |
81 | init_unicode(); |
82 | |
83 | if (ENABLE_INCLUDE_SUSv2) { |
84 | /* Turn any numeric options into -w options. */ |
85 | int i; |
86 | for (i = 1; argv[i]; i++) { |
87 | const char *a = argv[i]; |
88 | if (*a == '-') { |
89 | a++; |
90 | if (*a == '-' && !a[1]) /* "--" */ |
91 | break; |
92 | if (isdigit(*a)) |
93 | argv[i] = xasprintf("-w%s", a); |
94 | } |
95 | } |
96 | } |
97 | |
98 | getopt32(argv, "bsw:", &w_opt); |
99 | width = xatou_range(w_opt, 1, 10000); |
100 | |
101 | argv += optind; |
102 | if (!*argv) |
103 | *--argv = (char*)"-"; |
104 | |
105 | do { |
106 | FILE *istream = fopen_or_warn_stdin(*argv); |
107 | int c; |
108 | unsigned column = 0; /* Screen column where next char will go */ |
109 | unsigned offset_out = 0; /* Index in 'line_out' for next char */ |
110 | |
111 | if (istream == NULL) { |
112 | exitcode = EXIT_FAILURE; |
113 | continue; |
114 | } |
115 | |
116 | while ((c = getc(istream)) != EOF) { |
117 | /* We grow line_out in chunks of 0x1000 bytes */ |
118 | if ((offset_out & 0xfff) == 0) { |
119 | line_out = xrealloc(line_out, offset_out + 0x1000); |
120 | } |
121 | rescan: |
122 | line_out[offset_out] = c; |
123 | if (c == '\n') { |
124 | write2stdout(line_out, offset_out + 1); |
125 | column = offset_out = 0; |
126 | continue; |
127 | } |
128 | column = adjust_column(column, c); |
129 | if (column <= width || offset_out == 0) { |
130 | /* offset_out == 0 case happens |
131 | * with small width (say, 1) and tabs. |
132 | * The very first tab already goes to column 8, |
133 | * but we must not wrap it */ |
134 | offset_out++; |
135 | continue; |
136 | } |
137 | |
138 | /* This character would make the line too long. |
139 | * Print the line plus a newline, and make this character |
140 | * start the next line */ |
141 | if (option_mask32 & FLAG_BREAK_SPACES) { |
142 | unsigned i; |
143 | unsigned logical_end; |
144 | |
145 | /* Look for the last blank. */ |
146 | for (logical_end = offset_out - 1; (int)logical_end >= 0; logical_end--) { |
147 | if (!isblank(line_out[logical_end])) |
148 | continue; |
149 | |
150 | /* Found a space or tab. |
151 | * Output up to and including it, and start a new line */ |
152 | logical_end++; |
153 | /*line_out[logical_end] = '\n'; - NO! this nukes one buffered character */ |
154 | write2stdout(line_out, logical_end); |
155 | putchar('\n'); |
156 | /* Move the remainder to the beginning of the next line. |
157 | * The areas being copied here might overlap. */ |
158 | memmove(line_out, line_out + logical_end, offset_out - logical_end); |
159 | offset_out -= logical_end; |
160 | for (column = i = 0; i < offset_out; i++) { |
161 | column = adjust_column(column, line_out[i]); |
162 | } |
163 | goto rescan; |
164 | } |
165 | /* No blank found, wrap will split the overlong word */ |
166 | } |
167 | /* Output what we accumulated up to now, and start a new line */ |
168 | line_out[offset_out] = '\n'; |
169 | write2stdout(line_out, offset_out + 1); |
170 | column = offset_out = 0; |
171 | goto rescan; |
172 | } /* while (not EOF) */ |
173 | |
174 | if (offset_out) { |
175 | write2stdout(line_out, offset_out); |
176 | } |
177 | |
178 | if (fclose_if_not_stdin(istream)) { |
179 | bb_simple_perror_msg(*argv); |
180 | exitcode = EXIT_FAILURE; |
181 | } |
182 | } while (*++argv); |
183 | |
184 | fflush_stdout_and_exit(exitcode); |
185 | } |
186 |