blob: 9208d34ec487040bf3a7c40394f910cfe618367e
1 | /* vi: set sw=4 ts=4: */ |
2 | /* |
3 | * uniq implementation for busybox |
4 | * |
5 | * Copyright (C) 2005 Manuel Novoa III <mjn3@codepoet.org> |
6 | * |
7 | * Licensed under GPLv2 or later, see file LICENSE in this source tree. |
8 | */ |
9 | |
10 | /* BB_AUDIT SUSv3 compliant */ |
11 | /* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */ |
12 | |
13 | //usage:#define uniq_trivial_usage |
14 | //usage: "[-cdu][-f,s,w N] [INPUT [OUTPUT]]" |
15 | //usage:#define uniq_full_usage "\n\n" |
16 | //usage: "Discard duplicate lines\n" |
17 | //usage: "\n -c Prefix lines by the number of occurrences" |
18 | //usage: "\n -d Only print duplicate lines" |
19 | //usage: "\n -u Only print unique lines" |
20 | //usage: "\n -f N Skip first N fields" |
21 | //usage: "\n -s N Skip first N chars (after any skipped fields)" |
22 | //usage: "\n -w N Compare N characters in line" |
23 | //usage: |
24 | //usage:#define uniq_example_usage |
25 | //usage: "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n" |
26 | //usage: "a\n" |
27 | //usage: "b\n" |
28 | //usage: "c\n" |
29 | |
30 | #include "libbb.h" |
31 | |
32 | int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
33 | int uniq_main(int argc UNUSED_PARAM, char **argv) |
34 | { |
35 | const char *input_filename; |
36 | unsigned skip_fields, skip_chars, max_chars; |
37 | unsigned opt; |
38 | char *cur_line; |
39 | const char *cur_compare; |
40 | |
41 | enum { |
42 | OPT_c = 0x1, |
43 | OPT_d = 0x2, /* print only dups */ |
44 | OPT_u = 0x4, /* print only uniq */ |
45 | OPT_f = 0x8, |
46 | OPT_s = 0x10, |
47 | OPT_w = 0x20, |
48 | }; |
49 | |
50 | skip_fields = skip_chars = 0; |
51 | max_chars = INT_MAX; |
52 | |
53 | opt_complementary = "f+:s+:w+"; |
54 | opt = getopt32(argv, "cduf:s:w:", &skip_fields, &skip_chars, &max_chars); |
55 | argv += optind; |
56 | |
57 | input_filename = argv[0]; |
58 | if (input_filename) { |
59 | const char *output; |
60 | |
61 | if (input_filename[0] != '-' || input_filename[1]) { |
62 | close(STDIN_FILENO); /* == 0 */ |
63 | xopen(input_filename, O_RDONLY); /* fd will be 0 */ |
64 | } |
65 | output = argv[1]; |
66 | if (output) { |
67 | if (argv[2]) |
68 | bb_show_usage(); |
69 | if (output[0] != '-' || output[1]) { |
70 | // Won't work with "uniq - FILE" and closed stdin: |
71 | //close(STDOUT_FILENO); |
72 | //xopen(output, O_WRONLY | O_CREAT | O_TRUNC); |
73 | xmove_fd(xopen(output, O_WRONLY | O_CREAT | O_TRUNC), STDOUT_FILENO); |
74 | } |
75 | } |
76 | } |
77 | |
78 | cur_compare = cur_line = NULL; /* prime the pump */ |
79 | |
80 | do { |
81 | unsigned i; |
82 | unsigned long dups; |
83 | char *old_line; |
84 | const char *old_compare; |
85 | |
86 | old_line = cur_line; |
87 | old_compare = cur_compare; |
88 | dups = 0; |
89 | |
90 | /* gnu uniq ignores newlines */ |
91 | while ((cur_line = xmalloc_fgetline(stdin)) != NULL) { |
92 | cur_compare = cur_line; |
93 | for (i = skip_fields; i; i--) { |
94 | cur_compare = skip_whitespace(cur_compare); |
95 | cur_compare = skip_non_whitespace(cur_compare); |
96 | } |
97 | for (i = skip_chars; *cur_compare && i; i--) { |
98 | ++cur_compare; |
99 | } |
100 | |
101 | if (!old_line || strncmp(old_compare, cur_compare, max_chars)) { |
102 | break; |
103 | } |
104 | |
105 | free(cur_line); |
106 | ++dups; /* testing for overflow seems excessive */ |
107 | } |
108 | |
109 | if (old_line) { |
110 | if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */ |
111 | if (opt & OPT_c) { |
112 | /* %7lu matches GNU coreutils 6.9 */ |
113 | printf("%7lu ", dups + 1); |
114 | } |
115 | printf("%s\n", old_line); |
116 | } |
117 | free(old_line); |
118 | } |
119 | } while (cur_line); |
120 | |
121 | die_if_ferror(stdin, input_filename); |
122 | |
123 | fflush_stdout_and_exit(EXIT_SUCCESS); |
124 | } |
125 |